Potato native and wound periderms are differently affected by down-regulation of FHT, a suberin feruloyl transferase

Abstract

Potato native and wound healing periderms contain an external multilayered phellem tissue (potato skin) consisting of dead cells whose cell walls are impregnated with suberin polymers. The phellem provides physical and chemical barriers to tuber dehydration, heat transfer, and pathogenic infection. Previous RNAi-mediated gene silencing studies in native periderm have demonstrated a role for a feruloyl transferase (FHT) in suberin biosynthesis and revealed how its down-regulation affects both chemical composition and physiology. To complement these prior analyses and to investigate the impact of FHT deficiency in wound periderms, a bottom-up methodology has been used to analyze soluble tissue extracts and solid polymers concurrently. Multivariate statistical analysis of LC-MS and GC-MS data, augmented by solid-state NMR and thioacidolysis, yields two types of new insights: the chemical compounds responsible for contrasting metabolic profiles of native and wound periderms, and the impact of FHT deficiency in each of these plant tissues. In the current report, we confirm a role for FHT in developing wound periderm and highlight its distinctive features as compared to the corresponding native potato periderm.

INTRODUCTION

Potato (Solanum tuberosum L.) ranks as the fifth largest staple crop consumed worldwide, fulfilling essential needs in human nutrition and health (FAO, 2015). The complex dermal structure or periderm that covers plant tubers is critical to the quality, storage, and shelf life of potatoes. Its outermost layer, designated as the phellem, consists of several strata of dead cells with suberized walls. The phellem form the potato cork skin and provides the first line of constitutive defense for the tuber against dehydration, heat transfer, and pathogen infection, including both physical and chemical barriers. The chemical barrier consists of soluble phenolic compounds such as hydroxycinnamic acids and their amide derivatives, flavonoids, and glycoalkaloids, some of which display beneficial antioxidant properties (Akyol, et al., 2016). The physical barrier comprises the phellem cellulose cell wall in which the complex polyaliphatic and polyphenolic (lignin-like) suberin materials are deposited (Bernards, 2002).

Suberin is a fatty polyester that upon transesterification releases mainly soluble C₁₆–C₂₈ ω-hydroxyacids and α,ω-dicarboxylic acids as well as fatty acids (alkanoic acids), primary alcohols (1-alkanols), glycerol, and very small amounts of ferulic acid (Graça and Santos, 2001). It is deposited at the internal side of the cell wall facing the plasma membrane, forming an insoluble matrix within which are embedded a complex mixture of extractable lipids, collectively called wax and substantially similar to the released suberin monomers found in potato (Schreiber, et al., 2005). The more recalcitrant lignin-like portion of the polymeric material, composed principally of p-hydroxycinnamates (e.g., ferulate, p-coumarate, and sinapate) and their derivatives, is deposited within the polysaccharide primary cell wall (Lapierre, et al., 1995). The aliphatic and the phenolic polymeric structures are spatially separated (Yan and Stark, 1998; Gil et al., 1997), but they have been proposed to be linked covalently through ferulate ester bonds (Graça & Santos, 2007).

When the potato skin is broken, the tuber reacts rapidly to restore the barriers by forming a new periderm (wound periderm) that achieves impermeability and chemical defense for the newly exposed fleshy tissues. The healing process proceeds in two stages (Sabba & Lulai, 2002). First, cells adjacent to the wound surface exhibit an increase in polyamines and deposit suberin polymers prior to cell death, forming a wound closing layer within 1-3 days post-wounding. Secondly, a new periderm is formed internally by cell division during the next 5-12 days, where the time course is modulated by the environmental and physiological conditions under which the healing process takes place (Dean & Kolattukudy, 1976; Lulai & Corsini, 1998). Native and wound periderms have similar anatomical structures and undergo analogous maturation processes until they acquire resistance to excoriation (Sabba & Lulai, 2002). However, comparing mature wound and native periderms, the wound periderm is two orders of magnitude more permeable to water. At 21 days post-wounding, its released suberin and wax content is only 50-60% that of native periderm, but its wax fraction is proportionally enriched in alkyl ferulates (Schreiber, et al., 2005). Comparison of native mature periderm and 7-day immature wound periderms by solid-state ¹³C nuclear magnetic resonance (ssNMR) reveals that at this early-stage wound periderm has an enhanced hydrophilic–hydrophobic balance, lignin-like polymeric structures that are more resistant to degradation, and more flexible aliphatic chains, suggestive of a remodeled supramolecular organization (Serra, et al., 2014). Achieving a better metabolic understanding of the wound compared with native periderm has the potential to improve the outcomes of crop management.

A fatty ω-hydroxyacid/fatty alcohol hydroxycinnamoyl transferase), FHT, is the enzyme responsible for the formation of alkyl ferulates in potato periderm, whereby fatty ω-hydroxyacids and fatty alcohols are esterified to feruloyl moieties (Serra, et al., 2010). Down-regulation of ferulate ester synthesis by FHT-RNAi silencing leads to a reduction in alkyl ferulates in both the hydrolyzable aliphatic suberin and the extractable wax from native periderms (Graça, 2015; Serra, et al., 2010). That is, native FHT-deficient periderms yield much lower quantities of ferulic acid (72% reduction), C18:1 ω-hydroxyacids (76% reduction), and most primary alcohols by suberin transesterification compared with wild-type (WT) native skins. However, because FHT-defcient tubers experience a concomitant increase in the periderm thickness, the total amount of hydrolyzable suberin, measured as μg-cm^-2, remains virtually unchanged. As regards the embedded wax fraction, ferulate and alkanes are greatly reduced (72% and 70%, respectively), but the overall amount of wax compounds is doubled due mainly to increases in fatty acids and 1-alkanols. Moreover in FHT-RNAi native tubers, transpiration via the tuber skin is enhanced 14-fold and the skin takes on a very russeted and brittle appearance, but notably, the typical lamellated ultrastructure of the cell wall remains intact (Serra, et al., 2010). Furthermore, the macromolecular organization and mechanical properties of the FHT-RNAi native periderm are compromised. Solid-state ¹³C NMR analysis reveals abundant aromatic constituents that resist transesterification whereas the aliphatic chains exhibit changes in flexibility that can be linked to both resistance to deformation and mechanical resiliency (Serra, et al., 2014). On the other hand, although an intriguing activation of the FHT promoter has been observed in WT wounded periderm (Boher, et al., 2013; Lulai & Neubauer, 2014), it is not yet known how FHT deficiency impacts the chemical composition or supramolecular organization of the wound tissues.

To gain a deeper understanding of native and wound periderms and to investigate the impact of FHT in healing tissue, the current work compares these two tissue types from FHT-RNAi and WT tubers in parallel using immature periderms from freshly-harvested tubers and healing discs. A bottom-up metabolomic approach, combining metabolite profiling with ssNMR, is used to compare the soluble metabolites and the insoluble cell-wall embedded structural moieties of the respective native and wound periderms. To add molecular insights for the lignin-like polymeric materials in suberized cell walls, the soluble phenolics obtained from a degradative thioacidolysis treatment are also compared. This ‘holistic’ analysis of native and wound-healing periderm both complements and augments the earlier investigations, yielding comprehensive and statistically robust information about the impact of knocking down the FHT gene in these protective plant tissues.

2. Results

Parenchyma-free phellem tissues, isolated by skinning freshly-harvested tubers and the wound tuber discs, were first extracted to obtain polar and non-polar soluble metabolites. The insoluble interfacial residue was then treated enzymatically to remove any unsuberized cell walls and remaining waxes before ssNMR analysis, as described in the Materials and Methods section. The following overview of our investigative strategy lays the groundwork for the specific findings presented in the subsequent sections.

Native and wound periderms contain diverse soluble metabolites of differing polarity, chemical class, and abundance. In this context, a broadly applicable biphasic extraction approach was used to concurrently separate the samples into soluble polar and non-polar extracts plus an insoluble solid suspension (Choi, et al., 2004; Kim, et al., 2010; Wolfram, 2006). For the soluble metabolites, multivariate statistics were used to conduct Principal Component Analysis (PCA) of liquid chromatography – mass spectrometry (LC-MS) and gas chromatography (GC-MS) data for the respective types of extracts. Then, to identify those metabolites that were unique or had notably increased or decreased relative abundance in a particular sample (potential biomarkers), an Orthogonal Partial Least Squares Discriminant Analysis (OPLS-DA,(Worley & Powers, 2013)) was performed. By displaying the covariance and correlation from the OPLS-DA model as a scatter plot (S-plot), we visualized both the magnitude of the enhanced abundance (P[1]) and the reliability of the effect (P(corr)[1]). Points in the S-plot ‘wings’ were then linked to the corresponding chromatographic retention times, m/z values, and particular chemical compounds by comparison with reference databases or published MS data including molecular ions and fragmentation patterns (Wiklund, 2008). The insoluble solid suspension containing the interfacial residue was analyzed to deduce types and relative numbers of the major carbon-containing moieties by high-resolution ¹³C ssNMR as well as examination of the ether-linked polyphenolic thioacidolysis breakdown products. First, we compared FHT-RNAi with WT for native and wound periderms (14 days post-wounding) with respect to their polar and non-polar soluble metabolites and the corresponding interfacial residues. Then, we compared native with wound periderm in each of the WT and FHT-RNAi tubers. Finally, to better understand the formation of the wound periderm, we analyzed extracts and solid suspensions from three other key points of the healing process (0 days, 3 days, and 7 days post-wounding), together with the wound periderm (14 days post-wounding), for both WT and FHT-RNAi issues.

2.1 FHT-RNAi vs WT native and wound periderm

2.1.1 The soluble polar extracts

Polar extracts from FHT-RNAi and WT native periderms were analyzed by LC-MS. In Fig. 1A, the score plot illustrates that the polar metabolites detected by negative-mode LC-MS (m/z 100-1300) from nine replicates each of FHT-RNAi and WT samples cluster very well but are clearly separated from each other with respect to the first principal component, indicating notable and consistent changes associated with FHT silencing. Fig. S1 (top) represents superimposed chromatograms from nine biological replicates each of WT and FHT-RNAi, which show excellent reproducibility and reveal clear differences in metabolite abundance between WT and FHT-RNAi, as expected from the PCA analysis. Analogous discrimination is observed in positive-mode LC-MS data from polar extracts (Fig. S2) and in NMR-based PCA score plots (Fig. S3, top). Using OPLS-DA (Fig. 2) and drawing on prior MS reports, we identified 25 of 33 differentially accumulated metabolites (potential biomarkers) that belong to various structural classes and discriminate between the FHT-RNAi and WT soluble polar extracts. Those compounds present at elevated abundance in FHT-RNAi extracts included phenolic amines such as feruloyltyramine (FT), feruloylputrescine (FP), caffeoylputrescine (CafP), and amide dimers such as dihydrocaffeoylputrescine (DHCafP), two grossamide (FT+FT) isomers, and dimers derived from feruloyloctopamine (FO-FO), dihydro-FT-FO, and FT-FO (Table 1). Conversely, phenolic acids related to caffeoyl quinic acid and glycoalkaloids such as α-chaconine, α-solanine, and protodioscin were highly abundant in WT (Table 1). To put these findings into context, all 35 polar identified metabolites are listed in Table S1.

Fig. 1.

PCA score plots for metabolomic analysis of soluble extracts from native potato periderms, with 9 biological replicate clusters color-coded for WT (black) and FHT-RNAi (gray) potato periderms. A: Polar extracts examined by negative-mode LC-MS. B: Non-polar extracts examined by GC-MS. Data were analyzed using SIMCA-P⁺ software; the x and y axes represent the score values of principal components 1 and 2, respectively.

Fig. 2.

Scatter plot (S-plot) highlighting compounds that contribute significantly to differing metabolite profiles for the wild-type (WT) and FHT-RNAi silenced native potato periderms derived from negative-mode LC-MS of their polar extracts. P[1] shows the magnitude of enhanced relative abundance for each metabolite, and P(corr)[1] indicates the reliability of the effect. MS ions at the extreme ‘wings’ of the plot (P(corr)[1] values greater than 0.80) were designated as potential biomarkers and checked for specificity to the cultivar type using a variable line plot (Fig. S4).

Table 1.

Differentially Accumulated Metabolites between FHT-RNAi and WT by Chemical Class and Structure for Polar Extracts from Native Potato Periderms

No.	Ions a	TOF b (error)	MS/MS Fragment iona	Compound Name, Formula	Molecular Structure	Ref.
Highly Abundant FHT-RNAi Metabolites
Phenolic amines
1	250.9 [M-H]-	251.1399 (0.8ppm)	129.0, 120.9, 93.0	dihydro-Caffeoylputrescine C13H20N2O3		Not reported
2	249.3c [M-H]-	249.1255 (4.4ppm)	207.1, 160.9, 148.1, 135, 107.1	Caffeoylputrescine C13H18N2O3		(Matsuda et al., 2005; Dastmalchi et al., 2014)
3	265.1d [M+H]+	265.1536 (-3.8ppm)	248.1, 177.0, 145.0	Feruloylputrescine C14H20N2O3		(Matsuda et al., 2005; Dastmalchi et al., 2014)
4	312.1c [M-H]-	312.1236 (0.9ppm)	297.1, 190.2, 178.8, 148.1	N-Feruloyltyamine (FT) C18H19NO4		(Matsuda et al., 2005; Dastmalchi et al., 2014)
5	623.4 [M-H]-	623.2405 (1.0ppm)	460.2, 445.4, 336.2, 324.2, 322.2	Grossamide (FT-FT dimer) C36H36N2O8		(Serra et al., 2010; Dastmalchi et al., 2014)
6	623.3 [M-H]-	623.2405 (1.0ppm)	460.2, 444.2, 623.4, 283.1, 336.1, 445.2	Grossamide (FT-FT dimer) C36H36N2O8
7	655.3 [M-H]-	655.2301 (0.6ppm)	655.3, 637.3, 619.4, 484.3, 476.3, 458.3, 415.4, 297.3	FO-FO dimer C36H36N2O10		Not reported
8	641.4 [M-H]-	641.2513 (-1.0ppm)	623.4, 489.2, 460.2, 336.2, 326.2, 312.2, 151.0	Dihydro-FT-FO dimer C36H38N2O9		Not reported
9	639.4 [M-H]-	639.2346 (0.6ppm)	621.5, 460.4, 297.3, 578.5, 486.3, 458.3, 415.4	FT-FO dimer C36H36N2O9		(King and Calhoun, 2010; Serra et al., 2010; Landgraf et al., 2014)
Incompletely Identified Constituents
10	721.2	721.1606	721.2, 706.3, 585.4, 570.3, 555.4, 529.3, 512.3, 407.3, 391.3, 348.2, 191,150	phenolic amines
11	803.2	803.3113	803.3, 641.3, 623.3, 489.3, 326.2, 312.2, 151.1	cinnamic acid derivative
12	602.3		602.2, 341.2, 328.2, 282.1, 250.1, 245.0, 231.0, 193.0, 151.0, 136.0
Highly Abundant WT Metabolites
Phenolic acids
13	353.1c [M-H]-	353.0870 (-0.8ppm)	191.0, 161.1, 135	caffeoyl quinic acid isomer C16H18O9		(Narváez-Cuenca et al., 2012; Dastmalchi et al., 2014)
14	375.0 [M+Na-H]-	375.0701 (0.8ppm)	201.0, 179.0, 161.1, 135	C16H17O9Na
15	353.1 [M-H]-	353.0870 (-0.8ppm)	191.0, 161.1, 135	caffeoyl quinic acid isomer C16H18O9
16	375.8 [M+Na-H]-	375.0701 (0.8ppm)	201.0, 179.0, 161.1, 135	C16H17O9Na
17	179.1 [M-H]-		161, 135.0, 107.0, 89	Caffeic acid C9H8O4		(Narváez-Cuenca et al., 2012)
Glycoalkaloids
18	866.2c [M-H]-	866.4918 (1.2ppm)	720.5, 704.5, 558.5	α-Solanine C45H73NO15		(Matsuda et al., 2004; Shakya and Navarre, 2006; Dastmalchi et al., 2014)
19	912.5c [M+HCOOH -H]-	912.4982 (2.3ppm)	866.5, 720.6, 704.5	α-Solanine+HCOOH C45H73NO15•HCOOH
20	850.1c [M-H]-	850.4993 (4.1ppm)		α-Chaconine C45H73NO14		(Matsuda et al., 2004; Shakya and Navarre, 2006; Dastmalchi et al., 2014)
21	896.5c [M+HCOOH -H]-	896.5046 (3.8ppm)	850.5, 704.5	α-Chaconine+HCOOH C45H73NO14•HCOOH
22	704.3 [M-H]-	704.4374 (0.7ppm)	558.6, 112.9	Solanidine-Glc-Rha C39H63NO10		(Matsuda et al., 2004)
23, 24	1049.8c [M-H]- 1095.7c [M+HCOOH -H]-	1049.5568 (2.9ppm)	903.4, 757.4, 741.4, 433.4	Dihydro-ASP-II C51H86O22		(Dawid and Hofmann, 2014)
25	1061.8	1061.5220 (4.3ppm)	915.4, 899.4, 881.4, 753.5, 735.5, 687.4, 573.5, 555.5	C51H84O23		Not reported
26, 27	1047.7 [M-H]-1093.7 [M+HCOOH -H]-	1047.5391 (0.9ppm)	901.5, 883.5, 755.4, 737.5	Protodioscin/neoprotodioscin C51H84O22		(Dawid and Hofmann, 2014)
Other Constituents
28	191.1 [M-H]-	191.0557 (2.1ppm)	147.0, 111.0, 87.0	Quinic acid C7H12O6		(Dou et al., 2007)
Incompletely Identified Constituents
29	1045.7	1045.5243 (1.7ppm)	1027.4,899.4, 881.4, 735.4, 555.4	C51H82O22
30	933.5 [M-H]-	933.2677 (-0.7 ppm)	753.2, 771.3, 449.1, 287.1	C43H50O23 Flavonoid
31	925.0	925.4343	717.4, 570.4, 553.1, 511.4
32	1063.6	1063.5422 (25 ppm)	1045, 903, 883.5, 721.4
33	1033.7 [M-H]- 1079.7 [M+HCOOH -H]-	1033.5624 (3.4ppm)	887.5, 869.5, 741.5, 723.5	C51H86O21	1049-O

^aIons and fragmentation data obtained from LC-MS and LC-MS² analysis using a 4000Q Trap instrument.

^bExact mass data obtained from TOF data analysis.

^cMetabolites that distinguish polar extracts from native periderms of four cultivars (Atlantic, Chipeta, Norkotah Russet, Yukon Gold) (Huang, et al., 2017).

^dPositive-mode LC-MS.

The soluble polar extracts were also analyzed in day-14 wound periderms. However, no significant differences between FHT-RNAi and WT could be detected by PCA analysis and no differentially accumulated metabolites could be identified (data presented below with the wound-healing time course).

2.1.2 The soluble non-polar extracts

Non-polar extracts were analyzed by GC-MS (m/z 45–600) and compared. The score plot in Fig. 1B and chromatograms in Fig. S1 (bottom) illustrate that the soluble non-polar metabolites detected by GC-MS from nine replicates of each FHT-RNAi and WT sample cluster very well, respectively. As observed for the corresponding polar extracts (Fig. 1, left), the non-polar extracts are clearly separated from each other with respect to the first principal component, indicating notable and consistent changes associated with FHT silencing. Analogous discrimination was found for the NMR-based PCA score plot (Fig. S3, bottom). Twenty-five of the 31 metabolites accumulated differentially by FHT-RNAi and WT could be selected by OPLS-DA (Fig. S4) as outlined above and then identified (Table 2). In FHT-RNAi the highly abundant metabolites included long-chain saturated fatty acids (C22, C23, C24, C27), long-chain primary alcohols (C22, C26, C27, C28, C29), methyl esters (C17:0, C19:2), 1-monohexadecanoyl glycerol and glucose. In WT the highly abundant metabolites included alkanes (C21, C23, C25, C27, C29), caffeic acid, fatty acids (C16:0, C18:0, C18:2), 2-hydroxydecanedioic acid and triacontanol. To put these findings into context, all 76 non-polar identified metabolites are listed in Table S2.

Table 2.

Differentially Accumulated Metabolites between FHT-RNAi and WT by Chemical Class and Structure for Non-polar Extracts from Native Potato Periderms

No.	M.W.	Formula	Compound Namea	Molecular Structure	Ref.
Highly Abundant FHT-RNAi Metabolites
1	326.6	C22H46O	Straight-chain primary alcohols		b, d, e, g
2	382.7	C26H54O			d, e, g
3	396.7	C27H56O
4	410.7	C28H58O			b, d, e, g
5	424.8	C29H60O			b, d
6	242.4	C15H30O2	Saturated fatty acids		c, d, g
7	340.6	C22H44O2			c, d, e, g
8	354.6	C23H46O2			b, d
9	368.6	C24H48O2			c, d, e, g
10	410.7	C27H54O2
11	270.5	C17H34O2	Methyl palmitate
12	294.5	C19H34O2	Methyl linoleate
13	180.1	C6H12O6	Glucose		d
14	330.5	C19H38O4	Palmitoyl glycerol (glycerol monoacyl fatty acid)		b, c, d
No.	M.W.	Formula	Name of Metabolite	Molecular Structure	Ref.
Highly Abundant WT Metabolites
1	296.6	C21H44	n-Alkanes		c, b, d
2	324.6	C23H48			d
3	352.7	C25H52			d
4	380.7	C27H56			b, d, e
5	408.8	C29H60			d
6	438.8	C30H62O	Triacontanol (Straight-chain primary alcohol)		b, e
7	256.4	C16H32O2	Saturated fatty acids		c, d, g
8	284.4	C18H36O2			c, d, g
9	, f 280.4	C18H32O2	Linoleic acid (Unsaturated fatty acid)		c, d
10	180.1	C9H8O4	Caffeic acid		b, d, e
11	218.2	C10H18O5	2-hydroxydecanedioic acid		c

^a25 out of 31 non-polar metabolites were identified from GC-MS data using GCMSsolution software to access the NIST 2008 and Wiley 2009 mass spectral libraries.

^bMetabolites identified by Huang, et al. (2017) in non-polar extracts from native periderms of four potato cultivars (Atlantic, Chipeta, Norkotah Russet, Yukon Gold).

^cMetabolites identified by Dastmalchi, et al. (2015) in non-polar extracts from wound periderms of four cultivars (Atlantic, Chipeta, Norkotah Russet, Yukon Gold).

^dMetabolites identified by Yang and Bernards (2007) in non-polar extracts from wound periderms (Russet Burbank).

^eMetabolites identified by Serra, et al. (2010) in wax compound from potato native periderm (Desirée)

^fMetabolites identified by Graça and Pereira (2000) in depolymerization compounds from potato native periderm.

^gUp-regulated metabolites identified in FHT-RNAi wound healing samples (Table 3).

Non-polar extracts from day-14 wound periderms were compared analogously. Unlike the polar wound extracts, the non-polar extracts showed distinct compositional features for FHT-RNAi and WT (Fig. 3). Differentially accumulated metabolites could be identified by the OPLS-DA method (Table 3). The up-regulated metabolites in FHT-RNAi 14-day wound periderm included saturated fatty acids (C8, C12, C14, C15, C17, C18, C22, C24, C26), long-chain primary alcohols (C20, C22, C24 C26, C28), alkanes (C21, C23, C25, C27, C29), glycerol monoacyl fatty acids (C18, C22) and a sterol. Only fatty acids (C16:0, C18:1, C18:2) were found as the most abundant metabolites characteristic of wound WT. All 76 non-polar identified metabolites appear in Table S2. Feruloyl esters could not be identified using the GC-MS protocol, so we used higher temperatures during GC in an effort to separate and detect these compounds. They were identified by GC-MS and quantified by GC-FID using 5 WT and 3 FHT-RNAi replicates, respectively. As expected in light of the established FHT enzymatic function as a feruloyltransferase, the current analyses confirmed a reduction of ferulate esters in the wound periderm and an increase of primary alcohols and fatty acids when compared with WT (Fig. 4) as reported for methanolysates from native periderm (Serra et al., 2010).

Fig. 3.

PCA score plots for metabolomic analysis of non-polar extracts (GC-MS) from WT and FHT-RNAi wound-healing periderms, with 9 biological replicate clusters color-coded for WT (black) and FHT-RNAi (gray). Wound-healing samples were obtained at day 14 post wounding. The x and y axes represent the score values of principal components 1 and 2, respectively.

Table 3.

Differentially Accumulated Metabolites^a by Chemical Class and Structure for Non-polar Extracts from Day-14 Wound Healing Periderms

No.	M.W.	Formula	Compound Name	Molecular Structure	Ref.
Highly Abundant FHT-RNAi Metabolites
1	298.3	C20H42O	Straight-chain primary alcohols		b, d
2	326.3	C22H46O			b, d, e, g
3	354.4	C24H50O			b, d, e
4	382.4	C26H54O			b, d, e, g
5	410.4	C28H58O			b, d, e, g
6	144.1	C8H16O2	Saturated fatty acids		b
7	200.2	C12H24O2			b, d
8	228.2	C14H28O2			b, c, d
9	242.2	C15H30O2			b, c, d, g
10	270.2	C17H34O2			b, c, d
11	284.2	C18H36O2			b, c, d, e
12	340.3	C22H44O2			b, c, d, g
13	328.3	C24H48O2			c, d, e, g
14	396.4	C26H52O2			b, d, e
15	324.4	C23H48	n-Alkanes		b, d
16	338.4	C24H50
17	352.4	C25H50			b, d
18	358.3	C21H42O4	Glycerol monoacyl fatty acid		b, c, d, f
19	414.4	C25H50O4			f
20			Sterols Beta-sitosterol Stigmasterol Cycloarlenol		a
Highly Abundant WT Metabolites
1	256.4	C16H32O2	Saturated fatty acids		b, d
2	280.4	C18H32O2			b, c, d
3	282.4	C18H34O2	Oleic acid (Unsaturated fatty acid)		b, c, d

^a23 out of 30 non-polar metabolites were identified from GC-MS data using GC-MS solution software to access the NIST 2008 and Wiley 2009 mass spectral libraries.

^bMetabolites identified by Huang, et al. (2017) in non-polar extracts from native periderms of four potato cultivars (Atlantic, Chipeta, Norkotah Russet, Yukon Gold).

^cMetabolites identified by Dastmalchi, et al. (2015) in non-polar extracts from wound periderms of four cultivars (Atlantic, Chipeta, Norkotah Russet, Yukon Gold).

^dMetabolites identified by Yang and Bernards (2007) in non-polar extracts from wound periderms (Russet Burbank).

^eMetabolites identified by Serra, et al. (2010) in wax compound from potato native periderm (Desirée).

^fMetabolites identified by Graça and Pereira (2000) in depolymerization compound from potato native periderm.

^gUp-regulated metabolites identified in FHT-RNAi native periderms (Table 2).

Fig. 4.

Relative amounts of compounds detected by GC-FID in non-polar extracts of day-14 wound-healing periderm from FHT-RNAi and WT tubers. A: Alkyl ferulate esters of primary alcohols. B: Alkanes, free fatty acids and primary alcohols (OH). The carbon chain length of the fatty-acid derived compound is shown (Cpm). Each value represents the mean ± SD for FHT-RNAi line 37 (n = 5) and WT (n = 3).

2.1.3 The insoluble interfacial residue

Insoluble suspensions at the interface of the polar and non-polar extracts, obtained as described in the Materials and Methods section, were treated to remove the unsuberized tissue and the embedded waxes, and then analyzed by direct-polarization (DPMAS) ¹³C NMR (Fig. 5). Referring to prior literature, we assigned the major structural groupings to chemical shift ranges in the spectra as follows: carboxyls or amides (COO or CONH = COX; 160-180 ppm), alkenes and arenes (92-160 ppm), partially resolved alkoxy groups (CHO, CH₂O, CH₃O; 44-92 ppm), and long-chain methylenes (-(CH₂)_n; 10-44 ppm) (Garbow, et al., 1989; Järvinen, et al., 2011; Pascoal Neto, et al., 1995; Serra, et al., 2014). Highly reproducible spectra, albeit with broad features attributable to the chemical heterogeneity and amorphous nature of these materials, were obtained using three biological replicates each of the WT and FHT-RNAi periderms.

Fig. 5.

150 MHz DPMAS ¹³C NMR spectra of suberin-enriched interfacial residues from native (A, B) and 14-day wound (C, D) periderms, color-coded for WT (black) and FHT-RNAi (gray) cultivars. The functional groups are assigned as carboxyl and amide groups (COO and/or CONH, 160-180 ppm); multiply bonded (alkenes and arenes, 92-160 ppm); oxygenated aliphatic carbon groups (CHO + CH₂O + CH₃O, 44-92 ppm); long-chain methylene groups (-(CH₂)_n^-, 10-44 ppm).

Notably different relative ¹³C NMR signal intensities and corresponding functional group proportions were evident from these quantitatively reliable spectra. To estimate the relative numbers of each major carbon structural type, peak area ratios were compared within each DPMAS spectrum so that any unintentional variations in instrumental setup or sample mass could be avoided. The compositional trends were informative regarding the compatibilities of the suberin polymers with either the waxy overlayer or the underlying cell walls. These results are summarized in Fig. 6 (left) as ratios of carboxyls and amides (COX), alkene and arene groups, or the CH_nO alkoxy groups with respect to the long-chain aliphatic methylene groups that are expected to interact hydrophobically with waxes to confer waterproofing. Compared to WT, the FHT-RNAi native residue displayed an enhanced hydrophilic-to-hydrophobic balance, validated by ratios of alkene-and arene-to-(CH₂)_n and alkoxy-to-(CH₂)_n that were elevated by 60% and ~80%, respectively. Moreover, the degree to which polymeric suberin is deposited within the cell wall is indicated in Fig. 6 (right) by ratios with respect to the principal polysaccharide CHO groups (65-78 ppm). The residues from FHT-RNAi displayed significantly diminished peak area ratios for carboxyls and amides (COX), alkenes and arenes, and -(CH₂)_n groups.

Fig. 6.

Ratios of carbon-containing functional groups in the dewaxed interfacial residue of WT (black) and FHT-RNAi (gray) potato periderms with respect to long-chain methylene aliphatics (left) and the principal polysaccharide CHO groups (65-78 ppm) (CHO) (right), derived from 150 MHz DPMAS ¹³C NMR spectra. Mean values and standard error bars are based on two measurements conducted on a single NMR sample.

The distinctive FHT-RNAi ratios involving arenes, which were also reported for intact dewaxed suberized periderms (Serra, et al., 2014), prompted us to look more closely at the aromatic suberin domain by analyzing the ether-linked breakdown products using thioacidolysis conducted after suberin depolymerization (Fig. 7, top). For both FHT-RNAi and WT, we detected guaiacyl (G) and syringyl (S) units but no p-hydroxyphenyl (H) units (Fig. 7, top), in agreement with thioacidolysis products analyzed previously by Lapierre et al. (1995) and with intact polymers examined by Yan and Stark (2000). For WT periderm, the estimated proportions of G (62%) and S units (38%) were also in rough accordance with the G (65%) and S unit (35%) content reported previously for potato native periderms using the same methodology (Lapierre, et al., 1995). FHT-RNAi produced more abundant G units and less abundant S units than WT (Fig. 7, middle), although the differences in the S units were not statistically significant. However, the FHT-RNAi G/S ratio was about twofold larger than in WT (Fig. 7, bottom), suggesting a diminished cross-linking capacity of aromatic suberin polymeric structures in the FHT genetically modified tissue.

Fig. 7.

Amounts and proportions of phenylpropanoid building blocks in native potato periderms from WT and FHT-RNAi tubers. Mean values and standard deviation error bars are based on the peak areas of the identified guaiacyl and syringyl units in GC-MS. Top: chemical structures of major aromatic groups released by thioacidolysis; Middle: amounts of guaiacyl and syringyl units per mg of dry periderm tissue; Bottom: ratio of guaiacyl to syringyl units in periderms.

As regards the day-14 wound periderms, the interfacial dewaxed residues exhibited resonances corresponding to the same major chemical groupings as the native suberized cell walls but modestly better resolution of the ¹³C NMR DPMAS spectra (Fig. 5, right). As in native periderms, the compositional trends in the FHT-RNAi wound residue displayed alkene and arene-to-(CH₂)_n and alkoxy-to-(CH₂)_n ratios that exceeded the values for WT (Fig. 6 (left)), indicating an enhanced hydrophilic-to-hydrophobic balance. However, only minor differences between WT and day-14 FHT-RNAi wound periderms were observed for the ratio of (CH₂)_n groups to the primary polysaccharide CHO group (65-78 ppm), in contrast to the trend for native periderm (Fig. 6, right).

2.2 Native versus wound periderm

Metabolites present in extracts from freshly harvested (immature) native and 14-day post-wounding discs were compared in each of the WT and FHT-RNAi varieties. With respect to the polar metabolites, a PCA-based comparison between native and wound periderms was precluded by the higher overall concentration and much larger number of detectable metabolites in the native extracts. However, direct comparison of the LC traces revealed distinctive features of wound periderm in terms of the major chemical classes represented among the metabolites (Fig. 8 and Table S1). Five metabolites predominate in both WT and FHT-RNAi wound polar extracts: caffeoylputrescine (retention time (RT) 11.8 min), feruloylputrescine (RT 14.8 min), α-chaconine (RT 28.5 min), α-solanine (RT 29.0 min), and a feruloyloctopamine-feruloyloctopamine (FO-FO) dimer (RT 31.7 min). Peaks with retention times of 21.3-25.5 min were significantly diminished in both varieties of wound periderms, corresponding to a paucity of flavonoid glycosides (aglycones of tetrahydroxyflavonoids at m/z 903 and 933 with fragments 449, 287, and 269; luteolin-7-O-rutinoside at m/z 593) and phenolic polyamines containing more than two cinnamic acid derivatives (dihydrocaffeoyl spermines at m/z 721 and 723 with fragments 529, 407, and 191). For the non-polar metabolites, PCA-based comparisons of native and wound extracts were again precluded by disparities in overall concentration and number of detectable metabolites. However, it was possible to compare the respective potential biomarkers that discriminate between WT and FHT-RNAi in native (Table 2) and wound (Table 3) periderms. With respect to the corresponding WT periderms, both FHT-RNAi native and wound periderms had the same families of up-regulated metabolites. However, whereas alkanes were up-regulated in WT native periderms, those metabolites were up-regulated in FHT-RNAi wound. Moreover, abundant sterol biomarkers were observed uniquely in the FHT-RNAi wound periderm (Table 3).

Fig. 8.

Overlay of typical LC-MS traces for native (gray) and day-14 wound (black) polar potato periderm polar extracts. A: WT native vs. WT day-14 wound; B: FHT-RNAi native vs. FHT-RNAi day-14 wound.

2.3 Wound healing time course in FHT-RNAi and WT potato discs

As noted in the Introduction, wound healing to restore the chemical and physical barriers in fruits and tubers is a complex process that proceeds in distinct steps. To compare the wound healing time course in WT and FHT-silenced tubers, we undertook parallel studies of polar and non-polar soluble metabolites along with polymeric solid composites at four key time points: immediately after wounding (day 0), when a closing layer has formed (day 3), when a new wound periderm is nascent (day 7), and when the new periderm has developed but is still immature (day 14) (Lulai, et al., 2016).

For polar soluble metabolites, Fig. 9 shows a score plot from LC-MS-based multivariate analysis. Clustering indicates good consistency of the compositional data from each set of biological replicates. The largest variations with time occurred for principal component 1 (PC1) from day 0 to day 3; smaller changes, primarily in PC2, were observed from day 3 to days 7 and 14. The FHT-RNAi and WT polar extracts exhibited distinct profiles at day 0 with respect to other time points, and each variety of exposed wound tissue exhibited a similar compositional progression during the healing process. OPLS-DA pairwise analysis during the course of healing (comparing each pair of time points) revealed no significant compositional differences between days 7 and 14 for either WT or FHT-RNAi. In contrast, when comparing day 3 and day 7, we found that both WT and FHT-RNAi day-7 extracts were richer in α-chaconine and α-solanine glycoalkaloids, whereas only day-3 WT extracts were richer in feruloyloctopamine.

Fig. 9.

PCA score plot of negative-mode LC-MS data used in metabolomic analysis of polar extracts from wound-healing potato parenchyma (day 0) and wound-healing tissues (days 3-14), showing 9 biological replicates color-coded for WT (black) and FHT-RNAi (gray) tubers. The numbers of days after wounding are denoted by squares (0), triangles (3), diamonds (7), and circles (14), respectively. Differences between the WT and FHT-RNAi day-0 extracts are not statistically significant (see text).

To evaluate the impact of FHT silencing on the wound-healing process, pairwise comparisons of WT vs. FHT-RNAi were made for polar extracts at each of days 0, 3, 7 and 14 by PCA and OPLS-DA methods. No statistically significant differences in the overall metabolite profiles could be verified despite the appearance of day-0 variations in PC2 (Fig. 9): a negative value of Q² in the PCA modeling showed that the variation among biological replicates was comparable to differences between the two types of wound-healing periderm. Neither were there compositional differences between any of the non-polar extracts, with the exception of day 14, for either WT or for FHT-RNAi tissues (data not shown).

The DPMAS ¹³C NMR spectra of the interfacial residues isolated at days 0 to 14 are shown in Fig. S5. At day 0, residues are essentially comprised of polysaccharides since the region between 45 and 108 ppm is contributed primarily by cell walls. At days 3, 7, and 14, key resonances attributable to the suberin aliphatic polyester appear with increasing prominence as reported previously (Stark, et al., 1994): chain methylene groups, carbonyl groups, alkenes, and arenes. A quantitative analysis of the corresponding DPMAS spectra yields ratios for day-3, day-7 and day-14 residues, each measured with respect to the major polysaccharide peak of the cell-wall CHO region (Fig. S6). At day 3, the spectral regions attributable to the suberin (10-44, 108-160, 160-180 ppm) exhibit larger relative contributions in FHT-RNAi. At day 7, the difference between FHT-RNAi and WT in the corresponding ratios diminishes, and at day 14 WT shows nearly the same ratios as FHT-RNAi (Fig. S6, Fig. 6 (right)).

3. Discussion

3.1 FHT deficiency in native vs wound periderms

By conducting the analysis of extracted metabolites and the solid interface holistically, the current work also complements prior studies of FHT deficiency in native periderms that were made by examination of soluble waxes, soluble monomeric suberin breakdown products, and polymeric solids (Graça, 2015; Serra, et al., 2014; Serra, et al., 2010). As expected from its demonstrated role as a feruloyl transferase in native periderm, FHT down-regulation in wound periderm leads to diminished amounts of non-polar alkyl ferulates (reaction products) and augmented amounts of primary alcohols (reaction substrates) (Fig. 3B; Tables 2 and 3), thus extending the role for this enzyme to the wound healing process. However, the impact of FHT silencing on the wound-healing response does not invariably conform to the findings for native periderm. Thus for the polar metabolites, higher levels of phenolic polyamines are observed preferentially only for the FHT-RNAi native periderm (Table 1). As regards the non-polar metabolites, in FHT-deficient tubers the saturated fatty acids and glycerol monoacyl fatty acids increase in both native (Table 2) and wound periderms (Table 3), although the wound periderm shows a broader distribution of fatty acid chain lengths. Conversely, alkanes are abundant in FHT-RNAi wound but diminished in the corresponding native periderm (Fig. 4, Tables 2 and 3) (Serra, et al., 2010), whereas sterols are more abundant only in FHT-deficient wound periderm (Table 3).

For the interfacial polymeric solids, both the native and day-14 wound FHT-deficient periderms have larger alkene/arene and alkoxy proportions with respect to long-chain methylene groups. The latter ratio (alkoxy/long-chain methylene groups) and diminished contributions from (CH₂)_n moieties demonstrates an augmented hydrophilic-hydrophobic balance (Fig. 6, left) and provides a rationale for the enhanced permeability of the FHT-RNAi native periderm(Serra, et al., 2010). On the other hand, in the native periderms a diminished propensity for suberin deposition within the cell wall of FHT-deficient periderms is inferred from the proportions of (CH₂)_n moieties with respect to CHO groups, confirming prior observations for unextracted samples (Fig. 6, right; (Serra, et al., 2014). The alkene- and arene-to-CHO ratio is also diminished in the native periderm solids, though the aromatic functional groups are retained strikingly in the recalcitrant residue remaining after suberin depolymerization (Serra, et al., 2014).

This last compositional trend is extended by the finding of relatively greater G-unit content for groups linked through β-O-4 (labile alkyl-aryl ether) bonds in native FHT-RNAi (Fig. 7), supporting facilitated linkage by oxidative coupling of ferulic-type units in the lignin-like polymeric structures and/or architectural alterations that promote the release of these G units upon thioacidolysis. Recent FT-IR analyses of wax-free periderms (Graça, 2015) have also shown enhanced G content in FHT-RNAi compared with WT. The greater proportion of labile β-O-4 linked G units in the FHT-RNAi polymer correlates additionally with the finding of larger amounts of feruloyl polyamines in the polar extracts. When incorporated within the cell wall, such structures result in reinforcement and stiffening (Bassard, et al., 2010), in agreement with the greater stiffness and the compromised mechanical strength of the FHT-RNAi periderm reported previously (Serra, et al., 2014). However, our analysis does not probe the enrichment in terminal or internal G-units or the presence of other thioacidolysis-resistant types of inter-unit bonds (β-1, β-β, β-5, 4-0-5 and 5-5 linkages) (Lapierre, et al., 1996; Négrel, et al., 1996), which could also affect the cross-linking capacity and associated mechanical properties of the FHT-RNAi periderm. Taken together, the changes induced by FHT down-regulation in native and wound periderms nonetheless demonstrate many common features of their biosynthetic pathways.

The current analyses, together with related prior work on aliphatic suberin breakdown products, allow us to infer both direct consequences of alkyl ferulate blockage and indirect effects of dehydration stress as depicted in Fig.10 (Bernard, et al., 2012; Bernards, 2002; Fraser & Chapple, 2011; Matsuda, et al., 2005; Serra, et al., 2010; Yang, et al., 2010; Yang & Bernards, 2006). Since FHT silencing blocks the esterification of ferulic acid to ω-hydroxyfatty acids or alcohols, elevated levels of primary long-chain alcohols in the non-polar extracts and up-regulation of suberin-associated waxes in FHT-RNAi native periderm (Serra, et al., 2010) are expected. However, the ω-hydroxyfatty acid substrates are not found abundantly. Thus, these latter substrates could react to form α,ω-diacids and then be converted efficiently to glycerol esters via the catalytic action of CYP86 family members, GPAT5, and GPAT7 enzymes, a hypothesis supported by our finding of a highly abundant glycerol monoacyl fatty acid in FHT-RNAi periderms (Tables 2 and 3). These intermediates can ultimately be incorporated into suberin, but their elevated levels can also have a negative feedback effect that decreases the level of ω-hydyroxyacids among the aliphatic suberin breakdown products (Serra, et al., 2010) and accounts for the abundant LC and VLC fatty acids in extracts from the FHT-RNAi periderms. It is also noteworthy that accumulation of alkanes is favored in FHT-RNAi wound periderm but diminished in FHT-RNAi native periderm, compared with the respective wild types.

Fig. 10.

Proposed biosynthetic pathways for suberized potato periderms. (Bernards, 2002; Matsuda et al., 2005; Yang and Bernards, 2006; Serra et al., 2010; Yang et al., 2010; Bernard et al., 2012; Li-Beisson et al., 2013; Graça, 2015). Reactions denoted by solid lines have been derived from the indicated references, while those denoted by broken lines are hypothetical or assumed.

Abbreviations: 4CL, 4-hydroxycinnamic acid; PHT, putrescine hydroxycinnamoyltransferase; FHT, fatty ω-hydroxyacid/fatty alcohol hydroxycinnamoyltransferase; THT, tyramine Nhydroxycinnamoyltransferase CYP, cytochrome P450 monooxygenase; FAR, alcohol-forming fatty acyl-CoA reductase; G3P, glycerol-3-phosphate; GPAT, glycerol 3-phosphate acyltransferase. * Saturated and unsaturated carbon chains

Turning to the feruloyl-CoA precursor that is blocked from esterification in FHT-RNAi periderms, the changes observed for metabolites in both soluble polar extracts and insoluble cell-wall polymers exemplify the possibility of ferulic acid uptake via multiple biochemical processes and support the cross-talk proposed for pathways involving alkyl ferulates, cell-wall monolignols and hydroxycinnamates, and hydroxycinnamic acid amines in potato periderms (Bernards, 2002; Graça & Santos, 2007; Matsuda, et al., 2005). In addition to the rerouting of feruloyl-CoA to feruloylputrescine and related amides, it is interesting to highlight the concurrent channeling of caffeoyl-CoA to caffeoylputrescine. Both reactions are catalyzed by the same putrescine hydroxycinnamoyltransferase enzyme (PHT) (Matsuda, et al., 2005), suggesting that the presence of feruloyl-CoA and/or the dehydration-induced stress can act as positive regulators of PHT in native periderms. The latter rationale also dovetails with our recent finding of elevated caffeoylputrescine accumulation in highly permeable native Norkotah Russet potato periderms, as compared with smooth Yukon Gold cultivars (Huang, et al., 2017).

3.2 Phenolic polyamines and russeted potato skins

The role of polyamines in wound periderm has been a matter of longstanding discussion. Phenolic amines can combine the growth and development regulatory functions of polyamines (Martin-Tanguy, 1985) with the toxic free radical scavenging, antioxidative, and wall-strengthening properties conferred by hydroxycinnamic acids (Volpert et al., 1995, among others). It was suggested by (Heng, et al., 2016) that the H₂O₂ generated during polyamine metabolism might contribute to russet formation on the exocarp of pear fruits. A role as cross-linked intermediates putatively involved in the suberization process was suggested by King and Calhoun (King & Calhoun, 2005, 2010) for feruloyl amides and cross-linked dimers in potato scab lesions.

FHT deficient tubers (cv. Desirée) show a distinctive russeted skin that is more rigid, fragile, and prone to microfissure and has increased water permeability in contrast with the smooth WT tubers (Serra et al, 2010). Moreover, the native FHT-deficient periderm accumulates phenolic amines such as caffeoylputrescine, feruloylputrescine and feruloyloctopamine compared with WT (Table 1). Notably, potatoes with russeted skin display increased water permeability (Ginzberg, et al., 2012; Lulai & Orr, 1994) and phenolic polyamines are notably abundant in native and wound periderms from russeted Norkotah and Atlantic potato cultivars (Dastmalchi, et al., 2014; Huang, et al., 2017). In this regard, (Landgraf, et al., 2014) have made an interesting observation: in tubers (cv. Desirée) silenced for an ABC transporter (StABCG1), a decrease in suberin amount and an increase in water permeance are associated with abundant levels of phenolic amines and russeted skin appearance. Taken together, the above considerations suggest that a certain degree of water loss inherent in the FHT-RNAi tuber skin may induce higher levels of phenolic polyamines and a russeted periderm appearance.

3.3 FHT deficiency affects soluble metabolites and polymeric solids formed during wound healing

One perspective on healing differences in FHT deficient and WT tubers comes from disparities in the progress of suberin deposition, reported by ratios of particular functional groups measured in solid-state DPMAS ¹³C NMR spectra of the solid interfacial layer. As shown in Fig. S6 with respect to the polysaccharide cell-wall CHO resonances, all structural moieties present in the interfacial solid accumulate more rapidly during the first 3 days in FHT-RNAi wound periderms. During the following days, suberin deposition occurs more vigorously for the WT until it “catches up” or surpasses the FHT-RNAi at day 14. Considering that FHT-RNAi wound periderm has knocked down transferase enzymatic activity for the aliphatic constituents, a diminished rate of development for both arene and aliphatic constituents of the suberin biopolymer should be expected. Nonetheless, the early deposition of suberin in the FHT-deficient wound periderm could be explained by a baseline condition of water stress in the tuber parenchyma prior to wounding, associated in turn with the greater rate of water loss of the FHT-RNAi periderm (Serra, et al., 2014). This state of hydric stress could induce a protective response that overlaps with the wound response. However other alternative explanations, such as the formation by day 3 of different polymeric structures in FHT-RNAi compared with WT, cannot be ruled out.

A contrasting perspective emerges from consideration of the pools of soluble metabolites in wound-healing tissues. Differences between the non-polar extracts are found only in the day-14 wound periderm, where the knocked down feruloyl transferase activity is evident in the abundant long-chain fatty acids and fatty alcohols (reactants) and the diminished ferulate esters (products). For the polar extracts, some differences between FHT-RNAi and WT are evident immediately after wounding (day 0) but subsequently, both wound periderms experience an overall compositional convergence (from day 3 onwards, Fig. 7). Surprisingly, at day 0 there is a greater accumulation of caffeoylputrescine in FHT-RNAi compared with WT. Given that the FHT protein does not accumulate and that the gene promoter is not expressed in the tuber parenchyma (Boher, et al., 2013), the abundance of phenolic polyamines in the newly exposed wound tissue (day 0) of the FHT-RNAi tubers should be unrelated to blockage of FHT but rather viewed as a constitutive feature of the FHT-RNAi tubers. Plausibly, the hydric stress produced by the 14-fold greater water loss in periderms of FHT-RNAi tubers (Serra, et al., 2010) causes a permanent stress within the inner fleshy tissues, which in turn could induce the synthesis of radical scavenging phenolic polyamines as a defense mechanism (Bassard, et al., 2010).

4. Conclusions

These ‘holistic’ analyses encompass soluble metabolites, solid polymer composites, and breakdown products from diverse chemical treatments of FHT-deficient immature native and wound periderms compared with WT. Both FHT-RNAi periderms show an augmented hydrophilic-hydrophobic balance A metabolic rerouting to enhance the relative amounts of phenolic polyamine structures occurs upon FHT-RNAi silencing only in native periderms, providing a rationale for their enhanced permeability (Serra, et al., 2010) but leaving the analogous permeability trend in wound periderms unexplained. Each of these observations is also shared to different degrees by the immature native periderm of russeted potato cultivars (Huang, et al., 2017) and by ABCG1-RNAi tubers (Landgraf, et al., 2014), as well as by the scabby lesions produced by Streptomyces scabies (King & Calhoun, 2005, 2010). With this comprehensive molecular-level information in hand, we are now better able to understand the impact of this key biosynthetic step on the protection against crop infection and dehydration that is provided by the periderm layer.

5. EXPERIMENTAL PROCEDURES

Plant materials

FHT-RNAi-silenced potato plants (line 37) (Solanum tuberosum cv. Desirée) were obtained as described by Serra et al. (2010). FHT-RNAi silenced plants and the corresponding wild-type (WT) cultivar were first propagated in vitro: stem cuttings were cultured in Murashige and Skoog media supplemented by 2% (w/v) sucrose and placed in growth cabinets at 22 °C with a light/dark photoperiod cycle of 16/8 h for 21 days. After transfer to soil, the plants were grown in a greenhouse to form tubers that were harvested from 9-week-old plants. Harvested tubers were rinsed with tap water to remove soil particles and the tuber skin was collected using a flat spatula, taking advantage of easy peeling at this immature developmental stage and avoiding the underlying parenchymatic flesh tissue to the extent possible. The samples were frozen in liquid nitrogen, ground using a mortar and pestle, lyophilized until dry, and stored in a vacuum-sealed bag at -80 °C Although the collected tuber skin corresponded strictly speaking to the phellem tissue, we designate the samples as native periderms (containing suberized phellem, phellogen, and phelloderm tissues) in conformance with previous reports (Franke, et al., 2005; Hammes, et al., 2006; Serra, et al., 2014; Stark, et al., 1994).

To prepare wound-healing samples from freshly harvested potato tubers (Dastmalchi, et al., 2014; Pacchiano Jr, et al., 1993; Yang & Bernards, 2006), internal flesh tissues about 5 mm thick were sliced longitudinally to obtain disks and placed on wet cellulose filter paper. The healing process was accomplished inside closed humidified plastic boxes equipped with wire netting supports at 25 °C in the dark. The new brown surface layer of wound-healing tissue was excoriated and collected with a spatula at 0, 3, 7 and 14 days after wounding, corresponding to each stage of suberization: fresh cut samples, early healing in which the suberized closing layer has formed, the developing wound periderm, and well-formed wound periderm. The 3-, 7- and 14-day old wound periderms, as well as the freshly harvested native periderm used in our analyses, were susceptible to skinning. Therefore, both periderms were compared before the skin-set process that marks the transition from immature to mature periderm has taken place. In wound periderm, this phenomenon occurs about 3 weeks post-wounding (Sabba & Lulai, 2002). Upon harvest of the wound-healing materials, samples were collected by the same process as described for the native periderms and stored as dry powders. Both WT and FHT-RNAi samples at the day-0, day-3, day-7, and day-14 time points were collected.

Chemicals

Solvents for soluble metabolite analysis were purchased as follows: analytical grade chloroform and HPLC grade methanol from Fisher Scientific (Pittsburgh, PA); LC-MS grade water and acetonitrile from J. T. Baker (Center Valley, PA); analytical grade hexane and anhydrous pyridine from EMD Chemicals (Gibbstown, NJ). Additional reagents used in these analyses included analytical grade formic acid from Sigma-Aldrich (St. Louis, MO) used in LC-MS experiments, N-Methyl-N-(trimethylsilyl)trifluoroacetamide (MSTFA) + 1% trimethylchlorosilane (TMCS) from Thermo Scientific (Bellefonte, PA) used in GC-MS, and a C7-C40 Saturated Alkanes Retention Index Standard from Supelco (Bellefonte, PA). For removal of cell-wall materials from the solid interfacial suspension between polar and non-polar extracts, cellulase (EC 3.2.1.4; CAS No. 9012-54-8) and pectinase (EC 3.2.1.15; CAS No. 9032-75-1) were purchased from Sigma-Aldrich.

Sample preparation

Dried native periderm (~10 mg) and wound-healing tissues (~20 mg) were weighed and transferred into 4 mL glass vials. Nine replicate extracts were prepared for each WT and FHT-RNAi native potato periderm and each wound-healing time point. Extraction followed a published protocol (Choi, et al., 2004; Kim, et al., 2010; Wolfram, 2006). Briefly, 1.54 mL methanol and 0.33 mL water were added to each dried sample together with 5 small zirconia oxide beads, and the mixture was vortexed for 60 s. After addition of 0.77 mL chloroform and an additional 60 s of vortexing, the samples were shaken in an incubator for 15 min at room temperature. After addition of another 1.54 mL portion of chloroform-water (1:1 v/v), the samples were sonicated for 15 min and shaken in an incubator for another 15 min. The extraction mixtures were centrifuged at 3000 rpm for 15 min at 4 °C; a lower non-polar soluble layer, an upper polar soluble layer, and a solid interfacial suspended residue were collected separately with glass Pasteur pipettes. The non-polar portion (~1.5 mL) was divided between two 2-mL brown vials and dried in a hood at room temperature. The polar portion (~2.0 mL) was also divided between two portions, one of which was dried under nitrogen and the other kept as a liquid without further processing. All samples were stored at -20 °C for subsequent analysis by GC-MS, LC-MS, solution-state NMR, and/or solid-state NMR.

Derivatization for GC-MS analysis was conducted by dissolving the dried non-polar extract in 50 μL of dry pyridine with 80 μL of MSTFA and 1% TMCS (Yang & Bernards, 2007), then shaking in an incubator at 50 °C for 1 h and equilibrating at room temperature.

The solid interfacial residues were washed twice with distilled water and dried overnight at 50 °C, then treated with hydrolytic enzymes to remove unsuberized cellulose and pectin associated with the cell wall. The following procedures, which have been developed for ‘whole’ periderm tissues that were not subjected to extraction (Pacchiano Jr, et al., 1993; Serra, et al., 2014), were used to produce suberin-enriched periderm residues. Aspergillus niger cellulase (0.1% w/v) and pectinase (0.4% v/v) were dissolved in 100 mM acetate buffers at pH 5.0 and pH 4.0, respectively. A 4-mL portion of cellulase buffer was added to each of the samples, which were shaken at 150 rpm and 37 °C for 48 h and then at 44 °C for 48 h. The completeness of cell-wall cellulose digestion using this two-step protocol was validated in two ways: by comparison of our yields (Dastmalchi, et al., 2015) with the >80% losses in mass reported after 5 cellulase treatments of wound potato parenchyma plus periderm (Pacchiano Jr, et al., 1993); and by observing that no further decreases in mass of peeled potato periderms occurred upon retreatment (B. Yu, unpublished results). After rinsing twice with distilled water to remove the cellulase, 4 mL of pectinase buffer was added to each of the samples, which were shaken at 28 °C and then 31 °C for 24 h each. After rinsing three times with distilled water, the residues were dried in an oven at 50 °C overnight. Next, 24-h Sox hlet solvent extraction treatments were used to remove soluble waxes and fatty acids using refluxing methanol, chloroform, and hexane successively. Finally, the samples were dried in an oven at 50 °C for 10 h.

Solution-state NMR analysis

The polar extracts were reconstituted using a 100 mM pH 7.4 phosphate buffer in D₂O containing 0.01 mg/mL of the internal standard 4,4-dimethyl-4-silapentane-1-sulfonic acid (DSS). NMR spectra were obtained using a Bruker AVANCE PLUS spectrometer (Bruker Biospin, Karlsruhe, Germany) operating at an 800 MHz ¹H frequency and fitted with a cryomicroprobe for 1.7 mm o.d. sample tubes (Norell, Landisville, NJ) and an automated NMR SAMPLEJET accessory. ¹H NMR data were acquired at 25 °C using TOPSPIN version 3.1 software and presaturation of the residual water signal set to 4.695 ppm, a constant receiver gain, 512 scans, and a 1.0-s recycle delay between transients. The spectrum after Fourier transformation and signal conditioning had a spectral window of 14 ppm defined by 32K data points.

Gas Chromatography-Mass Spectrometry (GC-MS) analysis

The derivatized non-polar samples were analyzed using a Shimadzu QP2010 plus instrument (Canby, OR) equipped with an AOC-5000 automatic injection system, a RESTEK (Bellefonte, PA) Rtx®-5MS low-bleed column (fused silica, 30 m × 0.25 μm i.d.), and GCMSsolution software version 2.53. The injection temperature was set to 250 °C. A 1 μL sample was injected in splitless mode with high purity helium as carrier gas at a constant flow rate of 1 mL/min and a pressure of 61.5 kPa. The temperature program was set as follows (Yang & Bernards, 2007): after a 5 min delay at 70 °C to allow the solvent t o purge the system, the oven temperature was increased to 310 °C at 5 °C /min, then held for 11 min at that temperature. The total GC run time was 64 min. Mass spectra were recorded in Electron Ionization (EI) mode from 8 to 64 min to avoid the peaks from solvent, pyridine and derivatization reagent. The scan time was 0.5 s and the scan range was m/z 45–600. Instrumental parameters included the ionization voltage (70 eV), ion source temperature (200 °C), and interface temperature (280 °C). The mass spectrometer was tuned with perfluorotributylamine (PFTBA) for mass calibration, performance optimization, and consistency. Three blank samples were run at the beginning, middle, and end of the experimental sequence to record the baseline, check the instrumental performance, and detect possible contamination, respectively.

Liquid Chromatography-Mass Spectrometry (LC-MS) analysis

LC-MS of the polar extracts, for both native periderm and wound-healing samples, was carried out using an Applied Biosystems Sciex4000 QTRAP mass spectrometer (Foster City, CA) equipped with a Shimadzu UFLC system (Canby, OR) and including two LC-20 AD pumps, an SIL-20AC automatic injector, a CBM-20A communication bus module and a CTO-20AC column oven. A 150×4.6 mm i.d. 3.0-μm reverse phase AscentisR™ C18 column (Supelco Corp., Bellefonte, PA) was used with the column oven set to 30 °C and injection volume of 10 μL and 20 μL for native and wound periderm samples, respectively. The mobile phase consisted of (A) water containing 0.1% formic acid (v/v) and (B) acetonitrile containing 0.1% formic acid (v/v). The flow rate was 0.5 mL/min with gradient elution conducted as follows: 0-1 min: 2% B; 1-30 min: 2% - 35% B; 30-45 min: 35% - 98% B; 45-51 min: 98% B; 51-52 min: 98% - 2% B. Electrospray ionization (ESI) mass spectra were acquired in both negative and positive ion modes for the range m/z 100-1300. Pressures of high-purity nitrogen gas (99.995 %) were set at 50 psi for the nebulizing gas and 40 psi for the heating gas, respectively. The source temperature was 300 °C. The optimized values of spr ay voltage, declustering potential, and entrance potential in negative mode were -4500 V, -140 V, and -10 V, while the values in positive mode were 5500 V, 80 V, and 10 V, respectively. Analyst 1.4.2 software was used for data processing. Tandem mass spectrometric analyses (LC-ESI-MS² and MS³) were conducted at four different collision energy levels for both negative mode (-35, -50, -65, and -80 eV) and positive mode (35, 50, 65 and 85 eV). Chlorogenic acid and rutin, compounds reported previously in potatoes, were used as reference standards. Blank samples were run as described for the GC-MS experiments (Dastmalchi, et al., 2014; Neubauer, et al., 2012).

Time-of-flight mass spectrometry (TOF-MS)

To obtain highly accurate masses of metabolites present in the potato skin, the polar extracts of potato native periderm were fractionated 4 times under the same chromatographic conditions used in the LC-MS experiments using an Agilent 1200 series HPLC system (Santa Clara, CA) equipped with a quaternary pump, helium degasser, column heater, diode array detector and analytical fraction collector. Fractions were collected from 6 to 45 min with 1 min intervals for a total time of 40 min. A Waters LCT Premier XE time-of-flight mass spectrometer (Milford, MA) was used to obtain high-resolution molecular masses, with 400 pg/L leucine enkephalin (Sigma-Aldrich, m/z 556.2771) as a standard. The flow rate of the reference solution was 100 μL/min. Mass spectra were acquired over the range m/z 100-1300 in both positive and negative modes. The desolvation gas flow was set to 800 L/h at a temperature of 350 °C, and the cone gas flow was 10 L/h. The cone and aperture voltages were set to 10 V and 25 V for negative mode, 25 V and 10 V for positive mode, respectively. The data acquisition rate was 0.15 sec/scan, with a 0.01 sec interscan delay using dynamic range enhancement (DRE) lock mass. Datasets were collected for 1 min each. The possible elemental compositions for each accurate mass were calculated with Masslynx 4.1 software using a monoisotopic mass type.

Processing of mass spectral data

The software package MZmine 2.4 (VTT Technical Research Center, Helsinki, Finland and Turku Center for Biotechnology, Turku, Finland; http://mzmine.sourceforge.net/) was utilized for processing of GC-MS and LC-MS chromatograms (Katajamaa, et al., 2006; Pluskal, et al., 2010) after conversion of the raw data into metabolomics data matrices suitable for multivariate analysis (Want & Masson, 2011). The raw GC-MS data files (.qgd) were converted into NetCDF by the software GCsolution; the raw LC-MS data (WIFF and SCAN files) were converted into mzXML files using the software Proteo Wizard (http://proteowizard.sourceforge.net/). The converted files were imported to MZmine for filtering of retention time and m/z range, peak detection including mass detection, chromatogram build and peak deconvolution, alignment and peak finder functions. Retention time ranges of 11-55 min for GC-MS and 5-45 min for LC-MS were selected to avoid artifacts from derivatization agents and baseline noise, respectively. Peak detection was optimized to reduce the complexity of the chromatograms, avoid excessive noise features, and create a list of peaks that were each denoted by a specific mass and retention time. Peak alignment was done subsequently to adjust for slight variations in retention time and m/z values for different runs (Katajamaa & Orešič, 2005; Piletska, et al., 2012). The parameters for Min time span, Min peak duration, m/z tolerance, retention time tolerance, and noise level were: (a) 0.1, 0.1, 0.2, 0.2 and 3 × 103 for GC-MS; (b) 0.1 0.1, 0.3, 0.3 and 1.5 × 105 for LC-MS negative mode; (c) 0.1, 0.1, 0.3, 0.5 and 5 × 105 for LC-MS positive mode, respectively. The deconvoluted and aligned datasets were exported to .csv files and normalized with EXCEL to reduce systematic error in the data comparisons.

Multivariate data analysis was performed with SIMCA-P+ version 13.0 (Umetrics, Umea, Sweden) for PCA and orthogonal partial least-squares discriminate analysis (OPLS-DA), which organized the observations, cultivar types, and variables according to their dependence on a small number of variables. The statistical results were visualized with a score plot that discriminated between the cultivars and among the wounding time points, establishing which values of m/z and retention time are responsible for these differences. As described previously (Dastmalchi, et al., 2015; Wiklund, 2008), the OPLS-DA results were visualized in a scatter plot, in which the variables at the extreme ends (e.g., MS ions) were designated as potential biomarkers and verified as specific to a particular sample type using a variable line plot (Wiklund, 2008). The statistical model was validated by calculating values of R² and Q², indicators of fitness to the data and predictive ability, respectively (Worley and Powers, 2013). The value of Q², which is derived from cross validation of the model when a portion of the data is removed, was found to exceed 0.5 and the difference R²-Q² was positive but small, as recommended (Wiklund, 2008).

Metabolite identification

GC-MS identification of compounds in the non-polar periderm extracts was accomplished by comparing the metabolite spectra with the National Institute of Standards and Technology (NIST) 2008 mass spectral library and Wiley Registry of Mass Spectral Data 9th library resident in the GCMSsolution software. Most of the metabolite identifications were based on the similarity score resulting from the comparison of our fragmentation patterns and the library data. Typically, spectral ions were assigned to a specific compound when the similarity score was higher than 80. The 2-monoacylglycerols were identified by comparing reported marked ions and EI fragmentation patterns, e.g., the fragments m/z 218, 203 and 191 that are characteristic for the β-isomer (Destaillats, et al., 2010). As families of alkane compounds have similar MS patterns and fragments, a retention index (RI) was used to assist with identification by comparing their retention times against a standard C7-C40 saturated series (Sigma-Aldrich) with retention index values from 700 to 4000.

The metabolites of the polar extracts of native periderm were identified by comparing LC retention times and tandem MS/MS/MS fragments from the 4000 QTRAP analysis, and accurate masses from TOF-MS, with available literature reports. Searches were conducted using on-line databases such as Scifinder, Pubchem, and Metfrag to match the current data to a reported spectrum. Additionally, some metabolites for which no spectrum had been reported nevertheless displayed similar fragments to identified compounds, allowing for provisional assignment based on MS/MS fragmentation patterns and subsequent confirmation using TOF-MS data.

Wax compositional analyses by GC

The non-polar extracts were also used to determine the wax composition including the alkyl ferulates. The extracts were dried under nitrogen flow and derivatized using 20 μl of N, O-bis(trimethylsilyl)trifluoroacetamide (BSTFA; Macherey-Nagel) and 20 μl of pyridine (Sigma-Aldrich). The wax compositional data was obtained by GC-FID analysis performed using a Shimadzu GC-2010 Plus instrument using a BP1 capillary column (30 m length, 0.25 mm i.d., 0.1 μm film thickness, Teknokroma). A split/splitless injector in the splitless injection mode was used (splitless time 1 min) with the injector temperature at 280 °C and the detector temperature maintained at 320 °C. Helium was used as the carrier gas at a constant flow rate of 1 ml min^-1. Initially the oven temperature was at 120 °C, then increased by 20 °C min^-1 up to 270 °C, then by 1.5 °C min^-1 up to 340 °C and held at 340 °C for 7 minutes. Chromatograms were processed using GC Solution software (version 2.41) from Shimadzu. The percentage of each compound in the extract was calculated based on the individual integrated peak area divided by the sum of the areas of all identified compounds.

The compounds were identified in a GC-MS analysis using a GC Trace instrument (ThermoFisher) coupled to an ion trap mass spectrometer (PolarisQ, ThermoFisher), a BPX-5 capillary column (30 m length, 0.25 mm i.d., 0.25 μm film thickness) (SGE, Australia & Pacific Region) and helium as a carrier gas at 1 mL min^-1. A split/splitless injector in the splitless injection mode was used (splitless time 1 min) with the injector temperature at 280 °C. The oven temperature program was the same described for the GC-FID analyses. MS analyses were carried out with electron impact ionization operating at 70 eV and the ion source was set at 225 °C. The transfer line was held at 290 °C and acquisition was performed in full-scan mode, with a scan range of 50-700 amu. Chromatographic data were analysed using Xcalibur 1.4 software.

Solid-state NMR analysis

The principal chemical moieties present in each of three dry biological replicate suberin extraction residues were determined using standard cross-polarization magic-angle spinning (CPMAS) ¹³C NMR experiments carried out on a 4-channel Varian (Agilent) DirectDrive I (VNMRS) NMR 600 MHz spectrometer (Palo Alto, CA) equipped with a 1.6 mm HXY FastMAS probe. The ¹³C resonance frequency was 150.757400 MHz, with a customary acquisition time of 25 ms, a delay time of 3 s between successive acquisitions, a CP contact time of 1.5 ms, and a 200 kHz ¹H decoupling strength achieved using the SPINAL method (Fung, et al., 2000). Typically, 5 mg samples were spun at 10,000 ± 20 Hz for approximately 7 h. The resulting CPMAS data were processed using VNMRJ (version 2.2C; Agilent) and/or ACD/NMR Processor Academic Edition 12 (Advanced Chemistry Development, Inc., Toronto, ON, Canada) with 100 Hz of exponential line broadening to improve the signal-to-noise ratio of the spectra. Chemical shift referencing was done externally using the methylene (-CH_2^-) group of adamantane (Sigma-Aldrich) set to 38.48 ppm.

To estimate ratios of the various carbon types, quantitatively reliable direct-polarization magic-angle-spinning (DPMAS) ¹³C NMR spectra of these amorphous plant materials (Serra, et al., 2014; Zlotnik-Mazori & Stark, 1998) were acquired with 3000 scans for 83 h. A customary acquisition time of 25 ms and delay time of 100 s between successive acquisitions were used. After applying exponential line broadening of 200 Hz, signal intensities in the DPMAS spectra were measured by cut-and-weigh-paper methods and by counting pixels using Photoshop software for the following regions: carboxyl groups (COO; 160-180 ppm), multiply bonded groups (aromatics and alkenes; 92-160 ppm), oxygenated aliphatic carbon groups (CHO + CH₂O + CH₃O; 44-92 ppm) and long-chain methylene groups (-(CH₂)_n-; 10-44 ppm). Ratios of integrated signal intensities for particular spectral regions were measured for each cultivar using both methods; error bars represent the difference between the two values.

Thioacidolysis analysis

Between 10 and 45 mg of native potato tuber periderm samples previously treated with 2% of cellulase and pectinase as described (Serra et al., 2010) were used as starting tissue for each replicate. Prior to thioacidolysis analysis the samples were treated with methanol:chloroform (1:1, v/v) overnight to remove the extractives and with methanol/boron trifluoride (~10% BF₃ in methanol; Fluka) overnight to remove the aliphatic suberin polyester as described in Serra et al. (2010). The remaining insoluble residue was subjected to thioacydolysis to determine the lignin-like composition by GC–FID of lignin-derived monomer trimethylsilyl derivatives (Lapierre, et al., 1986). H, G and S units were identified by comparison with commercially available standards. The amount of each unit was related to the weight of the dried starting material.

Highlights.

1. Soluble metabolite profiles of FHT and wild-type native periderms show significant differences.

2. A metabolic rerouting to increase phenolic amines occurs in FHT native periderm.

3. Differences between FHT and wild-type wound non-polar extracts are found only at day14.

4. FHT native and wound periderms show an augmented hydrophilic-hydrophobic balance.