Varying the position of phospholipid acyl chain unsaturation modulates hopanoid and sterol ordering

The cell membrane must balance mechanical stability with fluidity to function as both a barrier and an organizational platform. Key to this balance is the ordering of hydrocarbon chains and the packing of lipids. Many eukaryotes synthesize sterols, which are uniquely capable of modulating the lipid order to decouple membrane stability from fluidity. Ancient sterol analogs known as hopanoids are found in many bacteria and proposed as ancestral ordering lipids. The juxtaposition of sterols and hopanoids in extant organisms prompts us to ask why both pathways persist, especially in light of their convergent ability to order lipids. In this work, simulations, monolayer experiments, and cellular assays show that hopanoids and sterols order unsaturated phospholipids differently based on the position of double bonds in the phospholipid acyl chain. We find that cholesterol and diplopterol’s methyl group distributions lead to distinct effects on unsaturated lipids. In Mesoplasma florum, diplopterol’s constrained ordering capacity reduces membrane resistance to osmotic stress, unlike cholesterol. These findings suggest that cholesterol’s broader lipid-ordering ability may have facilitated the exploration of a more diverse lipidomic landscape in eukaryotic membranes.


Introduction
To support life, cell membranes must balance mechanical stability (sufficient to perform robustly as a barrier) against fluidity and deformability (sufficient to support its role as an organizational platform for bioactivity).However, for bilayer forming mixtures of lipids, stability is often gained at the expense of fluidity.Different solutions to this dilemma have evolved on different branches of the tree of life.For example, thermophilic archaea synthesize double headed bolalipids that maintain membrane integrity even at extreme temperatures (1).In many eukaryotes, sterols solve this problem, their twin faces simultaneously ordering hydrocarbon chains while promoting lateral diffusivity (2).By decoupling the local motion of acyl chains and lipid translational freedom of motion, sterols allow cells to build membranes that are mechanically stable enough to withstand environmental perturbations, but fluid enough to support diffusion-dependent biochemistry.
Most non-eukaryotic organisms cannot synthesize sterols and must rely on other mechanisms to modulate their membrane properties.Some bacteria utilize a family of compounds called hopanoids (3).Hopanoids are tri-terpenoids, whose sedimentary record as early as 1.64 billion years ago (4).
Since both families are synthesized from squalene, with homologous enzymes (squalene-hopene cyclase and oxidosqualene cyclase) (5), they share certain chemical similarities.Like sterols, hopanoids also reside within the membrane and modulate membrane robustness, fluidity, and resistance against abiotic stresses (3,6).For these reasons, hopanoids are considered both bacterial and ancient sterol analogs.But why does life need two analogous classes of ordering lipids?
Despite their similarities, hopanoids and sterols possess distinct properties.Both diplopterol (Dpop), a common hopanoid in bacteria, and Chol (Chol), a mammalian sterol, interact favorably with and condense saturated lipids, a diagnostic feature of lipid ordering (7).However, the condensing effect of Dpop is impaired when an acyl chain unsaturation is introduced, and in a manner that depends on the location of the double bond.In this brief report, we examined how double bond position influences sterol and hopanoid ordering.We find that methyl group distribution on hopanoid backbones restricts its ordering ability, resulting in pronounced effects on cellular robustness in the model organism Mesoplasma florum.Conversely, Chol ordering is comparatively insensitive to double bond position, allowing eukaryotic membranes more degrees of freedom in regulating their lipidomes.

Results
Since unsaturation changes the interaction between Chol/Dpop and phospholipids, we hypothesize that moving the position of double bonds along the phospholipid acyl chain can help us probe Chol/Dpop ordering, thus revealing key structural features needed for ordering.We also aim to draw a comparison between Chol and Dpop's ordering to showcase a constraint in the evolution of sterol-and hopanoid-containing lipidomes.
We first determined the ordering of phospholipid chains by measuring monolayer surface pressure vs area isotherms.Similar isotherms were obtained for PC phospholipids with a double bond at one of three positions: ∆6, ∆9, or ∆11 (chemical structures are shown in Fig. 1A), suggesting similar packing in the pure membranes, regardless of the isomer (Fig 1B).The isotherms differ, however, for binary mixtures of either Chol or Dpop, in a manner dependent on both terpenoid choice and double bond position.To better quantify this difference, we calculated the condensing effect of the terpenoid on each lipid, as well as the free energy of mixing (∆Gmix) the phospholipid and terpenoid (see Methods for definitions and details).A positive condensing effect is obtained when the terpenoid orders lipid chains and increases lipid packing, which Chol consistently exhibited regardless of the double bond position.In contrast, Dpop only condenses the ∆11 isomer, does not noticeably affect the ∆6 isomer, and has a negative condensing effect on ∆9 isomer.These observations are mirrored by ∆Gmix, which reflects the thermodynamic balance between lipid-lipid interactions and mixing entropy.Chol has favorable interaction with all 3 lipid isomers (∆Gmix < 0), which progressively increases as the double bond moves away from the headgroup.On the other hand, Dpop interaction with PC varies based on double bond positions.While Dpop was ideally mixed with ∆6-PC (∆Gmix = 0), the mixing of Dpop and ∆9-PC was unfavorable (∆Gmix >0), and only with ∆11-PC was the ∆Gmix <0.The distinction between Dpop interaction with ∆9-PC and ∆11-PC is intriguing, as the double bonds are just 2 carbons apart.
To gain molecular insight into differential ordering of ∆9 and ∆11-PC by Dpop we performed molecular dynamics simulations of bilayers of these binary mixtures.(Details in Methods) Figure 3 reports the 2H NMR chain order parameters (SCD) for several different binary mixtures of Dpop with phospholipids.SCD is defined in Eq. 2 in methods; larger values indicate more ordered chains.A model for Dpop compatible with the CHARMM36 family of force fields was built as described in Methods, and first tested by simulating binary mixtures with DPPC (Fig. 3A), obtaining an ordering effect similar to that observed in experiments on giant unilamellar vesicles (7).
We next considered binary mixtures with lipids with saturated chains at the sn-1 position and monounsaturated chains (∆19 or ∆11 isomers) at the sn-2 position (Fig 3B).As expected, the saturated chains were more ordered by Dpop than the unsaturated chains, and were ordered similarly regardless of which isomer was present in the other chain.Comparing the unsaturated chains, the ∆11 isomer was significantly more ordered than the ∆9 isomer.This trend was reproduced in PC with unsaturation in both acyl chains (Fig. 3C), where we again observed that the ∆11 isomer was significantly more ordered.In summary, the position of the double bond was critical in determining the acyl chain's order when interacting with Dpop, mirroring the observations in the monolayer experiments and supporting a model in which unsaturation position has a significant effect on condensation in fluid lipid bilayers.
In prior work, Martinez-Seara et al. found that the positioning of the double bond relative to the methyls protruding from cholesterol controls its condensing effect (8).We therefore performed a similar analysis, comparing the location of the PC lipid's double bond to the positions of methyl groups extending from the Dpop ring structure (annotated in Fig 1A) in our simulations.Figure 4 reports the distribution of double bonds of the hydrocarbon chains and Dpop methyl groups along the membrane normal, with z = 0 being the center of the bilayer.The distribution of ∆9-PC's double bond overlaps almost exactly with that of methyl group M2, while the ∆11-PC double bond falls in between the M3 and M4 methyl groups.The alignment between the  bond and the methyl group might explain the reduced condensing effect observed for ∆9-PC, while Dpop and ∆11-PC interacted more favorably (8).This interplay between lipid structure and membrane biophysics provides insights that could be extended to other sterols and hopanoids, possibly laying a path for predicting how hopanoid/sterol and phospholipid structure collectively influence membrane properties.
To investigate how this variation in lipid-lipid interaction might influence biomembrane function, we employed Mesoplasma florum as a living model system.M. florum is a Mollicute with no cell wall and a minimal genome (9).With limited machinery, Mesoplasma cannot synthesize its own lipids and relies on supplemented lipids from the media, offering a straightforward way to manipulate its lipidomes.By introducing either ∆9-or ∆11-PC to its lipid diet, we can create two identical biological membrane systems differing only in their unsaturation site.We then investigated this system to explore the effect of lipid-lipid interactions on a cellular scale.
Traditionally cultured with an undefined lipid diet in serums, we first test Mesoplasma's ability to grow in a defined lipid diet.Figures 5A and 5B report growth rate and cellular lipid content, respectively, suggesting Mesoplasma can both grow on and incorporate defined lipids into their membrane.Since sterol ordering has previously been associated with membrane robustness (2), we tested Mesoplasma membrane robustness with hypoosmotic shock.Live cells are subjected to hypoosmotic conditions, forcing cells to rapidly expand.As the membrane is stressed and ruptures, exposed cellular DNA is stained by propidium iodide.By quantifying the fluorescence intensity, we estimated the fraction of cells lysed due to osmotic shock, inferring membrane robustness.Figure 5C shows the cell's susceptibility to osmotic shock when supplied with different lipid diets.When cells were supplied with Chol, the addition of ∆9-or ∆11-PC did not produce significant changes in cellular robustness.However, cells fed with ∆9-PC and Dpop exhibited higher susceptibility to lysis than cells fed with ∆11-PC and Dpop.This data suggested that Dpop and ∆9-PC unfavorable interaction counteracts Dpop's ability to bolster membrane robustness.

Discussion
Chol and Dpop are both prevalent lipids, accounting for approximately 40% of their respective membrane lipidomes (10,11).We show that both of these tri-terpenoids interact with unsaturated lipids in a double bond position-dependent manner.While Chol condenses all unsaturated isomers comparably, Dpop shows the strongest condensing effect on ∆11-PC, albeit less potently than Chol.Simulations indicate that Dpop's interaction with unsaturated lipids is significantly hindered by having multiple methyl groups extending from both sides of the cyclic ring structure, similar to the biosynthetic precursor to Chol, lanosterol (12)(13)(14).Notably, when the double bond resides at the ∆9 position, and overlaps with Dpop's methyl group M2, it prevents effective Dpop-induced lipid packing.In combination with earlier simulation results for cholesterol from Martinez-Seara et al. (8), our observations highlight the power of using simulations to explore lipid-lipid interactions, especially in the case of less commercially available lipids like hopanoids.With careful consideration in model development, one can explore the chemical landscape of lipids and the consequences of their collective interactions.
From these observations, we hypothesized that the differentiation between ∆9 and ∆11 should be the most significant in hopanoid-bearing membranes.In Mesoplasma florum, the favorable interaction of ∆11-PC and Dpop enhanced membrane resilience to osmotic shock compared to ∆9-PC.This remarkable result highlights how subtle changes to lipid structure can have striking consequences for cellular robustness, and suggests a potential mechanism for osmoadaptation, for which hopanoids have been shown to play a critical role in soil and plant associated bacteria (15,16).Indeed, multiple hopanoid-bearing bacteria have ∆11 as the monounsaturation site (17), instead of ∆9 in eukaryotes.The hopanoid producing yeast, Schizosaccharomyces japonicus, also possess a ∆12-desaturase (18,19).Interestingly, in 2020, Chwastek et al. investigated the lipidome of a hopanoid-bearing organism Methylobacterium extorquens, and found that the main unsaturation position was ∆11 instead of ∆9, with an additional ∆5 unsaturation upregulated in coldadapted lipidome (11).Therefore, double bond position could represent a modifiable lipidomic feature that cells can employ to homeostatically fine-tune the ordering effects of hopanoids.
For Chol-bearing organisms, double bond position does not have such a pronounced effect on lipid ordering.This indifference might have alleviated evolutionary selection against ∆9 unsaturation in early sterol-bearing organisms, providing more flexibility to produce lipids with double bond positions optimized for orthogonal lipid-lipid or lipid-protein interactions.Our results, therefore, suggest that a transition from hopanoid to sterol-containing lipidomes could have widened the chemical landscape available for cells to explore for tuning membrane properties.

Materials
∆6-, ∆9-, ∆11-PC and egg sphingomyelin were purchased from Avanti Polar Lipids.Chol and palmitic acid were purchased from Sigma, and Dpop from Chiron.Stock concentrations of lipids were measured by phosphate assay.Chol and Dpop were weighed out on a precision scale and solubilized in a known volume of chloroform.

Methods
Monolayer.Chloroform solutions of pure lipids and mixtures were prepared at 0.2 mg/mL lipid concentrations.Monolayers were prepared by injecting 15-30 µL of lipid solution onto an aqueous subphase maintained at 20°C by a built-in temperature-controlled circulating water bath.The subphase was comprised of 10 mM HEPES, 150 mM NaCl, pH 7. Isotherms were recorded using a 70 cm 2 teflon Langmuir trough fitted with a motorized compression barrier equipped with pressure sensor (Kibron DeltaPi).
The mean molecular area (MMAs) for each mixture were estimated from the averages of isotherms from three monolayers that were prepared independently.Data were rounded down to the nearest neighbor for condensation effect and free energy calculation.All isotherms were fitted to a regression, and statistical significance was tested using manova with the 2 coefficients.During equilibration, double bonds were restrained in the cis configuration to prevent isomerization; these restraints are gradually reduced during the final three stages of the equilibration protocol.
Production simulations (NPT ensemble) were integrated with leapfrog using the Parinello-Rahman (25) barostat to control pressure (time constant 5 psec; compressibility 4.5e−5 bar−1; coupled anisotropically to allow independent fluctuation of the in-plane and normal directions) and temperature controlled using Nose-Hoover38,39 (time constant 1 psec) at a temperature of 298K.
Hydrogens were constrained with LINCS (expansion order 4), a 2 fsec timestep was used, short range electrostatics were computed directly within 1.2 nm, and long-range electrostatics were computed every timestep using particle mesh Ewald (27,28) with a grid spacing of 1 Å and cubic interpolation.Long range dispersion was smoothly truncated over 10-12 nm using a force-switch cutoff scheme.Simulations were performed with Gromacs 2020.4.
Calculation of simulation observables.The distribution of angles between either Dpop or Chol and the membrane normal was computed, defining the orientation of both by a vector from atom C24 to atom O3.The locations of methyl groups in Dpop or Chol along the direction normal to the membrane were recorded and compiled into histograms with a bin size of 0.87 Å. Deuterium order parameters were obtained from the simulations via where ϕ is the angle between the C-H bond vectors and the membrane normal at each position along the hydrocarbon chain.The area per lipid was calculated using the Voronoi construction implemented in MEMBPLUGIN (29).The location of each lipid is defined by the center of geometry of the C2, C21, and C31 atoms and the location of the Chol/ Dpop was defined by the O3 atom, and then a Voronoi construction is built around these points in the plane parallel to the membrane surface.Membrane incorporation.Cells were collected in early exponential stage and centrifuged (5000 rcf, 7 min, 30°C).Supernatant were discarded and cell pellet was washed with wash buffer (200 mM NaCl, 25 mM HEPES, 1% glucose, pH 7.0) and centrifuged (5000 rcf, 7 min, 30°C).The collected pellet was then subjected to a Bligh Dyer extraction (30).Briefly, the pellet was homogenized in a mixture of water:chloroform:methanol in 0.8:1:2 ratio and sonicated for 2 minutes.Subsequently, water and chloroform were added in 1:1 ratio.The mixture was centrifuged at 2000 rcf for 30 seconds in a mini centrifuge to promote phase separation.The lower, organic fraction containing lipids was collected and transferred to a fresh tube.The total lipid extract was then deposited on a silica gel plate (Supelco) and placed in a glass chamber.Chromatography was performed using chloroform as the running phase.After the run, the plate was dried and stained using 8 % copper sulphate in 3% phosphoric acid solution and heated until visible bands were observed.Images were captured using a GelDoc (Biozym Azure c600) and analysed using ImageJ.

Cell
Membrane osmotic shock.Cells were collected in early exponential stage and centrifuged (5000 rcf, 7 min, 30°C).The collected pellet was resuspended in a serial dilution of 0%, 20%, 40%, 60%, 80% and 100% of wash buffer (200 mM NaCl, 25 mM HEPES, 1% glucose, pH 7.0).The suspension was stained with 10 uM propidium iodide and added to a 96-well plate.Fluoresence emission was recorded using a Tecan Spark fluorescence reader, with excitation at 529 -549 nm and emission at 609 -629 nm.The fraction of cell lysed was calculated by normalizing the signal of each sample to the 0% and 100% wash buffer sample.Figure 5C represented the fraction of cell lysed in 80% of wash buffer.

PC isomers A
Where c = % condensation, Ao = the MMA of the lipid mixture, X1, X2 = the mole fraction of lipid 1 and 2 in the mix, and A1, A2 = the MMAs of lipid 1 and 2 at surface pressures 30 mN/m.Error bars were produced based on error propangation.The ΔG was calculated by integrating the areas of lipid mixtures over pressures Π = 5, 10, 15, 20, and 25 mN/m according to Grzybek et al(20).Error bars were produced based on error propagation.Diplopterol model development.A CHARMM compatible model for diplopterol (Dpop) was developed using the automated atom typing and parameter assignment pipeline CGenFF.30Charmm topology and parameter files are provided as Supplemental Material.Simulation composition and construction.Simulation systems contained either DOPC or POPC (unsaturation at either the ∆9 or ∆11 position) and one of either Dpop or Chol.All initial configurations were built using the CHARMM-GUI webserver(21)(22)(23).Systems containing atypical unsaturated chains (ie, ∆11) were generated by first building a binary mixture of the corresponding ∆9 lipid (DOPC or POPC) with either Dpop or Chol, then "mutating" the unsaturated chain(s) to move the double bond to the appropriate position, using a Charmm script provided by the Klauda Lab.All simulations contained approximately 550 lipids per leaflet and at least 50 TIP3P(24) water molecules per lipid.All lipids were modeled with the CHARMM36 force-field(23), except Dpop which was modeled using the CHARMM general force field, with atom types determined by the paramchem server.(Gromacs topology file provided in the Supplemental Information.)Initial dimensions in the membrane plane were about 17.5 nm x 17.5 nm, containing approximately 270,000 atoms.Four different binary mixtures were simulated: Δ9-DOPC: Dpop, Δ11-DOPC: Dpop, Δ9-POPC: Dpop, Δ11 POPC: Dpop.Each binary mixture was simulated at four different compositions: 95:5, 85:15, 70:30, 50:50.Each binary system was simulated for 500 nsec of production simulation as described below.Four additional controls were simulated without any sterol or hopanoid, each for 50 nsec of production simulation as described below: Δ9-DOPC, Δ11-DOPC, Δ9-POPC, Δ11-POPC.Equilibration and production simulations.Each system was prepared individually for production simulation through a series of 6 minimization and heating steps as provided by the CHARMM-GUI equilibration protocol: (i) steepest descent to minimize the initial configuration; (ii) 125,000 steps of leapfrog dynamics with a 1 fsec timestep and velocities reassigned every 500 steps; (iii) 125,000 steps of leapfrog dynamics with a 1 fsec timestep, pressure controlled by the Parinello-Rahman barostat(25) and velocities reassigned every 500 steps, then a total of 750,000 steps of leapfrog dynamics with a 2 fsec timestep and hydrogen positions constrained by LINCS(26), pressure controlled by the Parinello-Rahman barostat(25), and velocities reassigned every 500 steps.