The role of membrane physiology in sHSP Lo18-lipid interaction and lipochaperone activity

Abstract

To cope with environmental stresses, organisms, including lactic acid bacteria such as O. oeni, produce stress proteins called HSPs. In wine, O. oeni is constantly confronted by stress affecting its membrane fluidity. To survive through in these deleterious conditions, O. oeni synthesizes Lo18, a unique, small HSP which acts as a molecular chaperone and a lipochaperone. The molecular mechanism underlying its lipochaperone activity, particularly regarding membrane lipid composition, remains poorly understood. In this context, Lo18 lipochaperone activity and the associated modification in protein structure were studied during interaction with different liposomes from O. oeni cultures representing unstressed, stressed and stressed-adapted physiological states. The results showed that the presence of the membrane (whatever its nature) induces a modification of Lo18’s structure. Also, the presence of oleic acid and/or phosphatidylglycerol is important to favor Lo18-membrane interaction, allowing lipochaperone activity. This research enhances understanding of sHSP-membrane interactions in bacterial systems.

Introduction

Acidity and alcohol concentrations present in wine are the main causes of membrane damage found in wine-grown cells¹. Changes in the charge environment induced by low pH or -OH residues, which have a strong affinity for the polar heads of phospholipids, disrupt the organization and structure of the membrane, which has an impact on cell permeability and membrane fluidity^2,3. However, maintaining membrane fluidity is essential for the survival of micro-organisms. Any change in membrane fluidity disrupts the flux of nutrients/wastes across the membrane and dissipates the proton motive force^4,5, which can slow down cell metabolism, leading to cell growth arrest^6,7. Most of the key micro-organisms involved in the winemaking process have developed various strategies to maintain their membrane integrity.

For Oenococcus oeni, the main lactic acid bacteria involved in malolactic fermentation, one of these strategies is the synthesis of stress proteins, such as its unique small heat shock protein (sHSP) known as Lo18. sHSPs are small proteins of 12 to 42 kDa, present in almost all organisms. In oligomeric state, they are mainly known for their ability to prevent the aggregation of cellular proteins owing to their chaperone activity^8–11. However, for some sHSPs a second activity, known as molecular lipochaperone, has been described. This second activity consists in helping cells to maintain optimal membrane fluidity. Among sHSPs demonstrating this activity are: HSPB1 and HSPB5 from humans¹², HSPA from Synechococcus sp.^10,13, HSP17 from Synechocystis sp.^14,15, HSP17.8 from Arabidopsis thaliana¹⁶, HSP15.8 and HSP16 from Schizosaccharomyces pombe¹⁷, HSP18.55 from Lactiplantibacillus plantarum¹⁸ and Lo18^11,19–24. According to the literature, when oligomers dissociate into dimers close to the membrane, electrostatic and hydrophobic forces deform the protein, allowing it to partially penetrate into the membrane and maintain optimal fluidity²⁵. Regarding Lo18, previous studies have shown: (i) that the structure and oligomeric plasticity of the protein have an impact on lipochaperone activity^23,24, and (ii) that Lo18 preferentially interacts with membrane lipids not yet adapted to stress²³. However, the link between Lo18 lipochaperone activity, and more generally sHSPs, and the membrane lipid profile remain poorly described and understood.

The main aim of this study was to elucidate the links between the lipochaperone activity of Lo18 and the lipid composition and physical state of the membranes with which it interacts. First, the affinity of Lo18 for certain phospholipids constituting the membranes of O. oeni was investigated by immunolabelling, revealing a significant preference for lipids composed of phosphatidylglycerol and/or oleic acid. At the same time, the lipochaperone activity and structural changes in Lo18 were assessed during interaction with four types of liposomes representing the physiological states of unstressed, stressed and stress-adapted bacterial membranes. To do so, O. oeni bacterial membranes from (i) exponential and stationary growth phase culture, and (ii) 8% ethanol adapted culture, were collected to create liposomes. The results showed that membrane interaction induced a change in the secondary structure of Lo18, characterized by an increase in α-helices at the expense of β-sheets for the liposomes tested. Additionally, results indicated that the lipid composition of liposomes may influence the insertion of Lo18 into the membrane and lipochaperone activity.

Results

Favored lipid Lo18 substrates

Over the past decade, a growing number of studies have focused on the lipochaperone activity of Lo18^{11,20,23,24,26}. However, to date, the lipid domains or lipids specifically involved in this interaction have not been identified. The affinity between Lo18 and several common phospholipids was analyzed by means of immunolabeling. The phospholipids were chosen for their presence in O. oeni membranes and for their physical properties. Regarding the physical properties, phospholipids were composed of: (i) either unsaturated (dioleoyl, DO-) or saturated (dipalmitoyl, DP-) fatty acids, and (ii) neutral (phosphatidylethanolamine, -PE and phosphatidylcholine, -PC) or negative-charge head groups (phosphatidic acid, -PA and phosphatidylglycerol, -PG). The intensity of the signal obtained by chemiluminescence (ECL) after antibody hybridization with Lo18 was normalized with the quantity of phospholipids fixed on the PolyVinyliDene Fluoride (PVDF) membrane (obtained after staining with Nile Red) (Supplementary Data, Fig. S1). After this normalization step, Lo18 showed increased affinity for oleic acid (DOPG, DOPC and DOPE) and/or with the phosphatidylglycerol head group (DPPG and DOPG) (Fig. 1), visualized by a more intense ECL/Nile Red signal ratio. The unsaturation present in the carbon chains of oleic acid naturally induces a certain disorder in the membrane, which is often observed in membranes not yet adapted to fluidifying stresses²⁷. As described for HSP16 from S. pombe¹⁷ and human HSPB1^12,28, this suggests that Lo18 may preferentially interact with lipids on the membrane through both hydrophobic and electrostatic forces, induced by unsaturated and negatively charged fatty acids, respectively. These fatty acids are mainly found in membranes not yet adapted to fluidizing stress.

Figure 1.

Lipid affinity test of Lo18. Immunolabeling measurements were performed on Lo18 and several common phospholipids: unsaturated (dioleoyl, DO-) or saturated (dipalmitoyl, DP-) lipids with neutral (phosphatidylethanolamine, -PE and phosphatidylcholine, -PC) or negative (phosphatidic acid, -PA and phosphatidylglycerol, -PG) charged head groups. The intensity of the signal obtained after hybridization with Lo18 was normalized with the number of phospholipids fixed to the PVDF membrane. The data represent the means and SE of three independent experiments and were analyzed statistically by the Kruskal–Wallis nonparametric test.

Lipid membrane composition impacts Lo18 lipochaperone activity

Given the affinity of Lo18 for certain phospholipids, it therefore seemed reasonable to examine whether the lipid composition and physiological state of membranes could have an impact on the lipochaperone activity of Lo18. To investigate this point, we used liposomes composed of purified lipids from cultures of O. oeni at different growth phases (exponential, stationary, and grown in the presence of 8% ethanol) representing the physiological states of unstressed, stressed and stress-adapted bacterial membranes, respectively. Liposomes composed of purified lipids from exponential L. plantarum cultures were also used as controls to test the specificity of Lo18 for lipids from its producer organism. The lipid composition of these four types of liposomes was first determined by chromatography (Fig. 2, Table S1). The results showed that digalactosyldiacylglycerols (DGDG) were the main head group found in O. oeni regardless of the growth phase, representing 38%, 64%, and 57% of the total head groups in liposomes from cultures in the exponential and stationary phases, or in the presence of 8% ethanol, respectively. For liposomes derived from the exponential phase culture of O. oeni and L. plantarum, phosphatidylglycerol (PG) represented a predominant head group, accounting for 39% of total lipids for O. oeni and 67% for L. plantarum (Fig. 1A, Table S1). Lysyl-PG, cardiolipins (CL) and phosphatidylcholins (PC) were also present in O. oeni and L. plantarum membranes in very low quantities, which could be considered as negligible. Regarding the fatty acid chains, they ranged from myristic acid, C14:0, to lactobacillic and dihydrosterulic acid (respectively cycC19:0 n-7 and cycC19:0 n-9), abbreviated C19:1 (Fig. 2B, Table S1). Palmitic acid, C16:0 (saturated), and/or oleic acid, C18:1 (unsaturated), were the most abundant fatty acids in O. oeni and L. plantarum cells in the exponential and stationary phases, representing 37% (C16:0) and 26% (C18:1) in the exponential phase, 29% (C16:0) and 25% (C18:1) in the stationary phase in O. oeni cell culture. Conversely, proportionally, C18:1 was not the main fatty acid in L. plantarum, where it represented only 20% of total lipids. During adaptation to alcohol the proportion of unsaturated fatty acids was also significantly reduced, representing 12% of lipids compared with 28% and 29% in the exponential and stationary phases in O. oeni. This was mainly due to the decrease in the proportion of C18:1, which reached only 11% in the presence of ethanol adapted membrane. Thus, by examining the nature of the polar heads and the fatty acid chains, four distinct profiles based on the prevalence of phosphatidylglycerol and oleic acid could be distinguished. Liposomes derived from O. oeni cultures in the exponential growth-phase were rich in PG and oleic acid (PG + , DO +), while those from the stationary phase were low in PG but rich in oleic acid (PG-, DO +). Liposomes derived from alcohol-adapted cultures were low in PG and oleic acid (PG-, DO-), while those derived from a L. plantarum culture were high in PG but low in oleic acid (PG + , DO-). It should be recalled that phosphatidylglycerol and oleic acid both have a strong affinity with Lo18.

Figure 2.

Lipid composition of the liposomes used. Lipid profiles of (i) O. oeni liposomes in exponential, stationary, or 36 h ethanol-adapted culture, or (ii) L. plantarum liposomes. Proportion of polar head group type (A) and fatty acids (B) of each liposome. The data represent the means and SE of three independent experiments and were analyzed statistically by Kruskal–Wallis nonparametric test.

The variation of the membrane fluidity of these four liposomes was then measured by steady-state fluorescence anisotropy of a 1,6-diphenyl-1,3,5-hexatriene probe (DPH) in the absence or presence of Lo18 during thermal ramping from 15 °C to 65 °C (Fig. 3, Table S2). Higher anisotropy means lower membrane fluidity. The ability of the Lo18 protein to maintain membrane fluidity was specific to the physiological state of the cells used to produce liposomes. Indeed, Lo18 maintained the membrane fluidity in liposomes from exponential growth phase cultures of both O. oeni and L. plantarum. From 45 °C, fluidity reached 68% (O. oeni) and 62% (L. plantarum) of their initial anisotropy values, respectively, in the presence of Lo18, versus 52% (O. oeni), and 56% (L. plantarum) in the absence of Lo18. Regarding liposomes from O. oeni culture in the stationary phase, the lipochaperone activity of Lo18 appeared later at 65 °C. Finally, the lipochaperone activity of Lo18 over liposomes from ethanol-adapted O. oeni cultures was completely abolished in the temperature range 15 °C to 65 °C, where no difference was observed in the presence or absence of Lo18. Based on these results, it can be noted that the lipochaperon activity of Lo18 was diminished when the lipid composition of the membranes was lower in phosphatidylglycerol. This suggests that lipochaperon activity can take place via electrostatic interactions. However, it is important to take into account the difference in initial anisotropy, which is dependent on the lipid composition of the membranes used, and which may influence fluidization. This is particularly the case for membranes adapted to stress, which are more rigid from the outset.

Figure 3.

Lipochaperone activity of Lo18 according to the lipidic substrate. Measure of fluorescence anisotropy of DPH inserted into liposomes of: (i) O. oeni in the exponential phase (circle), the stationary phase (square), or 36 h ethanol-adapted culture (inverted triangle), or (ii) L. plantarum liposomes (triangle) during thermal ramping between 15 and 60 °C in the absence (white symbols) or presence of Lo18 (colored symbols). The data represent the means and SE of three independent experiments and were analyzed statistically by the Mann–Whitney U-test.

Impact of membrane lipid composition on the secondary structure of Lo18

The modification of Lo18 lipochaperone activity linked to the physiological state of the membrane used was further characterized by investigating the modification of Lo18 secondary structures. To do this, Synchrotron radiation circular dichroism (SRCD) experiments were performed on Lo18 alone or in interaction with the four types of liposomes during thermal ramping from 20 °C to 60 °C. Although the analysis was performed in this temperature range, previous results on the lipochaperone activity of Lo18 have shown that 45 °C is a key temperature in establishing the lipochaperone activity of the protein²⁶. Therefore, SRCD results obtained at this temperature during thermal ramping will be described in detail. At 45 °C, the proportion of α-helixes increased significantly when Lo18 interacted with liposomes, starting at 3% for the protein alone to 6%, 8%, 24% and 6% in the presence of liposomes from O. oeni cells in the exponential and stationary phases, and ethanol-adapted cultures, respectively, as well as with L. plantarum cells (Fig. 4). In contrast, the interaction with liposomes decreased the proportion of β-sheets. For Lo18 alone, the proportion of β-sheets was 42%, compared to 37%, 35%, 21%, and 31% upon interaction with liposomes from the exponential and stationary phases, the alcohol-adapted culture of O. oeni, and the culture of L. plantarum, respectively (Fig. 4). Thus, the presence of lipids modified the structure of Lo18 more significantly than lipid composition.

Figure 4.

Secondary structure modification of Lo18 in interaction with liposomes of (i) O. oeni in exponential phase, stationary phase or 36 h ethanol-adapted culture or (ii) L. plantarum liposomes. Proportion of secondary structure, α-helix and β-sheet (A) SRCD spectra on 190–240 nm wavelength at 15 °C (B) and 45 °C (C) of Lo18 alone (black) or in the presence of liposomes from O. oeni culture in exponential phase (purple), stationary phase (blue) or 36 h ethanol-adapted culture (turquoise) and L. plantarum culture (orange) (A). The data represent the means and SE of three independent experiments and were analyzed statistically by Kruskal–Wallis nonparametric test, pV < 0.05.

In addition, the insertion of Lo18 into the different types of liposomes was also monitored by SRCD. This insertion was highlighted by a shift in the spectrum of polarized light towards the lower frequencies (bathochromic effect) when Lo18 was in the presence of certain liposomes. A clearly visualized shoulder (Fig. 4B, in a rectangle box) appeared in the 195–210 nm range with liposomes derived from O. oeni cells in the exponential and stationary phases between 15 °C to 45 °C. For example, at 15 °C, the Δε value at 200 nm was -10.1 and -11.6 for these liposomes, respectively, compared to -14.7 for the protein alone. This shoulder was totally absent in the two other conditions (-15.7 and -15.8 for liposomes derived from an alcohol-adapted culture and from an L. plantarum culture, respectively). Both these latter liposomes were the poorest in proportions of oleic acid in contrast to the liposomes from the culture of O. oeni in exponential and stationary growth phases (Fig. 2). These differences in membrane composition, in particular the degree of fatty acid unsaturation, could influence membrane fluidity as well as the distribution of hydrophobic zones on the membrane surface. As Lo18 insertion is mainly observed in liposomes rich in oleic acid, it appears that this insertion may take place via hydrophobic interactions with relatively fluid membranes. The insertion could be observed at temperatures up to 45 °C (Fig. 4C), above which the structural change in Lo18 no longer allowed distinguishing this shoulder.

Discussion

Membranes are the first cellular component affected by environmental changes. The viscosity of the membrane must remain close to 0.1 Pa.s. However, as a function of time and the intensity of the stresses encountered, this viscosity will be modified, leading the cells to adapt by fluidizing or rigidifying their membrane.²⁹ To maintain optimum membrane fluidity, cells deploy a variety of mechanisms. In bacteria, adjustments such as changes in (i) saturated/unsaturated fatty acid ratios, (ii) the nature of unsaturation, (iii) the length of fatty acid carbon chains, or (iv) changes in polar heads are frequently observed^29–32. However, these response mechanisms can only regulate membrane fluidity over the long term. To cope rapidly with an alteration in membrane integrity, cells have the capacity to synthesize proteins that regulate membrane fluidity, such as sHSPs Lo18 synthetized by O. oeni^{11,20,21,23,24}.

Previous studies have shown that Lo18 has a stronger affinity for lipid substrates than for protein substrates, especially when the lipid composition of membranes is not yet adapted to stress. Until now, no specific affinities with phospholipids have been identified²³. In this context, the affinity between Lo18 and several phospholipids chosen for their physical properties was determined by means of immunolabeling. Although Lo18 interacts specifically with all the phospholipids tested, its affinity is most pronounced for lipids with a polar head group of PG and/or oleic acid fatty acid chains. No interaction with DOPA was evidenced, certainly due to its conical conformation that might be less accessible³³. Due to the physical properties of PG and oleic acid, i.e. negatively charged and unsaturated fatty acids, respectively, the results suggest that Lo18 interacts via electrostatic and hydrophobic forces. These lipids are particularly represented in fluid membranes, classically found before lipid modification during adaptation to stress^{29,32,34–36}. Indeed, both of these two phospholipid constituents are predominant in the membranes of both the exponential phase cultures (from O. oeni and L. plantarum) tested, mimicking unstressed membranes. This specific affinity for negatively charged lipids or fluid membranes has been reported for other sHSPs, such as human HSPB1^12,28, HSP16 from S. pombe¹⁷, HSP17 from Synechocystis³⁷ and HSP12 from Ustilago maydis³⁸. In all these cases, interaction appears to be mediated both by the state of fluidity of the membrane and by charges present at the lipid head and certain domains of the sHSP^12,28. More precisely, the hydrophobic domain of fatty acids could be used as an anchoring surface²⁸ and negative charges could attract the sHSPs on this surface.

To test the impact of these unsaturated (oleic acid) and negatively charged (PG) domains on Lo18/membrane interaction, Lo18 insertion into liposomes containing lipids from O. oeni or L. plantarum cultures at different physiological states (exponential growth, stationary phase or adaptation to 8% ethanol) was monitored by means of SRCD. The PG and oleic acid concentrations of these liposomes varied: rich in PG and oleic acid for membranes from O. oeni cultures in exponential phase, poor in PG and rich in oleic acid for those in stationary phase, poor in PG and oleic acid for those adapted to alcohol, and rich in PG but poor in oleic acid for membranes from L. plantarum cultures in exponential phase (Fig. 5A). A loss of negative absorption and a batho-chromic (red-) shift of the π-π*electronic transition at 208 nm translating protein insertion inability³⁹, were observed on SRCD spectra in the presence of ethanol-adapted liposomes and L. plantarum liposomes. Both these liposomes had the lowest content of oleic acids among the liposomes tested. This difference in fluidity mainly caused by the nature of the fatty acid chains, could change the insertion of Lo18 into the membrane, as was observed for membrane proteins, human sHSPs such as HSPB1 and HSPB5 or HSP17 from Synechocystis, which insert more efficiently into liquid crystalline phase membranes than into gel phase membranes^{12,28,38,40–42}. Indeed, the more rigid the membrane, the less mobile and more ordered the phospholipids are, thereby reducing the efficiency of protein insertion³². The decrease of protein insertion due to greater bilayer stability may have an impact on protein activity, particularly lipochaperone activity, without necessarily abolishing it⁴³.

Figure 5.

Part of the mechanisms by which membrane lipids interact with Lo18. Summary table of influence of lipid environment and physiological state of membranes on lipochaperone activity, secondary structure, interaction and insertion of Lo18 with membrane (A). Schematic illustration of Lo18-membrane interaction (B). During an association/dissociation cycle of Lo18 oligomeric subunits, the protein may be attracted to the membrane by electrostatic and/or hydrophobic forces, allowing the protein to chaperone and/or insert into the membrane.

In addition to the impact on Lo18 insertion, the lipid composition of membrane could also impact the structure of the protein⁴³. Indeed, a modification in membrane fluidity caused by the modification of the ratio between saturated/unsaturated fatty acids may also modify the attractive force around the membrane, leading to changes in the secondary structure of Lo18. The secondary structure of Lo18 was determined during Lo18-liposome interaction by SRCD measurement. After interaction with the four types of liposomes at 45 °C, an increase in the proportion of α-helices to the detriment of β-sheets was observed compared to the Lo18 protein alone. However, as this structural change occurred to a greater or lesser extent for all the liposomes tested, the presence of lipids (rather than their nature) was sufficient to cause the structural change in Lo18^26,44.

Both insertion and secondary structure modification due to membrane lipid composition may contribute to Lo18 lipochaperone activity²⁶. Consequently, Lo18 lipochaperone activity was monitored during the fluidization of four different types of liposomes, two of which represent optimal growth conditions and the other two stress conditions. In the presence of liposomes mimicking O. oeni membranes adapted to stress (stationary phase and ethanol adapted, the lipochaperone activity of Lo18 was established at higher temperature or abolished. Furthermore, these stress-adapted liposomes had the lowest phosphatidylglycerol content, suggesting that the electrostatic attraction of proteins by the membrane may be a more significant factor than membrane fluidity in lipochaperone activity. Conversely, for Lo18 membrane insertion, membrane fluidity appeared to be an important factor.

Indeed, although Lo18 is unable to insert into liposomes extracted from a culture of L. plantarum, having a lipid profile (for the main polar heads) similar to that of O. oeni culture in exponential phase, it retains lipochaperone activity. It is therefore plausible that Lo18, like human sHSPs HSPB1 and HSPB5¹² or HSP16 from S. pombe¹⁷, may be attracted first by polar head groups before interacting with specific membrane microdomains⁴⁵ via an as yet unknown mechanism, allowing it to reinforce its lipochaperone activity.

The results obtained in this study lead us to propose a model. During membrane fluidization, dimers from cytoplasmic oligomeric structures (which regularly undergo association/dissociation cycles) are attracted to the membrane. Hydrophobic and electrostatic forces could attract them towards lipids, in particular unsaturated and negatively charged phospholipids (notably oleic acid and phosphatidylglycerol) for which Lo18 has greater affinity. In particular, it appears that the contribution of electrostatic forces induced by the polar heads of membrane phospholipids may influence the lipochaperone activity of Lo18, whereas forces of a hydrophobic type would influence the insertion of Lo18 into membranes, which is consistent with current understanding of membrane protein insertion⁴⁰ (Fig. 5).

In conclusion, the present work contributes to a better understanding of the mechanisms of sHSPs and bacterial membrane interaction. Such understanding could allow the use the lipochaperone role of sHsps in different fields, such as in the agri-food industry, for example the resistance of L. plantarum to freeze drying processes⁴⁶ or in the medical field for the treatment of neurodegenerative diseases such as exudative retinopathy⁴⁷.

Material and methods

Media and growth conditions

O. oeni ATCC BAA-1163 was grown in modified FT80 medium⁴⁸ at pH 5.3 and 28 °C in the presence or absence of 8% ethanol. L. plantarum WFS1 was grown in MRS medium (Condalab, Spain) at pH 6.2 and 28 °C.

Preparation of liposomes

O. oeni and L. plantarum cultures were grown to the exponential and/or stationary growth phase before lipids were extracted and purified according to Bligh and Dyer⁴⁹. First, cells of O. oeni were incubated in anaerobic conditions at 28 °C for 24 h to reach the middle of the exponential phase (DO_600nm: 0.6), 48 h to reach the stationary phase (DO_600nm: 1.2) or 48 h in the presence of 8% ethanol (DO_{600nm :} 0,6). Cells of L. plantarum were growth at 28 °C for 3 h to reach the middle of the exponential phase (DO_{600nm :} 0,6). Then Liposomes were produced by adapting the method described by Maitre et al.¹¹. Briefly, the equivalent of 50 DO units, for each sample, was washed and re-suspended in 1 mL physiological water (NaCl 9 g/L). After the addition of 3.75 mL CHCl3/MeOH (1:2), the solution was incubated for an hour at room temperature. Next, 1.25 mL chloroform and 1.25 mL physiological water were added and the solution was centrifuged at 4 °C for 10 min at 3000 g. The lower phase containing lipids and chloroform was recovered and stored in a glass tube. The chloroform present in the lipid fraction was evaporated using a nitrogen flow. The lipids were then gently resuspended in 10 mL of 50 mM phosphate buffer, pH 7.0, previously heated to 55 °C. After being sonicated twice for 2 min (Branson Ultrasonics™ CPX-952-138R, Branson Ultrasonics, Brookfield, CT, US), the lipid solution was rehydrated for 4 h at 55 °C. The lipid particles were then extruded through a polycarbonate membrane with 1 μm-diameter pores to obtain the liposomes similar in size to O. oeni cells. The liposomes were stored at 4 °C for up to 1 week.

Fluidity measurements

Fluorescence anisotropy measurements (reflecting the fluidity state of the membrane) were performed using a FLUOROLOG-3 spectrofluorometer (Jobin Yvon Inc, USA). Each measurement was performed at excitation and emission wavelengths of 360 nm and 431 nm, respectively. Two hundred and fifty μL of liposomes (prepared as above corresponding to lipid concentrations at 250 µM) and 3 μM of 1,6-diphenyl-1,3,5 hexatriene (DPH) probe (Sigma Aldrich, St. Louis, US) were mixed in a quartz cuvette (10 mm optical path) and completed with phosphate buffer pH 7 solution to a final volume of 3 mL. Measurements were performed on the liposome suspension in the presence or absence of 10 µM Lo18 with a mass ratio of 1:2 (m/m) (Lo18/liposomes) following a temperature increase from 15 to 65 °C (2 °C per min increase) controlled by a Peltier system (QNW TC1 temperature controller, Quantum Northwest, Liberty Lake, WA, USA). For each experiment, an anisotropy value was obtained every 14.6 s and calculated according to Shinitzky and Barenholz 1978⁵⁰. Each experiment was performed in triplicate.

Synchrotron radiation circular dichroism (SRCD) spectroscopy

Circular dichroism spectra were collected on the DISCO beamline (Synchrotron SOLEIL, France) according to Evans et al., 2004⁵¹. The instrument was calibrated using 99% pure ( +) camphor-10-sulphonic acid (Sigma Aldrich, Saint-Louis, US) at 25 °C after each beam fill⁵². Lo18 at 140 µM was prepared in 50 mM sodium phosphate buffer, pH 7.0, alone or in the presence of liposomes at 250 µM. Then, 50 µL of sample was loaded into a 0.02 cm pathlength demountable cylindrical Suprasil quartz cell (Hellma, Germany) and subjected to a thermal scan from 15 °C to 65 °C with a step of 3 °C. Each dataset was collected from 262 to 176 nm, with an integration time of 1.2 s and a spectral bandwidth of 1 nm. Three scans of each sample and the equivalent baseline were collected. The sample spectrum was averaged across the three repeats, and the averaged baselines (buffer without protein) subtracted. Baseline-subtracted spectra were zeroed between 250 and 260 nm. All the spectra were normalized according to the protein concentration and pathlength using the CDToolX software and deconvoluted using BeStSel online software⁵³.

Lipid Protein overlay assay

The interactions between lipids and Lo18 or modified proteins were observed by immunostaining, according to Dowler et al., 2002⁵⁴. A drop of 2 µL of dipalmitoyl phosphatidylethanolamine (DPPE), dioleoyl phosphatidylethanolamine (DOPE), dioleoyl phosphatidic acid (DOPA), dipalmitoyl phosphatidylcholine (DPPC), dioleoyl phosphatidylcholine (DOPC), dipalmitoyl phosphatidylglycerol (DPPG), and dioleoylphosphatidylglycerol (DOPG) (Avanti polar, France) diluted in chloroform at 1.5 mM was deposited on a PVDF membrane (Fisher scientific, USA). The membrane was blocked in TBS/Regilait 5%, 1 h at 70 rpm and washed three times for 5 min with TBS. It was then incubated with Lo18 at 7.5 µg/mL for 1 h at 70 rpm and washed again three times for 5 min each. Afterward, the protein bound to the lipids was detected with antibodies against Lo18 (1/750), produced by Eurogenetec with mice, and against the primary antibody (1/2000) against mice antibodies, by incubating the membrane for 1 h at 70 rpm for each antibody. Chemiluminescence was detected using the ECL reagent kit (Thermo Fisher, USA). A normalization step was added to quantify the interaction between Lo18 and the phospholipids. To do this, the proportion of phospholipids fixed to the PVDF membrane was measured by fluorescence imaging after 15 min incubation with 50 mM of Nile Red staining. Then the signal intensity obtained for the immunolabeling measurement was adjusted to the signal obtained for the Nile Red staining.

Lipidomic analysis

Total fatty acids analysis by GCMS-NCI

Bligh and Dyer lipid extracts (50 µL) were mixed with 25 µL of a fatty acid internal standard mix (Avanti Polar Lipids, France) containing: 1146 ng of myristic acid-d3, 4973 ng of palmitic acid-d3, 3703 ng of stearic acid-d3, 3174 ng of linoleic acid d4, 45.8 ng of arachidic acid-d3, 1632 ng of arachidonic acid-d8, 47.6 ng of behenic acid-d3, 476.1 ng of DHA-d5, 22.9 ng of Lignoceric-d4, and 17.6 ng of cerotic acid-d4. Total fatty acids were quantitated after alkaline hydrolysis by GCMS-NCI, as previously described⁵⁵.

Lipid analysis by LCMS2

Bligh and Dyer lipid extracts (200 µL) were dried under vacuum and mixed with an internal standard mix (50 µL) (CDN Isotopes, France and Cayman, France), (14:0)4CL (800 ng), (12:0)2DG (8 µg), and (21:0)2PC (50 ng). Sample (1 µL) was analyzed by LCMSMS using a Zorbax®Eclipse Plus C18 1.8 µm, 2.1 × 100 mm column maintained at 55 °C (Agilent Technologies).

Briefly, digalactosyldiacylglycerols (DGDG), monogalactosyldiacylglycerols (MGDG), and diacylglycerols (DG) were separated on a 1260 Infinity LC system (Agilent Technologies) with a gradient of mobile phase A (acetonitrile/water/1 M ammonium formate (60/39/1 v/v/v) with 0.1% formic acid), and mobile phase B (isopropanol/acetonitrile/1 M ammonium formate (90/9/1 v/v/v) with 0.1% formic acid)⁵⁶ at a flow rate of 0.4 mLmL/min set as follows: 1 min hold at 50% B; 50–60% B in 4 min; 60–85% B in 10 min; 85–99% B in 1 min; 2 min hold at 99%, 99%-50% ramp-down in 0.1 min and maintained at 50% B for 3.9 min. Acquisition was carried out on a 6460 Triple Quadrupole (Agilent Technologies) equipped with an ESI Jet stream source (temperature 250 °C, nebulizer 20 L/min, sheath gas 11 L/min, sheath gas temperature 220 °C, capillary 3500 V, nozzle 1000 V) operating in positive Single Reaction Monitoring (SRM) mode (fragmentor 148 V, collision energy 23 V). Transitions were set as the neutral loss of 359 Da for DGDG or 197 Da for MGDG from their respective [M + NH4] + ions⁵⁷ DG (as NH4 + adducts) were quantified according to the sum of the signal resulting from the neutral loss of either their sn-1 or sn-2 fatty acid. Finally, lipid concentrations were determined by calculating their relative response to (12:0)2DG used as internal standard.

Phosphatidylglycerols (PG) and lysyl-phosphatidylglycerols (lysyl-PG) were separated with the same mobile phases and column as for DGs, and the elution gradient used was set as follows: 2 min hold at 50% B; 50–99% B in 14 min; 2 min hold at 99%, 99%-50% ramp-down in 0.1 min; return to initial conditions in 1.9 min. Acquisition was carried out on a 6460 Triple Quadrupole (Agilent Technologies) equipped with an ESI Jet stream source (temperature 200 °C, nebulizer 20 L/min, sheath gas 11 L/min, sheath gas temperature 220 °C, capillary 3500 V, nozzle 1000 V) operating in positive Single Reaction Monitoring (SRM) mode (fragmentor/collector 116 V/13 V and 300 V/34 V for PG and Lysyl-PG respectively). Transitions were set as the neutral loss of 189 Da or 300 Da for [PG + NH4] + and [Lysyl-PG + H] + ions respectively. Lipid concentrations were determined by calculating relative response ratios with regards to (12:0)2DG used as internal standard.

The analysis of cardiolipins (CL) and phosphatidylcholines (PC) was performed as previously described^58,59 except that a Vanquish LC system, coupled to a triple-stage quadrupole (TSQ) Altis mass spectrometer equipped with a heated electrospray ionization source (Thermo Scientific) was used (sheath gas, 50 arb; auxiliary gas, 10 arb; sweep gas, 1 arb; ion transfer tube temperature, 325 °C; vaporizer tempropylerature, 350 °C and ion spray voltage, 3500 V ( +), and 2500 V (-)). Lipid concentrations for all the quantified lipid classes were expressed in pmol/mg of Bligh and Dyer extract.

Statistical analysis

For each condition tested on fluorescence anisotropy, immunolabelling, lipid profile, and secondary structure analysis were performed using three independent measurements. The anisotropy values obtained were normalized to be expressed in percentage compared to the initial value at 15 °C. Statistical analyses were performed by the statistical RStudio software (version 1.2.5033). The normality of the distribution and homogeneity of the variances of each condition were tested by the Shapiro–Wilk test and the Bartlett test, respectively. Then, a non-parametric Kruskal–Wallis test was used to compare the samples with a significance level of α = 0.05. All the statistical tests were considered significant at a P-value < 0.05.

Data availability

The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Author contributions

T. B. and S. W: Conceptualization, T. B., and S. W: Methodology, T. B. Formal analysis, T. B.and S. W.: Investigation, T. B., S. W and F. W.: Resources, T. B. and S. W : Writing – Original Draft, T.B, S. W., F. W., P. F.-V. and S. C. : Writing – Review & Editing; S. W.: Validation, Project administration and Funding acquisition.

Competing interests

The authors declare no competing interests.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-67362-6.