Knocking Down of Isoprene Emission Modifies the Lipid Matrix of Thylakoid Membranes and Influences the Chloroplast Ultrastructure in Poplar

Isoprene emission in plants is correlated with structural organization and function of plastidic membranes.

Abstract

Isoprene is a small lipophilic molecule with important functions in plant protection against abiotic stresses. Here, we studied the lipid composition of thylakoid membranes and chloroplast ultrastructure in isoprene-emitting (IE) and nonisoprene-emitting (NE) poplar (Populus × canescens). We demonstrated that the total amount of monogalactosyldiacylglycerols, digalactosyldiacylglycerols, phospholipids, and fatty acids is reduced in chloroplasts when isoprene biosynthesis is blocked. A significantly lower amount of unsaturated fatty acids, particularly linolenic acid in NE chloroplasts, was associated with the reduced fluidity of thylakoid membranes, which in turn negatively affects photosystem II photochemical efficiency. The low photosystem II photochemical efficiency in NE plants was negatively correlated with nonphotochemical quenching and the energy-dependent component of nonphotochemical quenching. Transmission electron microscopy revealed alterations in the chloroplast ultrastructure in NE compared with IE plants. NE chloroplasts were more rounded and contained fewer grana stacks and longer stroma thylakoids, more plastoglobules, and larger associative zones between chloroplasts and mitochondria. These results strongly support the idea that in IE species, the function of this molecule is closely associated with the structural organization and functioning of plastidic membranes.

Isoprene is globally the most abundant biogenic hydrocarbon constitutively emitted from many plant species (Guenther et al., 2012). It has been proposed that leaf isoprene emission is an important adaptation for plants, conferring tolerance to different environmental constraints (Vickers et al., 2009; Loreto and Schnitzler, 2010; Loreto and Fineschi, 2014). However, biogenic isoprene emission represents a nontrivial carbon loss in plants, particularly under stress conditions (Fang et al., 1996; Brilli et al., 2007; Teuber et al., 2008; Ghirardo et al., 2014), and the reason(s) why plants emit isoprene are still ambiguous, and the true role of isoprene emission remains elusive. Different approaches and techniques have been used to determine whether and how the cost of this expensive carbon emission is matched by the accomplishment of the physiological function in planta. It has been shown that isoprene might quench and/or regulate reactive oxygen and nitrogen species formation (Behnke et al., 2010a; Velikova et al., 2012), thereby indirectly providing a general antioxidant action (for review, see Vickers et al., 2009; Loreto and Schnitzler, 2010) and stabilizing thylakoid membrane structures due to the lipophilic properties of this molecule (Sharkey et al., 2001; Velikova et al., 2011).

Protein and pigment-protein complexes are assembled and embedded in a lipid matrix, which has a unique lipid composition. The thylakoid lipid bilayer of chloroplasts is characterized by a high proportion of galactolipids with one (monogalactosyldiacylglycerol [MGDG]) or two (digalactosyldiacylglycerol [DGDG]) Gal molecules (Joyard et al., 2010). MGDGs are the primary constituents (approximately 50%) of thylakoid membrane glycerolipids, followed by DGDGs (approximately 30%), sulfoquinovosyldiacylglycerol (approximately 5%–12%), and phosphatidylglycerol (approximately 5%–12%; Kirchhoff et al., 2002). Galactolipids contain a large proportion of polyunsaturated fatty acids, and consequently, the thylakoid membrane is a relatively fluid system (Gounaris and Barber, 1983) compared with other biological membranes. The fluidity of the thylakoid membrane is essential for photosynthetic processes.

The thylakoid membranes are highly organized internal membrane chloroplast systems that conduct the light reactions of photosynthesis. These membranes comprise pigments and proteins organized in complexes. Thylakoid membranes are arranged into stacked and unstacked regions called grana and stroma thylakoids, respectively, differentially enriched in PSI and PSII complexes (Mustárdy et al., 2008). The spatial separation of the PSI and PSII complexes in the stacked and unstacked membrane regions and the macromolecular organization of PSII in stacked grana thylakoids are self-organizing processes and important features to maintain the functional integrity of the photosynthetic apparatus (Kirchhoff et al., 2007).

It is not known how changes in the lipid matrix affect lipid-protein interactions and vice versa, and how membrane macroorganization ensures the efficient diffusion of protein complexes associated with plant adaptation to the changing environment remains elusive. The isoprene impact on thylakoid intactness and functionality has been assessed using different biophysical techniques (Velikova et al., 2011). Thermoluminescence data demonstrated that the position of the main peak (Qb peak; Sane, 2004) was upshifted approximately 10° in isoprene-emitting (IE) plants, suggesting modifications in the lipid environment due to the presence of isoprene in heterologous Arabidopsis (Arabidopsis thaliana) plants expressing the isoprene synthase gene from poplar (Populus spp.). It was also shown that isoprene improves the stability of PSII light-harvesting complex II (LHCII) through the modification of pigment-protein complex organization in thylakoid membranes (Velikova et al., 2011). Moreover, we recently showed that knocking down isoprene emission in poplar remodels the chloroplast proteome (Velikova et al., 2014). The lack of isoprene resulted in the down-regulation of proteins associated with the light reactions of photosynthesis, redox regulation, and oxidative stress defenses and several proteins responsible for lipid metabolism (Velikova et al., 2014).

In this study, we focused on the lipid composition of thylakoid membranes in IE and nonisoprene-emitting (NE) poplar (Populus × canescens) leaves. Specifically, we determined whether the translational suppression of isoprene synthase in NE leaves influences the lipid matrix of thylakoids and how this phenomenon affects membrane structure and function. Here, we provide evidence that the suppression of isoprene biosynthesis in poplar (1) reduced total galactolipids, phospholipids (PLs), and linolenic fatty acid (18:3), (2) altered the chloroplast ultrastructure, and (3) stimulated photoprotective mechanisms.

RESULTS

Lipid, Fatty Acid, and Malondialdehyde Analyses

Chloroplast membranes isolated from NE poplar had significantly lower (−53%; P < 0.01) lipid contents than the membranes of IE plants (Fig. 1). In both NE and IE plants, the major molecular species of MGDG were 18:2-18:3 and 18:3 dimers. In DGDG, the major molecular species were n16:0-18:3 and 18:3 dimers. Linolenic acid (18:3) was the major fatty acid of both IE and NE chloroplast membranes, but the content of this fatty acid was consistently much lower in NE than in IE plants (Table I; Supplemental Fig. S1). The fatty acid analysis also revealed significantly lower palmitic (n16:0), linoleic (18:2), and stearic (n18:0) acid contents in NE plants compared with IE plants. In the fraction of the PLs, the phosphatidic acid (18:1) levels were lower in NE plants.

Figure 1.

Lipid contents in isolated chloroplasts of IE (wild type [WT] and empty vector [EV]) and NE (RNA interference transgenic lines, RA1 and RA2) poplar. Error bars display the se (n = 4). Asterisks indicate significant differences from the wild type: **, P < 0.01.

Table I. Fatty acid composition (μg mg⁻¹ chlorophyll) of the main lipid classes in C₁₆ to C₂₃ saturated (:0) and unsaturated (:1, :2, and :3) compounds in chloroplasts of IE (wild type [WT] and empty vector [EV]) and NE (RA1 and RA2) poplar plant lines.

Fatty acids are designated as the total number of carbon atoms followed by the number of double bonds and their location (omega) after the colon: n16:0, palmitic acid; 16:1, palmitoleic acid; n18:0, stearic acid; 18:1, oleic acid; 18:2, linoleic acid; 18:3, linolenic acid; and n23:0, tricosanoic acid. Saturated straight-chain fatty acids are indicated with an n. Means ± se are shown; n = 4. Asterisks and boldface indicate significant differences from the wild type: *, P < 0.05; **, P < 0.01; and ***, P < 0.001.

Line	Lipid Classes	n16:0	16:1	n18:0	18:1	18:2	18:3	20:3	n23:0
WT	MGDG	21.1 ± 3.9		2.8 ± 0.6		61.8 ± 13.6	411.1 ± 29.0
	DGDG	77.5 ± 22.1		13.2 ± 5.4		10.7 ± 3.3	265.1 ± 17.6	1.4 ± 0.6
	PLs	17.7 ± 3.6	3.5 ± 1.2	4.1 ± 0.5	12.9 ± 4.7	7.6 ± 1.3	151.9 ± 16.5		4.0 ± 1.0
EV	MGDG	19.5 ± 6.0		2.6 ± 0.9		54.9 ± 19.5	350.3 ± 75.2
	DGDG	56.0 ± 15.9		8.7 ± 2.7		8.8 ± 3.2	230.3 ± 59.4	1.4 ± 0.6
	PLs	17.8 ± 2.2	2.4 ± 0.4	4.3 ± 0.2	13.8 ± 6.4	9.1 ± 0.5	120.9 ± 3.0		4.1 ± 1.9
RA1	MGDG	12.7 ± 1.7*		1.7 ± 0.2		38.0 ± 8.9*	253.2 ± 20.0*
	DGDG	37.8 ± 9.0**		5.7 ± 1.5*		5.7 ± 1.6*	154.2 ± 30.1***	0.5 ± 0.2*
	PLs	9.1 ± 2.7	1.5 ± 0.5	3.3 ± 0.5	3.7 ± 0.7	5.7 ± 0.8	66.1 ± 9.0*		0.7 ± 0.1*
RA2	MGDG	11.9 ± 1.1**		1.9 ± 0.4		35.9 ± 4.9*	236.0 ± 8.8*
	DGDG	35.0 ± 1.8**		5.4 ± 0.4*		4.7 ± 0.7*	139.5 ± 1.5***	0.5 ± 0.1*
	PLs	11.5 ± 1.2*	0.9 ± 0.3	2.7 ± 0.3**	3.6 ± 1.3	3.2 ± 0.4	72.6 ± 6.6**		2.5 ± 0.5

The concentration of malondialdehyde (MDA), the principal product of polyunsaturated fatty acid peroxidation, was lower in chloroplasts isolated from NE plants than in those isolated from IE poplar (Fig. 2), associated with a lower concentration of polyunsaturated fatty acids in NE chloroplasts (Table I; Supplemental Figure S1).

Figure 2.

MDA levels in isolated chloroplasts of IE (wild type [WT] and empty vector [EV]) and NE (RA1 and RA2) poplar. Error bars display the se (n = 4). Asterisks indicate significant differences from the wild type: *, P < 0.05.

Chloroplast Ultrastructure Observations and Protein Abundance in Photosynthetic Membranes

To determine whether the different lipid concentrations and changes in lipid composition affect the chloroplast structure, thin leaf segments obtained from the middle region of IE and NE leaves were subjected to transmission electron microscopy (TEM) analyses. Representative micrographs of chloroplasts from IE and NE specimens are shown in Figures 3 and 4.

Figure 3.

A to D, Transmission electron micrographs of representative chloroplast cross sections taken from the intact leaves of IE (wild type [WT] and empty vector [EV]; A and B) and NE (RA1 and RA2; C and D) poplar. E and F, Height-length ratio (E) and average number of starch grains (F) in IE and NE chloroplasts. CW, Cell wall; GT, granal thylakoids; M, mitochondrion; P, plastoglobuli; S, stroma; SI, starch gain. Bars = 1 μm at 6,300× magnification.

Figure 4.

A to D, Transmission electron micrographs of representative chloroplast cross sections taken from the intact leaves of IE (wild type [WT] and empty vector [EV]; A and B) and NE (RA1 and RA2; C and D) poplar. E and F, Average number of stacks per chloroplast (E) and correlation between PSII-RCII protein abundance (10 peptides; for protein accession nos., see “Materials and Methods”) and number of stacks (F). CW, Cell wall; GT, granal thylakoids; M, mitochondrion; P, plastoglobuli; S, stroma; SI, starch gain. Bars = 500 nm in A and C at 10,000× magnification; bars = 200 nm in B and D at 20,000× magnification.

The typical elliptic shape of mesophyll chloroplasts was more oval in NE than in IE specimens (Fig. 3). The mesophyll cells of IE leaves are characterized by a well-developed inner membrane system, comprising grana of different sizes and relatively long stromal thylakoids. IE chloroplasts contained single, midsize starch granules and less numerous peroxisomes, and these organelles were associated with relatively small-sized mitochondria (Figs. 3, A and B, and 4, A and B).

Conversely, the chloroplasts of NE plants were characterized by a less developed membrane system, with shorter and fewer grana stacks and longer stroma thylakoids (Fig. 4, C–E). NE chloroplasts contained more plastoglobules and smaller starch grains than IE chloroplasts (Fig. 3, C and D). NE chloroplasts were also in close structural contact with mitochondria through relatively large associative regions (Fig. 4C). A relatively large number of NE chloroplasts were undeveloped (data not shown).

To further understand how the structural changes were related to the protein enrichment in photosynthetic membranes, we extracted mass spectrometry (MS) data from our recent proteome study (Velikova et al., 2014). The concentrations of PSI-PSI reaction centers (RCI) and PSII-PSII reaction centers (RCII) were strongly decreased in NE chloroplasts compared with IE chloroplasts (Fig. 5, A and B). Lower protein abundance of PSII-RCII correlated with fewer stacks (Fig. 4F). Chlorophyll concentrations in the NE lines RA1 and RA2 were also significantly reduced (Fig. 5C).

Figure 5.

Protein abundance of PSI-RCI (four peptides; for protein accession nos., see “Materials and Methods”; A), PSII-RCII (10 peptides; for protein accession nos., see “Materials and Methods”; B), and chlorophyll content (C) in IE (wild type [WT] and empty vector [EV]) and NE (RA1 and RA2) poplar plants. Protein abundance represents the sum of MS data extracted from our proteome study (Velikova et al., 2014). Error bars display the se (n = 4). Asterisks indicate significant differences from the wild type: **, P < 0.01; and ***, P < 0.001.

Chlorophyll Fluorescence

We measured light- and dark-adapted states of chlorophyll fluorescence in IE and NE poplar plants grown under ambient greenhouse conditions (Fig. 6). There was no significant difference in maximal PSII activity between IE and NE plants (data not shown), suggesting that the efficiency of PSII, when all reaction centers were open, was similarly high in both groups of plants. However, NE leaves exhibited significantly lower photosystem II photochemical efficiency (Φ_PSII) and PSII redox state (qL) and higher nonphotochemical quenching (NPQ) and energy-dependent quenching (qE; Fig. 6). Importantly, the true efficiency of PSII (Φ_PSII) was lower in NE compared with IE, indicating that a smaller fraction of the absorbed light energy was used for photochemistry. Indeed, the accurate indicator of the PSII redox state, qL (Baker, 2008), was significantly lower in NE, suggesting that the fraction of open PSII reaction centers was much lower in these mutants (Figs. 5 and 7; Supplemental Fig. S2).

Figure 6.

Φ_PSII (A), redox state of PSII (qL; B), NPQ (C), and qE (D) of IE (wild type [WT] and empty vector [EV]) and NE (RA1 and RA2) poplar plants at growth conditions. Values represent means of five to seven different plants out of three independent experiments (n = 15–21; se is given). Photosynthetic parameters are described in “Materials and Methods.” Asterisks indicate significant differences from the wild type: *, P < 0.05; **, P < 0.01; and ***, P < 0.001.

Figure 7.

Score (A), loading (B), and correlation coefficient plots (C) of orthogonal partial least squares (OPLS) of lipid classes, fatty acid composition, and MDA contents in isolated chloroplasts, chlorophyll fluorescence parameters measured in intact leaves (NPQ, Φ_PSII, qE, and qL), and chloroplast proteins related to photosynthesis and proteins with structural activity. A, IE plants (wild type [WT] and empty vector [EV]), gray circles; NE plants (RA1 and RA2), white triangles. B, Each parameter is indicated with a different symbol: dark-gray circles, MGDG; dark-gray squares, DGDG; gray triangles, PL; dark-gray circles with a dot, MGDG fatty acids; dark-gray squares with a dot, DGDG fatty acids; gray triangles with a dot, PL fatty acids; green diamonds, MDA; red stars, NPQ; red down triangles, qE; blue squares, qL; blue stars, proteins with structural activity; and green stars, proteins related to photosynthesis. C, Parameter colors match those of the symbols in B. Only discriminant data with variable of importance for the projection (VIP) > 1 (for all except proteins) and VIP > 0.5 (proteins) are presented. Model fitness: Q²(Y) = 84%, R²(X) = 44%, r² = 93%, and R²(Y) = 100% using one principal component; P = 0.00061, cross-validated ANOVA.

Multivariate Data Analyses

We examined the involvement of lipid content and fatty acid composition in NE and IE chloroplasts, compared with chlorophyll fluorescence measurements, MDA contents, and previously described proteomic differences (Velikova et al., 2014). Principal component analysis (PCA) showed that the isoprene emission traits reflected the largest variance of the measured data, indicated by the separation between NE and IE samples in the first two principal components (Supplemental Fig. S2; explained variance, PC1 = 51%, PC2 = 16%), where Q² is the fraction of Y variation predicted by the X model and according to cross-validation, and R²(X) and R²(Y) are the fraction of X variation modeled, using the X and Y model, respectively. Additionally, there were no appreciable differences between the two groups, as both RA1 and RA2 lines and WT and EV lines clustered together in the NE and IE groups, respectively.

We performed a discriminant analysis to determine which of the analyzed parameters were significantly affected by the suppression of isoprene biosynthesis and to assess the relative importance of these parameters in distinguishing NE from IE plants (Fig. 7). The OPLS analysis indicated, overall, that differences between NE and IE chloroplasts reflect the lipid composition, fatty acid and MDA contents, chlorophyll fluorescence parameters, and chloroplastic proteins associated with photosynthesis or cell structure. Each singular factor had a different importance (Fig. 7C). Specifically, the most important (high VIP values) variables negatively correlated with NE (positive and high correlation coefficient values) were MDA, qL, PL, saturated and unsaturated fatty acids, MGDG and DGDG, and photosynthetic proteins (Fig. 7, B and C). Importantly, the unsaturated fatty acid 18:3 (linolenic acid) was strongly negatively correlated with NE in all lipids (MGDG, DGDG, and PL). Additionally, linolenic acid was well correlated with the lipid degradation product MDA, detected in both MGDG and DGDG. The PL content was highly correlated with photosynthetic proteins, namely PSI proteins, ATP synthase, cytochrome b₆f, and an intrinsic protein of PSII oxygen-evolving complex, PsbP. Conversely, NPQ and qE were strongly and positively correlated with the NE genotype.

The computed OPLS model was reliable, resulting in a significant (P = 0.00061, cross-validated ANOVA) cross-validated predictive ability of Q²(Y) = 84% to distinguish NE from IE samples and a cross-validated goodness of R²(X) = 44%, r² = 93%, R²(Y) = 100% using only the first principal component.

DISCUSSION

Suppression of Isoprene Biosynthesis Decreases the Chloroplastic Lipid Content and Alters Chloroplast Ultrastructure

One of the proposed biological functions of isoprene is the stabilization of thylakoid membrane structures through modification of the lipid environment and organization of the pigment-protein complexes in thylakoid membranes (Velikova et al., 2011). Indeed, several clear alterations were evident in thylakoid membrane lipids and fatty acid composition due to the translational suppression of isoprene synthase activity in poplar plants. The most important changes in NE chloroplasts were the absolute decrease in the contents of galactolipids (MGDG and DGDG) and PLs through the down-regulation of the unsaturated fatty linolenic acid (18:3). The functional role of MGDGs in the bioactivity of various membrane proteins is well known (Lee, 2003, 2004): a mutant with a defective MGDG SYNTHASE1 is unable to produce photosynthetically active membranes (Kobayashi et al., 2013). MGDGs are essential for the efficient activity of violaxanthin deepoxidase (Yamamoto and Higashi, 1978). The ability of the lipid mixture to segregate into bilayer and nonbilayer phases might regulate the protein content in chloroplast membranes (Garab et al., 2000). Indeed, we could demonstrate that the proteins related to photosynthesis were strongly down-regulated in NE compared with IE plants (Figs. 7 and 8; Velikova et al., 2014).

Figure 8.

Schematic overview of the changes in chloroplast ultrastructure, lipid composition, protein abundance, and PSII photochemical efficiency triggered by the suppression of isoprene biosynthesis and emission in poplar plants. VDE, Violaxanthin deepoxidase.

DGDGs are the predominant bilayer lipid species in thylakoid membranes of higher plants (Joyard et al., 2010). They exert structural functions and improve the thermal stability of membranes, particularly at high temperatures (Krumova et al., 2010). DGDGs bind to PSII (Loll et al., 2007) through the formation of hydrogen bonds with Tyr in PSII (Gabashvili et al., 1998), and DGDGs are also important for binding of extrinsic proteins required for the stabilization of the oxygen-evolving complex (Sakurai et al., 2007). Our data clearly indicate that the suppression of isoprene biosynthesis significantly diminished the level of DGDGs in poplar chloroplasts, which was accompanied by a reduction of the RCI and RCII concentrations (Fig. 5, A and B), PsbP and PsbQ protein subunits of PSII, and light-harvesting complex I and LHCII (Velikova et al., 2014). When the thylakoid membrane protein complexes were resolved by blue native-PAGE, the protein patterns of the two groups of poplar lines looked quite similar in content and intensity of the individual bands (supplemental figure S3 in Velikova et al., 2014). However, semiquantitative analysis of the individual protein bands showed that the levels of PSI, the PSII dimer, ATP synthase, the PSII monomer, and the cytochrome b₆f complex were slightly reduced in NE compared with IE chloroplasts (supplemental figure S3 in Velikova et al., 2014).

Decreased lipid and protein levels were associated with changes in the ultrastructure of the chloroplasts from NE plants, suggesting a role for isoprene biosynthesis in the structural organization of plastidic membranes. These results are consistent with previous studies that indicated a role for MGDGs and DGDGs in the structure of thylakoid membranes (Dörmann et al., 1995; Jarvis et al., 2000). In this study, we observed a significant reduction of grana stacks per chloroplast in NE compared with IE poplar lines (Fig. 4, E and F), which was related to an important decrease in PSI and PSII proteins (Fig. 5, A and B). Alterations in protein stoichiometry could exert a direct influence on the thylakoid membrane ultrastructure (Pribil et al., 2014), in particular, the assembly of PSII and LHCII into supercomplexes and megacomplexes (Kouřil et al., 2012). It was demonstrated (Labate et al., 2004) that the constitutive expression of the pea (Pisum sativum) Lhcb1 gene in transgenic tobacco (Nicotiana tabacum) plants leads to increased grana stacking, indicating that increased concentrations of LHCII result in more stacking.

Typically, chloroplast membranes have a unique lipid composition characterized by a high proportion of galactolipids containing a large portion of triunsaturated fatty acids (C₁₆ or C₁₈; Joyard et al., 2010). The high content of triunsaturated fatty acids guarantees the high fluidity of the thylakoid membranes and the precise allocation of the photosynthetic machinery to efficiently acquire light energy (Gounaris and Barber, 1983). The level of membrane viscosity is an important factor for photosynthetic performance (e.g. providing optimal conditions for the diffusion of hydrophobic molecules, such as plastoquinol; Kirchhoff et al., 2000, 2002) or membrane intrinsic protein complexes (e.g. during state transitions; Allen and Forsberg, 2001; Tikkanen et al., 2008). The low linolenic acid (18:3) content in all lipid fractions from NE chloroplasts indicates that, in the absence of isoprene, the thylakoid membrane fluidity is reduced, which in turn negatively affects the efficiency of PSII photochemistry (Fig. 6). A low level of unsaturation in thylakoid membranes makes PSII extremely susceptible to photoinhibition and causes a significant reduction in the content of the D1 protein (the reaction center protein) at high irradiance (Kanervo et al., 1995), suggesting that membrane fluidity is a critical factor for PSII D1 protein turnover. Moreover, we detected in NE chloroplast lower amounts of phosphatidic acid (18:1), an important intermediate in lipid biosynthesis (Joyard et al., 2010) with functions as a signaling lipid (Testerink and Munnik, 2005, 2011; Horváth et al., 2012; McLoughlin and Testerink, 2013).

We observed that the lower level of linolenic acid (18:3) detected in NE chloroplasts was associated with significantly lower MDA chloroplast content in NE compared with IE poplar (Fig. 2; Table I). Previous studies have reported that MDA is primarily derived from triunsaturated fatty acids in chloroplasts (Yamauchi et al., 2008; Schmid-Siegert et al., 2012). MDA can be used as an oxidative stress marker when plants are exposed to unfavorable conditions (Esterbauer et al., 1991) but is also present in healthy plants (Weber et al., 2004; Mène-Saffrané et al., 2007, 2009). At the whole-leaf extract level, MDA levels were higher in NE poplar (Behnke et al., 2010b), which agrees with their higher concentrations of linolenic acid (Way et al., 2013). The production of MDA from triunsaturated fatty acids serves to adsorb a portion of the reactive oxygen species (Mène-Saffrané et al., 2009); therefore, MDA is a by-product in the mechanism of cell protection.

Another remarkable observation in this study was the increased abundance of plastoglobules in NE compared with IE chloroplasts. Plastoglobules are lipoprotein particles containing isoprenoid-driven metabolites (primarily prenylquinones, including plastoquinone and phylloquinone), tocopherols (Vidi et al., 2006), and structural proteins (plastoglobulins; Bréhélin et al., 2007). The increased number of plastoglobules in NE compared with IE chloroplasts might reflect the higher levels of α-tocopherol in leaves of these lines, as demonstrated previously (Behnke et al., 2010b).

Electron tomography revealed that plastoglobules are physically coupled to thylakoid membranes via a half-lipid bilayer, providing a direct lipid conduit for metabolite channeling between plastoglobules and thylakoid membranes (Austin et al., 2006). Moreover, plastoglobules are involved in different secondary metabolism pathways, stress responses, and the development of thylakoids (Bréhélin et al., 2007). In a previous study, we showed that the lipid-associated protein fibrillin content, comprising lipid-associated proteins and fibrillins, is negatively correlated with the NE plants (Velikova et al., 2014). This observation suggests the involvement of isoprene in the maintenance of thylakoid membranes.

Interestingly, we observed larger associative zones between chloroplasts and mitochondria in NE plants. Mitochondria are instrumental for the generation of metabolic energy in eukaryotic cells, and these organelles deliver intermediates to support different metabolic pathways, including photosynthesis (Jacoby et al., 2012). One of the important benefits of mitochondria-chloroplast interactions is the optimization of photosynthetic carbon assimilation through the coordinated production and utilization of ATP and NADPH, the induction of photosynthesis, the activation of enzymes, and the maintenance of metabolite levels (Raghavendra and Padmasree, 2003). We propose that the larger associative zones between chloroplasts and mitochondria in NE plants reflect a higher demand for assimilatory power (ATP and NADPH) compared with IE plants. Indeed, the down-regulation of the cytochrome b₆f complex in NE chloroplasts indicates the inhibition of ATP production, associated with the down-regulation of extrinsic subunits of ATP synthase in isoprene-suppressed lines (Velikova et al., 2014). Because isoprene functions as a protective molecule against oxidative stress (Loreto and Schnitzler, 2010), the isoprene suppression in NE plants might be balanced by enhancing other compensatory protective mechanisms such as photorespiration and oxidative electron transport, which are both mediated by mitochondria.

Recent analyses demonstrated that the suppression of isoprene biosynthesis dramatically reduces carbon fluxes throughout the methylerythritol 4-phosphate pathway (Ghirardo et al., 2014), followed by the reallocation of carbon to other pathways, which in turn induces profound metabolic changes, particularly in lipid biosynthesis (Way et al., 2013; Kaling et al., 2015). Thus, at the cellular level, lipid metabolism is up-regulated in NE leaves, whereas at the subcellular level, as shown here, low levels of galactolipids and PLs comprise the structure of NE chloroplasts. These results suggest that the crucial needs of NE plants to maintain the correct fluidity of thylakoid membranes induces the up-regulation of lipid metabolism, including lipid intermediates, likely compensating for the low levels of galactolipids and PLs packed into chloroplast membranes. Thus, isoprene might (1) directly improve the fluidity of thylakoid membranes in synergy with galactolipids or (2) indirectly affect lipid biosynthesis or trafficking into the chloroplast. Whether the lack of isoprene function or the alteration of the plastidic isoprenoid pathway itself induces changes in the chloroplast lipid levels, thereby affecting membrane fluidity, should be examined in future studies.

Functional Changes Relate to Structural Alterations in NE Chloroplasts

We measured light- and dark-adapted states of chlorophyll fluorescence in IE and NE poplar plants grown under unstressed conditions in order to assess whether the structural changes have functional significance with regard to the differences in the ability to emit isoprene. Our results showed that Φ_PSII was lower in NE plants than in IE plants, consistent with previous observations that the proteins involved in photosynthetic processes are down-regulated in NE plants, potentially decreasing the efficiency of the photochemistry of photosynthesis (Fig. 5; Velikova et al., 2014). The lower Φ_PSII values in NE plants were negatively correlated to NPQ, a protective mechanism for the removal of excess excitation energy within pigment complexes and the inhibition of the formation of free radicals (Demmig-Adams and Adams, 2006).

Higher NPQ levels in concert with restricted electron transport rate between both photosystems and a reduced plastoquinone pool have been shown (Härtel et al., 1998) to be accompanied by DGDG modifications in the Arabidopsis mutant (dgd1). Moreover, in this mutant, PSI showed an increased capacity for cyclic electron transfer and a reduced capacity for state transitions (Ivanov et al., 2006). Similar to the dgd1 mutant (Dörmann et al., 1995), the NE poplar plants showed a lower DGDG content, modified chloroplastic ultrastructure, increased NPQ, restricted electron transport rate (Behnke et al., 2007), and decreased total chlorophyll content (Fig. 5C; Behnke et al., 2013; Way et al., 2013; Ghirardo et al., 2014).

The NPQ comprises qE (i.e. dependent on the energization of thylakoid membranes), state transitions (Minagawa, 2011), photoinhibition quenching (Müller et al., 2001), and zeaxanthin-dependent quenching (Nilkens et al., 2010). qE is the most important and well-characterized component of NPQ. This transition is triggered through the acidification of the thylakoid lumen (Ruban et al., 2012), which in turn leads to the protonation of violaxanthin deepoxidase, for the conversion of violaxanthin to zeaxanthin, and PsbS, a polypeptide of the PSII-associated light-harvesting complex (Kiss et al., 2008; Murchie and Niyogi, 2011).

Here, we showed higher values of qE in NE plants, which might reflect a particular conformation of the LHCII complex resulting from chlorophyll and/or xanthophyll-protein interactions (Horton et al., 2005). Indeed, we observed that many proteins associated with photosynthesis are less abundant in NE chloroplasts (Figs. 5 and 7; Velikova et al., 2014). This lack of photosynthetic proteins could lead to specific conformational changes, which in turn could determine the higher qE in NE poplar. However, the supramolecular organization of the PSII antenna involves numerous interactions between proteins, suggesting that the changes in these interactions (Garab and Mustardy, 1999; Horton et al., 2005) could be responsible for the increase in NPQ we observed in NE plants. Indeed, with circular dichroism spectroscopy, it has been shown that isoprene deficiency inhibits the formation of the chirally organized macrodomains. This effect in turn decreases the thermal stability of thylakoid membranes (Velikova et al., 2011). We also observed the significant down-regulation of the cytochrome b₆f complex in NE lines (Velikova et al., 2014), which might inhibit the production of ATP in isoprene-suppressed plants. The increase of qE in NE lines might reflect the optimization of electron transport and ATP synthesis through the modulation of the cyclic electron transfer around PSI, the activation state of ATP synthase, and the partitioning of the proton-motive force between ΔpH and the membrane electrical potential (Horton et al., 2005).

CONCLUSION

The proposed biological functions of isoprene in plants have been associated with the ability of this molecule to affect thylakoid membrane organization and reduce the formation of reactive oxygen species, conferring tolerance to heat and oxidative stress. It has been hypothesized that isoprene improves the thermal stability of thylakoid membranes by affecting the membrane lipid composition (Velikova et al., 2011). Here, we provided direct evidence of the relationship between isoprene emission and the level of main lipid classes and their fatty acid composition, and we characterized the structural organization of the photosynthetic machinery in IE and NE poplar genotypes. The suppressed isoprene production in NE plastids was associated with the reduced amount of galactolipids and PLs, the lower level of the major fatty acid (18:3), and the altered chloroplast ultrastructure (Fig. 8). The suppression of isoprene biosynthesis causes considerable metabolic changes, particularly in lipid biosynthesis (Way et al., 2013; Kaling et al., 2015), and significant alterations in the chloroplast proteome (Velikova et al., 2014). The majority of the plastidic and mitochondrial proteome is encoded in the nuclear genome, and there is a continuous exchange of forward information from nucleus to organelle (anterograde) and of backward information from organelle to nucleus (retrograde; Pfannschmidt, 2010). According to the retrograde signaling concept, based on the available experimental data, signals originating in chloroplasts and/or mitochondria modulate nuclear gene expression (Leister, 2012). These signals originate from carotenoid biosynthesis, reactive oxygen species, photosynthetic redox processes, and changes in the pool of metabolites (Pfannschmidt, 2010; Leister, 2012). The plastidic signals identified so far have been associated with specific stress conditions. It is likely that the comprehensive changes in the metabolome (Way et al., 2013; Kaling et al., 2015), liposome, proteome (Velikova et al., 2014), and ultrastructure of the chloroplasts in NE poplar (Fig. 8), as well as the distinct physiological behavior of these plants, reflect finely tuned retrograde signaling. The precise mechanisms for the transmission of the changes in chloroplast to the nucleus in NE plant cells remain elusive.

MATERIALS AND METHODS

Plant Material

In this study, we used the same gray poplar (Populus × canescens; syn. Populus tremula × Populus alba) genotypes as utilized in previous chloroplast proteome research (Velikova et al., 2014), namely, two IE lines (WT and EV) and two NE lines (RA1 and RA2). The empty vector line was used to ensure that the differences between NE and IE plants reflected specific alterations in the isoprene synthase gene and not a more general genetic manipulation effect. The plants were grown in a greenhouse as described previously (Velikova et al., 2014). Briefly, the ambient temperature was 25°C/20°C with a relative humidity of 50%/60% and a photoperiod of 16 h of day/8 h of night. The plants were fertilized weekly with Triabon (Compo) and Osmocote (Scotts Miracle-Gro; 1:1 [v/v]; 10 g L⁻¹ soil).

Four-month-old plants were used for the experiments. Fully expanded leaves (ninth node from the apical meristem) from six to seven different plants, considered as biological replicates, were used for physiological, biochemical, and structural studies. The chloroplasts were isolated as described previously (Velikova et al., 2014) and used for lipid and MDA analyses.

Lipid Extraction Procedure

The total lipids from chloroplasts were extracted according to Bligh and Dyer (1959). All procedures were performed in dim light using chilled solvents (containing 0.01% [w/v] butylated hydroxytoluene) and glassware. The chloroplast samples (0.5 mL) were mixed with chloroform:methanol (1:2 [v/v]; 1.9 mL) for approximately 2 min; subsequently, 0.625 mL of chloroform and 0.625 mL of distilled water were added. The lower chloroform phase, containing the lipids, was removed, and aliquots were transferred into vials and exsiccated under N₂. The residues were weighed and calculated for total lipids.

Gas Chromatography-MS Analysis of PL Fatty Acid Composition

PL fatty acids were analyzed as described previously (Behnke et al., 2013; Way et al., 2013). Briefly, the PL fatty acids were separated from other lipids using a silica-bonded phase column (MEGA-BE-SI, 2 g 12 mL⁻¹, 20/PK, Bond ELUT; Agilent Technologies). Fatty acid methyl esters were obtained after mild alkaline hydrolysis. Myristic acid was used as an internal standard for gas chromatography analysis. Unsubstituted fatty acid methyl esters were measured using the 5973MSD gas chromatograph-mass spectrometer (Agilent Technologies) coupled with a combustion unit to an isotope ratio mass spectrometer (DeltaPlus; Thermo Electron) and identified using the established fatty acid libraries and characteristic retention times of pure standards. The fatty acids were named according to the total number of carbon atoms and double bonds. Saturated straight-chain fatty acids are indicated with an n.

Ultraperformance Liquid Chromatography Electrospray Time-of-Flight MS of Galactolipids

Lipids were dissolved in 1 mL of liquid chromatography-MS-grade methanol (Fluka). MGDG and DGDG contents were analyzed using the ultra-performance liquid chromatography electrospray time-of-flight MS system (maXis; Bruker). Aliquots of 2.5 µL of each sample were analyzed in three technical replicates in randomized order.

The chromatographic separation was achieved on a C₁₈ ACQUITYUPLC BEH column, 50 mm, 2.1 mm, and 1.7 µm (Waters), using a gradient elution. The composition was changed from 50% to 92% B for 10 min and maintained for an additional 10 min, then changed to 100% B for 1 min and maintained for 5 min. The flow was set to 0.4 mL h⁻¹. Mobile phase A comprised water:isopropyl alcohol (95:5, v/v), and mobile phase B comprised acetonitrile:isopropyl alcohol (95:5, v/v). A format of 0.001 mm sodium (Sigma-Aldrich) was added to both mobile phases. This method has been published previously for profiling photosynthetic glycerol lipids (Xu et al., 2010).

MGDGs and DGDGs were detected as sodium adducts through positive electrospray ionization. The instrument was calibrated with ESI Tune Mix (Agilent Technologies). Acquired spectra were internally calibrated and exported to GENEDATA software for chromatographic alignment and peak picking. MGDGs and DGDGs were identified based on the retention times and detected exact masses (mass error < 0.01 D).

MGDG and DGDG standards (Larodan) were used to evaluate the analytical performance and determine the quality control, which was injected 10 times in the beginning for column conditioning and after every 10th sample to validate the measuring performance.

MDA Content

The lipid peroxidation level in extracted chloroplast samples was quantified after measuring the MDA content using the thiobarbituric acid-reactive substances assay according to Hodges et al. (1999). The chloroplast sample (0.100 mL) was mixed with 1.2 mL of 80% (w/w) ethanol (containing 0.01% [w/v] butylated hydroxytoluene) and sonicated in a water bath sonicator for 3 min, followed by centrifugation at 5,000g for 10 min at 4°C. An aliquot of the obtained supernatant (0.5 mL) was mixed with the same volume of 0.65% (w/v) thiobarbituric acid solution containing 20% (w/v) TCA. Another aliquot of the supernatant (0.5 mL) was mixed with 0.5 mL of 20% (w/v) TCA, representing the zero control. The mixture was heated at 95°C for 30 min. The reaction was terminated after incubation in an ice bath. The cooled mixture was centrifuged at 10,000g for 10 min at 4°C, and the absorbance of the supernatant was measured at 532, 600, and 440 nm (Perkin Elmer). MDA equivalents were calculated according to Hodges et al. (1999):

(1)

(2)

(3)

Protein and Chlorophyll Analyses

For the calculation of the abundance of reaction center proteins in PSI and PSII, we used the chloroplast proteome data published by Velikova et al. (2014). Peak intensities of peptides identified as RCI (protein accession nos. POPTR_0008s15100.1, POPTR_0006s27030.1, POPTR_0003s14870.1, and POPTR_0002s25510.2) and RCII (protein accession nos. POPTR_0011s03390.1, POPTR_0004s03160.1, POPTR_0005s22780.1, POPTR_0002s05660.1, POPTR_0005s01430.1, POPTR_0005s27800.3, POPTR_0002s05720.1, POPTR_0002s25810.1, POPTR_0001s44210.1, and POPTR_0006s26270.1) were summed and expressed per mg of chlorophyll.

The chlorophyll content was measured in isolated chloroplast suspension after extraction with 80% (v/v) ice-cold acetone. Absorbance at 663 and 646 nm was detected to determine chlorophyll a and b concentrations, calculated according to Porra et al. (1989).

Chlorophyll Fluorescence Measurements

The chlorophyll fluorescence parameters were measured on intact leaves using a MINI-PAM Photosynthesis Yield Analyzer (Heinz-Walz). The leaves were dark adapted for 15 min prior to the determination of the minimal (F_o) and maximal (F_m) chlorophyll fluorescence, and subsequently, the leaves were exposed to actinic light (430 μmol m⁻² s⁻¹). After steady-state fluorescence was obtained, a saturating pulse was applied to determine the maximum fluorescence in the light (F_m′). The Φ_PSII was calculated from (F_m′ – F′)/F_m′ (Genty et al., 1989). The redox state of PSII was assessed based on the parameter qL = (F_q′/F_v′)/(F_o′/F′), where F′ is the fluorescence emission from the light-adapted leaf, F_v′ is variable fluorescence from the light-adapted leaf, and F_q′ is the difference in fluorescence between F_m′ and F′ (Baker, 2008). F_o′ was estimated using the following equation: F_o′ = F_o/[(F_v/F_m) + F_o/F_m′] (Oxborough and Baker, 1997). The NPQ was calculated as NPQ = (F_m − F_m′)/F_m′ (Bilger and Björkman, 1991). The NPQ relaxation kinetics in the dark was used to calculate qE. qE was assigned as a fast-relaxing component (within the first 2 min of dark relaxation after switching off the actinic light), calculated as qE = (F_m′′ – F_m′)/F′′_{m, 2min dark} (Zaks et al., 2013).

TEM

Leaf segments (1 mm²) were cut from the middle of the leaves for TEM analyses. The segments were fixed in 2.5% (v/v) glutaraldehyde (electron microscopy grade) in 0.1 m sodium cacodylate buffer, pH 7.4 (Science Services), postfixed in 2% (v/v) aqueous osmium tetraoxide (Dalton, 1955), dehydrated in an ethanol gradient (30%–70%), stained with uranyl acetate (2% in 70% ethanol), dehydrated in an ethanol gradient (70%–100%) and propylene oxide (100%), embedded in Epon (Merck), and cured for 24 h at 60°C. Semithin sections (300 nm) were cut and stained with Toluidine Blue. Ultrathin sections of 50 nm were collected onto 200-mesh copper grids before examination using TEM (Zeiss Libra 120 Plus; Carl Zeiss). The images were acquired using a Slow Scan CCD camera and iTEM software (Olympus Soft Imaging Solutions).

Statistical Analyses

Correlation analyses between different data sets of PL and galactolipid (MGDG and DGDG) contents, fatty acid compositions, chlorophyll fluorescence parameters, MDA content, the data groups, and IE or NE genotypes were performed using PCA and OPLS from the software package SIMCA-P (version 13.0.0.0; Umetrics). In addition, we included the chloroplastic protein contents associated with photosynthesis and structure (Velikova et al., 2014) to correlate lipids to proteins. Because the proteomic data originated from three samples for each plant genotype (containing six to seven different leaves each of the three samples), multivariate analyses were performed using only the data matching the same three samples used for both proteomic and PL analyses. Galactolipids, MDA, and chlorophyll fluorescence measurements were obtained from more and different extracts; therefore, only the data from three samples were taken randomly and used for these analyses. We added the means of all biological replicates to examine the correlations between genotypes using data originating from different leaf extracts. The resulting matrix size, therefore, was 78 × 16 (variables × observations). Thus, our analyses could correlate any data value with an isoprene emission trait (IE and NE plant genotype), but correlations between/within variables could be achieved only using data from the same leaf material (i.e. within PL and proteins, MGDG, DGDG, and MDA, and within chlorophyll fluorescence data).

The multivariate data analyses followed the established procedures to analyze MS data as described previously (Ghirardo et al., 2005, 2012; Kreuzwieser et al., 2014; Vanzo et al., 2014; Velikova et al., 2014). The isoprene emission trait was selected as the Y variable for the OPLS analysis by setting NE = 1 and IE = 0. The X variables were centered, and each type of data was block-wise scaled with 1 sd⁻¹, considering the different number of X variables in each group of data. Each calculated significant principal component was validated using full cross validation, with 95% confidence level on parameters. The regression model OPLS was further tested for significance using cross-validated ANOVA (Eriksson et al., 2008). Variables showing VIP values greater than 1 and jack-knifing method uncertainty bars smaller than the respective VIP values were defined as discriminant variables to distinguish IE from NE samples. For the proteomics data, containing a much higher number of variables, the VIP threshold was set to 0.5. The statistical significance of the differences between the means of discriminant variables and the functional and structural parameters measured in NE and IE plants was additionally evaluated using Student’s t test and at an α level of 0.05, unless otherwise stated.

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Lipid content and fatty acid composition in isolated chloroplasts of IE (WT and EV) and NE (RA1 and RA2) poplar.

• Supplemental Figure S2. Score and loading plots of all PCA parameters analyzed (lipid and fatty acid composition, MDA, NPQ, Φ_PSII, qE, qL), proteins related to photosynthesis, and proteins with structural activity.

Glossary

MGDG

monogalactosyldiacylglycerol

DGDG

digalactosyldiacylglycerol

LHCII

light-harvesting complex II

IE

isoprene-emitting

NE

nonisoprene-emitting

PL

phospholipid

MDA

malondialdehyde

TEM

transmission electron microscopy

MS

mass spectrometry

NPQ

nonphotochemical quenching

qE

energy-dependent quenching

Φ_PSII

photosystem II photochemical efficiency

PCA

principal component analysis

OPLS

orthogonal partial least squares

VIP

variable of importance for the projection