Protein assembly and heat stability in developing thylakoid membranes during greening

Abstract

The development of the thylakoid membrane was studied during illumination of dark-grown barley seedlings by using biochemical methods, and Fourier transform infrared and spin label electron paramagnetic resonance spectroscopic techniques. Correlated, gross changes in the secondary structure of membrane proteins, conformation, composition, and dynamics of lipid acyl chains, SDS/PAGE pattern, and thermally induced structural alterations show that greening is accompanied with the reorganization of membrane protein assemblies and the protein–lipid interface. Changes in overall membrane fluidity and noncovalent protein–lipid interactions are not monotonic, despite the monotonic accumulation of chlorophyll, LHCII [light-harvesting chlorophyll a/b-binding (polypeptides) associated with photosystem II] apoproteins, and 18:3 fatty acids that follow a similar time course with highest rates between 12–24 h of greening. The 18:3 fatty acid content increases 2.8-fold during greening. This appears to both compensate for lipid immobilization by membrane proteins and facilitate packing of larger protein assemblies. The increase in the amount of protein-solvating immobile lipids, which reaches a maximum at 12 h, is caused by 40% decrease in the membranous mean diameter of protein assemblies at constant protein/lipid mass ratio. Alterations in the SDS/PAGE pattern are most significant between 6–24 h. The size of membrane protein assemblies increases ≈4.5-fold over the 12–48-h period, likely caused by the 2-fold gain in LHCII apoproteins. The thermal stability of thylakoid membrane proteins increases monotonically, as detected by an increasing temperature of partial protein unfolding during greening. Our data suggest that a structural coupling between major protein and lipid components develops during greening. This protein–lipid interaction is required for the development and protection of thylakoid membrane protein assemblies.

If a leaf is grown in darkness, proplastids develop into etioplasts with three-dimensional semicrystalline arrays of internal membranes forming the prolamellar bodies, which contain the chlorophyll (Chl) precursor protochlorophyll(ide) and a remarkably high amount of monogalactosyldiacylglyceride (MGDG) (1–3). When exposed to light, protochlorophyll(ide)s are rapidly converted in a matter of minutes to Chl, which is paralleled with the disorganization of the prolamellar bodies and formation of planar thylakoid membranes (primary lamellae; refs. 4–6). Because the subsequent synthesis, assembly, and modification of pigment, lipid, and protein components may last for days until mature thylakoids appear (2), this process (the visible greening) is a very useful model to study the light-controlled development of the thylakoid membrane (2). Of the pigment-proteins, most abundant in mature thylakoids is LHCII [light-harvesting Chl a/b-binding (polypeptides) associated with photosystem II; ref. 7]. MGDG, with >90% of 18:3/18:3 fatty acids (8, 9), accounts for >50% of the total lipid content in this membrane (1, 9, 10). LHCII has a high specificity toward MGDG and certain other lipids that stabilize it toward thermal denaturation (11, 12), demonstrating the significance of strong, reciprocal protein–lipid interactions in native thylakoids (13).

The techniques used here complement each other because electron paramagnetic resonance (EPR) detects only the spin label in these experiments and has a time window that is optimal for lipid chain dynamics, whereas Fourier transform infrared (FTIR) detects all lipid and protein vibrations simultaneously on a fast time scale. FTIR spectroscopy is widely used to study protein secondary structure (14, 15) and lipid chain vibrations (16–18). Motional coupling between large, relatively immobile membrane proteins and mobile fluid lipids results in motionally restricted protein-solvating first-shell lipids (19). This appears as “immobile” component in EPR spectra of spin-labeled lipid analogues as demonstrated in numerous membrane systems (20), including thylakoids (21). By quantitating these lipids, we have estimated changes in the size of protein assemblies of major thylakoid membrane protein complexes. The partitioning of 2,2,6,6,-tetramethylpiperidine-1-oxyl (TEMPO) between membranous and aqueous environments depends on lipid packing and fluidity in the lipid bilayer and can be measured with EPR (22). A drop in the membranous TEMPO signal (in EPR) and a simultaneous increase in the CH₂ vibration frequency (in FTIR) diagnosed a lipid transition that was interpreted as the formation of a nonbilayer phase. The temperature of this transition was found to vary during greening. The present study provides data on the time course of the thermal behavior of, and on the protein–lipid rearrangements in the thylakoid membrane during development.

Materials and Methods

Materials.

Barley (Hordeum vulgare) L. cv. Triangel plants were germinated and grown in vermiculate in complete darkness at 25°C. After 6 days of growth seedlings were illuminated with white light at a photon flux of 100 μmol⋅m⁻²⋅s⁻¹. Samples were taken at 0.5, 6, 12, 24, and 48 h of illumination. Plastids were isolated in dim light at 4°C in a medium containing 0.4 M sorbitol, 10 mM NaCl, 5 mM MgCl₂, 1 mM MnCl₂, 1 mM phenylmethylsulfonyl fluoride (PMSF), and 50 mM Mes (pH 6.5). Plastids were disrupted by osmotic shock in a medium containing 40 mM sucrose, 10 mM NaCl, 5 mM MgCl₂, 1 mM PMSF, and 20 mM Tricine-KOH buffer (pH 7.8). The spin labels, D₂O (of 99.9% purity), and molecular weight standards were purchased from Sigma.

Fatty Acid Composition and SDS/PAGE.

Chl content was determined from 80% acetone extract of 0.2–0.4 g of leaf segments according to ref. 23. Extraction and analysis of lipids was carried out as described (24). Fatty acid composition was determined by gas chromatography after extracting total lipids (25). Briefly, lipids were transmethylated in the presence of absolute methanol containing 5% HCl at 85°C for 2 h and the fatty acid methyl esters were separated on an HP 5890 (Hewlett–Packard) gas chromatograph. Peaks were identified using authentic methyl ester standards (from Sigma). The total protein content was determined according to ref. 26. SDS polyacrylamide gels were prepared as described (27). SDS/PAGE densitograms were integrated along the lanes over the selected regions after background subtraction using nih image (National Institutes of Health, Bethesda) and igor (WaveMetrics, Lake Oswego, OR).

Spin Label EPR Spectroscopy.

Five microliters of 5-SASL [(4′,4′-dimethyloxazolidine-N-oxyl)stearic acid spin label] solution in ethanol (1 mg/ml) or 10 μl of TEMPO solution in water (10 mM) were added to 100-μl thylakoid suspension (corresponding to 1.35 ± 0.15 mg protein). This was incubated for 10 min in the dark at room temperature, repeatedly vortexed, filled into a 1 mm (i.d.) glass capillary, and concentrated in a bench top centrifuge (≈4,000 × g, 10 min). EPR spectra were recorded at a subsaturating microwave power by using an X-band ECS 106 spectrometer (Bruker, Rheinstetten, Germany) equipped with a thermostat and a standard TE102 cavity using first harmonic detection at 100 kHz field modulation with an amplitude of 0.1 mT and 0.03 mT for 5-SASL and TEMPO, respectively. Scan range was 12 mT for 5-SASL and 8 mT for TEMPO spectra. Data analysis was performed using winepr (Bruker) and igor. TEMPO spectra were simulated using software based on ref. 28 and written by us.

FTIR Spectroscopy.

Fifty-microliter aliquots of thylakoid membrane suspension (corresponding to 600–750 μg protein) were pelleted (≈12,500 × g, 10 min) and resuspended in 500 μl of buffer, which was the same as above but D₂O-based, and centrifuged again. The 0.4 unit difference between pH and pD was taken into account. The pellet was placed between CaF₂ windows by using a Teflon spacer of 20 μm. FTIR spectra were recorded on a Philips PU9800 FTIR spectrometer equipped with a homemade thermostated sample holder. The temperature was measured on the surface of the window. Data processing was performed using spserv spectrum handling software (Cs. Bagyinka, Institute of Biophysics, Biological Research Centre, Szeged, Hungary) and igor. Six and seven Lorentzians were least-squares fitted to the ≈2,800–3,000 cm⁻¹ and ≈1,760–1,530 cm⁻¹ regions, respectively (18, 29, 30).

Results

Composition of Thylakoid Membranes During Greening.

The first samples taken after 30 min of illumination already represent primary thylakoid membranes (2, 4). Chl accumulation defines the course of greening and serves as a reference when comparing different plants and experiments (27). It displays a sigmoidal dependence on the duration of illumination (Fig. 1A). Whereas most fatty acids change little (not shown), the concentration of the three most abundant ones 16:0, 18:0, and 18:3 change grossly and their summed fraction remains between 70 and 78% during greening (Table 1). The fraction of the saturated chains 16:0 and 18:0 decrease and that of the polyunsaturated chain 18:3 increases during greening. The SDS/PAGE pattern of whole thylakoids, shown in Fig. 1B as a function of greening, agrees with earlier reports (31). Two families of early light-inducible proteins (ELIPs) identified at 13–14 and 17–19 kDa by (32) are related to etioplast-to-chloroplast transformation (32, 33). The major LHCII apoproteins are located between 25 and 28 kDa (34, 35), and the fractional weight of the corresponding band rises ≈4.3-fold during greening (Table 1). The assayed total protein/lipid ratio was found to be 5 mg/mg during greening and 6 mg/mg for the mature green plants.

Fig 1.

Chl (a + b) content (A) and SDS/PAGE (B) of thylakoid membranes of dark-grown barley seedlings after different periods of illumination. The solid line in A is a best-fitting sigmoid function. The densitometric data for the 16–21- and 21–29-kDa SDS/PAGE regions is presented in Table 1.

Table 1.

Quantities related to the lipid and protein composition of thylakoid membranes during greening

Greening, h	Fatty acid, %			Fraction of molecular weight (kDa) bands, %
	16:0	18:0	18:3	16–21	21–29
0.5	43	15	15	29	9
6	42	9	19	28	12
12	37	6	31	25	19
24	31	5	40	21	30
48	29	7	42	17	38

^*Fractional weight of given fatty acids (error is ±2% of individual values).

^† Values are given to significant digits.

^‡ Fractional intensities of given molecular mass regions from density analysis of the SDS/PAGE (Fig. 1).

Spin Label EPR Spectroscopy of Acyl Chain Mobility and Lipid Packing.

EPR spectra of 5-SASL and TEMPO in thylakoid membrane dispersions are shown in Fig. 2. 5-SASL is a stearic acid analogue with the nitroxyl located on C-5; hence, it monitors acyl rotational dynamics close to polar–apolar membrane interface (22). The spectra of 5-SASL (Fig. 2A) indicate limited rotational mobility (22) and gradual collapse of the hyperfine anisotropy with increasing temperature both in the primary lamellae and fully developed state. The degree of motional averaging of the hyperfine anisotropy can be quantitated with the dynamic orientational order parameter (S) from the outer and inner splittings A_II and A_⊥ defined in Fig. 2A. The method described in detail in ref. 36 was used. S decreases without transition-like changes upon heating at any given periods of greening (not shown). Values of S at 25°C are 0.75, 0.74, 0.77, 0.76, and 0.74 (± 0.005) at 0.5, 6, 12, 24, and 48 h of greening, respectively.

Fig 2.

EPR spectra of 5-SASL and TEMPO (A and B, respectively) incorporated into primary lamellae (dotted lines) and mature (solid lines) thylakoid membranes of barley. Spectra are scaled to the same amplitude. Hyperfine splittings used to calculate order parameters (S) are indicated on the 46°C 5-SASL spectrum. The dashed lines, shifted above the 34°C spectra, are perfectly fitting line shapes simulated as described in the text. The membranous and aqueous TEMPO components are indicated on the high-field peak as M and W, respectively, on the spectrum recorded at 10°C (B).

The solubility of TEMPO in membranes is lower than that of 5-SASL and it depends on the lipid packing (22). Its partitioning between the aqueous phase and the membrane results in two spectral components; these are indicated as “W” and “M,” respectively, in Fig. 2B. Best fitting simulated line shapes (dashed line above the 34°C spectrum in Fig. 2B) were calculated to determine the membranous TEMPO fraction (f_m,TEMPO). Our program simulates and combines the two components and results in excellent fits between experimental and simulated spectra. f_m,TEMPO is presented in Fig. 3 both as a function of temperature for primary lamellae (0.5 h) and mature (48 h) thylakoids (Fig. 3A) and as a function of greening measured at 25°C (Fig. 3B). It increases monotonically with increasing temperature until 45°C, but drops at ≈45–48°C. This was true for all periods of greening. A dramatic change in f_m,TEMPO can also be seen at different periods of greening when looked at fixed temperatures (Fig. 3B) because f_m,TEMPO decreases to ≈50% during the first 12 h then rises almost 2-fold for the green plant, relative to the primary lamellae state.

Fig 3.

The fraction of TEMPO spin probe partitioned into primary lamellae (filled circles) and mature (open circles) thylakoid membranes (f_m,TEMPO) of barley as a function of temperature (A) and the duration of greening at 25°C (B). f_m,TEMPO was determined by spectral simulations as shown in Fig. 2B. Error bars in B were due to errors among different preparations (fitting errors were negligible).

Rearrangements of the Protein–Lipid Interface.

The low- and high-field hyperfine lines of the 5-SASL spectra recorded in a temperature range of 22–40°C are too broad to originate from single component spectra (Fig. 2A). In addition, A_II varies nonmonotonically during greening when comparing spectra recorded at the same temperature (not shown). This behavior can only originate from a linear combination of a minimum of two (mobile and immobile) components with different A_II and A_⊥ values (see, e.g., ref. 37). Two-component EPR spectra in thylakoid membranes were reported earlier for phospholipid and stearic acid C-14 nitroxyl labels over a wide temperature range. The immobile component was assigned to protein-solvating interfacial lipids (38, 39). Because the outer hyperfine splitting of the components differed by more than 5 G, we could determine the immobile fraction (f_i,SASL) by applying an intersubtraction technique (21, 40) on the 34°C spectra, aided with nonlinear least-squares optimization. The intersubtraction leads to excellent fits between the linear combination of the two components and the experimental spectra (Fig. 2A) justifying the use of two components. Values of f_i,SASL are 58%, 58%, 74%, 59%, and 38% (± 2%) for 0.5, 6, 12, 24, and 48 h of greening, respectively.

The big, nonmonotonic change in f_i,SASL during greening, at relatively constant lipid/protein ratio (N_t) and lipid headgroup composition (2), can only be explained if changes in the intramembranous size of protein assemblies are considered. Assuming, for simplicity, membrane proteins of the same size, the number of interfacial lipids per monomer protein (N_b) is related to the number of monomers in the protein assembly (oligomer), because of the exclusion of lipids from protein–protein contact regions. A simple model (9, 21) can be used to estimate the relative mean oligomer dimensions of these proteins from f_i,SASL (20, 41). Assuming circular intramembranous intersection for both lipids and proteins of an nxn oligomer, N_b is reduced approximately by 1/n relative to that of a single monomer (for n > 1), neglecting lipid-trapping sites unavailable for bulk lipids and the label. The change in the mean diameter of oligomers relative to that of the monomer (i.e., n) can be estimated from the change in N_b, because

where K_r is the association constant of SASL relative to the host lipid (20). N_t is obtained from the assayed protein/lipid ratio. A mean molecular mass of MW_p = 623 kDa and MW_l = 0.9 kDa for monomeric protein (assemblies) and lipid, respectively, can be taken according to refs. 9 and 38. The association constant of SASL relative to MGDG in thylakoid membranes can be estimated from that of SASL and MGDG observed relative to phosphatidylglycerol (21, 39) to be K_r = ≈6.5. With these estimates the values of n are 1.0, 1.1, 0.6, 1, and 2.7 for 0.5, 6, 12, 24, and 48 h of greening, respectively. One also obtains with the above formula that changes in n are rather insensitive to K_r, MW_p, and MW_l. The mean oligomer number (i.e., nxn) decreases ≈2.8-fold over the 0.5–12-h period and increases ≈20-fold over the 12–48-h period paralleled with a loss in early light-inducible proteins (ELIPs) and a large gain in LHCII, respectively (Fig. 1, Table 1). LHCII is known to be in the monomeric form in the early phases of greening (35).

FTIR Spectroscopy of Thylakoid Membranes.

Fig. 4 illustrates band-fitting analysis of FTIR spectra of green plants at 6°C (Fig. 4A) and 64°C (Fig. 4B) for the amide I and II vibration regions. Upon heating, the band at 1,632 cm⁻¹ (band I/B) characteristic of β-sheets shifts to 1,620 cm⁻¹, a region assigned to intermolecular β-structures (42). The band at 1,657 cm⁻¹ (band I/A), characteristic of α-helices, shifts to 1,651 cm⁻¹ and gains significant intensity upon heat treatment, indicating formation of disordered polypeptide segments (14). An increase in intensity at 1,685 cm⁻¹ and 1,621 cm⁻¹ can be related to the appearance of aggregated structures (15). These are in agreement with the enhanced exchange of amide NH groups to ND in the deuterated buffer reported by the intensity loss in the single 1,550 cm⁻¹ band. The band at ≈2,850–2,854 cm⁻¹ (not shown) of symmetric CH₂ vibrations (16, 43) measures the conformational disorder of lipid acyl chains (30). The middle frequencies of the bands ν_symCH₂ and the amide I/A vibrations (≈1,648–1,658 cm⁻¹) are plotted in Fig. 5 together with the intensity of the amide II (≈1,550 cm⁻¹) band, as a function of temperature for thylakoids at 12 and 48 h of greening. Protein vibrations and amide II intensity change abruptly at ≈35–45°C, followed by an upward shift in ν_symCH₂ at ≈48–52°C. These values vary during greening and indicate unfolding of proteins and increasing conformational disorder of acyl chains, respectively. Protein unfolding is only partial because thermal denaturation happens at much higher temperatures (44).

Fig 4.

Band fitting analysis of FTIR spectra of thylakoid membranes of green barley seedlings in the amide I and II region. The spectra were recorded at 6°C (A) and 64°C (B). Experimental spectra are shown after baseline subtraction together with fitted Lorentzian component bands, their best fitting superposition (dotted lines), and residuals (above the spectra). The two major amide I components and the amide II band are indicated with dashed lines.

Fig 5.

Temperature dependence of characteristic protein and lipid FTIR component bands in barley thylakoid membranes after 12 and 48 h of greening (full and open symbols, respectively). The changes are scaled to aid comparison. In the case of amide I/A (squares) and ν_symCH₂ (circles) vibration bands, y axis means the wave number of maximum absorbance, whereas in the case of amide II vibration band (triangles), it means the integrated intensity. The units 0 and 1 correspond to the minimum and maximum values (given in the text).

The characteristic temperatures of the middle frequency for the two major amide I bands (obtained at 25°C) and ν_symCH₂ (obtained at 34°C) are reported in Table 2 as a function of greening. The characteristic temperatures of the amide I/A and I/B components attributed to partial protein unfolding shift monotonically from 35 and 40°C to 44 and 43°C, respectively, with the progression of greening. For green plant, these values agree with those obtained in pea thylakoids (29). When compared at 25°C, the middle frequencies of these bands shift also monotonically to higher wave numbers (Table 2), indicating a gain in α-helix and β-strand structures. The characteristic temperature of the transition in ν_symCH₂ shifts nonmonotonically from 52°C to 48°C, from the primary lamellae to the fully developed state.

Table 2.

Temperature values of steepest changes of FTIR band shifts recorded as function of temperature and middle frequencies of selected vibration bands at different periods of greening

Greening, h	Amide I
	T, °C		frequency, cm−1		νsymCH2
	A	B	A	B	T, °C	frequency, cm−1
0.5	35	40	1,654	1,624	52	2,850.2
6	35	40	1,656	1,626	53	2,849.6
12	39	41	1,656	1,629	55	2,849.8
24	41	41	1,656	1,630	52	2,850.3
48	44	43	1,658	1,632	48	2,850.2

^*Temperature of steepest changes on the parameter vs. temperature plots (cf. Fig. 5).

^† Values are given to significant digits.

^‡ Middle frequencies of amide (at 25°C) and ν_symCH₂ (at 34°C) bands.

Discussion

The development of the photosynthetic membrane was studied in whole thylakoid preparations because no significant differences in spin label EPR spectra of stacked and destacked thylakoids were observed earlier (21). It is demanding to consider the effect of protein–lipid interactions on the spectral parameters. Interfacial lipids possess higher dynamic order than those in the bulk phase (20); hence, the overall membrane fluidity depends on the size of the protein–lipid interface. The acyl conformational disorder depends on the molecular details of protein–lipid interactions (18, 30). TEMPO certainly prefers the more mobile bulk phase to the protein-solvating lipid shell. The fractional increase of interfacial “rigidifying” lipids (20) and the accumulation of the 18:3 “fluidizing” lipids (see, e.g., ref. 45) has an opposite effect on overall membrane fluidity (18).

Keeping these relations in mind, our data on thylakoid development can be consistently interpreted as a sequence of correlated molecular events grouped into three phases: onset, rearrangement, and maturation. Fig. 6 schematically represents changes in the size of protein assemblies, the amount protein-solvating lipids, fatty acid composition, and protein secondary structures in these phases. Greening starts with the onset phase (<6 h), where the Chl content increases at a very low rate, LHCII apoproteins are almost absent but ELIPs have their maximum density in the SDS/PAGE, and accumulation of polyunsaturated lipids sets on. Concerning protein–lipid interactions, most striking is that greening is not monotonic showing significant changes during the rearrangement phase (6–24 h), where ≈67% of the total Chl content is gained. All relevant parameters show that the thylakoid membrane is least fluid in the rearrangement phase. Taking into account that the lipid composition of the prolamellar body and thylakoid are very similar (1, 2), this is caused initially by a conversion/disassembly of protein assemblies to smaller units that are subsequently replaced by larger assemblies of newly synthesized proteins. These result in larger and smaller protein–lipid interface, at 12 and 48 h of illumination, respectively, relative to the primary lamellae state. The 2.8-fold increase of the 18:3 fatty acid content during greening, ≈2-fold in the rearrangement phase, is in part likely to compensate for this immobilization. This is certainly required to maintain mobility of membrane components, especially in this phase. The protein/lipid mass ratio increases only during the maturation phase (≈24–48 h) in which LHCII apoproteins dominate the SDS/PAGE, in agreement with ref. 35. The growth of certain SDS/PAGE bands, most notably LHCII, which binds most of Chl in mature thylakoids (9), and the 18:3 fatty acid content follow a time course similar to that of the Chl accumulation. This suggests that the activity of related fatty acid desaturases (10, 46, 47), Chl synthesis, and expression of certain proteins are synchronized. The relative amount of α-helix and β-sheet structural forms increase monotonically, on the cost of less regular structures, during greening as new thylakoid proteins are synthesized/assembled.

Fig 6.

Schematic representation of the onset (A), rearrangement (B), and maturation (C) phases of the development of the thylakoid membrane during greening based on our data on protein–lipid interaction, fatty acid composition, and protein thermal stability. The larger objects are protein assemblies. Lipids are shown with small circles. Protein-solvating “immobile” lipids are shown as black circles. The fraction of these lipids, the protein/lipid ratio, and the size of membranous protein assemblies reflect relative changes observed. The different patterns for the protein assemblies refer to changes in their composition, increasing amount of α-helical and β-sheet forms and improving thermal stability during greening. Those for the lipids indicate monotonic increase in polyunsaturated (18:3) fatty acid content.

The lack of phase transition-like changes by 5-SASL (Fig. 2A) and f_m,TEMPO (Fig. 3A) was expected in a membrane whose major lipid [MGDG with 18:3/18:3 acyl chains (8, 9)] adapts an inverted hexagonal (H_II) phase under normal conditions (48–50). However, because f_m,TEMPO and ν_symCH₂ are to increase upon increasing fluidity (16, 22), the upward shift in ν_symCH₂ (Fig. 5) and the simultaneous drop in f_m,TEMPO (Fig. 3A) upon heating demands explanation. ν_symCH₂ is rather high (≈2854 cm⁻¹) in the nonbilayer (H_II) phase of MGDG (16). Considering the similar values obtained after the sharp increase in ν_symCH₂ above 48°C (Fig. 5), the transition can only be interpreted as the formation of the H_II phase by MGDG, in agreement with refs. 29, 51, and 52. It also explains the sharp decrease in f_m,TEMPO (Fig. 3A), assuming that TEMPO partitions badly into the H_II phase of MGDG because of the tightly packed headgroup region in this structure (16). The temperature of this transition is known to decrease with increasing double bond content (52, 53), but protein–lipid interactions may have an opposite effect by stabilizing MGDG in the bilayer (54, 55). The characteristic temperature of this transition is highest (55°C) in the rearrangement phase in agreement with the protein-stabilizing structural role of MGDG (13, 54–56) and the largest protein–lipid interface. Exclusion of MGDG with its “fluidizing” chains leaves the bilayer in a more rigid state, reducing the damaging effect of elevated temperature. This happens 16°C higher than partial protein unfolding at the beginning of the rearrangement phase. This difference decreases to 4–5°C for the green plant because of monotonically improving heat stability of newly synthesized/assembled thylakoid proteins, a further increase in the 18:3 acyl species, and probably stronger protein–lipid structural coupling. The exclusion of MGDG (and formation of nonbilayer phase) in mature thylakoids seems therefore to be initiated by partial thermal unfolding of proteins for their own protection.

In conclusion, the protein–lipid interaction and the fluidity of the thylakoid membrane is determined to a large extent by nonmonotonic changes in protein assemblies and monotonic progression of fatty acid desaturation. The latter is important for the control of overall membrane fluidity and the nonbilayer propensity of MGDG. It appears that a structurally balanced interplay between LHCII and MGDG develops gradually into a fully functional mechanism in mature thylakoid. This allows rearrangement of protein–protein and protein–lipid interactions—hence, adaptation to external conditions during development—and serves both structural flexibility and stability. Similar studies on photosystem I- and II-enriched preparations are to be combined with direct structural data (e.g., refs. 56 and 57) for constructing more detailed models of the lateral (re)organization of protein assemblies during chloroplast development.

Abbreviations

• Chl, chlorophyll

• EPR, electron paramagnetic resonance

• FTIR, Fourier transform infrared

• LHCII, light-harvesting Chl a/b-binding (polypeptides) associated with photosystem II

• MGDG, monogalactosyldiacylglyceride

• SASL, (4′,4′-dimethyloxazolidine-N-oxyl)stearic acid spin label

• TEMPO, 2,2,6,6,-tetramethylpiperidine-1-oxyl