Hopanoids As Functional Analogues of Cholesterol in Bacterial Membranes

Hopanoids as functional analogues of cholesterol in bacterial membranes

James P. Sáenza,1, Daniel Grossera, Alexander S. Bradleyb, Thibaut J. Lagnya, Oksana Lavrynenkoa, Martyna Brodaa,c, and Kai Simonsa,1

aMax Planck Institute of Molecular Cell Biology and Genetics, 01307 Dresden, Germany; bDepartment of Earth and Planetary Sciences, Washington University, St. Louis, MO 63130; and cDepartment of Biotechnology, University of Wroclaw, 50-383 Wroclaw, Poland

Contributed by Kai Simons, August 18, 2015 (sent for review July 8, 2015; reviewed by Damien Devos and Felix Goni) The functionality of cellular membranes relies on the molecular order fluid but mechanically robust plasma membrane (12). Early invest- imparted by lipids. In eukaryotes, sterols such as cholesterol modulate igations showed that both sterols and hopanoids exhibit the ability membrane order, yet they are not typically found in prokaryotes. The to condense lipids (13, 14). We demonstrated that hopanoids are structurally similar bacterial hopanoids exhibit similar ordering prop- indeed bacterial sterol surrogates with respect to their ability to erties as sterols in vitro, but their exact physiological role in form a liquid ordered phase, and that they promote liquid-liquid living bacteria is relatively uncharted. We present evidence that phase separation in membranes (15). This observation decouples hopanoids interact with glycolipids in bacterial outer membranes to the evolution of ordered biochemically active liquid membranes form a highly ordered bilayer in a manner analogous to the interac- from the requirement for molecular oxygen and suggests that the tion of sterols with sphingolipids in eukaryotic plasma membranes. ability to subcompartmentalize membranes could have preceded Furthermore, multidrug transport is impaired in a hopanoid-deficient the evolution of sterols. Subsequently, it was shown that hopanoid- Methylobacterium extorquens mutant of the gram-negative ,which based ordering can be tuned by structural modifications of their introduces a link between membrane order and an energy-depen- ring structure or polar side chain (16, 17). dent, membrane-associated function in prokaryotes. Thus, we reveal It is clear that hopanoids in vitro exhibit an ability to order a convergence in the architecture of bacterial and eukaryotic mem- synthetic membranes in a sterol-like manner (15). However, it is branes and implicate the biosynthetic pathways of hopanoids and not known whether hopanoids impart molecular order within other order-modulating lipids as potential targets to fight patho- genic multidrug resistance. bacterial membranes in vivo, and with which lipids they interact to achieve such order. Hopanoids have been identified in many membrane order | hopanoids | multidrug efflux | outer membrane | gram-negative bacteria (18, 19), but their exact physiological role Methylobacterium is unclear. Deletion of hopanoid synthesis is nonlethal in the bacteria that have been studied so far, but hopanoid-deficient mutants have been shown to exhibit increased sensitivity to an- terols (e.g., cholesterol; Fig. S1) are ubiquitous eukaryotic tibiotics and detergents as well as susceptibility to stresses, in- Smembrane lipids with a planar geometry that endows them – with a propensity to constrain, and thereby order, lipid bilayers. cluding variation in pH, temperature, and osmotic pressure (20 The principal terms contributing to lipid order are the rotational 24). It is not understood precisely how hopanoids are linked to freedom of motion and lateral packing of the lipids within the antibiotic resistance and tolerance to stress, but understanding plane of the bilayer. Lipid order is directly linked to essential the role of hopanoids in shaping membrane properties would membrane properties, including fluidity, permeability, lateral provide an important step toward bridging this gap. segregation, and the propensity for membranes to bind and in- tegrate other biomolecules (1). High plasma membrane order is Significance a fundamental property shared across the domains of modern PHYSIOLOGY life, and it may have been a key factor in the selective fitness of The function of the cell membrane as a barrier and a matrix for primitive life (2). However, sterols as the primary membrane- biochemical activity relies on the properties imparted by lipids. ordering lipid present a conundrum because sterol biosynthesis In eukaryotes, sterols are crucial for modulating the molecular requires molecular oxygen, but life was present on Earth at least order of membranes. Sterol ordering provides the basis for a billion years before cyanobacteria first enriched the atmo- membrane lateral segregation and promotes a fluid, mechanically sphere with oxygen (3, 4). Are there lipids that could have pro- robust plasma membrane. How do organisms that lack sterols moted membrane ordering before sterol synthesis emerged? determine membrane order? Hopanoids are bacterial membrane All three domains of life share isoprenoid synthesis pathways lipids that have been demonstrated to have sterol-like properties that give rise to a broad suite of structurally homologous lipids, in vitro. We now explore the distribution of hopanoids and their including sterols and hopanoids (e.g., diplopterol; Fig. S1). Hopa- effect on membranes in Methylobacterium extorquens.Wefind noids are some of the most ubiquitous cyclic isoprenoidal lipids in that hopanoids determine bacterial outer membrane order in a the sedimentary record, and they have been used as molecular manner analogous to sterol ordering in the eukaryotic plasma proxies for ancient microbial life (5). Importantly, hopanoid syn- membrane, and that their deletion impairs energy-dependent thesis does not require molecular oxygen, and hopanoids have been multidrug efflux. reported in sediments predating the enrichment of oxygen in Earth’s atmosphere (6, 7). Their discovery led to the proposal Author contributions: J.P.S. and K.S. designed research; J.P.S., D.G., A.S.B., T.J.L., O.L., and M.B. performed research; A.S.B. and O.L. contributed new reagents/analytic tools; J.P.S. that they might serve as sterol surrogates in bacteria (8), espe- analyzed data; and J.P.S. wrote the paper. cially because hopanoids and sterols share common structural Reviewers: D.D., Uni. Pablo de Olavide, Sevilla, Spain; and F.G., Unidad de Biofisica features and are cyclized by closely related enzymes (9, 10). Leioa Spain. One of the properties of sterols in eukaryotes is their ability to The authors declare no conflict of interest. interact preferentially with lipids such as sphingomyelin (SM) to 1To whom correspondence may be addressed. Email: [email protected] or simons@mpi- form liquid ordered phases in lipid membranes (11). This inter- cbg.de. action provides the mechanistic basis for the underlying intercon- This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. nectivity supporting membrane lateral segregation and promoting a 1073/pnas.1515607112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1515607112 PNAS | September 22, 2015 | vol. 112 | no. 38 | 11971–11976 Downloaded by guest on September 24, 2021 To understand how the ordering capacity of hopanoids con- Polar Apolar tributes to membrane functionality, it is necessary to examine IM OM IM OM membrane physiology in vivo. Here, we have explored the physiology of a hopanoid-producing organism from a physicochemical perspective by measuring the distribution of hopanoids and their effect on membranes in the gram-negative plant-associated bacterium Methylobacterium extorquens. We demonstrate that hopanoids

preferentially interact with outer membrane (OM) glycolipids to Diplopterols produce highly ordered membranes, thus revealing a convergent

strategy for the functional ordering of bacterial and eukaryotic Phospholipids surface membranes. We also identify impaired multidrug efflux as a phenotype of hopanoid deletion, which we propose accounts for sensitivity to chemical stresses. Our findings imply a link between membrane order and protein function in prokaryotes. They also suggest a possible lipid target to address bacterial multidrug resistance, further implicating isoprenoidal lipid biosynthesis as a bacterial Achilles heel (25, 26). hopanoids Polar

Results and Discussion Fig. 1. Hopanoids are enriched in the OM. A TLC plate shows phospholipid Hopanoids Are Localized in the OM and Determine Membrane Order. and hopanoid distributions in OM and inner membrane (IM) fractions. The The major hopanoids in M. extorquens are diplopterol and its major hopanoids detected are diplopterols (diplopterol and 2-methyl-dip- methylated derivative 2-methyl-diplopterol (Fig. S1), which together lopterol; Fig. S1) and polar hopanoids (BHT-GCE and BHT-CE; Fig. S1). make up most of the total hopanoid content (27). M. extorquens also Structures were confirmed by MS. produces extended side-chain polar hopanoids, known as bacter- iohopanepolyols (BHPs), predominately composed of a bacter- conclude that hopanoids, as well as cholesterol itself, have the po- iohopanetetrol cyclitol ether (BHT-CE) and its guanidine-modified tential to determine order in the outer bacterial membrane. derivative (BHT-GCE) (28) (Fig. S1). Trace quantities (<1% of total hopanoid) of BHT and adenosylhopane can also be detected, Hopanoids Are Predicted to Interact Preferentially with Lipid A in the but these trace quantities are most likely just biosynthetic inter- OM. It remained unclear which lipids hopanoids interact with in mediates. Using lipid thin-layer chromatography (TLC), we found the OM to determine order. In eukaryotes, sterols interact with that the major hopanoid diplopterol and its methylated derivative saturated sphingolipids to form a membrane that is highly ordered. comprised roughly 19 mol% of the total lipids (phospholipids, LPS, We demonstrated that hopanoids also interact with sphingolipids to and diplopterols), therefore representing a substantial component of form liquid ordered membranes (15). However, with a few excep- the cellular lipidome (Fig. S2). tions, bacteria do not produce sphingolipids (32). The hopanoid- Previous work in a closely related organism, Methylobacterium containing bacterial OM is characterized by an asymmetric bilayer organophilum, identified lipids with hopanoid-like properties (e.g., in which the inner leaflet contains mostly phospholipids and the nonsaponifiable) in the OM; however, the structural identity of outer leaflet contains lipid A, which is the conserved core of LPS these lipids was not confirmed (29). To determine the intracellular (33). In M. extorquens, virtually all of the phospholipids are un- distribution of hopanoids in M. extorquens, we purified membranes saturated, with roughly 90% containing double bonds in both acyl using gradient centrifugation and analyzed fractions by TLC, re- chains and the remaining 10% containing both saturated and un- vealing that the major hopanoids are highly enriched in the OM saturated acyl chains (Fig. S4). In contrast, lipid A has a highly fraction (Fig. 1). We confirmed hopanoid identity by MS frag- conserved structure that bears many features similar to sphingoli- mentation studies. To characterize the effect of hopanoids on the membrane, we pids, most notably the saturated acyl chains, amide-linked back- measured order in OM fractions purified from the WT and a bone, and hydroxylations (Fig. 3). We therefore hypothesized that hopanoid-deficient mutant of M. extorquens, ΔSHC (squalene– hopanoids interact with lipid A in a manner analogous to the in- hopene cyclase deletion mutant). This mutant was constructed teraction of sterols with sphingolipids to promote order in the OM. by knocking out the squalene-hopene cyclase gene essential for To examine which type of lipids hopanoids would preferen- hopanoid synthesis. OM fractions from WT and ΔSHC were tially interact with, we compared the favorability of interaction of analyzed using a lipophilic fluorescent probe, 6-dodecanoyl-2- diplopterol with lipid A, SM, and synthetic phospholipids with methylcarboxymethylaminonaphthalene (C-laurdan), that has been varying degree of unsaturation (Fig. 3) by measuring the Gibbs Δ ex extensively used in studying membrane packing and fluidity (30, 31). excess free energy of mixing ( G ) on a Langmuir trough. We We observed a significant difference in order between the WT made the same comparison using cholesterol in place of dip- Δ ex and ΔSHC OMs, with significantly lower membrane order in the lopterol. The G is a quantitative measure of the interaction hopanoid-deficient mutant (Fig. 2). This effect was confirmed with between lipids, with negative values indicating a favorable/ another fluorescent probe, Di-4-ANEPPDHQ (Di-4), that responds attractive interaction and positive values indicating a repulsive similarly to membrane ordering (31) (Fig. S3). To confirm that the interaction. We revealed a key difference between diplopterol decrease in order was due to the absence of hopanoids and not and cholesterol. Whereas cholesterol exhibits a favorable in- to other compositional changes in the membrane of the mutant, we teraction with phospholipids of varying degrees of unsaturation, tested the effect of depleting hopanoids from the WT OM and, diplopterol exhibits a repulsive interaction with unsaturated conversely, the effect of loading hopanoids into the ΔSHC OM phospholipids. Furthermore, the ΔGex values for the diplopterol- using methyl-β-cyclodextrin. Hopanoid depletion decreased the lipid A and cholesterol-SM are nearly identical and negative, order of the WT OM, and, analogously, loading the membranes thus confirming that interactions between both pairs of these with diplopterol increased the order of the ΔSHC OM, demon- lipids are favorable. This finding is consistent with our previous strating that perturbed membrane order is reversible and based results showing that diplopterol orders lipid A but does not order on ordering by diplopterol. Surprisingly, loading the hopanoid- unsaturated phospholipids (15). Thus, we show that diplopterol in- mutant OM with cholesterol also increased its order. Therefore, we teracts favorably only with saturated lipids and that the interactions

11972 | www.pnas.org/cgi/doi/10.1073/pnas.1515607112 Sáenz et al. Downloaded by guest on September 24, 2021 caused by hopanoid deletion is also linked to the impairment of 0.35 active processes, such as the energy-dependent transport of compounds out of the cell. To examine the extent of detergent sensitivity as an indicator 0.30 of membrane barrier function, we compared the influence of a nonionic detergent, Triton X-100 (TX-100), on WT and ΔSHC viability using a spot assay (Fig. 4). The lethal concentration of 0.25 TX-100 was more than 1,000-fold lower in ΔSHC than in WT. Importantly this phenotype is also one of the classic symptoms of an impaired multidrug transport system (37). 0.20 We examined whether hopanoid deletion affected membrane permeability and transport by monitoring the accumulation of the lipophilic dye 1-N-phenylnaphthylamine (NPN) in the cell. NPN 0.15 fluorescence is low while in solution and high when the dye is Membrane Order (c-laurdan GP) (c-laurdan Order Membrane WT WT ∆SHC ∆SHC ∆SHC present in the membrane environment (within the cell). WT Depleted +Dip +Chol M. extorquens showed an initial increase of fluorescence signal, followed by a decrease of signal, pointing toward the active efflux Fig. 2. Hopanoids determine OM order. Membrane order was determined of the dye, as typically observed in healthy bacteria (38) (Fig. 4). by C-laurdan GP of OM fractions from WT (gray), hopanoid-depleted WT Δ (WT Depleted), hopanoid-deficient ΔSHC (red), and ΔSHC loaded with dip- The SHC mutant, however, accumulated dye steadily until an lopterol (ΔSHC + Dip) or cholesterol (ΔSHC + Chol). Data represent average equilibrium level was reached. This result is typical for bacteria values from biological replicates (n = 3) measured at 30 °C and pH 7.4. One- that have been treated with ATP synthesis-blocking cyanide (38). tailed P values for WT vs. WT Depleted (P = 0.047) and WT vs. ΔSHC (P = The slope of NPN accumulation of the ΔSHC mutant is 2.2-fold 0.006) were made by an unpaired t test using Prism software (GraphPad). higher than for the WT, indicating higher susceptibility to dye penetration of the mutant OM. However, membrane permeability changes alone cannot account for the more than 1,000-fold re- of diplopterol with lipid A and cholesterol with SM are thermody- duction in tolerance to detergent that we observed (37). Therefore, namically analogous. we hypothesized that these observations pointed toward impaired multidrug transport in the hopanoid-deficient mutant. Deletion of Hopanoids Impairs Energy-Dependent Cellular Efflux. To test directly whether energy-dependent multidrug trans- Having established that hopanoids determine the order of the port was impaired in the hopanoid-deficient mutant, we used a OM, we examined the impact of hopanoid deletion on the standard multidrug efflux assay utilizing the dye Hoechst 33342 physiology of M. extorquens. The OM serves as the first barrier in (H33342). H33342 is a substrate for a broad range of bacterial gram-negative bacteria (33, 34). Reduced membrane order could multidrug transporters (39) and fluoresces only when bound to lead to lower resistance to bilayer-disrupting agents (e.g., deterg- nucleic acids. Thus, its fluorescence intensity is a proxy for cel- ents), membrane permeability, and impaired membrane protein lular uptake and efflux (40–42). We simultaneously provided function (e.g., passive and active transporters) (35). In line with this cells with H33342 and energy-depleted them using the ionophore reasoning, previous studies in other species of hopanoid-producing carbonyl cyanide 3-chlorophenylhydrazone (CCCP) in the ab- bacteria have shown that hopanoid deficiency sensitizes bacteria to sence of an utilizable carbon source (succinate) to inhibit energy- antimicrobial compounds and membrane-disrupting agents, such as dependent efflux. Upon addition of succinate and withdrawing bile salts, peptides, and detergents (20, 22–24, 36). These observa- CCCP, ATP production was restored and the WT cells were tions have generally been interpreted as evidence that hopanoids rapidly able to efflux H33342, reducing fluorescence to levels maintain low permeability and mechanical robustness of the OM. observed in untreated cells (Fig. 4). In contrast, upon restoring Δ However, it remained to be tested whether the chemical sensitivity ATP synthesis in SHC cells, H33342 fluorescence exhibited PHYSIOLOGY

KDO-Lipid A Sphingomyelin (SM) Di[3-deoxy-D-manno-octulosonyl]-lipid A N-stearoyl-D-erythro-sphingosylphosphorylcholine

1000 HO 1000 HO OH O N+ O HO HO -O O OH O- - O O P O O O O Cholesterol O- O H

O O HO NH O P O 0 -O O O 0 H O HO P HO O O OH O NH O- O NH O O O J/mol J/mol O Diplopterol O HO O HO ex O ex G G -1000 -1000 Unsaturation

-2000 -2000

DOPC POPC DPPC 1,2-dioleoyl-sn-glycero-3-phosphocholine 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine 1,2-dipalmitoyl-sn-glycero-3-phosphocholine + N + N + 1000 1000 1000 N O P O O O - O P O O O - P O O O - H O O H H O O O O O 0 O 0 0 O O O O O J/mol J/mol J/mol ex ex ex G G -1000 -1000 G -1000

-2000 -2000 -2000

Fig. 3. Diplopterol interacts preferentially with saturated lipids (e.g., lipid A). The ΔGex is shown for the interaction of diplopterol or cholesterol with lipid A, SM, and phospholipids with various degrees of unsaturation (indicated by a red arrow) from 1:2 mixtures (by molarity) of diplopterol/lipid or cholesterol/lipid. Positive values indicate a repulsive interaction, whereas negative values indicate an attractive/favorable interaction. The ΔGex values represent the average of replicates (n = 3) measured by Langmuir trough at room temperature and pH 7.4. Original data used for these calculations are included in Tables S1–S3.

Sáenz et al. PNAS | September 22, 2015 | vol. 112 | no. 38 | 11973 Downloaded by guest on September 24, 2021 A Sensitivity to non-ionic detergent Triton X-100 B 1-N-phenylnaphthylamine (NPN) uptake WT SHC Dilution Dilution 100 10-1 10-2 10-3 10-4 10-5 100 10-1 10-2 10-3 10-4 10-5 4 1006 Untreated

Cells treated with Triton X-100 for 1 hour (% w/v) 3 1006 SHC 0.0001 % 2 1006 0.001 %

0.01 % 1 1006 WT

0.1 % 420 nm at Emission NPN 0 1 % 100 Minutes 5 %

10 %

C Multidrug efflux activity assay with Hoechst 33342 WT SHC

1 1006 1 1006 Control 1 05 05 ATP depleted with CCCP 8 10 8 10 Added Glucose (non-utilizazble)

6 1005 6 1005 Experiment ATP depleted with CCCP 05 05 4 10 4 10 Added Succinate (utilizable)

2 1005 2 1005 Control 2 Untreated (no CCCP) H33342 Emission at 460 nm at H33342 Emission 0 0 Added Succinate (utilizable) 100 100 Minutes Minutes succinate / glucose succinate / glucose

Fig. 4. Energy-dependent multidrug transport is impaired by hopanoid deletion. (A) Detergent sensitivity of WT and ΔSHC was determined by spot assay. (B) Membrane permeability and transport were assessed by monitoring the accumulation of the lipophilic dye NPN in the cell (n = 3 biological replicates). (C) ATP-dependent multidrug transport activity was examined by H33342 fluorescence assay (n = 3 biological replicates).

much slower reduction, indicating impaired efflux. This result are diverse OM efflux proteins that interact with different types of shows that energy-dependent multidrug transport is deficient in transport systems to facilitate cellular efflux (52–55). The activity of the hopanoid ΔSHC mutant. an OM efflux protein might potentially be impaired by altered diffusivity or defective gating associated with a change in mem- Summary and Outlook brane order (52). For instance, it has been shown that the substrate We demonstrate that hopanoids can determine order in the affinity of proteins and receptors in the eukaryotic plasma mem- bacterial OM through their interaction with lipid A, analogous to brane is modulated by cholesterol ordering (35, 56). Given the the interaction of cholesterol with sphingolipids in eukaryotic architectural convergence between the bacterial OM and the plasma membranes. Hopanoids have been identified in the OM eukaryotic plasma membrane, similar mechanisms could account of diverse bacteria (29, 43–46), and two recent studies have reported for the functional role of hopanoids in bacteria. However, a com- a covalently linked hopanoid-lipid A compound in rhizobial plant- plete understanding of the mechanistic link between hopanoids associated bacteria (47, 48). These observations suggest that hopa- and multidrug transport represents a complex problem that re- noid-lipid A ordering may be widespread among hopanoid-pro- mains to be explored. Nonetheless, our results raise the interesting ducing bacteria. Interestingly, hopanoids have also been observed in possibility that targeting the synthesis of bacterial lipids that determine association with other specialized membranes, such as the thylakoid membrane order may provide a novel approach to addressing membranes in cyanobacteria and in the vesicle envelope of the root antibiotic resistance. nodule symbionts Frankia spp. (44, 45, 49, 50). The exact role of Our observations raise several key questions surrounding the hopanoids in these diverse bacterial membranes is not fully un- functional significance of hopanoids in membrane physiology derstood. Furthermore, bacteria that lack hopanoids may use other, and the role of the OM in bacterial cellular organization. It will possibly unknown, ordering lipids in their membranes. For instance, be important to understand how lipid order influences the dy- it has been shown that Borrelia burgdorferi, the causative agent of namics of OM proteins, such as their structure, insertion into the Lyme disease, can maintain membrane order by incorporating membrane, lateral diffusion, and dimerization. Furthermore, it cholesterol from its host (51). It has also been proposed that pro- remains unknown if the OM is capable of lateral compartmental- teins could serve a role in maintaining membrane order in bacteria ization as has been proposed to occur in other bacterial mem- that lack ordering lipids (2). Thus, although our understanding of branes (57). Do hopanoids promote such membrane organization? the diverse mechanisms that bacteria use to modulate molecular Recently, it was demonstrated that the bacterial actin homolog order in the membrane remains incomplete, we reveal hopanoid/ MreB contributes to the organization of the membrane (58). Are sterol ordering of saturated lipids (e.g., lipid A, sphingolipids) as there links between hopanoids, OM compartmentalization, and a strategy for achieving ordered cell-surface membranes that are bacterial cytoskeletal analogs? Our work provides a foundation for found from bacteria to eukaryotes. exploring these fundamental questions by introducing a bacterial Our data show that multidrug efflux in M. extorquens is de- system for studying the effect of lipid ordering on membrane pendent on hopanoids. This finding may explain the basis for function in vivo. previous observations linking hopanoids to antibiotic resistance Materials and Methods in the hopanoid-producing pathogen Burkholderia (22–24). One Materials. SM, kdo-lipid A, DOPC (1,2-dioleoyl-sn-glycero-3-phosphocholine), explanation for impaired efflux could be that reduced membrane DPPC (1,2-dipalmitoyl-sn-glycero-3-phosphocholine), POPC (1-palmitoyl-2-oleoyl- order affects the functionality of some component of the multidrug sn-glycero-3-phosphocholine), and cholesterol were purchased from Avanti transport system. However, it is also possible that hopanoids in- Polar Lipids. Diplopterol was purchased from Chiron AS. C-laurdan was a gift teract directly with certain proteins, modulating their action. There from B. R. Cho Korea University, Seoul, South Korea. Stock concentrations

11974 | www.pnas.org/cgi/doi/10.1073/pnas.1515607112 Sáenz et al. Downloaded by guest on September 24, 2021 of lipids were measured by phosphate assay. Cholesterol and diplopterol GP = ðICh1 − ICh2Þ=ICh1 + ICh2. were weighed out on a precision balance and solubilized in chloroform/ methanol (2:1). Methods for data presented in Figs. S1–S8 are described in SI Materials Monolayers and ΔGex Calculations. Isotherms of monolayers of synthetic and and Methods. purified lipids prepared as described previously (63) were recorded using a 70-cm2 Teflon Langmuir trough fitted with a motorized compression barrier Media, Growth Conditions. All Methylobacterium strains were grown at 30 °C equipped with a pressure sensor and Wilhelmy plate (Nima Technnology). The (unless otherwise stated) in a minimal medium (59), referred to hereafter as mean molecular areas for each mixture were estimated from the averages of Hypho medium, with 15 mM succinate as the sole carbon source. Escherichia isotherms from three monolayers that were prepared independently. The coli strains were grown at 37 °C on solid LB. Suitable antibiotics were used for ΔGex was calculated by integrating the areas of lipid mixtures over pressures selection: 50 μg/mL ampicillin, 20 μg/mL chloramphenicol, 50 μg/mL kanamycin, Π = 5, 10, 15, 20, and 25 mN/m according to Grzybek et al. (63), and as de- 50 μg/mL rifamycin, 35 μg/mL streptomycin, and 10 μg/mL tetracycline. scribed in detail in SI Materials and Methods.

Construction of Plasmids and Generation of Hopanoid-Deficient Mutant ΔSHC. Detergent Sensitivity Spot Assay. A spot assay was used to determine the A hopanoid-deficient mutant strain of M. extorquens PA1 was constructed sensitivity of the WT and ΔSHC to detergent. Cells in exponential growth using a modified procedure described previously (60). Details are described adjusted to an OD of 0.2 were treated with varying concentrations of TX- in SI Materials and Methods. Regions of the chromosome, including the 600 deleted sequence, were amplified by PCR and sequenced to confirm gen- 100 for 1 h. Cells were washed twice with medium to remove TX-100. After μ eration of the desired, unmarked deletion. The deletion of hopanoid syn- thoroughly mixing, 5 L of each dilution was applied on a Hypho agar plate thesis was further confirmed by the absence of hopanoids in a total lipid in a single drop, starting from the lowest to the highest dilution. The plates extract visualized by TLC (Fig. S6). were incubated for 2–3 d at 30 °C.

– Membrane Separation from M. extorquens WT and ΔSHC. A membrane sepa- NPN Uptake. M. extorquens cultures were grown overnight to an OD600 of 0.2 ration protocol was adapted from a protocol previously established for 0.3 and harvested by centrifugation (5,000 × g for 10 min). The pellets were

Methylobacterium (29). The method is described in detail in SI Materials washed twice with medium and adjusted to a final OD600 of 0.5 (8,000 × g for and Methods. 2 min). One hundred eighty microliters per well of cells was transferred to a 96-well plate (black, clear, flat bottom; Sarstedt). The emission of NPN was recorded Hopanoid Distribution in OM Fractions. OM fractions were extracted using the using a plate reader [Perkins Elmer Envison; filters (wavelength/bandwidth): procedure of Bligh and Dyer (61). The resulting extracts were loaded on a excitation = 340/25 nm, emission = 450/8 nm]. Readouts were taken every silica gel plate (HPTLC Silica gel 60 with concentrating zone; Merck) and minute, with 55 s of shaking between readouts. The background signal was resolved with chloroform/acetic acid/methanol/water [80:15:2:4 (vol/vol/vol/ measured for 5 min before NPN was added to a final concentration of 5 μM(5μL vol)] for phospholipids and polar hopanoids or with chloroform for dip- of 185 μM NPN solution) per well. The uptake of dye was recorded for 90 min by lopterols. The identity of the polar hopanoids (BHT-CE and BHT-GCE) and measuring its emission under the same conditions. The entire assay was per- the diplopterols (diplopterol and 2-methyl-diplopterol) was confirmed by formed at room temperature. Succinate was present at all times at 15 mM. MS fragmentation of bands that were scraped from the plates and reex- tracted from the silica gel using the procedure of Bligh and Dyer (61), as described in SI Materials and Methods. H33342 ATP-Dependent Efflux Assay. M. extorquens cultures were grown overnight to an OD600 of 0.2–0.3 in Hypho medium at 30 °C. Cells were har- × Methyl-β-Cyclodextrin Depletion and Loading of Membrane Fractions. The vested by centrifugation (5,000 g for 10 min) and washed once with medium depletion and loading of hopanoids into membranes were achieved using a lacking an utilizable carbon source (succinate). All following wash steps and method utilizing methyl-β-cyclodextrin (MβCD) (62). For depletion, membrane resuspensions were performed using succinate-free Hypho medium containing amounts were adjusted using the intensity of a lipid bilayer scattering peak at 5 μMH33342(8,000× g for 2 min). The OD600 was adjusted to 1.0 before μ 425 nm (excitation wavelength λex = 385nm)(2)to1mMlipidandthende- freshly prepared CCCP in DMSO was added to a final concentration of 100 M pleted with 10 mM MβCD for 2 h on ice. For loading, a loading solution of 10:1 [∼1% (vol/vol) DMSO] to abolish ATP synthesis. The 1% DMSO did not inhibit (by molarity) MβCD/cholesterol or diplopterol was prepared by equilibrating growth of the WT or ΔSHC mutant. The mixtures were incubated in darkness β 20 mM M CD with 2 mM cholesterol or diplopterol overnight at 30 °C. Membrane for 1 h, washed twice to remove CCCP, and then resuspended to an OD600 of fraction amounts were then adjusted to ca. 2 mM lipid and loaded with 20 mM

1.0. One hundred eighty microliters of the suspension was transferred to wells PHYSIOLOGY β β 10:1 M CD/diplopterol or 10:1 M CD/cholesterol and gently shaken for 2 h at 4 °C. of a 96-well plate. The initial uptake/equilibrium of H33342 was recorded us- To remove MβCD, the membrane fractions were pelleted by centrifugation at ing a plate reader (Envison) for 45 min [filters (wavelength/ bandwidth): ex- 70,000 × g for 90 min and then washed twice with buffer M; they were eventually citation = 340/25 nm, emission = 405/8 nm], and readings were taken for 45 pelleted and resuspended in sterile Hypho medium for subsequent analysis. cycles (every minute), with 55 s of shaking between readouts. Afterward, ei- ther glucose or succinate was added to a final concentration of 20 mM (5 μLof C-Laurdan and Di-4 Spectroscopy of Membrane Fractions. Membrane fractions 740 mM solution) per well and the change in H33342 emission was recorded of WT and ΔSHC OMs from three independent cultures were prepared as described above. The membrane fractions were subjected to sonication for 5 min for an additional 60 min (60 cycles, 55-s delay with shaking). Glucose was used to promote formation of unilamellar membranes. The presence of bilayers and as a negative control because M. extorquens is unable to utilize it. All steps of estimation of membrane amount were assessed in unlabeled membrane frac- the assay were performed at room temperature. tions by the intensity of lipid bilayer scattering. Membrane fractions adjusted to ca. 200 μM lipid were stained with 400 nM C-laurdan or Di-4 (0.2 mol%) and ACKNOWLEDGMENTS. We thank Michal Surma, Robert Ernst, Ilya Levental, incubated on ice for 20 min. Labeled fractions were then equilibrated at 30 °C Michal Grzybek, Unal Coskun, Helena Jambor, and Jeff Woodruff for helpful for 10 min. Spectra were recorded at a resolution of 1 nm on a Fluoromax-3 comments and discussions. We also thank Prof. Martin Pos for advice and help with implementing a multidrug efflux assay. A.S.B. thanks Chris Marx and Ann fluorescence spectrometer (Horriba) at a constant temperature of 30 °C. Ex- Pearson for support and useful discussions. This material is based upon work citation of C-laurdan was 385 nm, and excitation of Di-4 was 497 nm. The supported by the US National Science Foundation (Grant EAR-1024723) and general polarization (GP) values were calculated from two emission bands: International Research Fellowship Program (Grant 1064754), the Alexander 400–460 nm (Ch1)and470–530 nm (Ch2) for C-laurdan and 525–580 nm (Ch1) von Humboldt Foundation, the Simons Foundation (Simons Collaboration on and 655–750 nm (Ch2) for Di-4, according to the following equation (30, 31): the Origins of Life Postdoctoral Fellowship), and the Max Planck Society.

1. van Meer G, Voelker DR, Feigenson GW (2008) Membrane lipids: Where they are and 5. Ourisson G, Albrecht P (1992) Hopanoids. 1. Geohopanoids: The most abundant how they behave. Nat Rev Mol Cell Biol 9(2):112–124. natural products on earth? Acc Chem Res 25(9):398–402. 2. Kaiser H-J, et al. (2011) Molecular convergence of bacterial and eukaryotic surface 6. Briggs DEG, Summons RE (2014) Ancient biomolecules: Their origins, fossilization, and order. J Biol Chem 286(47):40631–40637. role in revealing the history of life. BioEssays 36(5):482–490. 3. Ohtomo Y, Kakegawa T, Ishida A, Nagase T, Rosing MT (2014) Evidence for biogenic 7. Kannenberg EL, Poralla K (1999) Hopanoid biosynthesis and function in bacteria. graphite in early Archaean Isua metasedimentary rocks. Nat Geosci 7(1):25–28. Naturwissenschaften 86(4):168–176. 4. Brocks JJ, Logan GA, Buick R, Summons RE (1999) Archean molecular fossils and the 8. Ourisson G, Rohmer M, Poralla K (1987) Prokaryotic hopanoids and other poly- early rise of eukaryotes. Science 285(5430):1033–1036. terpenoid sterol surrogates. Annu Rev Microbiol 41(1):301–333.

Sáenz et al. PNAS | September 22, 2015 | vol. 112 | no. 38 | 11975 Downloaded by guest on September 24, 2021 9. Wendt KU, Poralla K, Schulz GE (1997) Structure and function of a squalene cyclase. 40. Shapiro AB, Ling V (1995) Reconstitution of drug transport by purified P-glycoprotein. Science 277(5333):1811–1815. J Biol Chem 270(27):16167–16175. 10. Frickey T, Kannenberg E (2009) Phylogenetic analysis of the triterpene cyclase protein 41. Shapiro AB, Corder AB, Ling V (1997) P-glycoprotein-mediated Hoechst 33342 trans- family in prokaryotes and eukaryotes suggests bidirectional lateral gene transfer. port out of the lipid bilayer. Eur J Biochem 250(1):115–121. Environ Microbiol 11(5):1224–1241. 42. Kim M, et al. (2002) The multidrug resistance transporter ABCG2 (breast cancer re- 11. Mouritsen OG, Zuckermann MJ (2004) What’s so special about cholesterol? Lipids sistance protein 1) effluxes Hoechst 33342 and is overexpressed in hematopoietic 39(11):1101–1113. stem cells. Clin Cancer Res 8(1):22–28. 12. Lingwood D, Simons K (2010) Lipid rafts as a membrane-organizing principle. Science 43. Doughty DM, et al. (2011) The RND-family transporter, HpnN, is required for hopa- 327(5961):46–50. noid localization to the outer membrane of Rhodopseudomonas palustris TIE-1. Proc 13. Kannenberg E, Blume A, McElhaney RN, Poralla K (1983) Monolayer and calorimetric Natl Acad Sci USA 108(45):E1045–E1051. studies of phosphatidylcholines containing branched-chain fatty-acids and of their 44. Doughty DM, Hunter RC, Summons RE, Newman DK (2009) 2-Methylhopanoids are interactions with cholesterol and with a bacterial hopanoid in model membranes. maximally produced in akinetes of Nostoc punctiforme: Geobiological implications. Biochim Biophys Acta 733(1):111–116. Geobiology 7(5):524–532. 14. Poralla K, Kannenberg E, Blume A (1980) A glycolipid containing hopane isolated 45. Simonin P, Jürgens UJ, Rohmer M (1996) Bacterial triterpenoids of the hopane series from the acidophilic, thermophilic Bacillus acidocaldarius, has a cholesterol-like from the prochlorophyte Prochlorothrix hollandica and their intracellular localiza- – function in membranes. FEBS Lett 113(1):107 110. tion. Eur J Biochem 241(3):865–871. 15. Sáenz JP, Sezgin E, Schwille P, Simons K (2012) Functional convergence of hopanoids 46. Jahnke LL, Stan-Lotter H, Kato K, Hochstein LI (1992) Presence of methyl sterol and – and sterols in membrane ordering. Proc Natl Acad Sci USA 109(35):14236 14240. bacteriohopanepolyol in an outer-membrane preparation from Methylococcus cap- 16. Wu C-H, Bialecka-Fornal M, Newman DK (2015) Methylation at the C-2 position of ho- sulatus (Bath). J Gen Microbiol 138(8):1759–1766. panoids increases rigidity in native bacterial membranes. eLife, 10.7554/eLife.05663. 47. Silipo A, et al. (2014) Covalently linked hopanoid-lipid A improves outer-membrane 17. Poger D, Mark AE (2013) The relative effect of sterols and hopanoids on lipid bilayers: resistance of a Bradyrhizobium symbiont of legumes. Nat Commun 5:5106. – When comparable is not identical. J Phys Chem B 117(50):16129 16140. 48. Komaniecka I, et al. (2014) Occurrence of an unusual hopanoid-containing lipid A among 18. Rohmer M, Bouvier-Nave P, Ourisson G (1984) Distribution of hopanoid triterpenes in lipopolysaccharides from Bradyrhizobium species. JBiolChem289(51):35644–35655. – prokaryotes. J Gen Microbiol 130:1137 1150. 49. Rexroth S, et al. (2011) The plasma membrane of the cyanobacterium Gloeobacter 19. Pearson A, Flood Page SR, Jorgenson TL, Fischer WW, Higgins MB (2007) Novel ho- violaceus contains segregated bioenergetic domains. Plant Cell 23(6):2379–2390. – panoid cyclases from the environment. Environ Microbiol 9(9):2175 2188. 50. Berry AM, et al. (1993) Hopanoid lipids compose the Frankia vesicle envelope, pre- 20. Welander PV, et al. (2009) Hopanoids play a role in membrane integrity and pH sumptive barrier of oxygen diffusion to nitrogenase. Proc Natl Acad Sci USA 90(13): homeostasis in Rhodopseudomonas palustris TIE-1. J Bacteriol 191(19):6145–6156. 6091–6094. 21. Kulkarni G, Wu C-H, Newman DK (2013) The general stress response factor EcfG 51. LaRocca TJ, et al. (2013) Proving lipid rafts exist: membrane domains in the pro- regulates expression of the C-2 hopanoid methylase HpnP in Rhodopseudomonas karyote Borrelia burgdorferi have the same properties as eukaryotic lipid rafts. PLoS palustris TIE-1. J Bacteriol 195(11):2490–2498. Pathog 9(5):e1003353. 22. Malott RJ, Steen-Kinnaird BR, Lee TD, Speert DP (2012) Identification of hopanoid 52. Zgurskaya HI, Krishnamoorthy G, Ntreh A, Lu S (2011) Mechanism and Function of the biosynthesis genes involved in polymyxin resistance in Burkholderia multivorans. Outer Membrane Channel TolC in Multidrug Resistance and Physiology of Enter- Antimicrob Agents Chemother 56(1):464–471. obacteria. Front Microbiol 2:189. 23. Malott RJ, et al. (2014) Fosmidomycin decreases membrane hopanoids and potenti- 53. Oswald C, Pos KM (2011) Drug resistance: A periplasmic ménage à trois. Chem Biol ates the effects of colistin on Burkholderia multivorans clinical isolates. Antimicrob 18(4):405–407. Agents Chemother 58(9):5211–5219. 54. Koronakis V, Eswaran J, Hughes C (2004) Structure and function of TolC: The bacterial 24. Schmerk CL, Bernards MA, Valvano MA (2011) Hopanoid production is required for exit duct for proteins and drugs. Annu Rev Biochem 73:467–489. low-pH tolerance, antimicrobial resistance, and motility in Burkholderia cenocepacia. 55. Du D, et al. (2014) Structure of the AcrAB-TolC multidrug efflux pump. Nature J Bacteriol 193(23):6712–6723. 509(7501):512–515. 25. López D, Kolter R (2010) Functional microdomains in bacterial membranes. Genes Dev 56. Lingwood D, et al. (2011) Cholesterol modulates glycolipid conformation and re- 24(17):1893–1902. ceptor activity. Nat Chem Biol 7(5):260–262. 26. Lorenzen W, Ring MW, Schwär G, Bode HB (2009) Isoprenoids are essential for 57. Bramkamp M, Lopez D (2015) Exploring the existence of lipid rafts in bacteria. fruiting body formation in Myxococcus xanthus. J Bacteriol 191(18):5849–5853. – 27. Bradley A, Pearson A, Sáenz J, Marx C (2010) Adenosylhopane: The first intermediate Microbiol Mol Biol Rev 79(1):81 100. in hopanoid side chain biosynthesis. Org Geochem 41(10):1075–1081. 58. Strahl H, Bürmann F, Hamoen LW (2014) The actin homologue MreB organizes the 28. Renoux JM, Rohmer M (1985) Prokaryotic triterpenoids. New bacteriohopanetetrol bacterial cell membrane. Nat Commun 5:3442. cyclitol ethers from the methylotrophic bacterium Methylobacterium-organophilum. 59. Lee MC, Chou HH, Marx CJ (2009) Asymmetric, bimodal trade-offs during adaptation – Eur J Biochem 151(2):405–410. of Methylobacterium to distinct growth substrates. Evolution 63(11):2816 2830. 29. Hancock IC, Williams KM (1986) The outer membrane of Methylobacterium organo- 60. Marx CJ (2008) Development of a broad-host-range sacB-based vector for unmarked philum. Microbiology 132(3):599–610. allelic exchange. BMC Res Notes 1(1):1. 30. Parasassi T, De Stasio G, d’Ubaldo A, Gratton E (1990) Phase fluctuation in phos- 61. Bligh EG, Dyer WJ (1959) A rapid method of total lipid extraction and purification. – pholipid membranes revealed by Laurdan fluorescence. Biophys J 57(6):1179–1186. Can J Biochem Physiol 37(8):911 917. 31. Sezgin E, Sadowski T, Simons K (2014) Measuring lipid packing of model and cellular 62. Christian AE, Haynes MP, Phillips MC, Rothblat GH (1997) Use of cyclodextrins for – membranes with environment sensitive probes. Langmuir 30(27):8160–8166. manipulating cellular cholesterol content. J Lipid Res 38(11):2264 2272. Ł ł 32. Olsen I, Jantzen E (2001) Sphingolipids in bacteria and fungi. Anaerobe 7(2):103–112. 63. Grzybek M, Kubiak J, ach A, Przyby o M, Sikorski AF (2009) A raft-associated species 33. Nikaido H, Vaara M (1985) Molecular basis of bacterial outer membrane permeability. of phosphatidylethanolamine interacts with cholesterol comparably to sphingomye- Microbiol Rev 49(1):1–32. lin. A Langmuir-Blodgett monolayer study. PLoS One 4(3):e5053. 34. Nikaido H (2003) Molecular basis of bacterial outer membrane permeability revisited. 64. Karkhanis YD, Zeltner JY, Jackson JJ, Carlo DJ (1978) A new and improved microassay Microbiol Mol Biol Rev 67(4):593–656. to determine 2-keto-3-deoxyoctonate in lipopolysaccharide of Gram-negative bac- 35. Coskun Ü, Grzybek M, Drechsel D, Simons K (2011) Regulation of human EGF receptor teria. Anal Biochem 85(2):595–601. by lipids. Proc Natl Acad Sci USA 108(22):9044–9048. 65. Schindelin J, et al. (2012) Fiji: An open-source platform for biological-image analysis. 36. Welander PV, et al. (2012) Identification and characterization of Rhodopseudomonas Nat Methods 9(7):676–682. palustris TIE-1 hopanoid biosynthesis mutants. Geobiology 10(2):163–177. 66. Ejsing CS, et al. (2009) Global analysis of the yeast lipidome by quantitative shotgun 37. Nikaido H (1996) Multidrug efflux pumps of gram-negative bacteria. J Bacteriol mass spectrometry. Proc Natl Acad Sci USA 106(7):2136–2141. 178(20):5853–5859. 67. Herzog R, et al. (2012) LipidXplorer: A software for consensual cross-platform lipidomics. 38. Loh B, Grant C, Hancock RE (1984) Use of the fluorescent probe 1-N-phenylnaphthyl- PLoS One 7(1):e29851. amine to study the interactions of aminoglycoside antibiotics with the outer membrane 68. Gibson DG (2011) Enzymatic assembly of overlapping DNA fragments. Methods in of Pseudomonas aeruginosa. Antimicrob Agents Chemother 26(4):546–551. Enzymology 498:349–361. 39. van den Berg van Saparoea HB, Lubelski J, van Merkerk R, Mazurkiewicz PS, Driessen AJM 69. Delaney NF, et al. (2013) Development of an optimized medium, strain and high- (2005) Proton motive force-dependent Hoechst 33342 transport by the ABC transporter throughput culturing methods for Methylobacterium extorquens. PLoS One 8(4): LmrA of Lactococcus lactis. Biochemistry 44(51):16931–16938. e62957.

11976 | www.pnas.org/cgi/doi/10.1073/pnas.1515607112 Sáenz et al. Downloaded by guest on September 24, 2021