Atom-scale depth localization of biologically important chemical elements in molecular layers

Edited by David A. Weitz, Harvard University, Cambridge, MA, and approved July 5, 2016 (received for review March 18, 2016)
August 8, 2016
113 (34) 9521-9526

Significance

Interfacial molecular layers are a major component of all biological matter and also play key roles in most biotechnological applications. The understanding of important biological processes involving molecular layers typically relies on detailed structural insight. We demonstrate that standing-wave X-ray fluorescence enables the label-free, element-specific structural investigation of molecular layers at atom-scale resolution perpendicular to the interface. The present work establishes a promising approach to the comprehensive structural analysis of complex biological and biotechnologically relevant surfaces.

Abstract

In nature, biomolecules are often organized as functional thin layers in interfacial architectures, the most prominent examples being biological membranes. Biomolecular layers play also important roles in context with biotechnological surfaces, for instance, when they are the result of adsorption processes. For the understanding of many biological or biotechnologically relevant phenomena, detailed structural insight into the involved biomolecular layers is required. Here, we use standing-wave X-ray fluorescence (SWXF) to localize chemical elements in solid-supported lipid and protein layers with near-Ångstrom precision. The technique complements traditional specular reflectometry experiments that merely yield the layers’ global density profiles. While earlier work mostly focused on relatively heavy elements, typically metal ions, we show that it is also possible to determine the position of the comparatively light elements S and P, which are found in the most abundant classes of biomolecules and are therefore particularly important. With that, we overcome the need of artificial heavy atom labels, the main obstacle to a broader application of high-resolution SWXF in the fields of biology and soft matter. This work may thus constitute the basis for the label-free, element-specific structural investigation of complex biomolecular layers and biological surfaces.
Nanometric biomolecular layers in interfacial geometries are key components of biological matter. Biological membranes, for example, are 2D molecular structures composed mainly of lipids and proteins in an aqueous environment (1, 2). Interfacial biomolecular layers are also important from a biotechnological perspective, where surfaces are often designed to selectively promote or prevent the adsorption of bio(macro)molecules from solution (3). To investigate the diverse behavior of biomolecules at interfaces, well-defined planar model systems of biological and biotechnologically relevant interfaces have been established (4, 5). Processes involving biomolecules at interfaces are typically accompanied by a spatial reorganization of molecules, changes in molecular conformations, or the adsorption of molecules in a chemically heterogeneous environment (6). Structural insight on the nanometer scale is therefore of paramount importance.
X-ray and neutron scattering are uniquely suited for the structural investigation of biomolecular systems. Specular reflectometry reveals matter density profiles perpendicular to a planar interface. However, it is generally difficult to elucidate molecular conformations and elemental distributions from such “global” density profiles. In contrast, standing-wave X-ray fluorescence (SWXF) allows determining element-specific density profiles across an interface. The SWXF technique is based on the element-characteristic fluorescence induced by a standing X-ray wave. The latter is formed by interference of the incident and reflected waves and its shape depends on the angle of incidence of the X-ray beam with respect to the interface. The angle-dependent fluorescence intensity thus contains information on the interfacial element distribution (7). This principle has been exploited for the study of interfacial phenomena involving solid, liquid, and gas phases (816). High resolution perpendicular to the interface is implied by SWXF experiments in Bragg reflection configuration, in which case planar, nanometric multilayers are used to create strongly modulated standing waves above the terminal surface close to the Bragg condition (9, 10). However, such high-resolution studies have thus far dealt only with the fluorescence of comparatively heavy elements. Even models of biological surfaces have been labeled, in one way or another, with heavy elements that are not naturally abundant in biological matter (7). Light chemical elements have so far only investigated in a grazing-incidence configuration with low spatial resolution (15, 16).
In the present work, we demonstrate that light, biologically relevant chemical elements such as P and S can be localized also in Bragg reflection configuration and thus with atom-scale resolution. This approach enables the label-free, high-resolution, element-specific structural investigation of biomolecular layers. We work with representatives of the three most important classes of biomolecules: lipids, saccharides in the form of glycolipids, and proteins. Lipids with phosphatidylcholine (PC) headgroups are among the most abundant lipid classes in eukaryotic cells and the dominant class in animals (17). Glycolipids bearing a sulfate group, such as the sulfoglycolipid 3-O-sulfo-d-galactosyl-β1–1′-N-heptadecanoyl-d-erythro-sphingosine (SGS), are found in eukaryotic cell membranes, especially in the nerve system and in photosynthetic membranes (18, 19). Element-specific studies as presented here can thus reveal the structural details of biological multicomponent membranes. The protein human serum albumin (HSA), on the other hand, is the most abundant protein in the human blood serum and plays an important role in protein adsorption to biomaterial surfaces and in the course of foreign body reactions (20). The structural investigation of HSA interaction with surfaces is therefore important for the rational design of protein resistant and thus biocompatible surfaces (21).

Results and Discussion

Fig. 1 A and B shows the chemical structures of the molecules of which the studied solid-supported layers are composed: SGS, the phospholipid 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC), the lipopolymer 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-5000] (PEG-lipid), and HSA. SGS (Fig. 1A, Top) roughly consists of a sulfated galactose headgroup connected to two hydrophobic fatty acid tails via a compact linker region. DSPC (Fig. 1A, Middle) has a similar architecture, but the headgroup is formed by phosphocholine comprising one P atom. PEG-lipid (Fig. 1A, Bottom) is similar to DSPC but the tertiary amine of choline is replaced by an amide-bonded PEG chain comprising 114 monomers. HSA (Fig. 1B) is a globular protein with about 600 amino acids and mass mw = 67 kDa and comprises 41 S atoms contained in its cysteine and methionine amino acids (www.uniprot.org/uniprot/P02768). The center of mass position of the S distribution approximately coincides with the one of the whole molecule (22).
Fig. 1.
(A) Chemical structures of SGS (Top), DSPC (Middle), and PEG-lipid (Bottom). Hydrocarbon tails (see main text) are denoted with R. (B) Structure of the protein HSA. (C) Setup and measurement geometry (top view) of the SWXF experiments. The X-ray beam illuminates the sample surface at an incident angle θ. At each incident angle, fluorescence spectra are recorded with an energy-sensitive fluorescence detector oriented perpendicular to the incident beam and close to the sample surface.
The planar multilayers used for the SWXF experiments had 20 repetitions of Al (10 nm)/Ni (10 nm) alternating layers on top of sapphire single crystal wafers, so that the terminal layer is Al, which forms a thin layer of amorphous aluminum oxide at the outer surface. The measurement geometry is schematically illustrated in Fig. 1C. The X-rays via photoelectric ionization induce characteristic fluorescence of the target elements. Fluorescence spectra (see Fig. S1 for a representative example) were measured for various incident angles θ in angular scans around the multilayer Bragg angle θB = 0.645°. For each θ, the amplitudes of the fluorescence peaks were determined. The angle-dependent fluorescence intensity of target element j, Ij(θ), is proportional to the spatial integral over the product of the elemental density profile perpendicular to the interface, ρj(z), and the known angle-dependent standing wave (SW) intensity Φ(θ, z)
Ij(θ)=AΦ(θ,z)ρj(z)dz.
[1]
Eq. 1 thus allows reconstructing ρj(z) from the angle-dependent characteristic fluorescence. A is a prefactor determined by fluorescence yield and detection efficiency, and in general also depends on the incident angle (7). In Eq. 1, we safely neglected the depth dependence of the fluorescence attenuation, because in the present study the target elements are confined in nanometric layers. For a given incident angle θ, Φ(θ, z) follows from the interfacial scattering length density (SLD) profile and can be computed from a suitable slab model representation of the SLD via the phase-correct summation of all reflected and transmitted partial waves (23), as has been described previously (14). As shown in the inset of Fig. 2, close to the Bragg condition of the periodic structures, strong beam reflection occurs, giving rise to a strongly modulated SW with period Λ = d/n, where d is the multilayer repeat and n is the Bragg peak order (n = 2 in the present work). The main panel of Fig. 2 shows SWs (solid, dashed, and dotted lines) computed for three different angles around θB indicated with vertical straight lines in the inset. As θ is increased through the strong reflection condition, the nodes and antinodes of the standing wave gradually move toward the interface by half a SW period (Fig. 2) and probe structures with high spatial resolution. With Φ(θ, z) at hand, the angle-dependent fluorescence intensities of the target elements, Ij(θ), were then modeled according to Eq. 1 using a suitable parameterization of ρj(z)
ρj(z)=ρmaxe(zzj)2/(2σj2).
[2]
In this Gaussian representation, the amplitude ρmax, the center position zj, and the width σj are adjustable fitting parameters. The weak angle dependence of the prefactor A in Eq. 1 arises due to geometrical effects and was approximated linearly in a narrow interval around the Bragg angle, as A(θ) = 1 + (θθB)C with an adjustable fitting parameter C.
Fig. 2.
(Inset) Measured angle-dependent X-ray reflectivity R(θ) of a multilayered planar solid (symbols) around the Bragg angle θB and modeled reflectivity curve (solid line). (Main panel) X-ray standing wave (solid, dashed, and dotted lines) above the terminal surface computed for three different angles around θB indicated with vertical straight lines of the same line styles as in the Inset.
Fig. S1.
(Left) X-ray fluorescence spectrum of a solid-supported double SGS monolayer (Fig. 4) exhibiting distinct fluorescence peaks of the chemical elements of interest S and K, as well as of chemical elements found in the multilayer substrates. (Right) Comparison between sample and reference spectra in the region of interest, highlighting the S and K fluorescence from the double SGS monolayer itself.
Column B of Figs. 3 and 4 and column A of Fig. 5 show angle-dependent Kα fluorescence intensities, Ij(θ), from P, S, and K atoms in the biomolecular layers studied together with the simulated intensities according to the best matching model parameters. It is seen that the shapes of the experimental data are well captured by the simulated intensities. The well-defined positions of the minima and/or maxima in the curves, together with the intensity levels on the two sides of these extrema, encode the center position zj. The associated uncertainty was estimated as ±2 Å (Supporting Information and Fig. S2). The width σj is primarily encoded in the relative amplitude of the modulation of Ij(θ) with respect to the “baseline” further away from the Bragg condition. When σj is much smaller than Λ, in practice σmin ∼ Λ/10 (Supporting Information and Fig. S3), then the amplitude of this modulation saturates and is no longer sensitive to the precise value of σj. In the present work, d ∼ 200 Å and with n = 2, we obtain Λ ∼ 100 Å and σmin ∼ 10 Å. Finally, ρmax merely acts as a scale factor. The best-matching values of zj and σj are summarized in Table 1.
Fig. 3.
Experimental results on lipid single monolayers on bare Al oxide. An SGS monolayer (Top), a DSPC monolayer (Middle), and a DSPC/PEG-lipid mixed monolayer (Bottom). (A) Chemical details highlighting the elements of interest P and S. (B) Measured angle-dependent P and S fluorescence intensities (symbols) together with the simulated intensities (solid lines) corresponding to the best-matching models. Vertical dashed lines indicate the Bragg condition. (C) Sketches of the layers studied. The axis perpendicular to the sample surface is denoted with z. (D) Illustration of the best-matching P and S distributions according to Eq. 2, where the width parameter σj was set equal to the topographic surface roughness σtop.
Fig. 4.
Experimental results on an SGS double monolayer on top of an OTS-functionalized substrate. (A) Chemical details highlighting the elements of interest S and K. (B) Measured angle-dependent S (circles) and K (squares) fluorescence intensities together with the simulated intensities (solid lines) corresponding to the best-matching models. Vertical dashed lines indicate the Bragg condition. (C) Sketch of the double monolayer architecture. The axis perpendicular to the sample surface is denoted with z.
Fig. 5.
Experimental results on surface-adsorbed protein layers: HSA adsorbed to bare Al oxide (Top), to OTS functionalized Al oxide (Middle), and to bare Al oxide at full hydration under water (Bottom). (A) Measured angle-dependent S fluorescence intensities (symbols) together with the simulated intensities (solid lines) corresponding to the best-matching models. Vertical dashed lines indicate the Bragg condition. (B) Sketches of the protein layers configurations. The axis perpendicular to the sample surface is denoted with z.
Table 1.
Best-matching values of zj and σj for the P, S, and K distributions as obtained in fits to the respective angle-dependent fluorescence intensities from the studied biomolecular layers
SampleElementCenter position (Å)Width (Å)
SGS monolayer on Al oxideSzS = 0σS ≲ 10
DSPC monolayer on Al oxidePzP = 4 ± 2σP ≲ 10
DSPC monolayer with 5 mol% PEG-lipid on Al oxidePzP = 3 ± 2σP ≲ 10
SGS double monolayer on OTSSzS = 58 ± 2Δz ≲ 19*
 KzK = 59 ± 2Δz ≲ 19*
HSA on Al oxideSzS = 17 ± 2σS ≲ 10
HSA on OTSSzS = 34 ± 2σS ≲ 10
HSA on Al oxide under waterSzS = 37 ± 5σS = 20 ± 5
*
Distance between two peaks in a bimodal distribution.
Fig. S2.
Angle-dependent P Kα fluorescence intensity from a DSPC single monolayer (symbols) together with modeled intensities (solid lines) for various shifts ΔzP in the center position zP of the P distribution. The dotted line in each panel represents the best-matching value of zP.
Fig. S3.
Exemplary set of fluorescence curves modeled for Gaussian elemental distributions with different widths σ. Λ is the standing wave period.

Lipid Single Monolayers.

Fig. 3 A and C, Top, schematically illustrates the chemical details and the architecture of a single SGS monolayer on top of bare Al oxide. As a result of the Langmuir–Blodgett (LB) transfer, the hydrophilic monosaccharide headgroup is oriented toward the solid surface. Fig. 3B, Top, shows the angle-dependent S Kα fluorescence intensity IS(θ) from the SGS sulfate groups together with the modeled intensity. The intensity minimum is located at (θθB) ∼ 0.005°. This feature, together with the significantly different intensity levels on the two sides of the minimum, sharply defines the value for the center position of the S distribution, zS. Because the sulfate group is chemically attached close to the tip of the molecule and can be assumed to be in close contact to the surface, we use the obtained value as zero position of our coordinate system, zS = 0 (Fig. 3D, Top).
Fig. 3, Middle, summarizes the results obtained with a single DSPC monolayer on Al oxide. The P Kα fluorescence intensity IP(θ) (Fig. 3B, Middle) from the phosphate groups in the DSPC monolayer (Fig. 3 A and C, Middle) exhibits a sharp minimum similar to the one of the S Kα fluorescence from the SGS monolayer. However, the minimum is located very close to the Bragg condition, (θθB) ∼ 0.002°, and thus shifted slightly but significantly with respect to IS(θ) of the SGS monolayer. Moreover, the intensity levels on the two sides of the minimum are quite similar. The best-matching center position of the P distribution according to these features is zP = 4 ± 2 Å (Fig. 3D, Middle). The obtained height difference of about Δz ∼ 4 Å between the DSPC phosphate layer and the SGS sulfate layer appears plausible in view of the different headgroup structures of the two molecules: while the sulfate group in SGS is located at the tip of the headgroup, the phosphate group in DSPC is not the terminal moiety of the molecule. Instead the terminal (CH2)2N(CH3)3+ (choline) groups can act as spacer layer with a thickness of a few Ångstroms between the substrate surface and the phosphate layer.
Fig. 3, Bottom, shows the results obtained with a DSPC monolayer incorporating 5 mol% PEG-lipid on Al oxide. The molecular organization of this DSPC/PEG-lipid mixed layer is not obvious a priori. Several scenarios with similar overall layer thicknesses may occur in principle. When deposited onto hydrophobic substrates in water, the used DSPC/PEG-lipid mixtures form dense lipid monolayers displaying a highly hydrated PEG brush on top of the hydrophilic headgroups (24). One may thus expect a similar architecture also on the solid surface, with a PEG layer between the Al oxide surface and the lipid monolayer. However, also intermixed phases of PEG and lipids (25), as well as the escape of PEG from the space between oxide surface and lipid layer (26), have been reported. In the present work, SWXF puts us in the position to accurately localize P atoms and thus directly discriminate between different scenarios. Fig. 3B, Bottom, shows the P Kα fluorescence from the phosphate groups in the mixed DSPC/PEG-lipid monolayer together with the modeled intensity. With zP = 3 ± 2 Å, the best-matching center position of the P distribution is undistinguishable from that in the pure DSPC monolayer. This result clearly rules out the presence of any significant spacer layer formed by PEG between Al oxide and the lipid layer. Instead, it suggests an architecture in which the PEG portion is mainly on the backside of the lipid monolayer, as depicted schematically in Fig. 3C, Bottom. This structure can be attributed to the more favorable interactions of the positively charged Al oxide with the polar lipid headgroups, especially with negatively charged phosphatidyl-ethanolamine moieties of PEG-lipid, than with PEG. As discussed in Supporting Information, only a small fraction of PEG chains may percolate the headgroup layer without affecting zP. Note that the transfer of the mixed monolayer onto Al oxide was confirmed in independent ellipsometry measurements (Supporting Information). For all studied lipid single monolayers the obtained widths of the P and S distributions, σP and σS, respectively, are below 10 Å. In fact, it can be assumed that the widths of these distributions are vastly dominated by the topographic roughness of the solid surface and therefore approximately coincide with its RMS roughness σtop ∼ 3 Å (Fig. S4 and Table S1). For the illustration of elemental distributions in Fig. 3D, σP = σS = σtop was therefore assumed.
Fig. S4.
Height profile of a cleaned multilayer substrate as obtained by AFM.
Table S1.
Characteristics of the surface topography according to the height profile in Fig. S4
ParameterValue
Z range6.352 nm
Mean−0.003 nm
Raw mean219.55 nm
RMS (Rq)0.321 nm
Mean roughness (Ra)0.244 nm
Box x dimension4.016 µm
Box y dimension4.016 µm

Lipid Double Monolayers.

Fig. 4 shows the results obtained with a double SGS monolayer on top of a surface hydrophobically functionalized with octadecyltrichlorosilane (OTS). The sample architecture is illustrated in Fig. 4 A and C. The two negatively charged headgroup layers are facing one another. Because for this sample the LS (Langmuir–Schaefer) and LB transfers were done from a 1 mM KCl solution, the negative charge is compensated solely by K+ counterions, which can be localized conveniently by SWXF. Fig. 4B shows the S Kα fluorescence from the sulfate groups along with the intensity of the K Kα fluorescence from the counter ions. Both exhibit pronounced maxima near the Bragg condition. This behavior is in clear contrast to the S fluorescence intensity measured with SGS single monolayer (Fig. 3B, Top) and reflects that the elements of interest are located at a substantially different height above the solid surface, as implied by the different sample architecture. Strikingly, IS(θ) and IK(θ) for the SGS double monolayer peak at virtually the same angle, (θθB) ∼ −0.003°, indicating that the distributions of S and K have a common center of mass position to good approximation, which can understood from the mirror symmetry of the double layer. The best-matching positions, zS = 58 ± 2 Å and zK = 59 ± 2 Å are identical within the error. Their precise common value zSK ∼ 58 Å can be rationalized in the following way. Starting from the solid surface at z = 0, the dense OTS layer, including silane and alkyl portions has a thickness of d1 ∼ 25 Å, as reported in previous reflectometry studies (27, 28). The thickness of the first (inner) monolayer of SGS with its C16/C17 alkyl chains and the saccharide headgroup can be estimated as d1 ∼ 20–25 Å. For symmetry reasons the centers of mass of the S and K distributions must be located at the midplane between the two SGS surfaces. The remaining height difference d3 = zSK – (d1 + d2) must thus be attributed to the separation between the two opposing monolayer surfaces, suggesting that the hydration layer between the surfaces has a significant thickness D = 2d3. Strong repulsion between the negatively charged surfaces can be expected when only monovalent counter ions are available, giving rise to substantial swelling with water even at moderate ambient air humidity. In fact, it was shown recently that the incorporation of negatively charged sulfoglycolipids at even small fractions dramatically extends the swelling range in glycolipid multilayers (29). Although the experimental data are well described by the model based on Eq. 2 with σj ≲ 10 Å, it should be noted that in the studied double monolayer neither S nor K distributions can be assumed as unimodal. As explained above, the shape of an elemental distribution cannot be resolved when its width, or more generally, its SD is below a certain threshold σmin. For a Gaussian distribution, SD and σj coincide. For a more suitable, bimodal description of the S and K distributions, comprising two Gaussian peaks of individual width σind separated by a distance Δz, the criterion for the SD to be below σmin is Δz2σmin2σind2. For σmin = 10 Å and with the rough approximation that σind coincides with the topographic roughness of the substrate, σind = σtop, we obtain Δz ≲ 19 Å (Table 1).

Protein Layers.

Fig. 5B, Top, schematically illustrates HSA adsorbed to a bare Al oxide surface. Fig. 5A, Top, shows the S Kα fluorescence from the S atoms contained in HSA's cysteine and methionine amino acids together with the modeled intensity. The intensity exhibits a minimum slightly below the Bragg angle (θθB ∼ −0.003°) and a maximum significantly above the Bragg angle (θθB ∼ 0.01°). These distinct features correspond to a center position of the S distribution at zS = 17 ± 2 Å. We recall that the center of mass of the S distribution and that of the whole HSA protein roughly coincide. The thickness dHSA of a HSA layer with an S distribution centered at an altitude Δz = zSzsurf above the surface can thus be approximated as dHSA ∼ 2Δz. For zS = 17 ± 2 Å and zsurf = 0, we obtain Δz = 17 ± 2 Å and dHSA ∼ 30–40 Å. Note that in this procedure the width of the S distribution is not invoked for the determination of dHSA. Serum albumin has been described as prolate ellipsoid with major and minor axes of 145 and 40 Å, respectively (30), or as heart-shaped molecule with dimensions (55 Å)2 × 90 Å (31). The thickness obtained here thus indicates that HSA assumes a side-on configuration in which d coincides with the shorter molecular axes. Our result, dHSA ∼ 30–40 Å, is somewhat smaller than the reported length of the shorter molecular axis, further indicating that HSA experiences significant deformation on the surface on adsorption and subsequent drying. The width of the S distribution obtained here is below 10 Å. Note, however, that the thickness of the protein layer more directly corresponds to the S-distribution's full-width-at-half-maximum (FWHM) = 22ln2 σS ≲ 24 Å. This result is consistent with dHSA ∼ 30–40 Å when considering that S atoms are not evenly distributed over the HSA molecule but preferentially found in its interior (22).
Fig. 5B, Middle, illustrates HSA adsorbed to an OTS-functionalized surface. The S Kα fluorescence (Fig. 5A, Middle) exhibits a maximum slightly above the Bragg angle at around θθB ∼ 0.004°. The fit to the data yields zS = 34 ± 2 Å. The thickness of the OTS layer is ∼25 Å (see above). The surface to which HSA is adsorbed is thus located at zsurf ∼ 25 Å so that Δz = zS − zsurf ∼ 10 Å and dHSA ∼ 2Δz ∼ 20 Å, indicating that the HSA layer adsorbed to the OTS-functionalized surface is even thinner than that on the bare Al oxide surface. In fact, more pronounced protein deformation on hydrophobic surfaces than on hydrophilic surfaces has been reported in several studies (3234) and has been attributed to exposure of hydrophobic portions to the surface (35). For the width of the S distribution, we consistently obtain σS ≲ 10 Å, corresponding to FWHM ≲ 24 Å.
HSA adsorbed to Al oxide surface was also investigated in a fully hydrated state under water, as illustrated in Fig. 5B, Bottom. In this configuration, where a polymer film is used to confine a thin water layer on top of the sample surface, the attenuation of the incident beam while traveling through the polymer film slightly reduces the fluorescence intensity and, thus, also signal statistics. Fig. 5A, Bottom, shows the S Kα fluorescence from the proteins under water together with the modeled intensity. The optimal value for the S distribution's center position is zS ∼ 37 Å. Due to the more scattered nature of the fluorescence intensities, we estimate the associated parameter error as 5 Å. With zS = 37 ± 5 Å, the center position obtained under water is shifted away from the surface substantially with respect to the dry case (zS = 17 ± 2 Å). Within the assumption of coinciding center of mass positions of the S distribution and of the whole protein, this indicates that the fully hydrated protein layer is as thick as dHSA ∼ 2Δz = 2zS ∼ 60–80 Å. For the width of the S distribution, we consistently obtain σS = 20 ± 5 Å, corresponding to an FWHM = 48 ± 12 Å. Adsorption of serum albumin to Al oxide has been previously investigated using zeta-potential and UV-Vis measurements, and a layer thickness of about 55 Å under water was reported and interpreted as side-on configuration (31). Thicker serum albumin layers of more than 100 Å and up to 70 Å were reported for mica/water (36) and air/water interfaces (37), respectively. Those results have been interpreted either as the result of tilted, or top-on configurations, in which the long protein axis is perpendicular to the surface, or as protein double layers. Our results point toward a tilted top-on configuration.
Important aspects of the configuration of proteins adsorbed to solid surfaces, such as their height and lateral extension, can also be obtained by atomic force microscopy (AFM) (38, 39). However, this approach is only sensitive to the outer protein shape and not applicable to proteins adsorbed to soft interfaces (24). In contrast, SWXF can equally probe rigid and soft interfaces and due to its element specificity has the potential to yield a comprehensive picture of protein configurations. Future studies optimized for the simultaneous localization of S, the low P amounts in phosphorylated amino acids, and metal ligands, can serve to unambiguously identify the orientation and internal conformation of proteins adsorbed to surfaces.
In the present work, the surface densities of P and S were at the order of 1 atom/nm2. Although with the current methodology densities as low as 1 atom/10 nm2 (corresponding to one or several S atoms per protein in a monolayer, depending on the protein size) already appear to be sufficient, one may expect that methodological improvements will further enhance the sensitivity. Especially for the most interesting but also most challenging measurements involving solid/liquid interfaces (Fig. 5, Bottom) such improvements may include (i) the use of thinner and chemically purer polymer films or nano-confinements and (ii) the use of multilayers with shorter periods and larger Bragg angles, thereby reducing beam absorption.

Conclusions

We showed that SWXF with planar solid multilayer substrates enables localization with near-Ångstrom precision of the light, biologically important chemical elements P and S in the direction perpendicular to an interface. By using this method, we structurally characterized nanometric layers of the most important classes of biomolecules. The measurements yield element-specific insight into the architecture of various lipid monolayer architectures and into the conformations of proteins adsorbed to planar surfaces under various conditions. The presented approach allows for the label-free investigation of complex biomolecular interfaces with great structural detail.

Materials and Methods

Materials and Sample Preparation.

Sulfoglycolipids (SGS, 3-O-sulfo-d-galactosyl-β1–1'-N-heptadecanoyl-d-erythro-sphingosine), phospholipids (DSPC, 1,2-distearoyl-sn-glycero-3-phosphocholine), and lipopolymers (PEG-lipid, 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-5000]) were dissolved in chloroform at 2 mg/mL. In addition, a mixture of DSPC and PEG-lipid with a PEG-lipid mole fraction of 5% at 2 mg/mL overall concentration was prepared. HSA was dissolved in water at 1 mg/mL Multilayer substrates were purchased from X’scitech. Further details about the materials are given in Supporting Information. The substrates were cleaned by washing with chloroform, acetone, ethanol, and water, followed by UV-ozone treatment. In some cases the surfaces were then rendered hydrophobic via covalent functionalization with OTS by immersion in freshly prepared solutions of OTS in hexadecane at a concentration of 1 mM for 1 h and subsequent rinsing in hexadecane and ethanol. This treatment resulted in water surface contact angles of >100°, in agreement with earlier studies dealing with the OTS functionalization of aluminum oxide surfaces (40). Single and double lipid monolayers on the surface of the multilayer substrates were prepared using the LB and/or LS transfer methods. For this purpose, lipid solutions (of SGS, DSPC, or DSPC/PEG lipid mixtures) in chloroform were first spread at the air/water interface in a Teflon Langmuir trough (Nima Technology) containing water or dilute aqueous salt solution. The amphiphilic lipid molecules immobilized at the interface were then compressed to a monolayer with a lateral pressure of 35 ± 1 mN/m. As can be seen in the compression isotherms shown in Fig. S5, the incorporation of 5 mol% PEG-lipid into DSPC had no significant influence on the area per lipid at this pressure. Single lipid monolayers were deposited via LB transfer onto bare Al oxide surfaces. For the preparation of double lipid monolayers, a first lipid layer was deposited onto OTS-functionalized surfaces via LS transfer. Subsequently the second layer was deposited via LB transfer on top of the first layer. HSA protein monolayers were prepared by letting HSA adsorb to bare or OTS-functionalized substrates from aqueous HSA solution for 1 h followed by rinsing with water. In the last step, the HSA layers were either dried or covered with a thin water layer under a 4-µm thin polymer film (Ultralene; SPEX SamplePrep). The water layer was stabilized by capillary forces and according to beam absorption measurements was no thicker than 1 µm.
Fig. S5.
Compression isotherms of pure DSPC and DSPC incorporating 5% PEG-lipid.

SWXF Experiments and Data Analysis.

SWXF experiments were carried out at the ID03 beamline of ESRF and at the DIFFABS beamline of Synchrotron SOLEIL with 7.0-keV beam energy. The sample plane was oriented vertically. X-ray fluorescence was measured using a silicon drift detector (either Ketek AXAS or Hitachi Vortex) placed perpendicular to the sample plane about 3 mm from the surface. To improve signal statistics without the risk of beam damage to the samples, the scans were repeated several times with laterally shifted beam position on the sample surface. For this purpose in the first step the reference spectra were analyzed. The relevant reference spectra for biomolecular layers on bare substrates, on OTS-functionalized substrates, and at the solid–liquid interfaces, respectively, are the spectra of the bare substrate, of the OTS-functionalized substrate, and of the bare substrate under the polymer film. After energy calibration they were modeled as linear combinations of elemental spectra following ref. 41. In the next step, the spectra of the biomolecular layers were modeled while taking into account the elemental amplitudes in their respective reference spectra. The angle-dependent SW generated by the multilayers was computed using Sergey Stepanov’s X-ray Server (sergey.gmca.aps.anl.gov), based on the structural parameters obtained in the reflectivity analysis (Supporting Information and Fig. S6).
Fig. S6.
X-ray reflectivity curve of a cleaned multilayer substrate (symbols) together with the best-matching fit (solid line).

X-Ray Fluorescence Spectra

The left panel of Fig. S1 shows the X-ray fluorescence spectrum of a solid-supported double SGS monolayer (Fig. 4). It exhibits the characteristic fluorescence peaks of the chemical elements of interest, S and K, as well as of some of the chemical elements found in the multilayer substrates and in the sample environment (Al, Ar, Ti, and Cr). Note that Ni is not visible because the incident beam energy is below its K edge. The right panel shows the same spectrum in the region of interest around the S and K peaks together with the reference spectrum of a bare OTS-functionalized multilayer substrate. The spectra are virtually identical apart from the S and K fluorescence originating from the double SGS monolayer itself.

Sensitivity of the SWXF Measurements to the Center Position zj of an Elemental Distribution

To illustrate the sensitivity of the SWXF measurements to the center position zj of the elemental distributions Fig. S2 exemplarily shows the angle-dependent P Kα fluorescence intensity from a DSPC single monolayer (symbols) together with modeled intensities for various assumptions of zP. The dotted line in each panel represents the best-matching value of zP. The solid lines correspond to various shifts ΔzP from the optimum, namely ΔzP = +2 Å (Top Left), ΔzP = +4 Å (Top Right), ΔzP = −2 Å (Bottom Left), and ΔzP = −4 Å (Bottom Right). It is seen that for |ΔzP| = 4 Å (Right), the deviation between experiment and model is substantial. For |ΔzP| = 2 Å (Left), the deviation can still be recognized but is visually at the limit of significance. |ΔzP| = 2 Å can therefore be considered a reasonable estimate for the uncertainty in the parameter zj.

Sensitivity of the SWXF Measurements to the Width σj of an Elemental Distribution

The width σj of an elemental distribution is encoded in the shape of the angle-dependent fluorescence intensity, primarily in the relative amplitude of the modulation with respect to the baseline further away from the Bragg condition (see Fig. S3 for an exemplary set of fluorescence curves modeled for Gaussian elemental distributions with different widths). When σj is much smaller than the standing wave period Λ, then the amplitude of this modulation saturates and is no longer sensitive to the precise value of σj. In the present work, in view of the limited statistics of the experimental data points and bearing in mind minor residual uncertainties in the background treatment, σmin ∼ Λ/10 can be considered a conservative estimate for the lower end of the σj range that can be determined reliably. The standing wave period close to the Bragg condition is determined by the multilayer period d and the order n of the Bragg peak: Λ = d/n. In the present work, the multilayer period was d ∼ 200 Å and a second-order Bragg peak was used (n = 2), so that Λ ∼ 100 Å and σmin ∼ 10 Å.

PEG Chains in the Headgroup Layer in DSPC/PEG-Lipid Mixed Monolayers on Al Oxide

In the following, we derive that only a small fraction of the PEG chains in a DSPC/PEG-lipid mixed monolayer may percolate the headgroup layer without significantly altering the height of the lipid layer with respect to the solid surface.
At a PEG-lipid mole fraction of 5% (x = 0.05), the PEG mass per lipid is xmPEG = 250 g/mol (with mPEG = 5,000 g/mol). The mass ratio between PEG and lipid is thus xmPEG/mlip ∼ 0.32 (with mlip = 790 g/mol). If we now safely assume that the density of the percolating PEG chains is at most as high as that of the lipids, the mass ratio can serve as a lower limit of the ratio between the thickness of the compact PEG layer dPEG and the thickness of the lipid layer dlip, dPEG/dlip ≳ 0.32. For an estimated dlip = 2.5–3.0 nm, we obtain dPEG ≳ 0.8 nm. In other words, even for the comparatively low PEG-lipid mole fraction used, the PEG amount is too high to get accommodated in the headgroup region without significantly elevating the lipid layer. Nonetheless, a small fraction of PEG may still percolate the headgroup layer.

Transfer of DSPC/PEG-Lipid Mixed Monolayers onto Al Oxide

The complete transfer DSPC/PEG-lipid mixed monolayers was confirmed in independent ellipsometry measurements on planar Al substrates covered with native Al oxide before and after monolayer transfer. In the data analysis, monolayers were approximated as homogeneous layers of refractive index n = 1.45, which is typical for organic materials (The exact value is not important for the following considerations.). For DSPC alone, the thickness of the organic layer as determined by ellipsometry was dDSPC = 2.8 ± 0.5 nm. For a DSPC/PEG-lipid mixture with 5% PEG-lipid, we obtained a significantly thicker layer with dmix = 4.1 ± 0.5 nm. The ratio between these two values, dmix/dDSPC = 1.5 ± 0.3, within the error is consistent with the ratio between the transferred amounts per unit area when a constant area per molecule is assumed. Expressed in terms of the mass, this ratio is (mlip + xmPEG)/mlip = 1.32. Here, mlip = 790 g/mol is the mass of the lipid anchor, mPEG = 5,000 g/mol is the mass of a PEG chain, and x = 0.05 is the fraction of lipids carrying a PEG chain.

Topographic RMS-Roughness of the Multilayer Substrate Outer Surface

The surface topography of a cleaned multilayer substrate was measured by AFM with a cantilever of tip radius of ∼10 nm. Fig. S4 shows the obtained height profile over a measurement area of 2 × 2 µm. Table S1 summarizes the characteristics of the topography.

Further Details About the Materials Used

Unless stated otherwise, all chemicals were purchased from Sigma and used without further purification. Water was purified and double-deionized (MilliQ). Sulfoglycolipids (SGS, 3-O-sulfo-d-galactosyl-β1–1'-N-heptadecanoyl-d-erythro-sphingosine), phospholipids (DSPC, 1,2-distearoyl-sn-glycero-3-phosphocholine), and lipopolymers (PEG-lipid, 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-5000]) were purchased from Avanti Polar Lipids. The design of the planar metal multilayer substrates used to generate the standing X-ray wave was guided by several simultaneous requirements. First, materials with strongly different electron densities were desirable to give rise to a strong Bragg reflection. Second, the materials used had to be free of fluorescence lines in the energy range relevant for the detection of the target elements. In addition, the period of the multilayer should be chosen to match the characteristic size of the system under investigation. Finally, the terminal surface had to be suited for the chemical functionalization with thin biomolecular layers.

Compression Isotherms of DSPC and DSPC/PEG-Lipid Mixtures at the Air/Water Interface

Fig. S5 shows compression isotherms of a pure DSPC monolayer and of a DSPC monolayer incorporating 5% PEG-lipid. The monolayers were compressed at a rate of 4 Å2/min per lipid. In both cases, the area per lipid Alip was rescaled such that at 30 mN/m it corresponds to the literature value for DSPC, Alip (30 mN/m) = 47 Å2 (42). It is seen that, although PEG-lipid has a strong influence on the compression isotherm for low pressures (Π ≳ 20 mN/m), for higher pressures (Π ≲ 20 mN/m) Alip is dominated by the largely incompressible lipid layer and therefore virtually unaffected by the presence of PEG-lipid. The lipid packing density at the transfer pressure of Π = 35 mN/m can therefore be assumed to be very similar in the presence of absence of 5% PEG-lipid.

Reflectivity of the Multilayer Substrates

Fig. S6 shows the measured X-ray reflectivity curve of a cleaned multilayer substrate together with the best-matching fit based on a 20-fold periodic structure with two layers (Al and Ni) per period on top of a semiinfinite sapphire support. The best-matching thickness parameters of Al and Ni layers are dAl = 9.93 nm and dNi = 9.71 nm, respectively. The best-matching RMS-roughness corresponding to the electron density transition between Al and Ni layers is σAl/Ni ∼ 1.3 nm.

Acknowledgments

We thank ESRF and Synchrotron SOLEIL for beam time allocation and the Partnership for Soft Condensed Matter (PSCM) laboratories of the Institut Laue-Langevin for support during sample preparation and precharacterization. We thank Alexis de Ghellinck and Anne Heilig for help with sample preparation and complementary AFM experiments, respectively, and Dominique Thiaudière and Thomas Dufrane for scientific and technical support during beamtimes. Financial support by the Deutsche Forschungsgemeinschaft via Emmy-Noether Grant SCHN1396/1 is gratefully acknowledged. E. Schneck acknowledges support from a Marie Curie Intra-European Fellowship (Grant 299676) within the European Commission seventh Framework Program.

Supporting Information

Supporting Information (PDF)
Supporting Information

References

1
R Lipowsky, E Sackmann Structure and Dynamics of Membranes: I. From Cells to Vesicles/II. Generic and Specific Interactions (Elsevier, New York, 1995).
2
OG Mouritsen Life—As a Matter of Fat (Springer, New York, 2005).
3
KL Prime, GM Whitesides, Self-assembled organic monolayers: Model systems for studying adsorption of proteins at surfaces. Science 252, 1164–1167 (1991).
4
E Sackmann, Supported membranes: Scientific and practical applications. Science 271, 43–48 (1996).
5
G Brezesinski, H Möhwald, Langmuir monolayers to study interactions at model membrane surfaces. Adv Colloid Interface Sci 100-102, 563–584 (2003).
6
JN Israelachvili Intermolecular and Surface Forces (Academic Press, London, 1991).
7
E Schneck, B Demé, Structural characterization of soft interfaces by standing-wave fluorescence with X-rays and neutrons. Curr Opin Colloid Interface Sci 20, 244–252 (2015).
8
JK Basu, et al., Direct probe of end-segment distribution in tethered polymer chains. Macromolecules 40, 6333–6339 (2007).
9
MJ Bedzyk, DH Bilderback, GM Bommarito, M Caffrey, JS Schildkraut, X-ray standing waves: A molecular yardstick for biological membranes. Science 241, 1788–1791 (1988).
10
MJ Bedzyk, GM Bommarito, M Caffrey, TL Penner, Diffuse-double layer at a membrane-aqueous interface measured with x-ray standing waves. Science 248, 52–56 (1990).
11
JM Bloch, et al., Adsorption of counterions to a stearate monolayer spread at the water-air interface: A synchrotron x-ray study. Phys Rev Lett 61, 2941–2944 (1988).
12
W Bu, et al., X-ray studies of interfacial strontium-extractant complexes in a model solvent extraction system. J Phys Chem B 118, 12486–12500 (2014).
13
V Padmanabhan, et al., Specific ion adsorption and short-range interactions at the air aqueous solution interface. Phys Rev Lett 99, 086105 (2007).
14
E Schneck, et al., Quantitative determination of ion distributions in bacterial lipopolysaccharide membranes by grazing-incidence X-ray fluorescence. Proc Natl Acad Sci USA 107, 9147–9151 (2010).
15
A Körner, et al., Quantitative determination of lateral concentration and depth profile of histidine-tagged recombinant proteins probed by grazing incidence X-ray fluorescence. J Phys Chem B 117, 5002–5008 (2013).
16
S Zheludeva, N Novikova, N Stepina, E Yurieva, O Konovalov, Molecular organization in protein-lipid film on the water surface studied by x-ray standing wave measurements under total external reflection. Spectrochim Acta B At Spectrosc 63, 1399–1403 (2008).
17
G van Meer, DR Voelker, GW Feigenson, Membrane lipids: Where they are and how they behave. Nat Rev Mol Cell Biol 9, 112–124 (2008).
18
L Boudière, et al., Glycerolipids in photosynthesis: Composition, synthesis and trafficking. Biochimica et Biophysica Acta (BBA)-. Bioenergetics 1837, 470–480 (2014).
19
I Ishizuka, Chemistry and functional distribution of sulfoglycolipids. Prog Lipid Res 36, 245–319 (1997).
20
CJ Wilson, RE Clegg, DI Leavesley, MJ Pearcy, Mediation of biomaterial-cell interactions by adsorbed proteins: A review. Tissue Eng 11, 1–18 (2005).
21
V Hlady, J Buijs, Protein adsorption on solid surfaces. Curr Opin Biotechnol 7, 72–77 (1996).
22
S Sugio, A Kashima, S Mochizuki, M Noda, K Kobayashi, Crystal structure of human serum albumin at 2.5 A resolution. Protein Eng 12, 439–446 (1999).
23
M Born, E Wolf Principles of Optics (Cambridge Univ Press, Cambridge, UK, 1999).
24
E Schneck, I Berts, A Halperin, J Daillant, G Fragneto, Neutron reflectometry from poly (ethylene-glycol) brushes binding anti-PEG antibodies: Evidence of ternary adsorption. Biomaterials 46, 95–104 (2015).
25
C Daniel, et al., Structural characterization of an elevated lipid bilayer obtained by stepwise functionalization of a self-assembled alkenyl silane film. Biointerphases 2, 109–118 (2007).
26
JP Talbot, DJ Barlow, MJ Lawrence, PA Timmins, G Fragneto, Interaction of cationic lipoplexes with floating bilayers at the solid-liquid interface. Langmuir 25, 4168–4180 (2009).
27
M Mezger, et al., High-resolution in situ x-ray study of the hydrophobic gap at the water-octadecyl-trichlorosilane interface. Proc Natl Acad Sci USA 103, 18401–18404 (2006).
28
M Pomerantz, A Segmüller, L Netzer, J Sagiv, Coverage of Si substrates by self-assembling monolayers and multilayers as measured by IR, wettability and X-ray diffraction. Thin Solid Films 132, 153–162 (1985).
29
B Demé, C Cataye, MA Block, E Maréchal, J Jouhet, Contribution of galactoglycerolipids to the 3-dimensional architecture of thylakoids. FASEB J 28, 3373–3383 (2014).
30
TJ Su, JR Lu, RK Thomas, ZF Cui, J Penfold, The conformational structure of bovine serum albumin layers adsorbed at the silica-water interface. J Phys Chem B 102, 8100–8108 (1998).
31
K Rezwan, et al., Bovine serum albumin adsorption onto colloidal Al2O3 particles:Aa new model based on zeta potential and UV-vis measurements. Langmuir 20, 10055–10061 (2004).
32
JR Lu, et al., The denaturation of lysozyme layers adsorbed at the hydrophobic solid/liquid surface studied by neutron reflection. J Colloid Interface Sci 206, 212–223 (1998).
33
TJ Su, JR Lu, RK Thomas, ZF Cui, J Penfold, The adsorption of lysozyme at the silica-water interface: A neutron reflection study. J Colloid Interface Sci 203, 419–429 (1998).
34
M Malmsten, Ellipsometry studies of the effects of surface hydrophobicity on protein adsorption. Colloids Surf B Biointerfaces 3, 297–308 (1995).
35
P Roach, D Farrar, CC Perry, Interpretation of protein adsorption: Surface-induced conformational changes. J Am Chem Soc 127, 8168–8173 (2005).
36
E Blomberg, PM Claesson, RD Tilton, Short-range interaction between adsorbed layers of human serum albumin. J Colloid Interface Sci 166, 427–436 (1994).
37
JR Lu, TJ Su, RK Thomas, Structural conformation of bovine serum albumin layers at the air-water interface studied by neutron reflection. J Colloid Interface Sci 213, 426–437 (1999).
38
M Fritz, et al., Imaging globular and filamentous proteins in physiological buffer solutions with tapping mode atomic force microscopy. Langmuir 11, 3529–3535 (1995).
39
DT Kim, HW Blanch, CJ Radke, Direct imaging of lysozyme adsorption onto mica by atomic force microscopy. Langmuir 18, 5841–5850 (2002).
40
D Wang, Y Ni, Q Huo, DE Tallman, Self-assembled monolayer and multilayer thin films on aluminum 2024-T3 substrates and their corrosion resistance study. Thin Solid Films 471, 177–185 (2005).
41
M Van Gysel, P Lemberge, P Van Espen, Description of Compton peaks in energy-dispersive x-ray fluorescence spectra. XRay Spectrom 32, 139–147 (2003).
42
A Hermelink, G Brezesinski, Do unsaturated phosphoinositides mix with ordered phosphatidylcholine model membranes? J Lipid Res 49, 1918–1925 (2008).

Information & Authors

Information

Published in

Go to Proceedings of the National Academy of Sciences
Go to Proceedings of the National Academy of Sciences
Proceedings of the National Academy of Sciences
Vol. 113 | No. 34
August 23, 2016
PubMed: 27503887

Classifications

Submission history

Published online: August 8, 2016
Published in issue: August 23, 2016

Keywords

  1. surfaces
  2. interfaces
  3. lipid membranes
  4. protein adsorption
  5. X-ray scattering

Acknowledgments

We thank ESRF and Synchrotron SOLEIL for beam time allocation and the Partnership for Soft Condensed Matter (PSCM) laboratories of the Institut Laue-Langevin for support during sample preparation and precharacterization. We thank Alexis de Ghellinck and Anne Heilig for help with sample preparation and complementary AFM experiments, respectively, and Dominique Thiaudière and Thomas Dufrane for scientific and technical support during beamtimes. Financial support by the Deutsche Forschungsgemeinschaft via Emmy-Noether Grant SCHN1396/1 is gratefully acknowledged. E. Schneck acknowledges support from a Marie Curie Intra-European Fellowship (Grant 299676) within the European Commission seventh Framework Program.

Notes

This article is a PNAS Direct Submission.

Authors

Affiliations

Emanuel Schneck1 [email protected]
Biomaterials Department, Max Planck Institute of Colloids and Interfaces, 14476 Potsdam, Germany;
Institut Laue-Langevin, 38000 Grenoble, France;
Ernesto Scoppola
Institut Laue-Langevin, 38000 Grenoble, France;
Institut de Chimie Séparative de Marcoule UMR 5257 Commissariat à l'énergie atomique/CNRS/École Nationale Supérieure de Chimie de Montpellier, Université Montpellier, 30207 Bagnols sur Cèze, France;
Jakub Drnec
European Synchrotron Radiation Facility (ESRF), 38000 Grenoble, France;
Cristian Mocuta
Synchrotron SOLEIL, 91192 Gif-sur-Yvette Cedex, France;
Roberto Felici
European Synchrotron Radiation Facility (ESRF), 38000 Grenoble, France;
Present address: Istituto SPIN (superconductors, oxides and other innovative materials and devices), I-00133 Roma, Italy.
Dmitri Novikov
Deutsches Elektronen-Synchrotron (DESY), 22607 Hamburg, Germany
Giovanna Fragneto
Institut Laue-Langevin, 38000 Grenoble, France;
Jean Daillant
Synchrotron SOLEIL, 91192 Gif-sur-Yvette Cedex, France;

Notes

1
To whom correspondence should be addressed. Email: [email protected].
Author contributions: E. Schneck, R.F., D.N., G.F., and J. Daillant designed research; E. Schneck, E. Scoppola, D.N., G.F., and J. Daillant performed research; J. Drnec, C.M., and R.F. contributed special experimental tools; E. Schneck and J. Daillant analyzed data; and E. Schneck, G.F., and J. Daillant wrote the paper.

Competing Interests

The authors declare no conflict of interest.

Metrics & Citations

Metrics

Note: The article usage is presented with a three- to four-day delay and will update daily once available. Due to ths delay, usage data will not appear immediately following publication. Citation information is sourced from Crossref Cited-by service.


Citation statements




Altmetrics

Citations

If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.

Cited by

    Loading...

    View Options

    View options

    PDF format

    Download this article as a PDF file

    DOWNLOAD PDF

    Get Access

    Login options

    Check if you have access through your login credentials or your institution to get full access on this article.

    Personal login Institutional Login

    Recommend to a librarian

    Recommend PNAS to a Librarian

    Purchase options

    Purchase this article to get full access to it.

    Single Article Purchase

    Atom-scale depth localization of biologically important chemical elements in molecular layers
    Proceedings of the National Academy of Sciences
    • Vol. 113
    • No. 34
    • pp. 9381-E5091

    Media

    Figures

    Tables

    Other

    Share

    Share

    Share article link

    Share on social media