Quantification and Discovery of Acyl-ACPs by LC-MS/MS

Acyl carrier proteins (ACPs) are the scaffolds for fatty acid biosynthesis in living systems, rendering them essential to a comprehensive understanding of lipid metabolism; however, accurate quantitative methods to assess individual acyl-ACPs do not exist. A robust method was developed to quantify acyl-ACPs at picogram levels. Acyl-ACP elongation intermediates (3-hydroxyacyl-ACPs and 2, 3-trans-enoyl-ACPs), and unexpected medium chain (C10:1, C14:1) and polyunsaturated long chain acyl-ACPs (C16:3) were also identified, indicating the sensitivity of the method and that descriptions of lipid metabolism and ACP function are incomplete. Such ACPs are likely important to medium chain lipid production for fuels and highlight poorly understood lipid remodeling events in the chloroplast. The approach is broadly applicable to Type II FAS systems found in plants, bacteria, and mitochondria of animal and fungal systems because it uses a strategy that capitalizes on a highly conserved Asp-Ser-Leu-Asp (DSLD) amino acid sequence in ACPs to which acyl groups are attached. This allows for sensitive quantification using LC-MS/MS with de novo generated standards and an isotopic dilution strategy and will fill a gap in understanding, providing insights through quantitative exploration of fatty acid biosynthesis processes for optimal biofuels, renewable feed stocks, and medical studies in health and disease.

unexpected medium chain (C10:1, C14:1) and polyunsaturated long chain acyl-ACPs 23 (C16:3) were also identified, indicating the sensitivity of the method and that descriptions 24 of lipid metabolism and ACP function are incomplete. Such ACPs are likely important to 25 medium chain lipid production for fuels and highlight poorly understood lipid remodeling 26 events in the chloroplast. The approach is broadly applicable to Type II FAS systems 27 found in plants, bacteria, and mitochondria of animal and fungal systems because it uses 28 a strategy that capitalizes on a highly conserved Asp-Ser-Leu-Asp (DSLD) amino acid 29 sequence in ACPs to which acyl groups are attached. This allows for sensitive 30 quantification using LC-MS/MS with de novo generated standards and an isotopic 31 dilution strategy and will fill a gap in understanding, providing insights through 32 quantitative exploration of fatty acid biosynthesis processes for optimal biofuels, 33 renewable feed stocks, and medical studies in health and disease. 34

INTRODUCTION 36
The synthesis of fatty acyl chains is essential for the production of storage and 37 membrane lipids and functional molecules that modulate gene expression or contribute to 38 protein activity. Studies on acyl chains span from disease and nutrition to demands for 39 renewable fuels and feedstocks (Uauy et al., 2000;Peralta-Yahya et al., 2012;Zhu et al., 40 2016;Vanhercke et al., 2019). The lengths of up to 18 carbons are produced during fatty 41 acid biosynthesis (Li-Beisson et al., 2013) on an acyl carrier protein scaffold (Overath 42 and Stumpf, 1964) which is connected to the acyl chain through linkage of a 4'-43 phosphopantetheine group to a serine residue of the protein (ACP; Fig. 1a). The acyl 44 chains are elongated by a series of four enzymatic steps that link ACP intermediates and 45 form a cycle that results in the addition of two carbons and four hydrogens. Two distinct 46 types of fatty acid synthase (FAS) complexes perform the repeated cycle of ketoacyl 47 synthesis, reduction, dehydration, and a second reduction producing long chain saturated 48 hydrocarbons (Fig. 1b). Yeast and mammals use a single large multifunctional FAS 49 protein (Type I) that includes multiple domains for the enzymatic steps and ACP region 50 (Bressler and Wakil, 1961;Hsu et al., 1965), whereas fatty acid synthesis in plants and 51 bacteria occur through the concerted action of individual proteins within a complex 52 including a distinct ACP that is 9-15 kDa in size (Alberts et al., 1963;White et al., 2005;53 Cronan, 2014). 54 In plants, the production of acyl chains take place predominantly in the 55 chloroplast, though also to a lesser extent in the mitochondria, using acyl-ACP 56 intermediates in both instances (Ohlrogge et al., 1979;Chuman and Brody, 1989). The 57 chain elongation process is terminated by thioesterase reactions that release a non-58 esterified fatty acid from the ACP and regenerate the ACP substrate pool for further fatty 59 acid biosynthetic reactions. The released fatty acid can then be activated to an acyl-CoA 60 to make glycerolipids for storage oil production, waxes, surface lipids or membrane 61 biogenesis (Li-Beisson et al., 2013). 62 Given the central role of acyl-ACPs in fatty acid biosynthesis, their quantitative 63 analysis can help our understanding of lipid metabolism; however, the acyl group is a 64 small percentage of the total acyl-ACP weight (usually significantly less than 3%) and 65 current methods are confined by biochemical approaches that analyze intact proteins or 66 spectrometer as described in the methods (Fig. 3a). The masses calculated from chemical 167 composition along with the multiple reaction monitoring (MRM) mode of analysis of the 168 tandem MS were used to confirm the products of acyl-ACP fragmentation and optimize 169 instrument performance. Two mass losses were consistently observed for all acyl-ACP 170 digestion products indicating that they were not specific to the changing size in acyl 171 chain, but instead components of the redundant 4'-phosphopantetheine and three amino 172 acid sequence (Fig. 1a). A product ion that differed from the precursor by 315.1 indicated 173 the loss of the three amino acids (DSL). The additional loss of phosphate increased the 174 difference in mass to 413.1 and resulted in the primary product ion observed by mass 175 spectrometry (Fig. 3b). Malonyl-ACP had an additional product ion with a difference in 176 mass of 457.1, that indicated the loss of the DSL, phosphate, and the carboxyl from the 177 malonyl-group and is the primary product ion for this molecule. The 15 N labeled 178 standards are 3 Da heavier giving a precursor-product ion difference of 318.1 and 460.1. 179 The changes in abundance were used to optimize the declustering potential (DP), 180 collision energy (CE) and collision cell exit potential from standards and acyl-ACPs 181 isolated from plant biomass (Fig. 3c). Optimal chromatography and detection on the mass 182 spectrometer of acyl-ACP digested products was indicated by spectra with well-defined 183 peaks that were also evenly spaced to represent the change in acyl mass by consistent 184 increments of C2H4. All acyl species from acetyl-ACP to 18 carbon acyl chains (i.e. 18:0-185 ACP) were identified in addition to non-esterified (i.e. holo) and apo-ACP. The dynamic 186 range (approximately 3-4 orders of magnitude) was established through comparison of 187 integrated peak areas for each acyl-ACP, with the exception of holo-and malonyl-ACP 188 due to dimerization and instability, from standards and Camelina samples and indicated 189 that femtomole (i.e. picogram of protein) quantities could be quantified accurately (Fig. 190 3c, Supplemental Figure S5). Since pooled acyl-ACPs are found at low concentrations 191 (10 ng/mg FW in spinach chloroplast (Ohlrogge et al., 1979)), saturation of the detector 192 is unlikely to be an issue for most preparations. Isotope dilution, a general strategy which relies upon internal standard addition 197 for derivation of calibration curves, is the benchmark for quantitation by mass 198 spectrometry. Isotopically labeled internal standards provide chemically equivalent ions 199 distinguished from sample metabolites, or analytes, by a change in mass corresponding to 200 the isotope label(s) (in this case the addition of three 15 N atoms in place of 14 N shift the 201 mass by ~3 Da). The addition of internal standards at an early stage of sample preparation 202 allows one to normalize data and rectify sample to sample variations in analyte recovery. 203 In addition, the chemical equivalency of isotopically labeled standards can account for 204 differences in observed analyte response factors and address the exposure to differential 205 matrix effects. Thus, an isotope dilution strategy is superior to other approaches that rely 206 on standards that are not chemically equivalent with differences in response factors and 207 subject to different matrix effects amongst analytes that are not accounted for resulting in 208 inaccuracies. Here, an isotope dilution strategy was used to quantify acyl-ACP levels in 209 Camelina seeds and leaves. Analyte acyl-ACPs from plant biomass were spiked with 210 internal standards before extraction and then quantified. Total ACP amounts detected in 211 seeds were ~26 pmol/mgFW composed predominantly of 18:1-ACP (Fig. 4). Camelina 212 leaves exhibited ~9-fold lower levels overall than seeds; consistent with calculated 213 estimates from the literature (~1-2 pmol/mgFW calculated from spinach leaves (Ohlrogge 214 et al., 1979)), and of comparable composition (Post-Beittenmiller et al., 1991). 215 216 Two acyl-ACP intermediates of the fatty acid biosynthetic cycle were identified and 217 previously undescribed unsaturated acyl-ACPs were discovered 218 The process of acyl chain extension in fatty acid synthesis is completed by a cycle 219 that generates three additional acyl-ACP elongation intermediates reflecting the ketoacyl 220 condensing, reducing, and dehydrating steps before the final reduced acyl-ACP. Though 221 these intermediates are believed to be present at low quantities we considered the 222 possibility of their detection. Since the intermediates are themselves ACP products that 223 differ only in the acyl structure and composition, the developed methods were presumed 224 applicable though potentially limited by the instrument sensitivity. 225 Analysis of seeds indicated several additional acyl-ACP peaks. We calculated the 226 molecular weights for the elongation intermediates assuming a loss of 413.1; identical to 227 the first intermediate in the fatty acid biosynthetic cycle (Fig. 1b); though this 231 intermediate was not detected. The 3-hydroxyacyl-ACPs in developing seeds were 232 present at similar amounts to most reduced acyl-ACPs ( Fig. 5a) with heightened levels of 233 C14-hydroxyacyl-ACP that may be involved in Lipid A-like molecule production in the 234 mitochondria (Wada et al., 1997;Li et al., 2011). Peaks for the 2, 3-trans-enoyl-ACP 235 were lower in abundance, generally less than 2.0% of total (C18-enoyl; Fig. 5a). 236 The 2, 3-trans-enoyl-ACP and equivalent chain length single desaturated acyl-237 ACPs are isomeric, therefore, for example, 16:1-ACP (cis-double bond) and C16-enoyl-238 ACP (2, 3-trans-double bond) cannot be discriminated by mass. Consequently, acyl chain 239 MRM transitions indicated a second peak, the 2, 3-trans-enoyl acyl-ACP, at a different 240 retention time. To conclusively identify the isomers, a standard of 18:1-ACP was 241 digested and processed resulting in a peak that eluted at the same retention time as one of 242 the two peaks in question (Supplemental Figure S6). This confirmed the designation of 243 each peak, with the 2, 3-trans-enoyl acyl-ACP characteristically eluting after the cis-244 unsaturated acyl-ACP as a much smaller peak. The 3-hydroxyacyl-ACP products that 245 differ in mass eluted ahead of the corresponding acyl-ACP as presented in a summary of 246 the retention times in Supplemental Table S3. 247 Interestingly, owing to the sensitivity of the method, we observed peaks 248 corresponding to several medium chain unsaturated acyl-ACPs (i.e. C10:1 and C14:1) 249 and a significant amount of polyunsaturated acyl-ACP (16:3) in seeds and leaves (Fig. 250 5b) that have not been previously described. The paired MS/MS transitions, retention 251 times, and peak shapes were all consistent with expectation for these species supporting 252 their identification. The retention time offset (Fig. 5b) was similar to measured 253 differences for 9-cis-18:1-ACP (Fig. S6) and 7-cis-16:1-ACP and indicated that these 254 unsaturated forms likely contain cis double bonds. Enoyl-ACPs contain trans double 255 bonds and have a near-linear predicted 3D structure and therefore more closely 256 approximate saturated ACPs with respect to their physical properties. cis-unsaturated-257 acyl groups are commonly generated within living systems, though are not believed to be 258 biosynthetic intermediates on fatty acid synthase assembly lines; thus, the roles in 259 metabolism for these molecules remains to be elucidated. This effectively eliminates protein size and sequence variation that convolute the 276 population of a single acyl-chain species across multiple possible carriers and enables 277 their detection in forms that are convenient for mass spectrometric analysis. 278 In comparison to described prior attempts the digestion and LC-MS/MS-based 279 analysis does not require antibodies, multiple urea-PAGE gels, or derivatization 280 techniques and requires only mg levels of biological material. Acyl-ACPs that are present 281 at picogram/mgFW quantities can be readily detected without interference from acyl 282 chains esterified to lipids or CoA species that are more abundant in biomass. In addition, 283 the use of the phosphopantetheinyl transferase Sfp, from B. subtilis, which tolerates a 284 wide variety substituents added onto its canonical substrate, 4'-phosphopantetheine, 285 enabled a one-step synthesis of acyl-ACPs starting with acyl-CoAs and ACP ( Fig. 2) 286 (Lambalot et al., 1996). The approach circumvented a more typical, two step enzymatic 287 synthesis beginning with production of holo-ACP, often with the E. coli 288 phosphopantethienyl transferase AcpS, followed by acylation with acyl-ACP synthetase 289 (AAS) to generate acyl-ACPs (Zornetzer et al., 2006). 290 291

Acyl-ACP profiling reveals unanticipated acyl-ACP groups with implications for 292 lipid metabolism in oilseeds 293
We quantified the absolute levels of individual acyl-ACPs containing up to 18 294 carbons for the first time. 18:1-ACP was present in the greatest quantities in seeds and 295 may indicate fatty acid biosynthesis is not a bottleneck for lipid production in oilseeds but 296 could be important to regulation (Andre et al., 2012). In addition, two intermediates of 297 the fatty acid biosynthetic cycle, 3-hydroxyacyl-ACP and 2,3-trans-enoyl-ACP were 298 identified. Interestingly, our approach elucidated other unsaturated acyl-ACPs in addition 299 to 18:1, including medium chain (C10:1 and C14:1) that have unknown function and that 300 have not been previously reported. Speculatively, it would be interesting to determine 301 whether plants or other species that have enhanced medium acyl chain lipid production 302 also have higher levels of these ACP-forms that could indicate differences in the fatty 303 acid biosynthetic machinery which would be targets for metabolic engineering. Higher 304 degrees of unsaturation in C16 acyl chains (i.e. 16:3-ACP) were also observed ( Fig. 5) 305 which may be relevant to lipid remodeling in the chloroplast. The detection of 306 unanticipated acyl-ACPs indicates that the method can serve as a sensitive and 307 quantitative tool for discovery aspects of fatty acid metabolism. 308 309 Absolute levels of Acyl-ACPs are greater in seeds than leaves of oilseeds 310 The levels of acyl-ACPs in seeds were greater than leaves. Oilseeds actively 311 produce storage lipids that can represent the largest fraction of seed biomass whereas 312 leaves make a small amount of lipids primarily for use in membranes (Fan et al., 2014). 313 Similar to more qualitative studies on spinach leaves (Post-Beittenmiller et al., 1991), our 314 method quantified high levels of long chain acyl-ACPs ( Fig. 4a), predominantly 18:1, in 315 seeds. The present method also detected an apo-ACP pool in camelina seed/leaf that was 316 not monitored in the spinach studies. The 18:1-ACP is a terminal product of fatty acid 317 synthesis and may indicate that acyl chain production is not limiting lipid biosynthesis. 318 The presence of an apo-ACP pool could also suggest that ACP abundance is not limiting 319 fatty acid production, but perhaps the activation to holo-form and the source of acetyl 320 groups impact the regulated rate of the fatty acid biosynthetic cycle. Future studies can 321 potentially capitalize on this method through transient isotopic labeling studies (Allen, 322 2016b) to assess synthetic and turnover characteristics in lipid metabolism more 323 rigorously. 324 19 325

The role of acetyl-ACP in plant lipid metabolism and regulation 326
In spinach leaves and seeds, 5-10% of the total ACP is in an acetylated form 327 (Post-Beittenmiller et al., 1991) which is comparable to numbers reported here for 328 Camelina tissues (4% and 11% for seed and leaf, respectively; Fig. 4). In vitro studies 329 have previously indicated that acetyl-ACP is inefficiently but preferentially used by KAS 330 I (as opposed to KAS III) condensation reactions (Jaworski et al., 1993) and it can also be 331  Plasmid pQE-60-eGFP was constructed by ligation of empty pQE-60 vector 377 (Qiagen) and plasmid pEGFP-NI (clontech) both of which had been double digested with 378 NheI and XhoI followed by gel purification. 379 Plasmid pT5Kan (Supplemental Figure S7a) construction began with digestion of 380 pQE60 (Qiagen) with XhoI and NheI which had previously been modified to contain 381 eGFP (pQE60-eGFP (unpublished)) at the NcoI/BamHI sites of the MCS. The gel-382 purified fragment was ligated to the PCR product of pET-28a (Novagen) amplified by 383 primers pET-28-XhoI-F and pET-28-NheI-R which had been similarly digested and gel 384 purified. The cloned product resulted in an intermediary plasmid that was further 385 modified to remove unnecessary intergenic regions in order to make the vector easier to 386 21 work with. The truncation was accomplished by PCR amplification of the intermediary 387 plasmid using primers pT5-Kan-NotI-F and pT5-Kan-NotI-R which amplified the lacI-388 MCS-KanR-ColEI region leaving only 300 bp separating the lacI and ori-ColEI 389 sequences as well as provided a NotI site for digestion and ligation of the amplified 390 fragment. The cloned product, named pT5-Kan was used for cloning and expression of 391 the E. coli acyl carrier protein gene, acpP. 392 Plasmid pT5-Kan-AcpP-His6 was constructed from the PCR product amplified 393 from genomic DNA isolated from E. coli MG1655 using primers Acp-His6-F and Acp-394 His6-R and after digestion with NcoI/BamHI followed by gel purification was ligated to 395 similarly digested and purified pT5-Kan. 396 Plasmid pET-Duet-1-Sfp was constructed from NdeI/KpnI digested PCR product 397 amplified from pUC-18 Sfp (a gift from Christopher T. Walsh (Lambalot et al., 1996)) 398 using primers Sfp-F and Sfp-R. 399 Plasmid pET-28a-Sfp-His6 (Supplemental Figure S7b) was constructed from 400 NcoI/BamHI digested and gel purified PCR product amplified from pET-Duet-1-Sfp 401 using primers Sfp-His6-F and Sfp-His6-R and similarly digested and purified pET-28a. Synthesized acyl-ACP standards were purified by TCA precipitation and the 528 reaction yields were determined. TCA was added to the reaction mixture to a final 529 concentration of 5% and the solution vortexed. The acyl-ACP was then pelleted by 530 centrifugation at 21,000 x g at 4°C, supernatant discarded and washed with one percent 531 TCA followed by centrifugation. Pellets were resuspended in MOPS pH 7.5 prior to use 532 as an internal standard. The yield of each acyl-ACP standard was determined by SDS-533 PAGE on 16.5% tris-tricine gels and confirmed by LC-MS/MS. 534 535

Isotope dilution-based quantitation of ACPs in biological samples 536
Acyl-ACPs from seed and leaf tissue were quantified using internal standards and 537 an isotope dilution-based approach. Unlabeled standards were serially diluted with a 538 constant concentration of 15 N3 acyl-ACP standards. The area ratios of unlabeled and 539 labeled peaks were plotted against the concentration of the unlabeled standards in order 540 26 to construct standard curves. Acyl ACPs from unlabeled camelina were spiked with 15 N3-541 labeled ACP standards then prepared and analyzed as described above with quantities 542 calculated using the aforementioned standard curves. All peak integration was performed 543 using Analyst 1.6 software (SCIEX). Standard curves and quantitative calculations were 544 performed using peak areas exported to Microsoft Excel.   The authors express no competing interests. 586 587 The author(s) responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantcell.org) is (are): Doug K. Allen (doug.allen@usda.gov) and Bradley S. Evans (bevans@danforthcenter.org).

FIGURE LEGENDS 588
Figure 1. Acyl-ACP structure and function. A) Acyl-ACP is a 9-15 kDa protein with 589 4'phosphopantetheine arm attaching a fatty acyl chain to the protein backbone. The 590 structure of acyl-ACP contains a highly conserved sequence of amino acids flanking the 591 serine that attaches to phosphopantetheine. Peptide hydrolysis at aspartate residues 592 produces an acyl molecular species that can be processed for characteristic fragments by 593 mass spectrometry that have precursor-product losses of 315.1 and 413.1. B Table I. Highly conserved amino acid sequence in ACPs. Type II ACPs frequently 637 contain a contiguous region of four amino acids (DSLD). The serine residue provides the 638 point of attachment to the acyl limb described in Figure 1A.      Data are presented on logarithmic scales. Y-axes, peak area analyte/peak area standard; x-axes, pmol analyte. See Figure 3 in the main text for further details. have the same molecular composition and therefore cannot be differentiated by mass alone as indicated in the upper panel that is a chromatogram extraction of the shared m/z. An isotopically labeled internal standard containing predominantly 18:1-ACP was used to assign the retention time for oleoyl-and the C18-enoyl-ACP isomer. As expected, the much smaller peak in the upper panel was the C18-enoyl-ACP isomer that is lower in concentration. Axes were redrawn for clarity.      Table S2. MRM list for acyl-ACP elongation intermediates. Masses based on composition and expected fragmentation patterns were used to examine the presence of acyl-ACP elongation intermediates. The optimized declustering potential (DP) and collision energy (CE) used were the same as described in Figure 3C.    Table S5. 15:0-ACP recovery. A 15:0-ACP internal standard was added at different points in the protein extraction to ensure reasonable recovery during the development of the extraction procedure.
Each peak area is the average of two replicates and indicates that there are not major losses at a particular step in the protein extraction protocol. Importantly, the isotope dilution approach accounts for any losses in a sample because of the presence of internal standards; however, the initial 15:0-ACP recovery here was used to help develop the method prior to absolute quantification.