bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 1 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 1 Original Research Article 5 6 7 2 Lipids in xylem sap of woody plants across 8 9 3 the angiosperm phylogeny 10 11 4 H. Jochen Schenk1,*, Joseph M. Michaud1, Kerri Mocko1, Susana Espino1, Tatiana 12 13 5 Melendres1, Mary R. Roth2, Ruth Welti2, Lucian Kaack3, and Steven Jansen3 14 15 6 1 16 Department of Biological Science, California State University Fullerton, 800 N. State College 17 18 7 Boulevard, Fullerton, CA 92831, USA; 2Kansas Lipidomics Research Center; Division of 19 20 8 Biology; Kansas State University; Manhattan, KS 66506, USA; 3Institute of Systematic Botany 21 22 9 23 and Ecology, Ulm University, Albert-Einstein-Allee 11, D–89081, Ulm, Germany. 24 25 10 *For correspondence. E-mail [email protected] 26 27 28 11 29 30 31 12 Short-running title: Lipids in xylem sap of angiosperms 32 33 13 Keywords: angiosperms, apoplast, cohesion-tension theory, galactolipids, lipidomics, 34 35 36 14 phospholipids, vessel volume, xylem, xylem sap, Laurus nobilis 37 38 39 15 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 1 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 2 of 54

1 2 Abstract 3 4 16 Lipids have been observed attached to lumen-facing surfaces of mature xylem conduits of 5 6 17 several plant species, but there has been little research on their functions or effects on water 7 8 18 transport, and only one lipidomic study of the xylem apoplast. Therefore, we conducted 9 10 19 lipidomic analyses of xylem sap from woody stems of seven plants representing six major 11 12 20 angiosperm clades, including basal magnoliids, monocots, and eudicots, to characterize and 13 14 21 quantify phospholipids, galactolipids, and sulfolipids in sap using mass spectrometry. Locations 15 16 22 of lipids in vessels of Laurus nobilis were imaged using TEM and confocal microscopy. Xylem 17 18 23 sap contained the galactolipids di- and mono-galactosyldiacylglycerol (DGDG and MGDG), as 19 20 24 well as all common plant phospholipids, but only traces of sulfolipids, with total lipid 21 22 25 concentrations in extracted sap ranging from 0.18 to 0.63 nmol / mL across all seven species. 23 24 26 Contamination of extracted sap from lipids in cut living cells was found to be negligible. Lipid 25 26 27 composition of sap was compared to wood in two species and was largely similar, suggesting 27 28 28 that sap lipids, including galactolipids, originate from cell content of living vessels. Seasonal 29 30 29 changes in lipid composition of sap were observed for one species. Lipid layers coated all lumen- 31 32 30 facing vessel surfaces of Laurus nobilis, and lipids were highly concentrated in inter-vessel pits. 33 34 31 The findings suggest that apoplastic, amphiphilic xylem lipids are a universal feature of 35 36 32 angiosperms. The findings require a reinterpretation of the cohesion-tension theory of water 37 38 33 transport to account for the effects of apoplastic lipids on dynamic surface tension and hydraulic 39 40 34 conductance in xylem. 41 42 43 35 44 45 36 46 47 48 49 50 51 52 53 54 55 56 57 58 2 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 3 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 37 INTRODUCTION 5 6 7 38 The lipidome of the plant apoplast has been characterized as a black box (Misra, 2016), because 8 9 39 there have been almost no lipidomic analyses of cell walls, intercellular spaces, and xylem 10 11 40 conduits and fibers, except for a few studies of surface lipids involved in the making of suberin, 12 13 41 cutin, and waxes. Studies of lipids in xylem conduits, including vessels, are especially limited. 14 15 42 Wagner et al. (2000) observed under transmission electron microscopy (TEM) a dark layer lining 16 17 43 vessel and tracheid surfaces of the resurrection plant Myrothamnus flabellifolia Welw. 18 19 44 (Myrothamnaceae) and after further studies concluded that it was composed of phospholipids 20 21 45 (Schneider et al., 2003). Dark layers on vessel walls that appeared under TEM after OsO4 22 23 46 fixation were also found in a few other species (Fineran, 1997, Zimmermann et al., 2004, 24 25 47 Westhoff et al., 2008). In fact, it had been shown much earlier that a thin coat of lipids remains 26 27 48 on the lumen walls of xylem conduits from live cell content after conduit maturation (Scott et al., 28 29 49 1960, Esau, 1965, Esau et al., 1966). 30 31 50 The presence of lipids in xylem conduits may seem difficult to reconcile with the cohesion 32 33 51 tension (CT) theory of water transport (Askenasy, 1895, Dixon and Joly, 1895), which posits the 34 35 52 existence of negative pressure (i.e., tension) in xylem sap. Lipids are either hydrophobic or 36 37 53 amphiphilic, either of which properties would appear to pose a problem for maintaining negative 38 39 54 pressure in the sap without forming gas bubbles on hydrophobic or amphiphilic surfaces. 40 41 55 Accordingly, Zimmermann et al. (2004) cited evidence for lipids in the xylem apoplast to bolster 42 43 56 their arguments against the CT theory. However, the CT theory is strongly supported by multiple 44 45 57 lines of evidence, both indirect and direct (Angeles et al., 2004, Wheeler and Stroock, 2008, 46 47 58 Jansen and Schenk, 2015, Venturas et al., 2017), so the existence of apoplastic xylem lipids 48 49 59 would require an explanation within the context of this theory (Schenk et al., 2017). 50 51 52 60 Any idea that there are no amphiphilic lipids in the xylem apoplast was put to rest when 53 54 61 Gonorazky et al. (2012) published an analysis of phospholipids in intercellular fluids and xylem 55 56 62 fluids of tomato plants. This was followed by reports of phospholipids in xylem sap and on 57 58 3 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 4 of 54

1 2 3 4 63 conduit surfaces of five woody angiosperm species from diverse phylogenetic backgrounds 5 6 64 (Schenk et al., 2017, Schenk et al., 2018). These findings raise questions about the origins of 7 8 65 these lipids, which could be remains from previous living vessel cell content (Scott et al., 1960, 9 10 66 Esau, 1965, Esau et al., 1966) or potentially be transported into vessels from conduit-associated 11 12 67 parenchyma cells (Morris et al., 2018b). Even more important are questions about their functions 13 14 68 in xylem (Schenk et al., 2015, Schenk et al., 2017). However, surely the first question about 15 16 69 these lipids regards their chemical nature. What are they? There is clear evidence for 17 18 70 phospholipids in the xylem apoplast, but what are their characteristics, and how about galacto- 19 20 71 and sulfolipids? 21 22 23 72 To answer these questions, we conducted a lipidomic analysis of xylem sap extracted from seven 24 25 73 species sampled across the angiosperm phylogeny, including five woody angiosperm species 26 27 74 included in previous studies (Schenk et al., 2017, Schenk et al., 2018), to quantify and 28 29 75 characterize all phospho- and galactolipids in xylem sap and also test for the presence of 30 31 76 sulfolipids in a subset of species. Because contamination from cut surfaces is a notorious 32 33 77 problem with all xylem sap analyses (Schurr, 1998), we analyzed contamination controls from 34 35 78 cut and cleaned xylem surfaces at the sap collection end of stems and compared their lipid 36 37 79 conentrations and compositions to xylem sap samples to test if lipids in extracted sap originate 38 39 80 from damaged living cells during extraction. We also used a lipid tracer to test for lipid 40 41 81 contamination from successive cuts of stems during xylem sap extractions. Because lipids in 42 43 82 xylem vessels have been reported to originate from lipid bilayer membranes in the cell content of 44 45 83 living vessels (Scott et al., 1960, Esau, 1965, Esau et al., 1966), we compared the lipid 46 47 84 composition of xylem sap for two species to that of wood from the same stems to test the 48 49 85 prediction that there would be no differences in composition. Lipids in sap sampled for one 50 51 86 species were compared in July and March to test for seasonal differences in lipid composition 52 53 87 that could indicate developmental effects, lipid transport into sap, or apoplastic enzyme activity. 54 55 56 57 58 4 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 5 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 88 In addition, locations of lipids in xylem were visualized via confocal and transmission electron 5 6 89 microscopy in bay laurel, Laurus nobilis, a species much studied for its hydraulic properties 7 8 90 (e.g., Tyree et al., 1999, Zwieniecki et al., 2001, Hacke and Sperry, 2003, Gascó et al., 2006, 9 10 91 Espino and Schenk, 2011, Nardini et al., 2017). 11 12 13 92 MATERIAL AND METHODS 14 15 93 Plant species 16 17 18 94 The study was conducted with seven plant species from six major angiosperm clades (APG III, 19 20 95 2009) and different growth forms: Liriodendron tulipifera L. (winter-deciduous tree, 21 22 96 Magnoliaceae, Magnoliales, magnoliid clade), Laurus nobilis L., (evergreen tree, Lauraceae, 23 24 97 Laurales, magnoliid clade), Bambusa oldhamii Munro (bamboo, Poaceae, Poales, commelinid 25 26 98 clade), Triadica sebifera Small (syn. Sapium sebiferum, winter-deciduous tree, Euphorbiaceae, 27 28 99 Malpighiales, fabid clade), Geijera parviflora Lindl. (evergreen tree, Rutaceae, Sapindales, 29 30 100 malvid clade), Distictis buccinatoria (DC.) A.H.Gentry (syn. Amphilophium buccinatorium 31 32 101 (DC.) L.G. Lohmann, evergreen liana, Bignoniaceae, Lamiales, lamiid clade), and Encelia 33 34 102 farinosa Torr. & A.Gray (drought-deciduous desert shrub, Asteraceae, Asterales, campanulid 35 36 103 clade). All grow in the Fullerton Arboretum or on the California State University Fullerton 37 38 104 campus in Fullerton, California, United States, except for Laurus nobilis, which was collected 39 40 105 from the Los Angeles County Arboretum in Arcadia, California, United States. Specimens of 41 42 106 Laurus nobilis for imaging studies were collected from the Botanical Garden at Ulm University, 43 44 107 Germany. All species will be referred to by their generic names in this paper. 45 46 47 108 Measurements were conducted in two stages: 1. Lipid composition of xylem sap and 48 49 109 contamination controls from all seven species: Stems for xylem sap extraction for the first stage 50 51 110 of measurements (n = 3 for each species) were collected in August 2017 from Liriodendron, 52 53 111 Triadica, Geijera, Distictis, and Encelia, and in May 2018 from Bambusa and Laurus. 2. 54 55 112 Comparisons of lipid composition in xylem sap and wood in a subset of three species, and 56 57 58 5 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 6 of 54

1 2 3 4 113 additional tests for potential contamination of xylem sap from cut surfaces using lipid tracers in 5 6 114 two species (Geijera and Distictis). Collections for the second stage of measurements (n = 4 for 7 8 115 each species) were done in October 2019 for Liriodendron, January 2020 for Distictis, and 9 10 116 March 2020 for Geijera. Some methods were common to both sets of measurements while others 11 12 117 differed, as indicated below. 13 14 15 118 Xylem sap extraction (all measurements) 16 17 119 Most collections were done at dawn to ensure full hydration of stems. The previously published 18 19 120 method for xylem sap extraction (Schenk et al., 2017) was modified to completely eliminate 20 21 121 contact of xylem sap with hydrophobic surfaces in order to avoid loss of lipids that could cling to 22 23 122 such surfaces (Fig. 1). Samples were only in contact with glass or, during stage 1 measurements, 24 25 123 with LoBind surfaces of microcentrifuge tubes (Eppendorf AG, Hamburg Germany), which 26 27 124 consist of a two-component polymer mix that creates a hydrophilic surface. Branches were cut 28 29 125 from the plants at a length exceeding the longest vessel for each species, which had been 30 31 126 previously measured via air injection (Greenidge, 1952). Stems were transported immediately to 32 33 127 the lab, and cut under water, with the final cuts made with a fresh razor blade. The bark was 34 35 128 removed from the proximal end for about 4 cm length to expose the xylem cylinder (Fig. 1). The 36 37 129 cut surface was thoroughly cleaned with deionized water using a high-pressure dental flosser 38 39 130 (WP-100 Ultra Water Flosser, Waterpik Inc., Fort Collins, CO, USA) for two minutes to remove 40 41 131 cell debris and cytoplasmic content from the surface. A control sample to determine the amount 42 43 132 of contamination from living cell remnants from the cleaned, cut surface was taken by inserting 44 45 133 the cut end into a glass vial containing 1 mL of nanopure water and leaving it there for one 46 47 134 minute. The liquid was then moved in a glass pipette into a pre-weighed 1.5 mL LoBind 48 49 135 Eppendorf microcentrifuge tube (for stage 1 extractions using partial freeze drying) or 2 mL 50 51 136 glass vials (for stage 2 extractions using a SpeedVac), immediately flash-frozen in liquid 52 53 137 nitrogen, and stored in a -20°C freezer until further processing, usually with 1-2 days. 54 55 56 57 58 6 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 7 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 138 Xylem sap was extracted under vacuum (Fig. 1). Tight-fitting latex tubing, about 2 cm long, was 5 6 139 fitted over about half of the exposed xylem cylinder next to the bark, leaving about 2 cm of 7 8 140 xylem cylinder freely exposed. The stem was then inserted from the top into a rubber flanged 9 10 141 stopper, avoiding any contact of the xylem with the stopper during insertion and creating a tight 11 12 142 seal between latex tubing and the stopper. The exposed xylem cylinder was then cleaned again 13 14 143 with deionized water using a high-pressure dental flosser and excess water removed with a 15 16 144 Kimwipe. Xylem sap was collected in a glass test tube embedded in ice inside a 1 L Buchner 17 18 145 flask, which was also embedded in ice. The flask was subjected to lab vacuum for about 30 19 20 146 seconds. The distal end of the branch was then cut back by about 4 cm, followed by successive 2 21 22 147 cm cuts, which were made to all the side branches of the stem until sap was observed dripping 23 24 148 into the test tube. Once dripping sap was observed, further 1 cm cuts were made every minute to 25 26 149 allow for slow, continuous removal of xylem sap. Depending on stem size, 1 to 2 mL of sap were 27 28 150 extracted from each stem, moved with a glass pipette into pre-weighed 1.5 mL LoBind 29 30 151 Eppendorf microcentrifuge tubes, (for stage 1 extractions using partial freeze drying) or 2 mL 31 32 152 glass vials (for stage 2 extractions using a SpeedVac), flash-frozen in liquid nitrogen, and stored 33 34 153 in the freezer until further processing. 35 36 37 154 Lipids were extracted using two different methods, because a SpeedVac evaporator instrument 38 39 155 had become available by the time stage 2 measurements were conducted, which allowed us to 40 41 156 simplify procedures. The two methods were compared and validated against each other 42 43 157 [Supplementary Information]. 44 45 158 Lipid extraction in the stage 1 measurements of seven species: Partial freeze-drying 46 47 48 159 In preliminary experiments we had found that lipid extraction from completely freeze-dried sap 49 50 160 samples was incomplete, most likely due to formation of insoluble aggregates. We therefore 51 52 161 developed a new protocol to partially freeze-dry samples for lipid extraction and thereby avoided 53 54 162 aggregate formation, which yielded about 8 times more lipids than complete freeze-drying (data 55 56 163 not shown). For this protocol, weighed cell contamination controls and xylem sap samples of 57 58 7 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 8 of 54

1 2 3 4 164 about 1 mL in pre-weighed LoBind Eppendorf microcentrifuge tunes were partially lyophilized 5 6 165 in a freeze-dryer (FreeZone 1 Liter Benchtop Freeze Dry System, Labconco, Kansas City, MO, 7 8 166 USA) and observed until only about 100 µL of sample remained. They were then taken from the 9 10 167 freeze-dryer and weighed again to determine the remaining volume of aqueous sample. The 11 12 168 remaining aqueous samples in two (for Encelia, Bambusa, and Laurus xylem sap and Bambusa 13 14 169 and Laurus controls) or four (all xylem sap and controls from all other species) LoBind 15 16 170 microcentrifuge tubes were combined into one sample originating from 2 mL or 4 mL (see 17 18 171 breakdown of species and controls above) of xylem sap or control using a glass pipette. 19 20 172 Methanol:chloroform 1:1 (both HPLC grader, Fisher Scientific) was added to the combined 21 22 173 aqueous sample to create an approximate 5:5:1 methanol:chloroform:water one-phase mixture. 23 24 174 The mixture was then vortexed, centrifuged, and the supernatant collected with a 1 mL glass 25 26 175 syringe with PrecisionGlide Needle. A new one-phase 5:5:1 mixture of 27 28 176 methanol:chloroform:water was then added to the residue, the mixture vortexed, centrifuged, and 29 30 177 the supernatant again collected and combined with the previously collected supernatant. Samples 31 32 178 were then dried in a desiccator outfitted with an in-line carbon filter capsule (model 6704-7500, 33 34 179 Whatman, GE Healthcare Life Sciences, UK) and sent to the Kansas Lipidomics Research 35 36 180 Center at Kansas State University for mass spectrometry analysis. 37 38 39 181 Lipid extraction in stage 2 measurements of three species: SpeedVac evaporation 40 41 182 The lipid extraction protocol at this stage was changed, because a SpeedVac evaporator designed 42 43 183 for chloroform extraction had become available, which allowed us to simplify the procedure. 44 45 184 Following xylem sap extraction, samples were immediately freeze-dried using a SpeedVac 46 47 185 Concentrator unit (model Savant SPD121P, Thermo Scientific, Waltham, MA, USA). Samples 48 49 186 were freeze-dried for 7 hours to permit complete drying of xylem sap in 2 mL glass vials. 50 51 187 Freeze-dried samples were then stored at -15 °C. 52 53 54 188 To isolate lipids from freeze-dried xylem sap samples, a 5:5:1 methanol:chloroform:water one- 55 56 189 phase mixture (341 µL methanol, 341 µL chloroform, 68 µL nanopure water) was added to 57 58 8 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 9 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 190 previously fully freeze-dried vials containing xylem sap. Following vortexing for 60 s, the 5 6 191 solution was transferred to new 5 mL glass centrifuge tubes via glass pipette and centrifuged at 7 8 192 3500 rpm for 8 min. The supernatant of the centrifuged sample containing lipids was then 9 10 193 collected using a glass pipette into a new 2 mL glass vial. This process was repeated to ensure 11 12 194 maximal collection of lipids from all glass surfaces, with supernatant of the repeated steps added 13 14 195 into the final 2 mL glass vial. Samples were then freeze-dried using the SpeedVac Concentrator 15 16 196 unit for 7 h to permit complete removal of all liquid in 2 mL glass vials. Freeze-dried lipid- 17 18 197 isolate samples were then stored on dry ice and sent to the Kansas Lipidomics Research Center at 19 20 198 Kansas State University for mass spectrometry analysis. 21 22 23 199 The partial freeze-drying and SpeedVac extraction methods were subjected to a methods 24 25 200 comparison and were found to yield similar results (Methods S1) [Supplementary 26 27 201 Information]. 28 29 202 Seasonal changes in lipid composition of xylem sap in Geijera 30 31 32 203 Xylem sap was extracted from Geijera in mid July 2019 (n = 8), and early March 2020 (n = 8). 33 34 204 At both times, half of the 8 samples were from lipid tracer experiments (see below), and because 35 36 205 no tracer was found in collected xylem sap for Geijera (see Results), samples from tracer 37 38 206 treatments and controls were combined to increase the statistical power of comparisons. Lipids 39 40 207 were extracted as described above, and lipid concentrations and composition compared between 41 42 208 sampling times to determine if there were changes between summer and winter collections. 43 44 45 209 Lipid extraction from wood in stage 2 measurements of three species 46 47 210 Lipids were extracted from wood by using a modification of methods developed originally for 48 49 211 lipid extractions from leaves (Shiva et al., 2018). Following xylem sap extractions (see above) 50 51 212 from the same stems, four thin wood sections of exposed xylem (~200 mg) were cut with a 52 53 213 sledge microtome (model GSL 1, Schweingruber, Switzerland) and immediately submerged in 54 55 214 0.4 mL of hot isopropanol (75°C) for 15 min in 2 mL glass vials to prevent enzymatic 56 57 58 9 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 10 of 54

1 2 3 4 215 degradation of lipids from cut cells. This procedure made it impossible to determine the precise 5 6 216 weight of cut sections, so it allows only determination of lipid composition, not lipid 7 8 217 concentrations. To approximate the average fresh weight of wood samples, separate sections 9 10 218 from the same stem were cut and weighed. To isolate lipids from wood samples, 0.54 mL 11 12 219 chloroform, 0.54 mL methanol, and 0.104 mL nanopure water were added to the vial and softly 13 14 220 shaken for 24 h on a compact orbital shaker (Cole-Parmer). The 5/5/1 15 16 221 chloroform/methanol/water solution was then transferred with a glass pipette into a new 2 mL 17 18 222 glass vial and freeze-dried using a Speedvac. 19 20 21 223 Lipid tracer studies to quantify lipid contamination from cut stem surfaces (stage 2 22 23 224 measurements) 24 25 225 To determine if cutting stems during vacuum extraction of xylem sap introduces lipids into 26 27 226 extracted sap, lipid tracer studies were conducted with Geijera (short vessels) and Distictis (long 28 29 227 vessels). The tracer was di17:0 phosphatidylethanolamine (di17:0 PE) (Avanti Polar Lipids, 30 31 228 Alabaster, AL, USA), because it was absent in previous measurements of xylem sap. All cutting 32 33 229 blades were cleaned vigorously with 1% Alconox solution (product No. 1104, Alconox, White 34 35 230 Plains, NY, USA) before each cut to prevent surface contamination with tracer from blades. Sap 36 37 231 was extracted from stems as described above until the remaining stems were 50 cm long. At this 38 39 232 point, vacuum was relaxed, and 0.1 mL of 0.07 mg/ml di17:0 PE lipid tracer in chloroform was 40 41 233 added to the freshly-cut distal stem surface (see Fig. 1). The chloroform evaporated immediately, 42 43 234 and the tracer was allowed to absorb into vessels for 15 s. Vacuum was reapplied to the stem 44 45 235 until approximately 0.5 mL xylem sap was collected. Two more cuts were done using the same 46 47 236 tracer application at 45 cm and 40 cm length, at which point the extraction was concluded. Lipids 48 49 237 were extracted from xylem sap as described above for stage 2 measurements. 50 51 52 238 Vessel volume analysis 53 54 55 56 57 58 10 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 11 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 239 Because lipid micelles are mostly too large to pass through pit membranes (Zhang et al., 2020), 5 6 240 we hypothesized that lipid concentrations in extracted xylem sap would be related to the volume 7 8 241 of vessels opened when a stem is cut for sap extraction. To determine the average opened vessel 9 10 242 volume for each species, we used silicone injections (Sperry et al., 2005, Wheeler et al., 2005). 11 12 243 Silicone compound (Rhodorsil RTV-141; Rhodia USA, Cranbury, NJ, USA) mixed with 0.41% 13 14 244 blue silicone dye (Blue Silc Pig; Smooth On, Easton, PA, USA) was injected from the proximal 15 16 245 end into flushed stems (n = 4 per species) that were between 0.9 and 1.7 cm in diameter and 25 17 18 246 cm in length, except 50 cm in length for Distictis. Injection took place at approximately 50 kPa 19 20 247 for 24 h, after which the silicone was allowed to dry for 72 h prior to sectioning. Cross sections 21 22 248 (cut at 0.7, 1.4, 2.9, 5.8, 8.3, 11.8, and 24 cm from the cut surface for all species, except at 0.7, 23 24 249 1.4, 2.8, 5.8, 11.6, 16.6, 23.6, and 48 cm for Distictis) were thin-sectioned using a sledge 25 26 250 microtome (model GSL 1, Schweingruber, Switzerland), dry-mounted on glass slides, and 27 28 251 imaged using a Leica MZ16 stereomicroscope with digital camera (model INFINITY 2-1C-IQ, 29 30 252 Lumenera, Ottawa, ONT, Canada) at 20× zoom and backlit using a transmitted light stage 31 32 253 connected to a halogen light source (model KL 1500 LCD, Schott, Germany). Adobe Photoshop 33 34 254 (version CS6) was used to blend images into one combined image for each cross-section (Fig. 35 36 255 S10) [Supplementary Information], and an analysis of filled vessels, vessels diameters, vessel 37 38 256 wall perimeters, and vessel density was completed using ImageJ (Schneider et al., 2012). 39 40 257 Silicone-filled vessels were counted in each image, and the area of silicone-filled vessels in each 41 42 258 image was quantified both in mm2 and as percent of the total xylem area. The silicone-filled 43 44 259 vessel volume was calculated from these data by linear interpolation between each cross-section 45 46 260 (Fig. S11) [Supplementary Information]. Counts of filled vessels (C) at each distance (d) from 47 48 261 the injected surface were used to estimate median vessel lengths, based on fitting the equation 49 50 262 ln C = a + b × d using the software TableCurve 2D (version 5.01, Systat, San Jose, CA, USA). 51 52 53 54 55 56 57 58 11 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 12 of 54

1 2 3 4 263 Mass spectrometry 5 6 7 264 Direct-infusion electrospray ionization triple-quadrupole mass spectrometry was performed 8 9 265 using a Waters Xevo TQS mass spectrometer or Applied Biosystems 4000 (Shiva et al., 2013). 10 11 266 Instrument-specific details can be found in Supplemental Tables S1 and S2. The dried samples, 12 13 267 each originating from 2 mL or 4 mL of xylem sap or cell contamination control, or from wood 14 15 268 samples (see breakdown of species above), were dissolved in 1 mL chloroform in a 2 mL glass 16 17 269 vial. After a test to determine the optimal amount for analysis, 50-1000 μL of each sample was 18 19 270 transferred to a 2-mL vial containing internal standards (Supplemental Table S2). 1.2 mL of 20 21 271 chloroform:methanol:300 mM ammonium acetate in water (30:66.5:3.5) was added and the 22 23 272 sample mixed. The sample was continuously infused to the mass spectrometer from a 1-mL 24 25 273 loop. 26 27 274 In addition to the scans described in Shiva et al. (2013), a scan for SQDG (Pre 261.1) in positive 28 29 275 ion mode was also acquired (except for samples from Laurus and Bambusa) (details in 30 31 276 Supplemental Table S2). Data were processed as described in Shiva et al. (2013) using the 32 33 277 LipidomeDB Data Calculation Environment at http://lipidome.bcf.ku.edu:8080/Lipidomics/ 34 35 278 (Zhou et al., 2011). Appropriate response factor for the biological galactolipid molecular species 36 37 279 in comparison to the saturated galactolipid internal standards were applied. The limit of 38 39 280 detection in the samples analyzed was 0.002 nmol for the stage 1 measurements of seven species. 40 41 281 And 0.0005 nmol for stage 2 measurements. 42 43 44 282 Confocal laser scanning microscopy 45 46 283 To test for lipids in xylem of Laurus, fluorescent FM1-43 dye (Molecular Probes, Life 47 48 284 Technologies, Eugene, OR, USA) was infiltrated into living xylem and imaged with laser 49 50 51 285 scanning confocal microscopy, as described in Schenk et al. (2018). FM1-43 is an amphiphilic 52 53 286 fluorophore that is virtually non-fluorescent in water and strongly fluorescent under cyan light 54 55 287 excitation when bound to lipids, including amphiphilic lipids, and biological membranes 56 57 58 12 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 13 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 288 (Jelínková et al., 2010). It was used as a 5 μg/mL solution in nanopure water with 488 nm 4 5 6 289 excitation and an emission window of 578 to 618 nm. Woody stems, approximately 5-8 mm in 7 8 290 diameter and about 10 cm long, were cut from the plants and submerged in DI water. About 2.5 9 10 291 cm were cut under water from each end of the stem to remove most bubbles in the xylem that 11 12 may have been induced by the cutting. The cut ends were then recut with new razor blades to 13 292 14 15 293 create clean and smooth surfaces. Bark was removed from about 1 cm length at the distal end to 16 17 294 expose the xylem. The proximal end of each stem was connected to a vacuum flask using tight- 18 19 295 fitting latex tubing around the stem inside Tygon tubing. A short piece of Tygon tubing was 20 21 22 296 fitted to the exposed xylem cylinder at the distal end, using latex tubing around the cylinder to 23 24 297 create a tight fit. The distal cut end was cleaned by washing it three times with 500 µL nanopure 25 26 298 water using a pipette. 1 mL of fluorophore solution was then added into the tubing and onto the 27 28 29 299 exposed xylem surface. Lab vacuum was turned on and turned off once the solution was 30 31 300 absorbed into the xylem, but no air was aspirated. The control was treated the same way but, 32 33 301 instead of the fluorophore solution, 1 mL of nanopure water was sucked through the stem. A 2 34 35 mm long stem piece was then cut along its transverse plane from the center of the stem segment 36 302 37 38 303 and the distal surface recut using a sledge microtome (Sledge microtome GSL 1, Schweingruber, 39 40 304 Switzerland). The stem piece was then placed onto a cell culture dish with a glass bottom of 175 41 42 305 µm (CELLviewTM dish, Greiner Bio-One, Germany) in a drop of water and observed using the 43 44 45 306 inverted microscope of a confocal laser scanning microscope (model TCS SP8 with a 46 47 307 HyVolution super-resolution detector, Leica Microsystems, Wetzlar, Germany). Lignin was 48 49 308 detected with 405 nm excitation and an emission window of 467 to 509 nm. Imaging was done 50 51 52 309 with a 63.0 × 1.20 objective and 3× digital zoom in water, using four scans per image at 1760  53 54 310 1760 resolution. 55 56 57 58 13 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 14 of 54

1 2 3 4 311 Transmission electron microscopy (TEM) 5 6 312 Small sectioning blocks of xylem from fresh Laurus stems were given two different treatments 7 8 313 as described in Schenk et al. (2017): (1) fixation with a solution containing 2.5% glutaraldehyde, 9 10 11 314 1% sucrose, and 0.1 M phosphate buffer at pH 7.3, and (2) fixation with glutaraldehyde and then 12 13 315 postfixation with a 2% aqueous osmium tetroxide (OsO4) solution for 2 hours at room 14 15 316 temperature. The samples were dehydrated through a gradual ethanol series and embedded in 16 17 317 Epon resin. Moreover, samples treated with glutaraldehyde only were observed with and without 18 19 20 318 post staining of TEM grids with aqueous uranyl acetate and lead citrate to test the potential 21 22 319 staining effect on vessel-vessel pit membranes (Ellis, 2014). The duration of the staining was 5 23 24 320 minutes for uranyl acetate and 1 minute for lead citrate. Transverse semi-thin sections were cut 25 26 27 321 with an ultramicrotome (Leica Ultracut UCT, Leica Microsystems, Vienna, Austria), stained 28 29 322 with 0.5% toluidine blue in 0.1 M phosphate buffer, and mounted on microscope slides using 30 31 323 Eukitt. Ultra-thin sections between 60 nm and 90 nm were mounted on copper grids (Athena, 32 33 34 324 Plano GmbH, Wetzlar, Germany) and observed with a JEM-1210 TEM (Jeol, Tokyo, Japan) at 35 36 325 120 kV. Digital images were taken using a MegaView III camera (Soft Imaging System, 37 38 326 Münster, Germany). 39 40 41 327 Data analysis 42 43 44 328 To test for the presence of di17:0 PE tracer in tracer experiments, concentrations of di17:0 PE in 45 46 329 xylem sap from stems treated with di17:0 PE tracer at cut stem surfaces were compared to 47 48 330 controls without tracer applications using two-tailed t-tests (n = 4) assuming equal variance in 49 50 51 331 Microsoft Excel. Mean lipid concentrations were compared to various anatomical parameters of 52 53 332 xylem for each species (incl. opened vessel volume, opened vessel volume relative to surface 54 55 333 area, mean vessel diameter and wall perimeter, both calculated from the diameter of a circle with 56 57 58 14 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 15 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 334 the same area as the vessel, and mean vessel density) via reduced major axis regression in the 4 5 6 335 statistical software package PAST (ver. 4.02) (Hammer et al., 2001). 7 8 336 To compare lipid compositions between samples, mass spectrometry data were analyzed using 9 10 11 337 the online MetaboAnalyst software package (https://www.metaboanalyst.ca). Any concentrations 12 13 338 below the detection limit were set to zero, and data were log-transformed to normalize 14 15 339 distributions. Compositions of polar lipids in xylem sap were compared in MetaboAnalyst as 16 17 18 340 normalized concentrations between seven species and between two sampling periods for Geijera 19 20 341 using principal component analysis (PCA) and heatmaps, based on Euclidean distance measures, 21 22 342 and a Ward clustering algorithm to visualize differences among the species or sampling times. 23 24 Unpaired t-tests, adjusted for false discovery rates (FDR) were used to identify lipids that 25 343 26 27 344 differed between sampling times. 28 29 30 345 Compositions of polar lipids in xylem sap from six species (one sample for Encelia was lost, 31 32 346 causing the species to be excluded from this analysis) were compared to those in paired 33 34 347 contamination controls to test the hypothesis that xylem sap lipids originate from damaged living 35 36 cells and therefore do not differ in lipid composition to contamination controls. For these 37 348 38 39 349 comparisons, concentrations were converted into percentages, data were log-transformed and 40 41 350 analyzed using PCA in MetaboAnalyst. 42 43 44 351 Polar lipid compositions were compared between wood and xylem sap for Geijera and Distictis 45 46 352 by converting concentrations to percentages, then excluding any measurements that were below 47 48 the detection limit, and using PCA (95% CI). Paired t-tests adjusted for FDR were used to 49 353 50 51 354 identify lipids that differed between sap and wood. Because wood samples included xylem sap 52 53 355 lipids, the two data sets were not independent, so PCA and t-tests were used qualitatively to 54 55 56 57 58 15 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 16 of 54

1 2 3 356 determine if there were differences between sap and wood and identified lipids that differed the 4 5 6 357 most. 7 8 9 358 RESULTS 10 11 12 359 Comparisons of xylem sap and contamination controls (seven species) 13 14 360 Polar lipid concentrations in cell contamination controls ranged from 1.1% of the amount 15 16 361 detected in xylem sap in Distictis to 9.0% in Laurus, averaging 4.49 (± 2.80 SD) percent across 17 18 362 species (Table 1). The complete lipidomics data set may be found in Table S2 for xylem sap and 19 20 363 Table S3 for cell contamination controls [Supplementary Information]. All contamination 21 22 364 controls were far more variable and distinctly different in polar lipid composition compared to 23 24 365 xylem sap, as shown in principal component analyses (PCA; Fig. S12), clearly demonstrating 25 26 366 that xylem sap lipids do not originate from cells cut open during sap extraction [Supplementary 27 28 367 Information]. 29 30 31 368 Lipid tracer studies to detect lipid contamination from cut surfaces (two species) 32 33 34 369 Concentrations of di17:0 PE in xylem sap of Geijera stems treated with di17:0 PE tracer at cut 35 36 370 distal stem surfaces and in controls without tracer applications were below 1 pmol/mL in both 37 38 371 treatments and not significantly different (p = 0.738). Concentrations of di17:0 PE in xylem sap 39 40 372 of Distictis stems treated with di17:0 PE tracer at cut distal stem surfaces were 0.9 pmol/mL (± 41 42 373 0.8 SE), 0.2 pmol/mL (± 0.1 SE) in controls without tracer applications, and not significantly 43 44 374 different (p = 0.131). 45 46 375 Xylem sap lipidomics (seven species) 47 48 49 376 Polar lipids extracted from xylem sap and cell contamination controls included the galactolipids 50 51 377 digalactosyldiacylglycerol (DGDG, Fig. S1) and monogalactosyldiacylglycerol (MGDG, Fig. 52 53 378 S2), as well as all common plant phospholipids. Total lipid concentrations for xylem sap varied 54 55 379 between 0.18 nmol / mL for Laurus and Liriodendron and 0.63 nmol / mL for Bambusa, 56 57 58 16 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 17 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 380 averaging 0.41 (± 0.17 SD) nmol / mL across species (Table 1; Fig. 2 A). Lipid concentrations in 5 6 381 sap were not statistically related to the vessel volume opened by cutting stems, but there was a 7 8 382 significant negative relationship to the total vessel perimeter at the xylem surface (Fig. 2 C). 9 10 11 383 The most common lipids grouped by their head groups (Fig. 2 A) were DGDG, MGDG, 12 13 384 phosphatidylcholine (PC, Fig. S4), phosphatidic acid (PA, Fig. S8), phosphatidylethanolamine 14 15 385 (PE, Fig. S5), phosphatidylinositol (PI, Fig. S6), with other phospholipids present in much lower 16 17 386 amounts, including phosphatidylserine (PS, Fig. S7), phosphatidylglycerol (PG, Fig. S3), 18 19 387 lysophosphatidylethanolamine (LPE, Fig. S9), lysophosphatidylcholine (LPC, Fig. S9), and 20 21 388 lysophosphatidylglycerol (LPG). Trace amounts of a contaminant that was also found in solvent 22 23 389 interfered with the detection of LPG, which was therefore excluded from all tables, figures, and 24 25 390 analyses for stage 1 measurements. (The problem was solved for stage 2 measurements by using 26 27 391 glass vials instead of LoBind Eppendorf tubes in all subsequent experiments.) 28 29 392 The most pronounced differences in chemical compositions between species were in the 30 31 393 percentage of galactolipids (Laurus 75.4%, Liriodendron 35.9%, Bambusa 57.0%, Triadica 32 33 394 18.8%, Geijera 10.6%, Distictis 20.4%, Encelia 7.1%). Sulfolipids were detected only in trace 34 35 395 amounts. In the PCA comparing the seven species, no species stood out as different in polar lipid 36 37 396 compositions from the other six species on the PC1 axis, which explained 46.3% of the variation, 38 39 397 while Laurus and Bambusa stood out on the PC2 axis, explaining 16.7% of the variation, as 40 41 398 distinctly different (Fig. 2 D). The most pronounced differences between species were in the 42 43 399 percentages contributed by DGDG, MGDG, PC, and PA (Table 1; Fig. 2 A). DGDG varied from 44 45 400 about 3.0% of all lipids in Encelia to 32.0% in Bambusa, with a mean across species of 12.4 (± 46 47 401 10.5 SD)%; MGDG varied from 4.1% in Encelia to 54.6% in Laurus, with a mean across species 48 49 402 of 19.8 (±17.8 SD)%; PC varied from 12.3% in Laurus to 30.1% in Triadica, with a mean across 50 51 403 species of 18.2 (±6.3 SD)%; PA varied from 1.9% in Laurus to 50.6% in Encelia, with a mean 52 53 404 across species of 28.7 (±17.6 SD)%. 54 55 56 57 58 17 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 18 of 54

1 2 3 4 405 The most common number of acyl chain carbon atoms were 34 and 36, with galactolipids mostly 5 6 406 having more 36-carbon combinations than phospholipids (Fig. 2 B, Fig. S1-2). Only on average 7 8 407 1.56 (± 0.70 SD)% of all lipids were fully saturated phospholipids, and the most common 9 10 408 numbers of double bonds were 1-3 for the 34 carbon combination of 2 acyl chains and 3-5 for the 11 12 409 36 carbon combination (Fig. 2 B, Fig. S3-9). Averaged across species, lipids with 34:1-3 (carbon 13 14 410 atoms:number of double bonds) and 36:3-5 accounted for 84.0 (± 4.9 SD)% of all apoplastic 15 16 411 xylem lipids. PA 34:2 was especially abundant in all species, except for Laurus, making up about 17 18 412 23% of all polar lipids in Encelia and 25% in Geijera. The lipid composition of the galactolipids, 19 20 413 with a total of four double bonds prominent, suggests that 18:2 fatty acids may be common. 21 22 23 414 Seasonal changes in lipid composition of xylem sap in Geijera 24 25 415 Total concentrations of lipids in xylem sap were not different between samples collected in July 26 27 416 and March, but concentrations of the galactolipids DGDG and MGDG were significantly higher 28 29 417 in sap collected in March (Table 2). Concentrations of PA tended to be lower in March, but 30 31 418 differences were not significant because of high variability among samples. Overall, 18 32 33 419 individual lipids differed significantly in their concentrations between July and March (Fig. 3 B), 34 35 420 including seven types of DGDG, four of MGDG, and the remaining ones phospholipids. In PCA 36 37 421 (Fig. 3 A), March samples were slightly more variable than July samples, and the 95% 38 39 422 confidence regions for the two periods were clearly separated on the PC1 axis, which explained 40 41 423 27% of the variation. Overall, there were clear seasonal differences in lipid composition. 42 43 44 424 Comparisons of lipid compositions in wood and xylem sap (two species) 45 46 47 425 Wood samples include xylem sap lipids, so the samples are not independent and can be 48 49 426 compared only qualitatively (Table 3). However, PCAs revealed some differences between wood 50 51 427 and sap, especially for Distictis (Fig. 4 A), where FDR-adjusted t-tests identified five 52 53 428 phospholipids (out of 125) as different between wood and sap. No headgroups, chain lengths, or 54 55 429 unsaturation levels were consistently different between wood and sap for Distictis, but several 56 57 58 18 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 19 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 430 types of PC stood out as being more abundant in sap than in wood, while several types of PS and 5 6 431 lysophospholipids showed the opposite trend (Fig. 4 B). For Geijera, only one polar lipid, PI 7 8 432 34:3, was identified by t-tests as different between wood and sap (Fig. 4 B), and PE was slightly 9 10 433 more abundant (by 3.3%) in wood than in sap, and lysoPC was slightly less abundant in wood 11 12 434 than in sap (by -0.14%). PA was just as abundant in wood as in sap (Table 3), even though wood 13 14 435 samples were treated immediately with hot isopropanol to inactive PLD enzyme activity. No 15 16 436 galactolipids were flagged as being different between wood and sap of either species. Overall, 17 18 437 differences in lipid composition between wood and sap were not pronounced. 19 20 21 438 Visualization of lipids in xylem vessels of Laurus 22 23 439 Treatment with FM1-43 showed continuous lipid layers on vessel surfaces of Laurus nobilis, 24 25 440 with high concentrations in inter-vessel pits (Fig. 5 A-B). Pit membranes were clearly visible 26 27 441 under confocal laser scanning microscopy as thick, pillow-shaped, black structures with lipid 28 29 442 coatings on either side and some faint labelling inside the membranes. Never-dried pit 30 31 443 membranes of Laurus were 1,252 nm (± 238 nm, standard deviation; n = 36 pit membranes) 32 33 444 thick under confocal microscopy (Fig. 5 A-B), but 601 nm (± 150 nm; n = 19) under TEM (Fig. 34 35 445 5 D-G), showing that the dehydration required for TEM by propanol could result in shrinkage of 36 37 446 the pit membrane. These observations were very similar to those published previously for five of 38 39 447 the other study species, including Liriodendron, Triadica, Geijera, Distictis, and Encelia 40 41 448 (Schenk et al., 2018). TEM samples not treated with OsO4 showed a dark and homogeneous 42 43 449 appearance (Fig. 5 D-E). After treatment with OSO4, the lipid layer appeared to be broken up 44 45 450 into aggregates under TEM (Fig. 5 F-G) but was found in a continuous layer in the same 46 47 451 locations observed under confocal microscopy on vessel and pit surfaces and on the pit 48 49 452 membrane (Fig. 5 A-B). 50 51 52 453 53 54 55 56 57 58 19 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 20 of 54

1 2 3 4 454 DISCUSSION 5 6 7 455 The most important take-home message of this study is that amphiphilic lipids exist in xylem 8 9 456 conduits across the angiosperm phylogeny, including magnoliids, monocots, and eudicots. The 10 11 457 available evidence based on our results indicate clearly that apoplastic lipids do not originate 12 13 458 from cells cut open during sap extraction, as shown by contamination controls, lipid tracer 14 15 459 experiments, and imaging. Phospholipids have been reported to occur in xylem sap (Gonorazky 16 17 460 et al., 2012, Schenk et al., 2017), but not galactolipids. These findings raise questions about the 18 19 461 chemical composition of xylem sap lipids, their origins, and functions of these lipids in 20 21 462 angiosperm vessels. 22 23 463 Chemical composition of xylem sap lipids 24 25 26 464 The concentrations of lipids in xylem sap were surprisingly consistent across species, with the 27 28 465 two species from the magnoliid clade, Laurus and Liriodendron, having about a third of the 29 30 466 concentrations of the other five species (Table 1; Fig. 2 A). Because lipid micelles or lipid- 31 32 467 coated gas bubbles in xylem sap are mostly larger than 50 nm in diameter (Schenk et al., 2017) 33 34 468 and therefore too large to pass through pit membrane pores, which do not exceed 20 nm in 35 36 469 hydrated pit membranes (Choat et al., 2003, Zhang et al., 2017, Zhang et al., 2020), it is likely 37 38 470 that the observed xylem sap lipids originate largely from vessels cut open for the sap extraction. 39 40 471 To account for this, we analysed the opened vessel volumes of the seven species and found no 41 42 472 significant relationship between lipid concentrations in sap and vessel volumes, but there was a 43 44 473 significant negative relationship with the total vessel wall perimeter at the cut xylem surface 45 46 474 (Fig. 2 C). It appears that the xylem functions like a filter during lipid extraction, so that xylem 47 48 475 with higher wall surface area, i.e., more and smaller vessels, traps more surface-active lipids, 49 50 476 leading to lower lipid yields in extracted sap. Because the open vessel volumes were typically 51 52 477 smaller than the 1 to 2 mL of xylem sap extracted (except for the liana species Distictis), it is 53 54 478 likely that the total amount of lipids in sap is underestimated due to a filtering and dilution effect. 55 56 57 58 20 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 21 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 479 Lipids are insoluble and in xylem conduits most are attached to surfaces (Fig. 4; Schenk et al., 5 6 480 2018), so the amounts extracted in sap may only be a small fraction of all apoplastic xylem 7 8 481 lipids. Quantification of all apoplastic lipids in xylem would require either complete extraction 9 10 482 from individual conduits with organic solvents and/or detergents, which could cause artifacts if 11 12 483 these substances penetrated into living vessel-associated cells (Morris et al., 2018b), or 13 14 484 quantification of lipids in vessel images via mass spectrometry imaging (Ellis et al., 2013). 15 16 17 485 The very low concentrations of lipids found in cell contamination controls from the basal xylem 18 19 486 surface show that on average at least 95% of the lipids in xylem sap samples did not originate 20 21 487 from cut parenchyma cells at basal xylem surfaces. Bambusa and Distictis, both of which have 22 23 488 phloem embedded with xylem (Distictis has intraxylary phloem) had the lowest percentages of 24 25 489 lipid concentrations in their control samples (4.3% in Bambusa and 1.1% in Distictis), 26 27 490 demonstrating that hardly any lipids in xylem sap samples originated from the phloem. Lipid 28 29 491 compositions of cell contamination controls were far more variable and different from those of 30 31 492 sap (Fig. S12), which also confirms that sap lipids, except for traces, do not originate from cut 32 33 493 living cells. Lipids in sap also clearly do not originate from cells cut at the distal end of stems, as 34 35 494 lipid tracer applied to distal stem surfaces during xylem sap extraction did not show up in sap 36 37 495 collections. Minute tracer amounts may have made their way through the exceptionally wide and 38 39 496 long vessels of Distictis, but not enough to raise tracer levels significantly. Lipids from cut cells 40 41 497 at the distal stem end are likely to attach to vessel surfaces as sap is extracted, or will be held up 42 43 498 by inter-vessel pit membranes, instead of moving through >40 cm of vessel length and showing 44 45 499 up in extracted sap. 46 47 500 The only other lipidomic analysis of xylem sap, for tomato (Gonorazky et al., 2012), did not 48 49 501 include analyses of galactolipids, but the chemical composition of phospholipids was roughly 50 51 502 comparable to our findings, with PC and PA being the most abundant phospholipids. The xylem 52 53 503 sap concentration of phospholipids in tomato (0.037 µmol kg-1) was similar to concentrations in 54 55 56 57 58 21 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 22 of 54

1 2 3 4 504 Laurus (0.044 µmol L-1), but much lower than the average phospholipid concentrations across all 5 6 505 species studied (0.296 ± 0.161 SD µmol L-1). 7 8 9 506 The presence of the galactolipids DGDG and MGDG in the xylem apoplast is especially notable, 10 11 507 because galactolipids are synthesized exclusively in plastids (Dörmann and Benning, 2002, Botté 12 13 508 and Maréchal, 2014) and are the most abundant lipids in chloroplasts, but they have never been 14 15 509 found in any apoplastic compartment, presumably because there have been so few studies of “the 16 17 510 black box of plant apoplast lipidomes” (Misra, 2016). Moreover, their physical properties, such 18 19 511 as interactions of the head groups with water molecules, are different from phospholipids 20 21 512 (Kanduč et al., 2017). MGDG have very small head groups and form bilayers only in mixture 22 23 513 with other lipids (Dörmann and Benning, 2002), where interactions between MGDG and DGDG 24 25 514 head groups affect the physical properties of galactolipids at gas-water interfaces (Bottier et al., 26 27 515 2007). It is also notable that sulfolipids, which are also abundant in plastids (Shimojima, 2011), 28 29 516 were only found in trace amounts in xylem sap. 30 31 517 Regarding acyl chains, it was observed that galactolipids mostly had 36-carbon combinations, 32 33 518 most likely consisting of two 18-carbon chains (Fig. 2 B). Some plants, termed “16:3 plants” can 34 35 519 incorporate 16C-fatty acids into galactolipids and desaturate them to 16:3 (via the “prokaryotic 36 37 520 pathway”), whereas others, termed “18:3 plants”, assemble the diacylglycerol backbone of 38 39 521 galactolipids only in the E.R. 16:3 plants contain a mixture of 34C and 36C galactolipids, 40 41 522 whether 18:3 plants contain only 36C galactolipids. Leaves of both Laurus nobilis and 42 43 523 Liriodendron tulipifera contain small amounts (slightly over 1%) of plastid-synthesized 16:3 44 45 524 (Mongrand et al., 1998). Triadica, Geijera, and Distictis leaves also may contain small amounts 46 47 525 of 16:3, but Bambusa and Encelia are likely to be 18:3 plants. Still, the majority of xylem sap 48 49 526 galactolipid backbones from all seven species are di-18C molecular species, consistent with 50 51 527 synthesis in the ER (Mongrand et al., 1998). Additionally, it has previously been shown that a 52 53 528 16:3 plant, such as Arabidopsis thaliana, favors the eukaryotic pathway in non-photosynthetic 54 55 529 tissue (Devaiah et al., 2006, Huynh et al., 2012). 56 57 58 22 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 23 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 530 Along the same lines, the bulk of the observed lipid species appear to be less unsaturated than 5 6 531 most leaf tissues, containing only minor amounts of 18:3 chains, which are the most prominent 7 8 532 fatty acids in leaves. Whole tissue data demonstrate that generally Arabidopsis shoots are very 9 10 533 high in 18:3, while roots and seeds are less unsaturated with more 18:2 and 18:1-containing 11 12 534 galactolipids (Devaiah et al., 2006, Huynh et al., 2012). Indeed, Gonorazky et al. (2011) 13 14 535 observed that 18:3-containing phospholipids in leaves were less likely to be found in the 15 16 536 extracellular fluid than in the whole leaf. The lipid composition of the xylem galactolipids, with 17 18 537 36:4 species prominent, suggests that 18:2 fatty acids may be common. Similarly, 34:2 species, 19 20 538 which are likely 16:0/18:2 combinations are very common in phospholipids, suggesting an 21 22 539 abundance of 18:2 in xylem sap. 23 24 25 540 Besides finding galactolipids in xylem sap, the presence of phosphatidic acid (PA) at high 26 27 541 concentrations in some species is also notable (Fig. 2 A). High concentrations of PA in plant 28 29 542 samples are usually attributed to phospholipase D (PLD) activity, which cleaves head groups 30 31 543 such as the choline head-group of phosphatidylcholine (PC) off phospholipids to create PA 32 33 544 (Christie, 1993, Welti et al., 2002). PLD is active when bound to cell membranes (Kolesnikov et 34 35 545 al., 2012), and cutting living cells causes PLD activation as part of a wounding response 36 37 546 (Bargmann et al., 2009). High PA concentrations can be artifacts of wounding, and thorough 38 39 547 cleaning of cut xylem surfaces is therefore essential for keeping fragments of cell membranes 40 41 548 containing PLD out of xylem sap collections. PLD can remain active even in organic solvents 42 43 549 such as chloroform:methanol (Christie, 1993), possibly because membrane fragments form 44 45 550 inverted vesicles in organic solvents that allow for continued PLD activity in the hydrophilic 46 47 551 interior of the vesicle, as has been suggested for phospholipase C (PLC) (Kates, 1957). 48 49 552 The high PA concentrations found in xylem sap of some species, such as Encelia, Geijera, and 50 51 553 Distictis (Fig. 2 A) were initially a concern, because it was suspected that they could be sampling 52 53 554 artifacts. However, treatments of cut xylem surfaces with the PLD inhibitors 1-butanol, 5-fluoro- 54 55 555 2-indolyl des-chlorohalopemide (FIPI) (Su et al., 2009), and a lipase inhibitor cocktail (Furse et 56 57 58 23 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 24 of 54

1 2 3 4 556 al., 2013) had no effect on PA concentrations in xylem sap of Distictis (data not shown). 5 6 557 Moreover, PLD or other phospholipases have never been detected in proteomic analyses of 7 8 558 xylem sap (Alvarez et al., 2006, Djordjevic et al., 2007, Aki et al., 2008, Krishnan et al., 2011, 9 10 559 Ligat et al., 2011, Dugé de Bernonville et al., 2014), which may not be surprising, because PLD 11 12 560 is often bound to membranes and therefore unlikely to occur naturally in xylem sap. Moreover, 13 14 561 thin wood samples also contained high amounts of PA (Table 3), even though the slices were 15 16 562 immediately dropped into hot isopropanol to inactivate PLD. 17 18 19 563 These lines of argument all support PA as a major natural ingredient of xylem saps, except for 20 21 564 that of Laurus (Fig. 2 A; Table 1). PA has also been found in relatively high concentrations in 22 23 565 xylem sap of tomato plants (Gonorazky et al., 2012). PA has many functions in plants, interacts 24 25 566 with many proteins, and has important roles in lipid metabolism and signalling (Testerink and 26 27 567 Munnik, 2011), wounding responses (Bargmann et al., 2009), and in responses to stress, such as 28 29 568 drought and salinity (McLoughlin and Testerink, 2013). The physical properties of PA are also 30 31 569 unique when compared to other amphiphilic lipids in having the smallest and negatively charged 32 2+ 2+ 33 570 head-group, binding of cations such as Ca and Mg (Ohki and Ohshima, 1985), and causing 34 35 571 PA to have lower surface tension in monolayers than other phospholipids (Weschayanwiwat et 36 37 572 al., 2005). Of all the various roles of PA, its interactions with cations and effects on surface 38 39 573 tension at gas interfaces are likely to be most important in the xylem apoplast environment. 40 41 574 The amounts of choline-containing phospholipids (PC and LPC) in xylem sap reported here were 42 43 575 much lower than estimated in previous research using an enzymatic assay (Schenk et al., 2017), 44 45 576 where all choline quantified in xylem sap was assumed to originate via PLD activity from 46 47 577 choline-containing phospholipids. However, that assumption appears to have been incorrect. 48 49 578 Choline has been found in xylem sap at molar concentrations far exceeding those of lipids (Lima 50 51 579 et al., 2017) and is likely to originate from sources other than phospholipids in sap. 52 53 54 55 56 57 58 24 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 25 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 580 Where do apoplastic xylem lipids come from? 5 6 7 581 There are two possible and not mutually exclusive origins for apoplastic xylem lipids: Lipids 8 9 582 have been shown to remain from the living cell content of conduits during conduit development 10 11 583 (Scott et al., 1960, Esau et al., 1966). If this includes lipids from plastids, then this would easily 12 13 584 explain the presence of galactolipids in xylem sap. Lipids could also be transported into conduits 14 15 585 from conduit-associated parenchyma cells (Czaninski, 1977, Morris et al., 2018b), but it seems 16 17 586 unlikely that plastid lipids would be moved into vessels. We compared the lipid composition in 18 19 587 sap to that of the wood that the sap was extracted from (Fig. 4) and found only minor differences 20 21 588 in composition and none in galactolipids. To some degree, this is expected, as wood includes 22 23 589 sap, so it is impossible for sap to include any lipids not found in wood samples. Out of 157 polar 24 25 590 lipids tested, only one for Geijera and five for Distictis stood out as different in relative amounts 26 27 591 between sap and wood (Table 2). Overall, the differences between sap and wood were small, so 28 29 592 it seems reasonable to conclude that sap and wood lipids are essentially the same, supporting the 30 31 593 origin of apoplastic xylem lipids, including galactolipids, from living vessel contents, as shown 32 33 594 previously in microscopic studies of developing vessels (Scott et al., 1960, Esau et al., 1966). 34 35 595 That said, there were clear seasonal differences in the polar lipid composition of Geijera xylem 36 37 596 sap between July and March (Fig. 3), with elevated galactolipid concentrations in March. 38 39 597 Seasonal changes could be a reflection of wood development, with newer vessels containing 40 41 598 different lipids than older ones, could be caused by apoplastic enzyme activities, or by lipid 42 43 599 transport from vessel-associated cells. This would require transport of lipids across a cell 44 45 600 membrane, the protective layer surrounding such cells, which consists mainly of polysaccharides 46 47 601 (Chafe and Chauret, 1974, Fujii et al., 1981, Wisniewski et al., 1991a, Wisniewski et al., 1991b, 48 49 602 Wisniewski and Davis, 1995), as well as across the pit membrane, which mainly consists of 50 51 603 cellulose. The transport could involve non-specific lipid transfer proteins (nsLTPs), which can 52 53 604 transport lipids across cell walls (Domínguez et al., 2015, Fich et al., 2016, Li et al., 2016, 54 55 605 Misra, 2016) and which have been found in xylem cell walls and xylem sap (Buhtz et al., 2004, 56 57 58 25 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 26 of 54

1 2 3 4 606 Djordjevic et al., 2007, Krasikov et al., 2011, Ligat et al., 2011). However, it is exceedingly 5 6 607 unlikely that plastid lipids would be transported into vessels, so seasonal differences in 7 8 608 galactolipids in xylem sap are more likely to be caused either by apoplastic enzyme activities or 9 10 609 by immobilization of galactolipids on vessel surfaces and in pit membranes, which would 11 12 610 prevent them from being extracted in sap. Further research will be required to understand these 13 14 611 seasonal changes. 15 16 17 612 Lipid transport or secretion of apoplastic lipid-altering enzymes from conduit-associated cells 18 19 613 would allow for changes in lipid concentration and composition in conduits seasonally or in 20 21 614 response to environmental conditions, which could possibly affect hydraulic functions and 22 23 615 embolism resistance (Charrier et al., 2018). Plants species vary widely in the amount and degree 24 25 616 of contact between conduits and living parenchyma cells in secondary xylem (Martínez-Cabrera 26 27 617 et al., 2009, Morris and Jansen, 2016, Morris et al., 2016, Morris et al., 2018a), and it could be 28 29 618 that some species undergo seasonal changes in apoplastic lipids while others do not. Seasonal 30 31 619 changes in the electron-density of pit membranes have been observed (Schmid and Machado, 32 33 620 1968), and this could potentially indicate lipid accumulation in conduits over time. 34 35 621 Apoplastic xylem lipids and functional implications 36 37 38 622 The presence of lipids on xylem conduit surfaces and pit membranes is incompatible with the 39 40 623 widely accepted assumption that gas bubble formation in xylem is prevented by the high surface 41 42 624 tension of water (Oertli, 1971, Sperry and Tyree, 1988, Tyree and Zimmermann, 2002, Stroock 43 44 625 et al., 2014). There is unequivocal evidence for negative pressure in xylem, and this aspect of the 45 46 626 CT theory is not in question, although many open questions remain (Jansen and Schenk, 2015, 47 48 627 Schenk et al., 2015, Schenk et al., 2017, Venturas et al., 2017). Molecular dynamics modelling 49 50 628 of lipid bilayers under negative pressure confirmed that cavitation in lipid bilayers and micelles 51 52 629 is unlikely to occur within the normal range of pressures (above -10 MPa) experienced in plants 53 54 630 (Kanduč et al., 2020). 55 56 57 58 26 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 27 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 631 However, layers of phospho- and galactolipids on conduit surfaces and in pit membranes (Fig. 5) 5 6 632 will reduce the surface tension at gas-water interfaces substantially below that of pure water, 7 8 633 with the dynamic surface tension depending on the local concentration of lipids at the surface 9 10 634 (Zuo et al., 2004, Kwan and Borden, 2010). Measurements of xylem lipids extracted from four 11 12 635 of the species studied in this paper found average equilibrium surface tensions to be about 25 13 14 636 mN/m, roughly a third of the surface tension of pure water (Yang et al., 2020). These findings 15 16 637 are entirely compatible with the air seeding pressure required to force air through pore 17 18 638 constrictions in angiosperm pit membranes. Pore constrictions rarely exceed 20 nm in diameter 19 20 639 (Zhang et al., 2020), and it would take about 7 MPa of pressure to force air through such a 21 22 640 constriction if surface tension were that of pure water (Kaack et al., 2019). Actual air seeding 23 24 641 pressures through pit membranes tend to be in the range of -0.4 to -2 MPa (Bartlett et al., 2016), 25 26 642 thereby supporting a role for lipids in pit membranes in the air seeding process. 27 28 29 643 In an earlier paper, we proposed a hypothesis to reconcile the presence of apoplastic xylem lipids 30 31 644 with the CT theory (Schenk et al., 2015) and later amended and modified that hypothesis in the 32 33 645 light of new data (Schenk et al., 2017). To summarize that hypothesis briefly, gas penetrating 34 35 646 from embolized conduits into pit membranes of sap-filled conduits may snap off nanobubbles 36 37 647 inside confined, lipid-coated pore spaces, resulting in the creation of lipid-coated nanobubbles 38 39 648 (Fig. 6), which have in fact been found and visualized in xylem sap (Schenk et al., 2017). Key to 40 41 649 this hypothesis is that both the low dynamic surface tension of lipids and snapping off of bubbles 42 43 650 at pore constrictions inside fibrous pit membranes would limit the bubbles’ sizes (Park et al., 44 45 651 2019) and keep them below a critical threshold that prevents bubble expansion into embolism 46 47 652 (Schenk et al., 2015, Schenk et al., 2017). This process could happen under normal xylem water 48 49 653 potentials and would allow xylem to operate without spreading embolisms via air seeding from 50 51 654 gas-filled conduits. The population of lipid-coated nanobubbles could also function as a gas 52 53 655 reservoir that serves as a buffer when dissolved gas in sap exceeds saturation under increasing 54 55 656 temperatures (Schenk et al., 2016). 56 57 58 27 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 28 of 54

1 2 3 4 657 Below a more negative air seeding pressure threshold, gas could potentially still penetrate 5 6 658 through the entire length of pit membrane pores, causing a growing bubble to emerge from the 7 8 659 pit membrane, where nothing would restrict the bubble’s growth and resulting in embolism. This 9 10 660 is the widely accepted air seeding process (Sperry and Tyree, 1988, Tyree and Zimmermann, 11 12 661 2002, Stroock et al., 2014, Brodribb et al., 2016), the underlying mechanism of which is still 13 14 662 poorly understood (Jansen et al., 2018), but which is supported by a large body of research, 15 16 663 including measurements of pit membrane pore constrictions and air seeding pressures (Kaack et 17 18 664 al., 2019). What the new lipid-nanobubble hypothesis could explain is how plants can operate 19 20 665 under negative pressure at all without forming embolism even under mildly negative pressure, a 21 22 666 feat that engineers have found almost impossible to replicate (Smith, 1994). If this hypothesis, 23 24 667 which is not contradictory but complementary to the cohesion-tension theory, is correct, then 25 26 668 lipids are required for water transport under negative pressure. In fact, Dixon (1914) already 27 28 669 suggested the colloids increase the tensile strength of xylem sap over that of pure water. 29 30 31 670 Conclusions 32 33 671 This study provides conclusive evidence for substantial amounts of phospho- and galactolipids in 34 35 672 xylem conduits across the angiosperm phylogeny, and these findings force major questions about 36 37 673 the functions of these lipids for plant water transport and the cohesion tension theory, as well as 38 39 674 about the origins of lipids. Research on the functions of lipids in water that is under negative 40 41 675 pressure and in the presence of gas-water interfaces in pit membranes and bubbles will benefit 42 43 676 from an interdisciplinary approach to involve surface scientists and physicists with expertise on 44 45 677 the behavior of lipids at surfaces such as pulmonary surfactant layers (Zuo et al., 2008, Zhang et 46 47 678 al., 2011, Kanduč et al., 2013, Kanduč et al., 2016). Moreover, possible interactions between 48 49 679 lipids and proteins, including enzymes, in xylem conduits need to be investigated. Even in the 50 51 680 absence of gas in conduits, lipids accumulated in pit membranes are likely to affect sap flow and 52 53 681 could be involved in the regulation of hydraulic conductance, the so-called ionic effect (Nardini 54 55 682 et al., 2011, Nardini et al., 2012). Finally, if lipids are crucial for plant water transport, then they 56 57 58 28 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 29 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 683 should be found as well in gymnosperms, ferns, and perhaps even moss hydroids. The discovery, 5 6 684 or rather rediscovery (Scott et al., 1960, Esau, 1965, Esau et al., 1966, Wagner et al., 2000), of 7 8 685 lipids in xylem conduits opens a vast field of new research questions. 9 10 11 686 12 13 687 FUNDING 14 15 16 688 The research was funded by research grants from the National Science Foundation (EAGER 17 18 689 IOS-1558108 and IOS- 1754850). SJ and LK acknowledge funding from the German Research 19 20 690 Foundation (DFG, project No. 383393940). 21 22 23 691 ACKNOWLEDGEMENTS 24 25 692 The authors thank Anne Basilio, Zoe Cuevas, Jessica Garcia, Ryan Cochoit, Alec Hunt, and Tilly 26 27 693 Duong for assistance with xylem sap extractions and vessel volume analyses, Jim Henrich at the 28 29 694 Los Angeles County Arboretum and Greg Pongetti at the Fullerton Arboretum for allowing 30 31 695 sampling of plants in their living collections, the Electron Microscopy Section of Ulm University 32 33 696 for preparing TEM samples, and the Core Facility for Confocal and Multiphoton Microscopy of 34 35 697 Ulm University for assistance with confocal imaging. 36 37 698 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 29 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 30 of 54

1 2 3 4 699 Table 1. Composition of xylem sap lipids (in nmol / mL) for the seven studied species determined 5 6 700 by mass spectrometry, arranged by head groups, and expressed as mean ± standard deviation (n = 7 8 701 3). LysoPG was excluded from these measurements, because trace amounts of a contaminant that 9 10 702 was also found in solvent interfered with the detection of LPG in stage 1 measurements. 11 12 Laurus Liriodendron Bambusa Triadica Geijera Distictis Encelia 13 Mean STD Mean STD Mean STD mean STD mean STD mean STD Mean STD 14 15 Xylem sap 16 DGDG 0.038 ± 0.013 0.015 ± 0.010 0.203 ± 0.046 0.051 ± 0.009 0.015 ± 0.006 0.041 ± 0.01 0.015 ± 0.008 17 MGDG 0.099 ± 0.052 0.050 ± 0.041 0.158 ± 0.022 0.042 ± 0.007 0.032 ± 0.012 0.050 ± 0.016 0.021 ± 0.010 18 PG 0.001 ± 0.000 0.002 ± 0.002 0.003 ± 0.001 0.007 ± 0.002 0.003 ± 0.003 0.003 ± 0.001 0.004 ± 0.004 19 LysoPC 0.000 ± 0.000 0.003 ± 0.003 0.006 ± 0.003 0.004 ± 0.002 0.001 ± 0.001 0.001 ± 0.001 0.007 ± 0.004 20 21 LysoPE 0.002 ± 0.002 0.001 ± 0.001 0.003 ± 0.002 0.002 ± 0.002 0.001 ± 0.001 0.001 ± 0.001 0.005 ± 0.005 22 PC 0.022 ± 0.003 0.040 ± 0.017 0.088 ± 0.046 0.148 ± 0.042 0.070 ± 0.031 0.087 ± 0.032 0.071 ± 0.053 23 PE 0.007 ± 0.001 0.016 ± 0.014 0.026 ± 0.020 0.042 ± 0.042 0.051 ± 0.016 0.046 ± 0.018 0.046 ± 0.055 24 PI 0.007 ± 0.003 0.009 ± 0.009 0.027 ± 0.009 0.068 ± 0.018 0.035 ± 0.027 0.051 ± 0.030 0.071 ± 0.081 25 PS 0.001 ± 0.000 0.004 ± 0.003 0.007 ± 0.003 0.011 ± 0.006 0.016 ± 0.007 0.013 ± 0.008 0.015 ± 0.015 26 27 PA 0.004 ± 0.002 0.041 ± 0.03 0.112 ± 0.049 0.120 ± 0.047 0.223 ± 0.125 0.153 ± 0.090 0.262 ± 0.144 28 29 Total 0.181 ± 0.071 0.179 ± 0.124 0.633 ± 0.166 0.494 ± 0.177 0.447 ± 0.213 0.447 ± 0.203 0.517 ± 0.371 30 31 SQDG n.d. 0.001 ± 0.001 n.d. 0.003 ± 0.000 0.003 ± 0.001 0.003 ± 0.001 0.009 ± 0.003 32 33 34 Contamination controls 35 DGDG 0.003 ± 0.001 0.005 ± 0.007 0.005 ± 0.004 0.001 ± 0.001 0.001 ± 0.002 0.002 ± 0.001 0.000 ± 0.000 36 MGDG 0.004 ± 0.002 0.004 ± 0.004 0.010 ± 0.009 0.001 ± 0.000 0.002 ± 0.001 0.002 ± 0.001 0.001 ± 0.000 37 38 PG 0.000 ± 0.000 0.000 ± 0.000 0.000 ± 0.001 0.000 ± 0.000 0.000 ± 0.000 0.000 ± 0.000 0.000 ± 0.000 39 LysoPC 0.000 ± 0.000 0.000 ± 0.000 0.001 ± 0.000 0.000 ± 0.000 0.000 ± 0.000 0.000 ± 0.000 0.000 ± 0.000 40 LysoPE 0.003 ± 0.002 0.000 ± 0.000 0.001 ± 0.001 0.000 ± 0.000 0.000 ± 0.000 0.000 ± 0.000 0.000 ± 0.000 41 PC 0.002 ± 0.002 0.003 ± 0.001 0.008 ± 0.013 0.008 ± 0.005 0.007 ± 0.001 0.003 ± 0.000 0.012 ± 0.004 42 PE 0.000 ± 0.000 0.000 ± 0.000 0.000 ± 0.000 0.002 ± 0.002 0.000 ± 0.000 0.000 ± 0.000 0.000 ± 0.000 43 44 PI 0.001 ± 0.001 0.000 ± 0.000 0.001 ± 0.001 0.002 ± 0.003 0.000 ± 0.000 0.000 ± 0.000 0.000 ± 0.000 45 PS 0.001 ± 0.001 0.000 ± 0.000 0.000 ± 0.000 0.000 ± 0.000 0.000 ± 0.000 0.000 ± 0.001 0.000 ± 0.000 46 PA 0.000 ± 0.000 0.007 ± 0.004 0.002 ± 0.002 0.011 ± 0.006 0.009 ± 0.006 0.002 ± 0.000 0.006 ± 0.001 47 48 Total 0.014 ± 0.008 0.020 ± 0.015 0.029 ± 0.029 0.025 ± 0.016 0.020 ± 0.008 0.009 ± 0.003 0.021 ± 0.003 49 50 51 SQDG n.d. 0.001 ± 0.001 n.d. 0.003 ±0.000 0.003 ± 0.001 0.003 ± 0.001 0.009 ± 0.003 52 53 703 Galactolipids: DGDG = digalactosyldiacylglycerol, MGDG = and monogalactosyldiacylglycerol. Phospholipids: 54 704 LysoPG = lysophosphatidylglycerol, LysoPC = lysophosphatidylcholine, LysoPE = lysophosphatidylethanolamine, PA 55 705 = phosphatidic acid, PC = phosphatidylcholine, PE = phosphatidylethanolamine, PI = phosphatidylinositol, PS = 56 706 phosphatidylserine, PG = phosphatidylglycerol. SQDG = Sulfolipids. 57 58 30 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 31 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 707 4 5 6 708 Table 2. Comparison of lipid composition categorized by headgroups in polar lipids in xylem sap 7 8 709 of Geijera (n = 8) sampled in July 2019 and March 2020. Mean concentrations (nmol / mL) 9 10 710 were compared between dates using FDR-adjusted t-tests in Metaboanalyst.ca. 11 12 13 Lipid type July 2019 March 2020 p-value 14 15 Mean ± STD Mean ± STD (FDR adjusted) 16 17 DGDG 0.0143 ± 0.0046 0.0589 ± 0.0284 0.00027 18 19 MGDG 0.0230 ± 0.0096 0.0472 ± 0.0193 0.03036 20 21 PG 0.0137 ± 0.0059 0.0174 ± 0.0097 n.s. 22 23 LysoPG 0.0074 ± 0.0025 0.0072 ± 0.0045 n.s. 24 25 LysoPC 0.0033 ± 0.0024 0.0018 ± 0.0004 n.s. 26 27 LysoPE 0.0018 ± 0.0011 0.0009 ± 0.0002 n.s. 28 29 PC 0.0851 ± 0.0407 0.0860 ± 0.0348 n.s. 30 31 PE 0.0550 ± 0.0368 0.0512 ± 0.0181 n.s. 32 33 PI 0.0185 ± 0.0114 0.0224 ± 0.0119 n.s. 34 35 PS 0.0148 ± 0.0068 0.0114 ± 0.0031 n.s. 36 37 PA 0.5329 ± 0.3504 0.3050 ± 0.1206 n.s. 38 39 Total Polar 0.7698 ± 0.4546 0.6094 ± 0.2025 n.s. 40 41 711 42 43 712 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 31 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 32 of 54

1 2 3 713 Table 3. Comparison of lipid composition (in %) in paired samples (n = 4) of xylem sap and 4 5 714 wood from the same stems collected in March 2020. All concentrations were converted to 6 715 percentages of total polar lipids. Wood and sap samples are not independent from each other, so 7 716 statistical comparisons were not possible. 8 9 Geijera Distictis 10 11 Sap Wood Sap Wood 12 13 mean STD mean STD Mean STD mean STD 14 15 DGDG 9.52 ± 0.84 7.23 ± 1.08 4.53 ± 1.93 6.05 ± 1.00 16 MGDG 7.74 ± 0.81 8.00 ± 1.54 4.33 ± 1.43 5.66 ± 0.78 17 18 PG 2.45 ± 0.55 5.78 ± 1.38 1.26 ± 0.30 2.73 ± 1.00 19 20 LysoPG 1.44 ± 1.66 0.16 ± 0.06 1.23 ± 0.75 0.50 ± 0.41 21 22 LysoPC 0.31 ± 0.15 0.18 ± 0.07 0.21 ± 0.07 0.50 ± 0.10 23 24 LysoPE 0.17 ± 0.08 0.15 ± 0.03 0.18 ± 0.08 0.30 ± 0.03 25 PC 11.64 ± 1.94 14.62 ± 5.38 23.55 ± 3.06 14.57 ± 3.23 26 27 PE 7.39 ± 2.40 10.66 ± 2.39 8.47 ± 1.37 9.01 ± 1.76 28 29 PI 3.73 ± 1.51 6.16 ± 0.83 7.35 ± 0.71 9.00 ± 0.87 30 31 PS 1.88 ± 0.62 1.57 ± 0.52 2.00 ± 0.77 1.68 ± 0.39 32 33 PA 53.74 ± 16.80 45.51 ± 4.85 46.88 ± 2.59 50.02 ± 3.85 34 Total 100 100 100 100 35 36 717 37 38 718 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 32 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 33 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 719 LITERATURE CITED 4 5 6 720 Aki, T., Shigyo, M., Nakano, R., Yoneyama, T. and Yanagisawa, S. (2008) Nano scale proteomics 7 8 721 revealed the presence of regulatory proteins including three FT-like proteins in phloem and xylem 9 10 722 saps from rice. Plant & Cell Physiol., 49, 767-790. https://doi.org/10.1093/pcp/pcn049 11 12 723 Alvarez, S., Goodger, J.Q.D., Marsh, E.L., Chen, S., Asirvatham, V.S. and Schachtman, D.P. (2006) 13 14 724 Characterization of the maize xylem sap proteome. J. Proteome Res., 5, 963-972. 15 16 725 https://doi.org/10.1021/pr050471q 17 18 726 Angeles, G., Bond, B., Boyer, J.S., Brodribb, T., Brooks, J.R., Burns, M.J., Cavender-Bares, J., 19 20 727 Clearwater, M., Cochard, H., Comstock, J., Davis, S.D., Domec, J.C., Donovan, L., Ewers, 21 22 728 F., Gartner, B., Hacke, U., Hinckley, T., Holbrook, N.M., Jones, H.G., Kavanagh, K., Law, 23 24 729 B., Lopez-Portillo, J., Lovisolo, C., Martin, T., Martinez-Vilalta, J., Mayr, S., Meinzer, F.C., 25 26 730 Melcher, P., Mencuccini, M., Mulkey, S., Nardini, A., Neufeld, H.S., Passioura, J., 27 28 731 Pockman, W.T., Pratt, R.B., Rambal, S., Richter, H., Sack, L., Salleo, S., Schubert, A., 29 30 732 Schulte, P., Sparks, J.P., Sperry, J., Teskey, R. and Tyree, M. (2004) The cohesion-tension 31 32 733 theory. New Phytol., 163, 451-452. https://doi.org/10.1111/j.1469-8137.2004.01142.x 33 34 734 Askenasy, E. (1895) Ueber das Saftsteigen. Verh. Nat.-med. Ver. Heidelb. N. F., 5, 325-345 35 36 735 Bargmann, B.O.R., Laxalt, A.M., Riet, B.T., Testerink, C., Merquiol, E., Mosblech, A., Leon-Reyes, 37 38 736 A., Pieterse, C.M.J., Haring, M.A., Heilmann, I., Bartels, D. and Munnik, T. (2009) 39 40 737 Reassessing the role of phospholipase D in the Arabidopsis wounding response. Plant Cell 41 42 738 Environ., 32, 837-850. https://doi.org/10.1111/j.1365-3040.2009.01962.x 43 44 739 Bartlett, M.K., Klein, T., Jansen, S., Choat, B. and Sack, L. (2016) The correlations and sequence of 45 46 740 plant stomatal, hydraulic, and wilting responses to drought. Proc. Natl. Acad. Sci., 113, 13098- 47 48 741 13103. https://doi.org/10.1073/pnas.1604088113 49 50 742 Botté, C.Y. and Maréchal, E. (2014) Plastids with or without galactoglycerolipids. Trends Plant Sci., 51 52 743 19, 71-78. https://doi.org/10.1016/j.tplants.2013.10.004 53 54 744 Bottier, C., Géan, J., Artzner, F., Desbat, B., Pézolet, M., Renault, A., Marion, D. and Vié, V. (2007) 55 56 745 Galactosyl headgroup interactions control the molecular packing of wheat lipids in Langmuir 57 58 33 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 34 of 54

1 2 3 4 746 films and in hydrated liquid-crystalline mesophases. Biochim. Biophys. Acta (BBA) - 5 6 747 Biomembranes, 1768, 1526-1540. https://doi.org/10.1016/j.bbamem.2007.02.021 7 8 748 Brodribb, T.J., Bienaimé, D. and Marmottant, P. (2016) Revealing catastrophic failure of leaf 9 10 749 networks under stress. Proc. Natl. Acad. Sci., 113, 4865-4869. 11 12 750 https://doi.org/10.1073/pnas.1522569113 13 14 751 Buhtz, A., Kolasa, A., Arlt, K., Walz, C. and Kehr, J. (2004) Xylem sap protein composition is 15 16 752 conserved among different plant species. Planta, 219, 610-618. https://doi.org/10.1007/s00425- 17 18 753 004-1259-9 19 20 754 Chafe, S.C. and Chauret, G. (1974) Cell wall structure in the xylem parenchyma of trembling aspen. 21 22 755 Protopl., 80, 129-147. https://doi.org/10.1007/bf01666355 23 24 756 Charrier, G., Delzon, S., Domec, J.-C., Zhang, L., Delmas, C.E., Merlin, I., Corso, D., King, A., 25 26 757 Ojeda, H. and Ollat, N. (2018) Drought will not leave your glass empty: Low risk of hydraulic 27 28 758 failure revealed by long-term drought observations in world’s top wine regions. Sci. Adv., 4, 29 30 759 eaao6969. https://doi.org/10.1126/sciadv.aao6969 31 32 760 Choat, B., Ball, M., Luly, J. and Holtum, J. (2003) Pit membrane porosity and water stress-induced 33 34 761 cavitation in four co-existing dry rainforest tree species. Plant Physiol., 131, 41–48. 35 36 762 https://doi.org/10.1104/pp.014100 37 38 763 Christie, W.W. (1993) Preparation of lipid extracts from tissues. In Advances in Lipid Methodology – 39 40 764 Two (Christie, W.W. ed. Dundee: Oily Press, pp. 195-213. 41 42 765 Czaninski, Y. (1977) Vessel-associated cells. IAWA Bull., 1977, 51-55 43 44 766 Devaiah, S.P., Roth, M.R., Baughman, E., Li, M., Tamura, P., Jeannotte, R., Welti, R. and Wang, 45 46 767 X. (2006) Quantitative profiling of polar glycerolipid species from organs of wild-type 47 48 768 Arabidopsis and a PHOSPHOLIPASE Dα1 knockout mutant. Phytochem., 67, 1907-1924. 49 50 769 https://doi.org/10.1016/j.phytochem.2006.06.005 51 52 770 Dixon, H.H. (1914) On the tensile strength of sap. Sci. Proc. Royal Dubl. Soc. N. S., 14, 229-234 53 54 771 Dixon, H.H. and Joly, J. (1895) On the ascent of sap. Phil. Trans. R. Soc. Lond. B, 186, 563-576 55 56 57 58 34 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 35 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 772 Djordjevic, M.A., Oakes, M., Li, D.X., Hwang, C.H., Hocart, C.H. and Gresshoff, P.M. (2007) The 5 6 773 Glycine max xylem sap and apoplast proteome. J. Proteome Res., 6, 3771-3779. 7 8 774 https://doi.org/10.1021/pr0606833 9 10 775 Domínguez, E., Heredia-Guerrero, J.A. and Heredia, A. (2015) Plant cutin genesis: unanswered 11 12 776 questions. Trends Plant Sci., 20, 551-558. https://doi.org/10.1016/j.tplants.2015.05.009 13 14 777 Dörmann, P. and Benning, C. (2002) Galactolipids rule in seed plants. Trends Plant Sci., 7, 112-118. 15 16 778 https://doi.org/10.1016/S1360-1385(01)02216-6 17 18 779 Dugé de Bernonville, T., Albenne, C., Arlat, M., Hoffmann, L., Lauber, E. and Jamet, E. (2014) 19 20 780 Xylem sap proteomics. In Plant Proteomics (Jorrin-Novo, J.V., Komatsu, S., Weckwerth, W. and 21 22 781 Wienkoop, S. eds): Humana Press, pp. 391-405. 23 24 782 Ellis, E.A. (2014) Staining sectioned biological specimens for transmission electron microscopy: 25 26 783 conventional and en bloc stains. In Electron Microscopy: Methods and Protocols (Kuo, J. ed. 27 28 784 New York: Humana Press, pp. 57-72. 29 30 785 Ellis, S.R., Brown, S.H., in het Panhuis, M., Blanksby, S.J. and Mitchell, T.W. (2013) Surface 31 32 786 analysis of lipids by mass spectrometry: More than just imaging. Progr. Lid. Res., 52, 329-353. 33 34 787 https://doi.org/10.1016/j.plipres.2013.04.005 35 36 788 Esau, K. (1965) Plant anatomy 2nd edn. New York: John Wiley & Sons. 37 38 789 Esau, K., Cheadle, V.I. and Gill, R.H. (1966) Cytology of differentiating tracheary elements. II. 39 40 790 Structures associated with cell surfaces. Amer. J. Bot., 53, 765-771. 41 42 791 https://doi.org/10.1002/j.1537-2197.1966.tb06832.x 43 44 792 Espino, S. and Schenk, H.J. (2011) Mind the bubbles: Achieving stable measurements of maximum 45 46 793 hydraulic conductivity through woody plant samples. J. Exp. Bot., 62, 1119-1132. 47 48 794 https://doi.org/10.1093/jxb/erq338 49 50 795 Fich, E.A., Segerson, N.A. and Rose, J.K.C. (2016) The plant polyester cutin: Biosynthesis, structure, 51 52 796 and biological roles. Annu. Rev. Plant Biol., 67, 207-233. https://doi.org/10.1146/annurev-arplant- 53 54 797 043015-111929 55 56 57 58 35 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 36 of 54

1 2 3 4 798 Fineran, B.A. (1997) Cyto- and histochemical demonstration of lignins in plant cell walls: an evaluation 5 6 799 of the chlorine water ethanolamine silver nitrate method of Coppick and Fowler. Protopl., 198, 7 8 800 186-201. https://doi.org/10.1007/BF01287568 9 10 801 Fujii, T., Harada, H. and Saiki, H. (1981) Ultrastructure of "amorphous layer" in xylem parenchyma 11 12 802 cell wall of angiosperm species. Mokuzai Gakkaishi, 27, 149-156 13 14 803 Furse, S., Liddell, S., Ortori, C.A., Williams, H., Neylon, D.C., Scott, D.J., Barrett, D.A. and Gray, 15 16 804 D.A. (2013) The lipidome and proteome of oil bodies from Helianthus annuus (common 17 18 805 sunflower). J. Chem. Biol., 6, 63-76. https://doi.org/10.1007/s12154-012-0090-1 19 20 806 Gascó, A., Nardini, A., Gortan, E. and Salleo, S. (2006) Ion-mediated increase in the hydraulic 21 22 807 conductivity of Laurel stems: role of pits and consequences for the impact of cavitation on water 23 24 808 transport. Plant Cell Environ., 29, 1946-1955. https://doi.org/10.1111/j.1365-3040.2006.01570.x 25 26 809 Gonorazky, G., Laxalt, A.M., Dekker, H.L., Rep, M., Munnik, T., Testerink, C. and de la Canal, L. 27 28 810 (2012) Phosphatidylinositol 4-phosphate is associated to extracellular lipoproteic fractions and is 29 30 811 detected in tomato apoplastic fluids. Plant Biol., 14, 41-49. https://doi.org/10.1111/j.1438- 31 32 812 8677.2011.00488.x 33 34 813 Greenidge, K.N.H. (1952) An approach to the study of vessel length in hardwood species. Amer. J. Bot., 35 36 814 39, 570-574. https://doi.org/10.1002/j.1537-2197.1952.tb13070.x 37 38 815 Hacke, U.G. and Sperry, J.S. (2003) Limits to xylem refilling under negative pressure in Laurus nobilis 39 40 816 and Acer negundo. Plant Cell Environ., 26, 303-311. https://doi.org/10.1046/j.1365- 41 42 817 3040.2003.00962.x 43 44 818 Hammer, Ø., Harper, D.A.T. and Ryan, P.D. (2001) PAST: Paleontological statistics software package 45 46 819 for education and data analysis. Palaeont. Etectron., 4, 9 47 48 820 Huynh, V.-B., Repellin, A., Zuily-Fodil, Y. and Pham-Thi, A.-T. (2012) Aluminum stress response in 49 50 821 rice: effects on membrane lipid composition and expression of lipid biosynthesis genes. Physiol. 51 52 822 Plant., 146, 272-284. https://doi.org/10.1111/j.1399-3054.2012.01622.x 53 54 55 56 57 58 36 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 37 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 823 Jansen, S., Klepsch, M., Li, S., Kotowska, M.M., Schiele, S., Zhang, Y. and Schenk, H.J. (2018) 5 6 824 Challenges in understanding air-seeding in angiosperm xylem. Acta Hort., 1222, 13-20. 7 8 825 https://doi.org/10.17660/ActaHortic.2018.1222.3 9 10 826 Jansen, S. and Schenk, H.J. (2015) On the ascent of sap in the presence of bubbles. Amer. J. Bot., 102, 11 12 827 1561-1563. https://doi.org/10.3732/ajb.1500305 13 14 828 Jelínková, A., Malínská, K., Simon, S., Kleine-Vehn, J., Pařezová, M., Pejchar, P., Kubeš, M., 15 16 829 Martinec, J., Friml, J., Zažímalová, E. and Petrášek, J. (2010) Probing plant membranes with 17 18 830 FM dyes: tracking, dragging or blocking? Plant J., 61, 883-892. https://doi.org/10.1111/j.1365- 19 20 831 313X.2009.04102.x 21 22 832 Kaack, L., Altaner, C.M., Carmesin, C., Diaz, A., Holler, M., Kranz, C., Neusser, G., Odstrcil, M., 23 24 833 Schenk, H.J., Schmidt, V., Weber, M., Zhang, Y. and Jansen, S. (2019) Function and three- 25 26 834 dimensional structure of intervessel pit membranes in angiosperms: a review. IAWA J., 40, 673- 27 28 835 702. https://doi.org/10.1163/22941932-40190259 29 30 836 Kanduč, M., Schlaich, A., de Vries, A.H., Jouhet, J., Maréchal, E., Demé, B., Netz, R.R. and 31 32 837 Schneck, E. (2017) Tight cohesion between glycolipid membranes results from balanced water– 33 34 838 headgroup interactions. Nat. Comm., 8, 14899. https://doi.org/10.1038/ncomms14899 35 36 839 Kanduč, M., Schlaich, A., Schneck, E. and Netz, R.R. (2016) Water-mediated interactions between 37 38 840 hydrophilic and hydrophobic surfaces. Langmuir, 32, 8767-8782. 39 40 841 https://doi.org/10.1021/acs.langmuir.6b01727 41 42 842 Kanduč, M., Schneck, E., Loche, P., Jansen, S., Schenk, H.J. and Netz, R.R. (2020) Cavitation in 43 44 843 lipid bilayers poses strict negative pressure stability limit in biological liquids. Proc. Natl. Acad. 45 46 844 Sci. USA, 117, 10733-10739. https://doi.org/10.1073/pnas.1917195117 47 48 845 Kanduč, M., Schneck, E. and Netz, R.R. (2013) Hydration interaction between phospholipid 49 50 846 membranes: Insight into different measurement ensembles from atomistic molecular dynamics 51 52 847 simulations. Langmuir, 29, 9126-9137. https://doi.org/10.1021/la401147b 53 54 848 Kates, M. (1957) Effects of solvents and surface-active agents on plastid phosphatidase C activity. Can. 55 56 849 J. Biochem. Physiol., 35, 127-142. https://doi.org/10.1139/o57-016 57 58 37 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 38 of 54

1 2 3 4 850 Kolesnikov, Y.S., Nokhrina, K.P., Kretynin, S.V., Volotovski, I.D., Martinec, J., Romanov, G.A. 5 6 851 and Kravets, V.S. (2012) Molecular structure of phospholipase D and regulatory mechanisms of 7 8 852 its activity in plant and animal cells. Biochem. (Moscow), 77, 1-14. 9 10 853 https://doi.org/10.1134/s0006297912010014 11 12 854 Krasikov, V., Dekker, H.L., Rep, M. and Takken, F.L.W. (2011) The tomato xylem sap protein 13 14 855 XSP10 is required for full susceptibility to Fusarium wilt disease. J. Exp. Bot., 62, 963-973. 15 16 856 https://doi.org/10.1093/jxb/erq327 17 18 857 Krishnan, H.B., Natarajan, S.S., Bennett, J.O. and Sicher, R.C. (2011) Protein and metabolite 19 20 858 composition of xylem sap from field-grown soybeans (Glycine max). Planta, 233, 921-931. 21 22 859 https://doi.org/10.1007/s00425-011-1352-9 23 24 860 Kwan, J.J. and Borden, M.A. (2010) Microbubble dissolution in a multigas environment. Langmuir, 26, 25 26 861 6542-6548. https://doi.org/10.1021/la904088p 27 28 862 Li, N., Xu, C., Li-Beisson, Y. and Philippar, K. (2016) Fatty acid and lipid transport in plant cells. 29 30 863 Trends Plant Sci., 21, 145-158. https://doi.org/10.1016/j.tplants.2015.10.011 31 32 864 Ligat, L., Lauber, E., Albenne, C., Clemente, H.S., Valot, B., Zivy, M., Pont-Lezica, R., Arlat, M. 33 34 865 and Jamet, E. (2011) Analysis of the xylem sap proteome of Brassica oleracea reveals a high 35 36 866 content in secreted proteins. Proteomics, 11, 1798-1813. https://doi.org/10.1002/pmic.201000781 37 38 867 Lima, M.R.M., Machado, A.F. and Gubler, W.D. (2017) Metabolomic study of Chardonnay 39 40 868 grapevines double stressed with esca-associated fungi and drought. Phytopathol., 107, 669-680. 41 42 869 https://doi.org/10.1094/phyto-11-16-0410-r 43 44 870 Martínez-Cabrera, H.I., Jones, C.S., Espino, S. and Schenk, H.J. (2009) Wood anatomy and wood 45 46 871 density in shrubs: Responses to varying aridity along transcontinental transects. Amer. J. Bot., 96, 47 48 872 1388-1398. https://doi.org/10.3732/ajb.0800237 49 50 873 McLoughlin, F. and Testerink, C. (2013) Phosphatidic acid, a versatile water-stress signal in roots. 51 52 874 Front. Plant Sci., 4, 525. https://doi.org/10.3389/fpls.2013.00525 53 54 875 Misra, B.B. (2016) The black-box of plant apoplast lipidomes. Front. Plant Sci., 7, 323. 55 56 876 https://doi.org/10.3389/fpls.2016.00323 57 58 38 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 39 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 877 Mongrand, S., Bessoule, J.-J., Cabantous, F. and Cassagne, C. (1998) The C16:3\C18:3 fatty acid 5 6 878 balance in photosynthetic tissues from 468 plant species. Phytochem., 49, 1049-1064. 7 8 879 https://doi.org/10.1016/S0031-9422(98)00243-X 9 10 880 Morris, H., Gillingham, M.A.F., Plavcová, L., Gleason, S.M., Olson, M.E., Coomes, D.A., Fichtler, 11 12 881 E., Klepsch, M.M., Martínez-Cabrera, H.I., McGlinn, D.J., Wheeler, E.A., Zheng, J., 13 14 882 Ziemińska, K. and Jansen, S. (2018a) Vessel diameter is related to amount and spatial 15 16 883 arrangement of axial parenchyma in woody angiosperms. Plant Cell Environ., 41, 245-260. 17 18 884 https://doi.org/10.1111/pce.13091 19 20 885 Morris, H. and Jansen, S. (2016) Secondary xylem parenchyma - from classical terminology to 21 22 886 functional traits. IAWA J., 37, 1-15. https://doi.org/10.1163/22941932-20160117 23 24 887 Morris, H., Plavcová, L., Cvecko, P., Fichtler, E., Gillingham, M.A.F., Martínez-Cabrera, H.I., 25 26 888 McGlinn, D.J., Wheeler, E., Zheng, J., Ziemińska, K. and Jansen, S. (2016) A global analysis 27 28 889 of parenchyma tissue fractions in secondary xylem of seed plants. New Phytol., 209, 1553-1565. 29 30 890 https://doi.org/10.1111/nph.13737 31 32 891 Morris, H., Plavcová, L., Gorai, M., Klepsch, M., Kotowska, M., Schenk, H.J. and Jansen, S. 33 34 892 (2018b) Vessel-associated cells in angiosperm xylem: Highly specialized living cells at the 35 36 893 symplast-apoplast boundary. Amer. J. Bot., 105, 153-162. https://doi.org/doi:10.1002/ajb2.1030 37 38 894 Nardini, A., Dimasi, F., Klepsch, M. and Jansen, S. (2012) Ion-mediated enhancement of xylem 39 40 895 hydraulic conductivity in four Acer species: relationships with ecological and anatomical features. 41 42 896 Tree Physiol., 32, 1434-1441. https://doi.org/10.1093/treephys/tps107 43 44 897 Nardini, A., Salleo, S. and Jansen, S. (2011) More than just a vulnerable pipeline: xylem physiology in 45 46 898 the light of ion-mediated regulation of plant water transport. J. Exp. Bot., 62, 4701-4718. 47 48 899 https://doi.org/10.1093/jxb/err208 49 50 900 Nardini, A., Savi, T., Losso, A., Petit, G., Pacilè, S., Tromba, G., Mayr, S., Trifilò, P., Lo Gullo, 51 52 901 M.A. and Salleo, S. (2017) X-ray microtomography observations of xylem embolism in stems of 53 54 902 Laurus nobilis are consistent with hydraulic measurements of percentage loss of conductance. 55 56 903 New Phytol., 213, 1068-1075. https://doi.org/10.1111/nph.14245 57 58 39 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 40 of 54

1 2 3 4 904 Oertli, J.J. (1971) The stability of water under tension in the xylem. Z. Pflanzenphysiol., 65, 195-209 5 6 905 Ohki, S. and Ohshima, H. (1985) Divalent cation-induced phosphatidic acid membrane fusion. Effect of 7 8 906 ion binding and membrane surface tension. Biochim. Biophys. Acta, 812, 147-154. 9 10 907 https://doi.org/10.1016/0005-2736(85)90532-2 11 12 908 Park, J., Go, T., Ryu, J. and Lee, S.J. (2019) Air spreading through wetted cellulose membranes: 13 14 909 Implications for the safety function of hydraulic valves in plants. Phys. Rev. E, 100, 032409. 15 16 910 https://doi.org/10.1103/PhysRevE.100.032409 17 18 911 Schenk, H.J., Espino, S., Rich-Cavazos, S.M. and Jansen, S. (2018) From the sap’s perspective: the 19 20 912 nature of vessel surfaces in angiosperm xylem. Amer. J. Bot., 105, 174-187. 21 22 913 https://doi.org/10.1002/ajb2.1034 23 24 914 Schenk, H.J., Espino, S., Romo, D.M., Nima, N., Do, A.Y.T., Michaud, J.M., Papahadjopoulos- 25 26 915 Sternberg, B., Yang, J., Zuo, Y., Steppe, K. and Jansen, S. (2017) Xylem surfactants introduce 27 28 916 a new element to the cohesion-tension theory. Plant Physiol., 173, 1177-1196. 29 30 917 https://doi.org/10.1104/pp.16.01039 31 32 918 Schenk, H.J., Espino, S., Visser, A. and Esser, B.K. (2016) Dissolved atmospheric gas in xylem sap 33 34 919 measured with membrane inlet mass spectrometry. Plant Cell Environ., 39, 944-950. 35 36 920 https://doi.org/10.1111/pce.12678 37 38 921 Schenk, H.J., Steppe, K. and Jansen, S. (2015) Nanobubbles: a new paradigm for air-seeding in xylem. 39 40 922 Trends Plant Sci., 20, 199-205. https://doi.org/10.1016/j.tplants.2015.01.008 41 42 923 Schmid, R. and Machado, R.D. (1968) Pit membranes in hardwoods—Fine structure and development. 43 44 924 Protopl., 66, 185-204. https://doi.org/10.1007/BF01252532 45 46 925 Schneider, C.A., Rasband, W.S. and Eliceiri, K.W. (2012) NIH Image to ImageJ: 25 years of image 47 48 926 analysis. Nat. Meth., 9, 671. https://doi.org/10.1038/nmeth.2089 49 50 927 Schneider, H., Manz, B., Westhoff, M., Mimietz, S., Szimtenings, M., Neuberger, T., Faber, C., 51 52 928 Krohne, G., Haase, A., Volke, F. and Zimmermann, U. (2003) The impact of lipid distribution, 53 54 929 composition and mobility on xylem water refilling of the resurrection plant Myrothamnus 55 56 930 flabellifolia. New Phytol., 159, 487-505. https://doi.org/10.1046/j.1469-8137.2003.00814.x 57 58 40 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 41 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 931 Schurr, U. (1998) Xylem sap sampling – new approaches to an old topic. Trends Plant Sci., 3, 293-298. 5 6 932 https://doi.org/10.1016/S1360-1385(98)01275-8 7 8 933 Scott, F.M., Sjaholm, V. and Bowler, E. (1960) Light and electron microscope studies of the primary 9 10 934 xylem of Ricinus communis. Amer. J. Bot., 47, 162-173. https://doi.org/10.2307/2439245 11 12 935 Shimojima, M. (2011) Biosynthesis and functions of the plant sulfolipid. Progr. Lid. Res., 50, 234-239. 13 14 936 https://doi.org/10.1016/j.plipres.2011.02.003 15 16 937 Shiva, S., Enninful, R., Roth, M.R., Tamura, P., Jagadish, K. and Welti, R. (2018) An efficient 17 18 938 modified method for plant leaf lipid extraction results in improved recovery of phosphatidic acid. 19 20 939 Plant Methods, 14, 14. https://doi.org/10.1186/s13007-018-0282-y 21 22 940 Shiva, S., Vu, H.S., Roth, M.R., Zhou, Z., Marepally, S.R., Nune, D.S., Lushington, G.H., 23 24 941 Visvanathan, M. and Welti, R. (2013) Lipidomic analysis of plant membrane lipids by direct 25 26 942 infusion tandem mass spectrometry. Methods Mol. Biol., 1009, 79-91. 27 28 943 https://doi.org/10.1007/978-1-62703-401-2_9 29 30 944 Smith, A.M. (1994) Xylem transport and the negative pressures sustainable by water. Ann. Bot., 74, 647- 31 32 945 651. https://doi.org/10.1006/anbo.1994.1166 33 34 946 Sperry, J.S., Hacke, U.G. and Wheeler, J.K. (2005) Comparative analysis of end wall resistivity in 35 36 947 xylem conduits. Plant Cell Environ., 28, 456-465. https://doi.org/10.1111/j.1365- 37 38 948 3040.2005.01287.x 39 40 949 Sperry, J.S. and Tyree, M.T. (1988) Mechanism of water stress-induced xylem embolism. Plant 41 42 950 Physiol., 88, 581-587. https://doi.org/10.1104/pp.88.3.581 43 44 951 Stroock, A.D., Pagay, V.V., Zwieniecki, M.A. and Holbrook, N.M. (2014) The physicochemical 45 46 952 hydrodynamics of vascular plants. Annu. Rev. Fluid Mechan., 46, 615-642. 47 48 953 https://doi.org/10.1146/annurev-fluid-010313-141411 49 50 954 Su, W., Yeku, O., Olepu, S., Genna, A., Park, J.-S., Ren, H., Du, G., Gelb, M.H., Morris, A.J. and 51 52 955 Frohman, M.A. (2009) 5-Fluoro-2-indolyl des-chlorohalopemide (FIPI), a Phospholipase D 53 54 956 pharmacological inhibitor that alters cell spreading and inhibits chemotaxis. Molec. Pharmacol., 55 56 957 75, 437-446. https://doi.org/10.1124/mol.108.053298 57 58 41 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 42 of 54

1 2 3 4 958 Testerink, C. and Munnik, T. (2011) Molecular, cellular, and physiological responses to phosphatidic 5 6 959 acid formation in plants. J. Exp. Bot., 62, 2349-2361. https://doi.org/10.1093/jxb/err079 7 8 960 Tyree, M.T., Salleo, S., Nardini, A., Lo Gullo, M.A. and Mosca, R. (1999) Refilling of embolized 9 10 961 vessels in young stems of laurel. Do we need a new paradigm? Plant Physiol., 120, 11-21. 11 12 962 https://doi.org/10.1104/pp.120.1.11 13 14 963 Tyree, M.T. and Zimmermann, M.H. (2002) Xylem structure and the ascent of sap 2nd edn. Berlin: 15 16 964 Springer-Verlag. 17 18 965 Venturas, M.D., Sperry, J.S. and Hacke, U.G. (2017) Plant xylem hydraulics: What we understand, 19 20 966 current research, and future challenges. J. Integr. Plant Biol., 59, 356-389. 21 22 967 https://doi.org/doi:10.1111/jipb.12534 23 24 968 Wagner, H.J., Schneider, H., Mimietz, S., Wistuba, N., Rokitta, M., Krohne, G., Haase, A. and 25 26 969 Zimmermann, U. (2000) Xylem conduits of a resurrection plant contain a unique lipid lining and 27 28 970 refill following a distinct pattern after desiccation. New Phytol., 148, 239-255. 29 30 971 https://doi.org/10.1046/j.1469-8137.2000.00755.x 31 32 972 Welti, R., Li, W., Li, M., Sang, Y., Biesiada, H., Zhou, H.-E., Rajashekar, C.B., Williams, T.D. and 33 34 973 Wang, X. (2002) Profiling membrane lipids in plant stress responses: Role of phospholipase Da 35 36 974 in freezing-induced lipid changes in Arabidopsis. J. Biol. Chem., 277, 31994-32002. 37 38 975 https://doi.org/10.1074/jbc.M205375200 39 40 976 Weschayanwiwat, P., Scamehorn, J. and Reilly, P. (2005) Surfactant properties of low molecular 41 42 977 weight phospholipids. J. Surfact. Deterg., 8, 65-72. https://doi.org/10.1007/s11743-005-0332-8 43 44 978 Westhoff, M., Schneider, H., Zimmermann, D., Mimietz, S., Stinzing, A., Wegner, L.H., Kaiser, W., 45 46 979 Krohne, G., Shirley, S., Jakob, P., Bamberg, E., Bentrup, F.W. and Zimmermann, U. (2008) 47 48 980 The mechanisms of refilling of xylem conduits and bleeding of tall birch during spring. Plant 49 50 981 Biol., 10, 604-623. https://doi.org/10.1111/j.1438-8677.2008.00062.x 51 52 982 Wheeler, J.K., Sperry, J.S., Hacke, U.G. and Huang, N. (2005) Inter-vessel pitting and cavitation in 53 54 983 woody Rosaceae and other vesseled plants: a basis for a safety vs. efficiency trade-off in xylem 55 56 984 transport. Plant Cell Environ., 28, 800-812. https://doi.org/10.1111/j.1365-3040.2005.01330.x 57 58 42 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 43 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 985 Wheeler, T.D. and Stroock, A.D. (2008) The transpiration of water at negative pressures in a synthetic 5 6 986 tree. Nature, 455, 208-212. https://doi.org/10.1038/nature07226 7 8 987 Wisniewski, M. and Davis, G. (1995) Immunogold localization of pectins and glycoproteins in tissues of 9 10 988 peach with reference to deep supercooling. Trees - Struct. Funct., 9, 253-260. 11 12 989 https://doi.org/10.1007/BF00202015 13 14 990 Wisniewski, M., Davis, G. and Arora, R. (1991a) Effect of macerase, oxalic acid, and EGTA on deep 15 16 991 supercooling and pit membrane structure of xylem parenchyma of peach. Plant Physiol., 96, 17 18 992 1354-1359. https://doi.org/10.1104/pp.96.4.1354 19 20 993 Wisniewski, M., Davis, G. and Schaffer, K. (1991b) Mediation of deep supercooling of peach and 21 22 994 dogwood by enzymatic modifications in cell-wall structure. Planta, 184, 254-260. 23 24 995 https://doi.org/10.1007/BF00197955 25 26 996 Yang, J., Michaud, J.M., Jansen, S., Schenk, H.J. and Zuo, Y.Y. (2020) Dynamic surface tension of 27 28 997 xylem sap lipids. Tree Physiol., 40, 433-444. https://doi.org/10.1093/treephys/tpaa006 29 30 998 Zhang, H., Fan, Q., Wang, Y.E., Neal, C.R. and Zuo, Y.Y. (2011) Comparative study of clinical 31 32 999 pulmonary surfactants using atomic force microscopy. Biochim. Biophys. Acta (BBA) - 33 34 1000 Biomembranes, 1808, 1832-1842. https://doi.org/10.1016/j.bbamem.2011.03.006 35 36 1001 Zhang, Y., Carmesin, C., Kaack, L., Klepsch, M.M., Kotowska, M., Matei, T., Schenk, H.J., Weber, 37 38 1002 M., Walther, P., Schmidt, V. and Jansen, S. (2020) High porosity with tiny pore constrictions 39 40 1003 and unbending pathways characterise the 3D structure of intervessel pit membranes in 41 42 1004 angiosperm xylem. Plant Cell Environ., 43, 116-130. https://doi.org/10.1111/pce.13654 43 44 1005 Zhang, Y., Klepsch, M. and Jansen, S. (2017) Bordered pits in xylem of vesselless angiosperms and 45 46 1006 their possible misinterpretation as perforation plates. Plant Cell Environ., 40, 2133-2146. 47 48 1007 https://doi.org/10.1111/pce.13014 49 50 1008 Zhou, Z., Marepally, S.R., Nune, D.S., Pallakollu, P., Ragan, G., Roth, M.R., Wang, L., Lushington, 51 52 1009 G.H., Visvanathan, M. and Welti, R. (2011) LipidomeDB data calculation environment: online 53 54 1010 processing of direct-infusion mass spectral data for lipid profiles. Lipids, 46, 879-884. 55 56 1011 https://doi.org/10.1007/s11745-011-3575-8 57 58 43 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 44 of 54

1 2 3 4 1012 Zimmermann, U., Schneider, H., Wegner, L.H. and Haase, A. (2004) Water ascent in tall trees: does 5 6 1013 evolution of land plants rely on a highly metastable state? New Phytol., 162, 575-615. 7 8 1014 https://doi.org/10.1111/j.1469-8137.2004.01083.x 9 10 1015 Zuo, Y.Y., Ding, M., Li, D. and Neumann, A.W. (2004) Further development of Axisymmetric Drop 11 12 1016 Shape Analysis-Captive Bubble for pulmonary surfactant related studies. Biochimica et 13 14 1017 Biophysica Acta (BBA) - General Subjects, 1675, 12-20. 15 16 1018 https://doi.org/10.1016/j.bbagen.2004.08.003 17 18 1019 Zuo, Y.Y., Keating, E., Zhao, L., Tadayyon, S.M., Veldhuizen, R.A.W., Petersen, N.O. and 19 20 1020 Possmayer, F. (2008) Atomic force microscopy studies of functional and dysfunctional 21 22 1021 pulmonary surfactant films. I. Micro- and nanostructures of functional pulmonary surfactant films 23 24 1022 and the effect of SP-A. Biophys. J., 94, 3549-3564. https://doi.org/10.1529/biophysj.107.122648 25 26 1023 Zwieniecki, M.A., Melcher, P.J. and Holbrook, N.M. (2001) Hydrogel control of xylem hydraulic 27 28 1024 resistance in plants. Science, 291, 1059-1062. https://doi.org/10.1126/science.1057175 29 30 31 1025 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 44 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 45 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 1027 FIGURE LEGENDS 4 5 6 1028 FIG. 1. Experimental setup for xylem sap extraction via the vacuum method designed to ensure 7 8 1029 that lipids in sap only come into contact with glass surfaces and are cooled immediately after 9 10 1030 extraction to inhibit enzyme activity. Also shown are the proximal xylem surface at the 11 12 1031 collection end of the stem from which cell contamination controls were collected after high- 13 14 1032 pressure cleaning of the surface and the successive distal cuts that open vessels for vacuum 15 16 1033 extraction. In lipid tracer (di17:0 PE) experiments, the tracer was added to the freshly cut distal 17 18 1034 surfaces of the last three cuts (50, 45, and 40 cm stem length). See Methods section for detailed 19 20 1035 explanations. 21 22 23 1036 FIG. 2. Polar lipid composition of xylem sap from seven angiosperm species (n = 3). A. Chemical 24 25 1037 composition of xylem sap lipids as determined by direct-infusion electrospray ionization triple- 26 27 1038 quadrupole mass spectrometry. See Table S1 for the standards used and Figures S1-9 and Tables 28 29 1039 S2-3 [Supplementary Information] for the complete data sets, including chains lengths and 30 31 1040 degrees of saturation. Sulfolipids were only found in trace amounts (Tables S2 and S3) and are 32 33 1041 not included in this figure. B. Most common numbers of acyl chain carbon atoms (mean ± SE) in 34 35 1042 xylem sap lipids (PL = phospholipids, GL = galactolipids) and their degree of unsaturation as 36 37 1043 determined by direct-infusion electrospray ionization triple-quadrupole mass spectrometry. See 38 39 1044 Table S2 [Supplementary Information] and Figures S1-9 for the complete data set. C. Total 40 41 1045 concentration of phospho- and galactolipids in xylem sap (mean ± SE) as a function of total 42 43 1046 vessel xylem perimeter at the cut xylem surface (mean ± SE). Numbers under the abbreviated 44 45 1047 generic names are the median vessel lengths in cm. Abbreviations: Bam = Bambusa, Dis = 46 47 1048 Distictis, Enc = Encelia, Gei = Geijera, Lau = Laurus, Lir = Liriodendron, Tri = Triadica. D. 48 49 1049 Polar lipid composition of xylem sap compared via principal component analysis (PCA), with 50 51 1050 the first two components shown that explain 63% of the variation among species. Shaded areas 52 53 1051 are 95% confidence regions. 54 55 56 57 58 45 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 46 of 54

1 2 3 4 1052 FIG. 3. Polar lipid composition compared between xylem sap (n = 8) of Geijera sampled in July 5 6 1053 2019 and March 2020. A. Results from a principal component analysis (PCA), with the first two 7 8 1054 components shown. Shaded areas are 95% confidence regions. B. Heatmaps for polar lipids in, 9 10 1055 based on Euclidean distance measures, a Ward clustering algorithm, and normalized data to 11 12 1056 visualize differences between sampling times. The lipids shown are the 17 most different 13 14 1057 between sampling times based on unpaired t-tests, adjusted for false discovery rates (FDR). All 15 16 1058 lipids in this heatmap were significantly different (p < 0.05) between dates after FDR adjustment. 17 18 19 1059 FIG. 4. Polar lipid composition compared between paired samples of xylem sap and wood of the 20 21 1060 same stems in Distictis and Geijera (n = 4). A. and C. show results from principal component 22 23 1061 analyses (PCA) for Distictis and Geijera, respectively, with the first two components shown. 24 25 1062 Shaded areas are 95% confidence regions. B. and D. are heatmaps for polar lipids in Distictis and 26 27 1063 Geijera, respectively, based on normalized data, Euclidean distance measures, and a Ward 28 29 1064 clustering algorithm to visualize differences between xylem sap and wood. The lipids shown are 30 31 1065 the 20 most different between sap and wood based on paired t-tests, adjusted for false discovery 32 33 1066 rates (FDR). Lipids indicated in bold were flagged as different at p < 0.05 between xylem sap 34 35 1067 and wood after paired t-tests with FDR adjustment. Note that lipids in xylem sap and wood 36 37 1068 samples are not independent of each other, so t-tests were used only as qualitative indicators. 38 39 1069 FIG. 5. Lipids in xylem and inter-vessel pits of Laurus nobilis. A-C: Super-resolution confocal 40 41 1070 laser scanning microscopy images (false color) showing lignified cell walls in blue based on 42 43 1071 lignin autofluorescence and FM1-43 dye bound to lipids in yellow. White arrows = intervessel 44 45 1072 pit membranes. A-B: With FM1-43, C: Control without FM1-43. D-G: Transmission electron 46 47 1073 micrographs of intervessel pits. D-E: without OsO4; showing unusually dark, electron-dense 48 49 1074 homogeneous pit membranes and linings on secondary walls, but no granular appearance. F-G: 50 51 1075 with OsO4, lipids visible as black aggregates lining the wall surface, pit chamber and inside the 52 53 1076 pit membrane. Abbreviations: PA = pit aperture, PM = pit membrane, SW = secondary wall. 54 55 56 57 58 46 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 47 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 1077 FIG. 6. Conceptual model of the formation of lipid-coated nanobubbles via snap-off inside pit 5 6 1078 membranes that are coated and infiltrated with surface-active lipids. The different sizes of 7 8 1079 micelles shown in this image reflect the wide range of nanoparticle sizes detected in xylem sap 9 10 1080 and presumed to be bilayer to multilayer micelles, vesicles, and lipid-coated nanobubbles 11 12 1081 (Schenk et al. 2017). 13 14 1082 15 16 1083 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 48 of 54

1 2 3 1084 SUPPLEMENTARY MATERIALS 4 5 1085 Supporting figures: 6 7 1086 Figure S1. . Xylem sap concentrations of digalactosyldiacylglycerol (DGDG) in seven 8 1087 angiosperms species. 9 10 1088 Figure S2. Xylem sap concentrations of monogalactosyldiacylglycerol (MGDG) in seven 11 12 1089 angiosperms species. 13 1090 Figure S3. Xylem sap concentration of phosphatidylglycerol (PG) in seven angiosperms species. 14 15 1091 Figure S4. Xylem sap concentration of phosphatidylcholine (PC) in seven angiosperms species. 16 17 1092 Figure S5. Xylem sap concentration of phosphatidylethanolamine (PE) in seven angiosperms 18 1093 species. 19 20 1094 Figure S6. Xylem sap concentrations of phosphatidylinositol (PI) in seven angiosperms species. 21 22 1095 Figures S7. Xylem sap concentrations of phosphatidylserine (PS) in seven angiosperms species. 23 24 1096 Figures S8. Xylem sap concentrations of phosphatidic acid (PA) in seven angiosperms species. 25 26 1097 Figure S9. Xylem sap concentrations of lysophosphatidylethanolamine (LPE) and 27 1098 lysophosphatidylcholine (LPC) in seven angiosperms species. 28 29 1099 Figure S10. Stems infiltrated with blue-colored silicone, cut at 0.7 cm from the injected surface. 30 The four spoke-like structures in Distictis buccinatoria xylem are areas of 31 1100 32 1101 interxylary phloem. 33 34 1102 Figure S11. Total vessel area filled with silicone as a function of distance from the infiltrated 35 1103 surface, shown for a sample of Triadica sebifera. The silicone-filled vessel volume 36 1104 was calculated from these filled areas by linear interpolation. 37 38 1105 Fig. S12. Principal component analyses (PCA) comparing polar lipid composition in xylem sap 39 1106 for six angiosperm species to paired cell contamination controls from the same 40 41 1107 stems. 42 1108 Supporting tables: 43 44 1109 Table S1. Standards used for mass spectrometry of xylem sap lipids. 45 46 1110 Table S2. Concentrations of polar lipids and sulfolipids in xylem sap. 47 48 1111 Table S3. Concentrations of polar lipids and sulfolipids in cell contamination controls for xylem 49 1112 sap. 50 51 1113 Supporting experimental procedures: 52 53 1114 Methods S1. Methods comparison of lipid extraction by partial freeze drying vs. SpeedVac. 54 55 1115 56 57 58 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 49 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 50 of 54

1 2 0.7 35 )

3 ‐ 1 L

A B 4 0.6 30 L) lipids

/ 5 sap (µmol 25 6 0.5 (µmol

7 sap 20 xylem in 0.4 8 lipids 9 15 polar sap 10 0.3 all 11 10 of

xylem 0.2 12

concentration 5 13 14 Total 0.1 Percent

Lipid 0 15 34:1 34:2 34:3 36:3 36:4 36:5 34:1 34:2 34:3 36:3 36:4 36:5 16 0.0 PL PL PL PL PL PL GL GL GL GL GL GL Laurus Liriodendron Bambusa Triadica Geijera Distictis Encelia 17 Total acyl C:Total C‐C double bonds DGDG MGDG PA PC PE PI PS Others 18 19 galactolipids phospholipids 20 D 21 0.8 22 r2 = 0.542; p = 0.0214

) )

1 C

‐ Bam 23 ‐ 1 0.7 L L

6.57 24 ‐ 25 0.6 (µmol (µmol

26 Dis 0.5 sap sap

27 Enc Tri 10.42 in in

28 3.02 0.4 4.25 Gei 29 3.28 30 0.3 31 32 0.2 concentration concentration

33 Lau Lir 34 0.1 2.99 1.32 Lipid 35 Lipid 36 0 37 0 500 1000 1500 38 Total vessel perimeter at the cut xylem surface (mm) 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 51 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 5 6 7 8 9 10 11 A B 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 July 2019 March 2020 28 lower higher 29 ‐1 0 1 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 52 of 54

1 2 3 A B 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Wood Sap 21 22 23 24 25 C D 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 Wood Sap 42 lower higher 43 44 ‐1 0 1 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not Page 53 of 54 certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 bioRxiv preprint doi: https://doi.org/10.1101/763771; this version posted October 12, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Page 54 of 54

1 2 3 4 5 6 7 Pit membrane 8 9 10 Sap Air 11 Lipid micelles 12 13 14 15 16 Nanobubble snap‐off 17 18 Lipid coat 19 20 Nanobubble 21 22 23 24 25 26 27 Cellulose microfibril 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60