Small-Angle X-Ray Scattering for Imaging of Surface Layers on Intact Bacteria in the Native Environment

Abstract

Crystalline cell surface layers (S-layers) represent a natural two-dimensional (2D) protein self-assembly system with nanometer-scale periodicity that decorate many prokaryotic cells. Here, we analyze the S-layer on intact bacterial cells of the Gram-positive organism Geobacillus stearothermophilus ATCC 12980 and the Gram-negative organism Aquaspirillum serpens MW5 by small-angle X-ray scattering (SAXS) and relate it to the structure obtained by transmission electron microscopy (TEM) after platinum/carbon shadowing. By measuring the scattering pattern of X rays obtained from a suspension of bacterial cells, integral information on structural elements such as the thickness and lattice parameters of the S-layers on intact, hydrated cells can be obtained nondestructively. In contrast, TEM of whole mounts is used to analyze the S-layer lattice type and parameters as well as the physical structure in a nonaqueous environment and local information on the structure is delivered. Application of SAXS to S-layer research on intact bacteria is a challenging task, as the scattering volume of the generally thin (3- to 30-nm) bacterial S-layers is low in comparison to the scattering volume of the bacterium itself. For enhancement of the scattering contrast of the S-layer in SAXS measurement, either silicification (treatment with tetraethyl orthosilicate) is used, or the difference between SAXS signals from an S-layer-deficient mutant and the corresponding S-layer-carrying bacterium is used for determination of the scattering signal. The good agreement of the SAXS and TEM data shows that S-layers on the bacterial cell surface are remarkably stable.

INTRODUCTION

Surface layers (S-layers) are a distinct cell surface decoration comprised of regularly arrayed, two-dimensional (2D) protein or glycoprotein crystals present on many prokaryotic organisms from almost all known phylogenetic branches (1–4). S-layers form upon self-assembly and exhibit nanometer-scale periodicity with oblique (p1, p2), square (p4), or hexagonal (p3, p6) lattice symmetry and species-specific lattice constant values in the range of 10 to 25 nm. The S-layer self-assembly system holds great promise as a bottom-up approach for organizing matter at the nanometer scale as required in the field of nanobiotechnology (5).

Because of their cell surface location, ultrastructural studies of S-layer lattices have been frequently performed by high-resolution transmission electron microscopic (TEM) methods such as ultrathin sectioning or heavy-metal shadow casting of freeze-etched or freeze-dried whole cells (6) as well as image analysis (7) or cryo-sectioning of frozen-hydrated cells (8). These electron microscopic approaches, except the latter, suffer from the drawback of the requirement for water-free specimens, implying a requirement for specific sample treatment prior to analysis. The standard procedures of chemical fixation, dehydration, and staining of cells for TEM analysis may affect cell morphology and ultrastructure and, thus, prevent the accurate determination of distinct S-layer features such as the thickness and lattice parameters. Newer methods such as atomic force microscopy (AFM) allow for the surface visualization of biological samples in aqueous environments and offer the additional possibility of improved height measurement. While most AFM analyses of S-layers have so far been performed on isolated, spread S-layer self-assembly products in vitro (9–11), we succeeded only recently in the noninvasive in vivo AFM imaging of the S-layer lattice such as is present on intact Tannerella forsythia cells after mechanical trapping of the bacteria on polycarbonate membranes (12). To provide complementary and/or supportive methods for visualization of structures under most nature-like conditions, efforts are made to expand the repertoire of techniques that allow for imaging of bacterial cell surface structures at high resolution in aqueous systems without massive sample pretreatment.

Small-angle X-ray scattering (SAXS) has the particular advantage of determining the structure of biological macromolecules in different environments under conditions close to those in physiological settings (13). It has, for instance, been shown that SAXS is suitable for determining the size, shape, and properties of magnetite crystals in the magnetotactic bacterium Aquaspirillum magnetotacticum (14) as well as the location of cell-bound metal precipitates of Pseudomonas fluorescens (15).

SAXS is sensitive to electron density differences within the material (16). The X-ray scattering intensities are described by analytical functions such as spheres, cylinders, or discs (17), by calculating the particle density distribution function by the inverse (18) or generalized inverse (19) Fourier transform, or by numerical methods, which reconstruct the 3D structure in real space from 1D scattering curves in reciprocal space (20). The latter method was developed in particular for proteins without any interaction in a dilute solution. In the case of interaction of the proteins, such as in lattices of noncovalently linked S-layer proteins, the structure factor value differs from 1, which considerably complicates procedures to gain information on the system. With regard to S-layers, SAXS has been used to describe the wet chemical metallization of S-layers under different chemical conditions (21). There, an approach was chosen to model the scattering intensity at low values of the scattering vector q by a mixture of mono- and bilayers and the scattering intensity at high q values by a square lattice as concluded from reconstruction of TEM images (21). Furthermore, protein-protein interaction within the S-layer was taken into account by combining molecular dynamics simulations and refining the model with experimental SAXS data (22), where the subunit also was modeled on the amino acid level (23). All reports on measuring SAXS from S-layers were based on concentrated suspensions of isolated S-layer protein self-assembly products after centrifugation.

The present work focuses on the exploitation of SAXS for imaging and characterizing of S-layers of intact bacteria in natural environments. This is a challenging task, as the scattering volume of the S-layer, which is generally only some nanometers in thickness, is low in comparison to the scattering volume of the bacterium itself. The main goals of this work are (i) to show that SAXS can provide structural information on S-layers on bacteria under nearly physiological conditions, exemplified with the model bacteria Geobacillus stearothermophilus ATCC 12980 and Aquaspirillum serpens MW5, and (ii) to use tetraethyl orthosilicate (TEOS) (24, 25) treatment (silicification) for enhancement of the scattering contrast of S-layer-covered cells or to use SAXS signals obtained from an S-layer-deficient mutant as a baseline for SAXS signal evaluation. Biogenic silica deposition in combination with SAXS measurement opens a wide field of applications, as this technique allows measurement of not only S-layers but also any templated surface in various environments.

MATERIALS AND METHODS

Cultivation and sample preparation.

Since S-layers can be present on Gram-positive as well as on Gram-negative bacterial cells, we have chosen for our experiments a representative of each group, namely, Geobacillus stearothermophilus ATCC 12980 (a Gram-positive bacterium) (26, 27) and Aquaspirillum serpens MW5 (a Gram-negative bacterium) (28).

A G. stearothermophilus ATCC 12980 wild-type strain and an S-layer-deficient mutant thereof were grown in S-VIII medium (29) for 5 h at 55°C and 150 rpm. The biomass was pelletized, washed once with water, and resuspended in phosphate-buffered saline (PBS) to yield a dense suspension of bacteria that was subjected to SAXS measurements.

A. serpens MW5 was cultivated in PSS medium [2 g bacteriological peptone, 1 g amber salt, 0.2 g (NH₄)₂SO₄, 0.2 g MgSO₄·7 H₂O, 200 μl of 2% FeCl₃·6 H₂O solution, 200 μl of 2% MnSO₄·H₂O solution, 200 ml of H₂O (adjusted to pH 6.8 with KOH; addition of 100 μl of 2% CaCl₂·2 H₂O solution after autoclaving)] at 28°C and 150 rpm for 24 h. The biomass was pelletized and washed once with MilliQ water. After resuspension in PBS, an aliquot of 1 ml was treated with TEOS solution for biogenic silica deposition according to a two-step method for silicification (30, 31). To obtain silicic acid monomers for immediate use, prehydrolysis of freshly prepared 1 M TEOS (Sigma, St. Louis, MO) was performed using 50% ethanol (Sigma) with 10 mM HCl (Riedel-de Haën, Seelze, Germany) upon stirring for 60 min at 25°C. For the condensation reaction, 1 ml of bacterial suspension in PBS (pH 7.0) was treated with 30 μl of the prehydrolyzed TEOS solution (final concentration of silica in the mixture, ∼30 mM). The reaction was stopped after 60 min of silicification by centrifugation at 20,000 × g followed by washing with PBS. Subsequently, the samples were suspended in a small volume of PBS, yielding a densely packed bed of bacteria for signal enhancement in SAXS measurements.

SAXS measurement.

The samples were filled into quartz glass capillaries (∼1.5-mm diameter and 10-μm wall thickness) using a syringe. The capillaries were then sealed followed by measurement of the samples in vacuum by SAXS. SAXS was performed with a rotating anode generator equipped with a pinhole camera (Nanostar; Bruker AXS, Karlsruhe, Germany) and CuKα radiation monochromatized and collimated from crossed Goebel mirrors. X-ray patterns were measured with a 2D position-sensitive detector (Vantec 2000; Bruker AXS) for at least 4 h for each sample. Examples of the effect of an increase of the measuring time to improve the signal-to-noise ratio can be found in Fig. S1 and S2 in the supplemental material. The scattering patterns were radially averaged to obtain the scattering intensity in dependence on the scattering vector q = 4 π/λ sin(θ), where λ = 0.1542 nm is the X-ray wavelength and 2θ the scattering angle.

For strain ATCC 12980, the intensities of the sample with (S+) and without (S−) the S-layer were normalized in the q-value range of about 2 nm⁻¹. It turned out that this also led to identical intensities in the low q-value range. The value for the sample with the S-layer was subtracted from the value for the sample without the S-layer. The difference in the scattering intensities was then attributed to the scattering of the S-layer alone.

For strain MW5, the samples to which silicification was applied (with [Si+]) and the wild-type samples (without silica [Si−]) were also normalized in the q-value range of ∼2 nm⁻¹. In similarity to ATCC 12980, where the value for S+ sample was subtracted, in the case of MW5, the value for the sample with the silica surface templating layer (Si+) was subtracted from the value for the sample without TEOS treatment (Si−). This is due to the fact that only electron density differences are responsible for the SAXS signal, and the electron density difference between the values of samples without surface templating or without the S-layer, respectively, and the solvent is higher than that for the corresponding values of samples with silica surface deposition or with the S-layer. The difference in the scattering intensities can then be attributed to the Si-layer, which is templating the S-layer on the surface of the bacterium.

Details about SAXS data evaluation can be found in the supplemental material.

Transmission electron microscopy.

Freeze-etching was carried out in a BAF 060 freeze-etching unit (Leica, Wetzlar, Germany). Fracturing and etching of frozen samples were done at −96°C for 90 s prior to platinum/carbon shadowing (6). Replicas were purified for 30 min in 70% sulfuric acid and then neutralized in MilliQ water and subsequently subjected to 14% sodium hypochloride solution (for 3 to 5 min), followed by three washing steps in MilliQ water and immobilization on 400-mesh TEM copper grids (Agar Scientific, Stansted, United Kingdom). Freeze-etched and negatively stained preparations were examined by transmission electron microscopy (TEM) (Tecnai G² 20 Twin transmission electron microscope; FEI, Eindhoven, The Netherlands) operating at 80 keV. Pictures were taken with an FEI Eagle 4 k charge-coupled-device (CCD) camera (4,096 by 4,096 pixels). The magnification was calibrated by using negatively stained catalase crystals (32).

RESULTS

TEM analysis of freeze-etched bacterial cells.

Freeze-etching and metal shadowing of intact cells of strain ATCC 12980 conducted in the course of the present study revealed an oblique S-layer lattice with lattice parameters of 9.4 nm and 5.8 nm and an angle γ of 82° (Fig. 1). On cells of strain MW5, an S-layer lattice with a lattice parameter of approximately 14.5 nm and an angle of approximately 59° was identified (28) (Fig. 2). These data are in agreement with the determination of a hexagonal S-layer lattice that is obtained from a curved cell surface. The lattice parameters determined for strains ATCC 12980 (26, 27) and MW5 (28) corroborate previously published values for both strains (Table 1). Generally, it should be considered that different sample preparation techniques might result in slightly different lattice parameters (33).

Fig 1.

Electron micrograph of a freeze-etched and metal-shadowed preparation of G. stearothermophilus ATCC 12980, revealing the outermost rectangular S-layer lattice with lattice parameters of a = 9.4 nm and b = 5.8 nm and a base angle of 82°.

Fig 2.

Electron micrograph of a freeze-etched and metal-shadowed preparation of A. serpens MW5, revealing a hexagonal S-layer lattice with a lattice parameter of ∼14.5 nm and an angle of 59°.

Table 1.

Overview of the lattice parameters for the S-layer of G. stearothermophilus ATCC 12980 and A. serpens MW5 according to different methods

Bacterium	Lattice parameter(s) and value(s) (reference)
	FEa-TEM	SAXS	Literature
G. stearothermophilus ATCC 12980	a = 9.4 nm	d1 = a = 9.0 nm	a = 9.9 nm
	b = 5.8 nm	d2 = b = 5.9 nm	b = 5.7 nm
	γ = 82°	γ = 90°	γ = 86°
		dth = 7.1 nm
	Lattice type: oblique	Lattice type: rectangular	Lattice type: oblique (self-assemblies, synchrotron) (46)
A. serpens MW5	a = 14.5 nm	a = 18.0 nm (d = 15.6 nm)	a = 18.5 nm (d = 16.0 nm)
	γ = 59°	γ = 60°	γ = 60°
		dth = 15.6 nm (single layer)
	Lattice type: hexagonal	Lattice type: hexagonal, double layer	Lattice type: hexagonal (TEM, negative stain) (28)

^aFE, freeze-etching.

SAXS measurements.

The inset in Fig. 3 shows the normalized scattering intensities of strain ATCC 12980 with the S-layer (S+; red line) and without the S-layer (S-; black line). Subtracting the intensities results in the curve in Fig. 3 (large image), and the dashed line corresponds to a fit with a function describing infinitely large plates with a Gaussian distribution of the thicknesses with a value of d_th = 7.1 nm ± 1.7 nm. The error bars were obtained from the program GNOM (34), based on the Monte-Carlo technique (35, 36). In a first approximation, it is common to use a formal factorization and describe the experimental scattering intensity by the product of a mean form factor, P(q), and an effective structure factor, S(q): I(q) ∼ P(q) S(q) (37). The ratio of the experimental intensities to the theoretical form factor of the plates then gives information on the structure factor S(q). Figure 4 shows this ratio, where two intensity peaks are visible with maxima at q₁ = 0.7 nm⁻¹ and q₂ = 1.07 nm⁻¹, corresponding to distances of d₁ = 9.0 nm and d₂ = 5.9 nm in real space, respectively. As no higher-order peaks are detected, the simplest assumption is that of a 2D lattice with two perpendicular axes. In this case, the two distances d₁ and d₂ would be the lattice parameters, i.e., the repeating unit of the lattice in two orthogonal directions.

Fig 3.

SAXS measurements of G. stearothermophilus ATCC 12980. Insert, normalized scattering intensities of strain ATCC 12980 with (red line) and without (black line) the S-layer. Large image, difference signal attributed to the S-layer (open circles) and fit function describing infinite plates with a Gaussian thickness distribution of d_th = 7.1 nm ± 1.7 nm (dashed line). a.u., arbitrary units.

Fig 4.

Ratio of experimental (exper.) scattering intensities to the intensities from the fit with a plate for strain ATCC 12980 (full black line and dashed line in Fig. 3). This visualizes the structure factor; two peaks are visible with maxima at q₁ = 0.7 nm⁻¹ and q₂ = 1.07 nm⁻¹, corresponding to lattice parameters of d₁ = 9.0 nm and d₂ = 5.9 nm in real space, respectively.

The same evaluation procedure was applied to the scattering intensities of strain MW5. However, the bacterium was exposed to silicification with TEOS solution for 1 h to template the surface for enhancement of the electron density contrast. Scattering curves obtained after 4 and 6 h were similar, and only the 6-h curve is shown here. Subtraction of the sample with the Si-templated S-layer from the native, nonderivatized sample (inset in Fig. 5, full and dashed lines) leads to the scattering intensity of the templated layer alone (full lines in the large image of Fig. 5). The dashed line is again a fit with a function describing plates with a broad Gaussian size distribution and a mean thickness of d_th = 3.3 nm ± 0.6 nm, which is the typical thickness of the Si-layer in this case. However, the value for the slope toward low q values (representing the shape of the object in this size range) is smaller than 2 (the value for a perfect plate). The slope is in this case the (fractal) dimension, and it is consistent with the interpretation of the Si-layer as a rough plate with a preferably 1D ribbed structure. Analogously to strain ATCC 12980, the ratio of experimental intensities to the fit for a plane results in an effective structure factor, which is shown in Fig. 6. The intensity modulations are attributed to a system of two layers with a hexagonal structure (28). If one layer (at a distance c) is shifted by one half-repeat along a (1, 0) lattice line, only reflections with even values (height [h] plus twice the length [2l]) remain. Thus, the (110) reflection is absent, whereas the (011) reflection is visible. All possible reflections are identified in Fig. 6 and are consistent with the model of two hexagonal layers with an in-plane lattice distance of 15.6 nm (center-to-center distance of units, 18.0 nm) and an out-of plane distance of c = 15.6 nm; i.e., the out-of-plane angles between the units incline at the same angle as the in-plane units do (Fig. 7).

Fig 5.

SAXS measurements of A. serpens MW5. Insert, normalized scattering intensities of strain MW5 with native, nonderivatized S-layer (black line) and Si-template S-layer (red line). Large image, difference signal attributed to Si-layer templating the S-layer (open circles) and fit function describing infinite plates with a Gaussian thickness distribution of d_th = 3.3 nm ± 0.6 nm (dashed line).

Fig 6.

Ratio of experimental scattering intensity to the intensity from fit with a plate for strain MW5 (full lines and dashed lines in Fig. 5). This visualizes the structure factor; indicated peak positions (dashed lines in Fig. 5) correspond to a hexagonal 2D lattice of two hexagonal planes with a lattice distance of d₁ = 15.6 nm (corresponding to a center-to-center spacing 18.0 nm), and one is shifted by half of the lattice line (9.0 nm).

Fig 7.

Schematic presentation of the double-layered S-layer of strain MW5. (a) Top view of the two superimposed layers. For clarity, the top layer (gray) is shown as not completely covering the bottom layer. (b) Cross section through the double-layer assembly along the dashed line in panel a to demonstrate the superposition of both layers. In similarity to the closely related strain A. serpens VHA (49), it was assumed that the two S-layers of strain MW5 are attached head to head.

DISCUSSION

Comparing TEM and SAXS data with regard to the investigation of S-layers, the main advantage of SAXS is that measurements can be performed in a natural environment such as a solution of arbitrary composition. Though the X-ray beam fully operates in a vacuum, the samples are in a solution: they are easily prepared by filling a solution into capillaries, which are then sealed vacuum tight. The values for the scattering signals from the sample in solution and the blank are subtracted from each other by performing an appropriate normalization to the scattering volume. Though this simplicity is rather striking, there are also some drawbacks of the method. (i) No direct image is obtained, and the information has to be extracted from (indirect) scattering data. Thus, the interpretation of the data may not be unambiguous. (ii) SAXS is an integral technique, and the scattering signal is dependent on the scattering volume. In the particular case of S-layers on intact bacterial cells, the S-layer volume on the surface is rather low in comparison to the volume for the whole organism. Consequently, the scattering volume and, thus, the scattering signal are rather weak. The difficulties arising therefrom are shown in the supplemental material for A. serpens MW5 by evaluation of data from different measurement times (compare with Fig. S1 and S2 in the supplemental material).

Two ways to extract the information of the S-layer on intact bacteria in natural environments are proposed in this work. These are (i) measurement of bacteria with and without the S-layer and subtraction of the X-ray signals and (ii) enhancement of the scattering contrast by deposition of an Si-layer only a few nanometers in thickness on the outermost surface of the bacteria. This was performed by submersion of the bacteria in an aqueous TEOS solution for 1 h at room temperature. Subsequently, as in the first approach, the values for the X-ray signals with and without a templated layer were subtracted to yield a value corresponding to the scattering of the surface layer alone. To avoid complete gelation in the present study, conditions using 30 mM TEOS were applied to the bacteria that were milder than the previously reported conditions for silicification (38). Thus, only small (1.7% total concentration [vol/vol]) amounts of ethanol were also added to the bacterial suspensions, preserving the viability of the bacteria (39, 40).

However, since the volume of the cell exceeds by far the volume of the S-layer, a rather strong scattering intensity of inhomogeneities within the cell in comparison to the layer is induced. This leads to rather long measurement times of at least 4 to 6 h, if a rotating anode generator is used, and can thus be seen as a challenging measurement even for modern laboratory X-ray sources. This is more a technical limitation than a limitation due to the method. A way to overcome these technical restrictions would be the use of a much more brilliant X-ray beam from third-generation synchrotron radiation sources, which would reduce the measurement times significantly and concomitantly improve the signal quality (41–43).

The obtained data were divided through the form factor to result in modulations of the X-ray intensities, which show an effective structure factor. For the form factor, the first approach is to choose for description the Fourier transform of a 2D layer with a certain thickness distribution. This thickness represents additional information which cannot be obtained by TEM. The structure factor obtained by this procedure exhibits only relatively small intensity variations of the X-ray signal, which indicates that the S-layer in natural environments is not a perfect 2D crystal. The coherence length, estimated for strain MW5 from the first peak by Scherrer's formula, gives a crystallite size of close to 300 nm, corresponding to roughly 20 in-line coherently scattering units. The line broadening and, concomitantly, the limited size of the ordering of the subunits within the layer might have been induced by several causes. One might have been the undulations of the surface, which—in similarity to properties of other 2D materials such as graphene (44)—could stabilize the structure. On the other hand, these undulations led to a decrease in the intensities of Bragg reflections. It may be speculated that the much more pronounced and intense Bragg reflections observed in a concentrated solution of purified S-layer self-assembly products (23) were induced by layer-to-layer interactions, which are also able to stabilize a 2D layer but which strongly enhance the crystallinity. The same effect of peak broadening could also have been the consequence of a local curvature of the cell surface, exhibiting the same effect as undulations. Another cause could be defects in the natural S-layer structure, either from missing subunits or from growth processes starting simultaneously from different positions on the surface and leading to a mosaic structure with grain boundaries (45). Still another explanation could be that, in the natural environment, cells originating from different growth stages, but not synchronized cells, are present. Thus, for an integral measurement method such as SAXS, the S-layer structure can be seen as representing an average from the early occurrence to the final stage, which would limit the mean size of the coherently scattering regions.

Even if the intensity modulations in Fig. 4 and 6 are small, the size of the lattice parameters measured by SAXS is in nearly perfect coincidence with those obtained from TEM analysis, which supports the validity of the chosen approach to evaluate the SAXS data. It should be noted that the corresponding TEM images are very important for a validation of the interpretation of the SAXS data. For strain ATCC 12980, a layer thickness with a Gaussian size distribution of d_th = 7.1 nm ± 1.7 nm was obtained by SAXS, where no equivalent value is available by TEM, but the in-plane arrangement of the subunits from SAXS resulted in a 2D lattice with lattice constants of d₁ = 9.0 nm and d₂ = 5.9 nm. Former synchrotron radiation experiments with self-assemblies of the S-layer protein SbsC from strain ATCC 12980 revealed lattice parameters of 9.9 and 5.7 nm, respectively (46). Additionally, an angle of 86° was measured between the two axes (46). This angle could not be measured by SAXS in our study, because only two peaks were visible in Fig. 4, and, therefore, a rectangular lattice was assumed for simplicity. Nevertheless, there is a nearly perfect coincidence between the structural parameters of the natural S-layer on the bacterial surface and those of the self-assembled S-layer in vitro. Comparing SAXS and TEM, freeze-etching experiments with these bacteria revealed an oblique lattice with lattice parameters of 9.4 nm and 5.8 nm and an angle of 82°. These results are also in agreement with the SAXS data. Minor deviations might occur due to curvature of the bacterial cell surface, which might particularly influence the angle enclosed by the two main axes.

According to the literature (28), the S-layer protein on strain MW5 is composed of two hexagonal sheets having p6 symmetry and a lattice distance of 16 nm (corresponding to a lattice parameter [center-to-center distance] of 18.5 nm). One sheet is staggered by one half-repeat along the lattice line of the p6 lattice relative to the second one so that, in projection, the pattern of the composite layer is a translational moiré pattern (28). Freeze-etching of whole bacterial cells in the course of this study revealed a hexagonal S-layer lattice with a lattice parameter of approximately 14.5 nm (Fig. 2). The evaluation of the X-ray intensities in Fig. 5 and 6 resulted in a mean thickness of the templating silica layer of d_th ∼ 3.3 nm. The value for the slope toward low q values in Fig. 5 is between 1 and 2, which corresponds to an object with a fractal dimension value between 1 and 2. This can be interpreted as a rough plate with a 1D ribbed structure, which is in accordance with the corresponding TEM images. The thickness of the Si-layer templating the S-layer on the cell surface grows in correspondence with the increases in the amount of time that the bacteria are subjected to silicification (47, 48).

The intensity modulations after division of the theoretical form factor of a plate visualize the structure factor and are attributed to two layers with an in-plane arrangement of the subunits in the form of a hexagonal 2D lattice with a lattice distance of 15.2 nm (center-to-center distance of units, 17.5 nm), evaluated from the first four peaks. An evaluation of all six peaks leads to a slightly larger lattice distance of a = 15.6 nm (center-to-center distance, 18.0 nm). The second plane has to be shifted by half the repeat length (i.e., 9.0 nm), which has the effect that only reflections with even (h + 2l) values appear and the index 1 value has to be 0 or 1 (see Fig. S3 in the supplemental material). As no additional (001) peak at lower q values is visible, this indicates that the distance c of the two layers is identical to the in-plane lattice constant and that the basic structural units incline at the same angle in plane and out of plane (Fig. 7). The distance of the two layers of 15.6 nm is additional information that is not accessible by surface-sensitive experimental methods. This head-to-head arrangement of the two S-layers of strain MW5 is comparable to previously reported data on the closely related strain A. serpens VHA (49, 50). Although the peaks visible in Fig. 6 are not very pronounced, the good coincidence of SAXS and TEM results supports the view that, in this case, the S-layer in the natural environment is not significantly different from the S-layer obtained after sample preparation for freeze-etching and subsequent shadowing with Pt and C (in vacuum and templating with Pt). Table 1 gives an overview on the S-layer lattice parameters for the two investigated bacteria, G. stearothermophilus ATCC 12980 and A. serpens MW5, as obtained from the different methods.

Summarizing, SAXS and TEM are complementary methods to determine the structure of the S-layer. Whereas TEM gives local information after templating, measuring in vacuum, and subsequent image processing, SAXS is able to measure the biological structure in natural environments. However, the interpretation of the SAXS patterns may be sometimes difficult; the signal is rather weak, and background subtraction is crucial. Background subtraction can be performed by measuring a bacterium with and without the S-layer. An extension of the method is to use silica surface deposition, which enables measuring the structure of the S-layer on any S-layer-covered bacterium in natural environments. It turns out that both methods, SAXS and TEM, lead to very similar results, which confirms the enormous stability of the S-layer.