Synchrotron X-Ray Microdiffraction Reveals Intrinsic Structural Features of Amyloid Deposits In Situ

Abstract

Amyloidoses are increasingly recognized as a major public health concern in Western countries. All amyloidoses share common morphological, structural, and tinctorial properties. These consist of staining by specific dyes, a fibrillar aspect in electron microscopy and a typical cross-β folding in x-ray diffraction patterns. Most studies that aim at deciphering the amyloid structure rely on fibers generated in vitro or extracted from tissues using protocols that may modify their intrinsic structure. Therefore, the fine details of the in situ architecture of the deposits remain unknown. Here, we present to our knowledge the first data obtained on ex vivo human renal tissue sections using x-ray microdiffraction. The typical cross-β features from fixed paraffin-embedded samples are similar to those formed in vitro or extracted from tissues. Moreover, the fiber orientation maps obtained across glomerular sections reveal an intrinsic texture that is correlated with the glomerulus morphology. These results are of the highest importance to understanding the formation of amyloid deposits and are thus expected to trigger new incentives for tissue investigation. Moreover, the access to intrinsic structural parameters such as fiber size and orientation using synchrotron x-ray microdiffraction, could provide valuable information concerning in situ mechanisms and deposit formation with potential benefits for diagnostic and therapeutic purposes.

Introduction

Amyloidoses constitute a heterogeneous group of acquired or inherited diseases characterized by the presence of amyloid deposits, mainly in the extracellular space, consisting of insoluble aggregated misfolded protein in the form of fibrils (1–3). To date, 27 different human proteins are known to have the ability to switch from a soluble structural form to insoluble fibrillar aggregates. Several additional components are ubiquitously present in all amyloid deposits, including glycosaminoglycans (GAG), apolipoprotein E (ApoE), and serum amyloid P-component (SAP). The fibril precursor proteins are the basis of the modern nomenclature of amyloidoses (4). Among these, serum amyloid A protein (SAA) is the precursor protein responsible for amyloid-associated (AA) amyloidosis, which sometimes complicates disorders that result from various chronic inflammations (5). The kidney is the organ most frequently affected in AA amyloidosis, which induces a progressive deterioration of renal function (6). Renal biopsy is widely used for the diagnosis, with immunohistochemical techniques that reveal the AA protein accumulation as amorphous matter inside various kidney compartments. Such deposits generally prevail in glomeruli and are also frequently present in vessels, and more rarely in interstitium and along the tubular basement membranes (7).

Amyloid deposits are identified by a characteristic fibrillar signature in electron microscopy images, a typical cross-β x-ray diffraction pattern, and histological staining reactions, with a strong affinity for the Congo red dye with apple green birefringence under polarized light. Although not fully understood, the fibrillar nature of the amyloid substance and its molecular configuration in the β-folded layer is supposed to be responsible for these characteristic tinctorial properties.

During the last two decades, major advances have been made in the field of amyloid structure characterization, partly thanks to technical developments. Recently, atomic force microscopy observation of scrapie-infected cells revealed a growth process of amyloid-like fibrils on the cell surface (8). Imaging the cell surface allowed characterization of the fibril distribution and their characteristic lateral and length dimensions, which were shown to be in agreement with in vitro structures. However, as noted by the authors, the observed structures may result from a cellular response to the scrapie infection by changing membrane composition or enhancing dendrite outgrowth, for example. Therefore, a characterization of the folded structure within the fibers is necessary to achieve an unambiguous demonstration of the cross-β structure and the amyloid nature of the deposits. At the molecular scale, using a combination of data from various in vitro experiments and molecular simulation procedures, many sophisticated structure models were built (9,10). Spectroscopic analyses, such as circular dichroism or infrared spectroscopy of fibrils, generated in vitro or extracted from tissues all lead to the same conclusion about the high β-sheet content of the fibrils. The particular insoluble form of this cross-β-secondary-structure-rich core is an admitted basic common structural characteristic of amyloid protein fibrils (11). The lack of solubility prevents the use of the classical high-resolution techniques of structure determination involving protein solutions, such as NMR or crystallography.

Therefore, electron microscopy and x-ray fiber diffraction are the most widely used techniques. They are often handled in a complementary way giving access to information at the fiber and molecular scale, respectively. Electron microscopy was the first technique to demonstrate the fibrillar nature of the deposits and is thus generally used to characterize the fiber morphology (12).

As reported throughout the literature, x-ray fiber diffraction reveals a cross-β secondary structure of amyloid materials (13). Current studies using this technique mainly deal with the detection of cross-β folding in amyloid structures formed in vitro using various peptides or proteins. These studies often aim to demonstrate the molecule propensity to form amyloid deposits. It has been suggested from various in vitro studies (14) that different proteins have the ability to form amyloids under appropriate conditions (15). These all share a common secondary-structure core, as evidenced by x-ray fiber diffraction (16). This molecular feature is nowadays the unique common information extractable from all fiber diffraction data. Additional data, such as high-resolution molecular structure, have recently been obtained on small designed amino acid sequences, but not on the longer ones, which are implied in amyloidosis (17).

Although very useful for molecular structure determination, the modes of preparation of such in vitro samples, which sometimes lead to highly ordered diffraction patterns, can nevertheless induce some biases on the fibers and their organization with respect to their natural state. Indeed, fibrillar peptides very likely interact with other tissue components during the fibrillation process. This state can be different from in vitro conditions in which the peptides undergo various purification and conditioning treatments (18). Up to now, the so-called in vivo amyloid structural studies have been carried out on amyloid material, which was extracted from the tissue and then purified. As with the fibers generated in vitro, they stain with thioflavin and Congo red in a similar manner. However, comparative studies of ultrastructural shapes of in situ isolated fibers using electron microscopy have concluded that a difference exists between their fine structure depending on the fixation used: cryofixation or glutaraldehyde fixation (18). Another study on AA amyloid isolated from a tissue used the classical Pras method (19), which consists of repeated washings, first with saline water and then with distilled water. It was shown that a macrofibril scaffold on which fibrils were formed was disrupted. The fibrils, whose diameter was initially 1–3 nm when associated with the microfibril scaffold, were shown to reorganize into structures resembling those observed in fibrils prepared in vitro. These observations raise questions about which fine structure reflects the actual in vivo state. The best answer would be given by in situ structure studies. However, to date, no technical approach that we know of has made it possible to acquire such information. Indeed, only a very few infrared studies have been carried out on in situ material using procalcitonin (20) and Aβ amyloids (21). All authors concluded that there was an increased content of the characteristic β-sheet amyloid signature in areas of tissue containing these deposits. Surprisingly, neither circular dichroism nor x-ray diffraction experiments have been performed with in situ fibers.

Here, we present the first study that we know of carried out on paraffin-embedded and frozen human renal tissue sections, with the aim of revealing the structural features of amyloid fibrils in deposits without the potential for structure modification induced by extraction. To this end, we used synchrotron-based microdiffraction, which provides access to micrometer-sized spatial resolution. This allowed us to detect variations in the molecular structure of the fibers along the tissue at the micron scale. In this way, fiber orientation could be followed along a glomerulus, and the variation inside and around the whole glomerulus unit could be monitored.

We further demonstrate that the in situ fibers in amyloid deposits are partly folded into cross-β sheets, exhibiting a diffraction feature at 4.7 Å as fibers formed in vitro. This validates the structural data from extracted fibers and strengthens the use of this diffraction signal as a main criterion of fibril formation in vitro. We provide the experimental conditions for detecting the characteristic 4.7-Å reflection inside tissue cuts. Working on the tissue allows accessing intrinsic information about the molecules, and we thereby show a correlation between the position inside the glomerulus and the partial orientation of the amyloid fibers.

Material and Methods

Sample preparation

Renal tissue samples of an individual with a systemic AA amyloidosis secondary to unknown hereditary disease were used for this study. They were obtained from a radical nephrectomy performed before renal transplantation. The presence of amyloid was established by the appearance of an apple green birefringence from alkaline Congo red staining under polarized light. AA amyloidosis was demonstrated by positive immunofluorescence staining with mouse anti-human Amyloid A monoclonal antibody (clone mc1; 1:10 dilution; DakoCytomation, Glostrup, Denmark). The samples were either frozen into liquid nitrogen or fixed in 4% buffered paraformaldehyde and embedded in paraffin according to the routine tissue processing for pathological examination.

For an initial examination of the amyloid deposit morphology, the renal tissue specimens were cut at 5 μm thickness and stained with Congo red. After verification, serial sections were cut from paraffin-embedded specimens 50 μm thick and frozen specimens 40 μm thick and inserted between two mica muscovite plates (cleaving planes [001]) for x-ray diffraction analyses.

X-ray microdiffraction mapping experiment

The experiments were performed at the European Synchrotron Radiation Facility (Grenoble, France) at the microfocus beamline ID13 (22). The high-intensity monochromatic beam (wavelength λ = 0.976 Å), obtained with an undulator source and a Si(111) double-crystal monochromator, was focused with an ellipsoidal mirror (focal spot 20 (h) × 40 (v) μm²) and then size-limited down to a 2-μm-diameter circular section by a collimator placed in the focal plane. A guard aperture (Pt-Ir, 10 μm in diameter) reduced diffuse scattering from the collimator exit. Samples were mounted with the cut perpendicular to the x-ray beam on a computer-controlled gantry coupled with a microscope, which permitted sample positioning with 0.1-μm resolution.

Experiments were carried out using a sample-detector distance of 149 mm. The distance calibration was done with mica Bragg spots whose first-order spacing is 4.48 Å. Thanks to the small size of the beam stop, two-dimensional x-ray scattering patterns were acquired from 0.008 to 0.4 Å⁻¹. After some trials to determine the best experimental parameters, each pattern was recorded with an exposure time of 4 s on a MAR-CCD camera (16-bit readout; 130-mm entrance window; 2048 × 2048 pixels; pixel size of 64.46 × 64.46 μm²). No radiation damage effects on the scattering features were detected for this exposure time, in agreement with recent observations on radiation effects on fibrous proteins (23).

Various zones along the whole tissue section were sampled randomly by mapping with large step sizes. When the patterns presented a significant signal, a more detailed map was acquired all around with step size of a few micrometers.

X-ray microdiffraction data treatment

The x-ray diffraction data were examined using the Fit2D software (24). The patterns were first rapidly analyzed by visual inspection to make a selection of the most interesting zones. A selection based on two criteria was used: the overall intensity of the pattern and the presence of a reflection at 4.7 Å. We then focused our analysis on one glomerulus to follow the variability of the diffraction patterns over the whole glomerulus unit and all the surrounding tissue.

It is well known from birefringence in polarized light, that the signal obtained from paraffin-embedded tissue contains a component from the paraffin overlaid with that of the fibrous proteins. In a similar way, diffraction patterns exhibited signals arising from paraffin embedding and also from the mica sheets used as sample holders. Therefore, the data-processing step involved subtraction of a scattering pattern of the mica plates and of the paraffin. However, it was not possible to completely remove the parasitic signal arising from these materials. Therefore these two signals will be present with a more or less intense contribution in the data presented here. No further data treatment was applied to minimize any distortion of the weak scattering features from the tissue.

For several selected patterns, the radial profiles were obtained over limited angular sectors to avoid intense parasitic reflections. This was achieved by circular integration within the same angle around the equator and around the meridian (encompassing the 4.7-Å extent of reflection) when the pattern was anisotropic, and over 360° when it was isotropic.

Results

Histological aspect of renal AA amyloidosis

The histological examination of the renal parenchyma revealed an abundant renal amyloid load, predominantly in glomeruli. All glomeruli indeed showed a nearly complete obliteration of glomerular tuft by amyloid deposits. Renal vessels were also involved in the amyloid deposits. Thus, arterioles frequently showed a complete replacement of the vessel walls by amyloid. Some amyloid deposits were also noted in interstitium and along tubular basement membranes. All these amyloid deposits were Congo-red-positive (Fig. 1 a) and strongly stained with mouse anti-human amyloid A monoclonal antibody (data not shown).

Figure 1.

(a) Massive renal AA amyloidosis clearly predominant in glomeruli (Congo red staining, original magnification 100×). (b and c) An x-ray diffraction zone is approximately indicated (b) with the whole produced diffraction patterns (c). (d) Amyloid fiber distribution based on the diffraction intensity of the 4.7-Å reflection arising from cross-β folded molecules within the kidney tissue.

The in situ amyloid molecular fibers display the same 4.7-Å ring as those formed in vitro

Fig. 1 b shows a nonstained image of a renal slice 50 μm thick. The zone indicated inside the rectangle was mapped by x-ray beam and at each point a whole diffraction pattern was recorded and mapped in Fig. 2 c.

Figure 2.

Examples of the three kinds of collected patterns in an amyloid laden renal cut. (a) Diffuse signal from nonorganized proteins inside and outside the glomerulus. (b) Superimposition of a diffuse signal and a well-defined diffraction amyloid fingerprint at 4.7 Å. (c) Diffraction pattern from in situ oriented amyloid fibers.

The scan of the sample allowed us to distinguish three types of diffraction patterns, depending on the location of the tissue section: those with no 4.7-Å reflection, from outside the glomerulus (Fig. 2 a); those with a thin reflection at 4.7 Å, from inside the glomerulus (Fig. 2 b.); or even some with a thin anisotropic arc (Fig. 2 c). The intensity of this peak was roughly estimated by the relative height of the peak after background subtraction. Indeed, as the sample was embedded in paraffin, as is customary in anatomopathological conditioning, the paraffin diffraction peaks were present at a higher intensity than the protein peaks. In such conditions, signal processing to remove the paraffin and extract the 4.7-Å intensity was not possible without a large disturbance of the values. The intensity distribution obtained in this way is shown in Fig. 1 d, where one can see the higher amyloid concentration inside the glomerulus, indicated by the higher signal intensity.

Going further inside the glomerulus, examination of the whole set of scattering data reveals two distinct categories of patterns. The first exhibits two broad isotropic rings at 10.5 Å and 4.2 Å (Fig. 2 b). These diffuse rings are characteristic of protein-based tissues. The one around 10 Å corresponds to the mean interchain distances, which are usually observed in concentrated proteins in a dense state without any specific organization at supramolecular scale. The broad reflection around 4.2 Å is characteristic of disordered or poorly ordered folding of polypeptide secondary chains. This reflection is generic to noncrystalline proteins; it is, for example, present in hair tissue superimposed on other characteristic structural features (25). In fact, the two broad reflections are similar to the signal arising from a hair follicle in the amorphous bulb zone, which has no specific relative organization between the proteins (26). This can be evidenced from the superposition of a signal extracted from amorphous keratin (Fig. 3). We can see, on this figure, the high similarity between the renal nonamyloid signal and a concentrated nonorganized protein signal. The diffraction peak from renal proteins around 10 Å is slightly wider, which may be related to a larger variability in its composition in proteins with regard to hair composition.

Figure 3.

Superimposition of the diffraction-extracted profile typical of areas without amyloid signal (black) on a profile acquired from hair-follicle diffraction in the nonorganized zone (gray).The two sharp peaks arise from paraffin embedding of the sample.

The second category of patterns inside the glomerulus exhibits a sharp ring at 4.7 Å (Fig. 2, b and c), which so far has only been observed for in vitro formed or extracted amyloid fibers. It occurs in most of the inner part of the glomerulus patterns and is the landmark of amyloid fibers. It is characteristic of cross-β folding, which is supposed to constitute the core of the fibers.

From the angular width of this arc, the coherent length of the core-β fibrils can be estimated at ∼85 Å. This value is of the same order as various amyloid fibers (10). In the small-angle x-ray scattering (SAXS) region, no particular signal is detectable; we find only a large diffuse halo.

Outside the glomerulus, the major and most significant difference is the absence of a 4.7-Å thin arc, in agreement with the absence of Congo-red-positive amyloid deposits in this area. Only the two broad diffuse isotropic rings at positions 10.5 Å and 4.2 Å are found to have a lesser intensity than inside the glomerulus. Their widths are narrower inside the glomerulus than outside. This indicates a better-defined and more homogeneous molecular architecture. In the SAXS region, a large diffuse anisotropic halo indicates the existence of randomly organized large objects. Its oblong shape reveals a preferential orientation, which varies across the sample.

Amyloid fibrils formed in vivo show a local intrinsic orientation inside the tissue

The characteristic cross-β-sheet signal at 4.7 Å is reinforced along two opposite directions in numerous patterns collected in the glomerulus region (an example is shown in Fig. 2 c).

This is indicative of a nonrandom orientation of the amyloid fibers inside the footprint of the beam (2 μm × 2 μm × 50 μm), since the 4.7-Å reflection from a cross-β folding appears along the fiber axis (meridian). A similar preferential orientation has already been observed in crystallized amylogenic oligomers or when the fibers have been oriented mechanically, for instance (see Jahn et al. (27) for a review). However, our patterns present no concomitant equatorial signal reinforcement along the equator around 10 Å, as expected for oriented fibers. This may reflect an important contribution of nonamyloid proteins to the diffracted intensity at 10 Å, preventing the detection of any small contribution of amyloid material. This effect may be reinforced by a poorly ordered lateral packing of the molecules in the fibrils. This point is discussed in detail in the Discussion section.

The anisotropic orientation seems to be globally linked to the glomerulus morphology

To investigate this anisotropy further, we collected x-ray microdiffraction patterns of a renal tissue section in scanning mode with sufficient resolution to allow establishment of a possible correlation with histological components. A large glomerulus zone and its surrounding (300 × 140 μm²) was scanned with a resolution of 7 × 7 μm² (Fig. 1 c). A surprising observation was that the mean direction of the intensity reinforcement of the 4.7-Å reflection is roughly the same over the whole glomerulus, which means that the amyloid fibers have a common average preferential orientation inside the glomerulus section. This result strongly supports a relationship between histological zones and the produced scattering features. Unfortunately, the experimental conditions did not allow us to establish a histological identification at the same scale; moreover, the degree of orientation is not high enough (see Fig. 4) to allow quantification of the degree of orientation. In Fig. 4, one can see that the difference between the meridian and the equatorial intensities is too small to be used for this purpose (Fig. 4, inset). However, the sensitivity of this diffraction technique allows one to evidence fibrous protein orientation on the two-dimensional pattern (Fig. 2 c).

Figure 4.

Two extracted profiles from an oriented pattern from the glomerulus. Integration over orthogonal sectors of 10° angular width along the meridian (gray) and the equator (black). The small shoulder (black arrow) arises from the sample supporting material, and the two sharp intense peaks come from paraffin embedding. (Inset) Representation of the difference between the meridian and equatorial profiles.

Fig. 4 also shows that the equator and meridian intensities at 10.5 Å are identical.

Discussion

The molecular structure is independent of sample preparation conditions

Various in vitro studies on amyloid fibrils revealed the existence of different molecular architectures from the same precursor protein, depending on physical or chemical treatments. For example, heating the prion-like Ure2p fibrils at 60°C changed the molecular organization within the fibrils (9). Our data are identical for the frozen and paraffin-embedded tissues. Therefore, the treatment for paraffin embedding, especially the heating to 60°C, has no effect on the fibril molecular organization. Indeed, the two characteristic reflections of cross-β structure for paraffin-embedded samples are present at the same positions in patterns of frozen samples (data not shown). Paraffin embedding therefore appears harmless compared with cryofixation or glutaraldehyde chemical preparation. Nevertheless, previous ultrastructural analyses of specimens treated with those procedures resulted in a helical or filamentous form, depending on the pretreatment (28). In this study, the structural similarity of the two sample treatments can be explained by the presence of the whole-deposit components, which may contribute to the structure's stability.

Similarities and differences between in vitro or extracted amyloid fibers and ex vivo ones

A cross-β core structure evidenced by the meridional 4.7-Å feature has already been shown to be common to all known amyloids, as much for in vitro generated fibers as for those extracted from a tissue. To the best of our knowledge, the only case where a cross-β structure was observed with minimal tissue treatment is for transthyretin (TTR), which is present in the vitreous fluid and throughout other organs of familial amyloid polyneuropathy patients. The structures of the TTR fibrils from the vitreous fluid (without Pras method extraction) and those extracted from kidney by the Pras method are identical (29). No data about in situ cross-β molecular structure have been reported so far. Therefore, the results presented here show, for the first time that we know of, that the molecular structures of amyloid material deposited in vivo inside a tissue share with in vitro or extracted amyloid fibers the common characteristic cross-β core structure. This is very important in validating the data obtained for in vitro samples, which are used to develop models based on the cross-β core structure. Indeed, a plethora of data from various proteins that have been coerced into making fibrils in the test tube have used this figure to support the conclusion that amyloid fibrils were actually produced.

On the other hand, differences between in vitro or extracted oriented fibers and in situ ones are apparent by looking at the equatorial 10.5-Å scattering region. We observe a nearly isotropic ring in the patterns, with an oriented peak at 4.7 Å. This is better illustrated when looking at the projections in Fig. 4. This reflection is sometimes slightly more intense around the equator, whereas it is more pronounced and anisotropic for fibers formed in vitro or extracted. The most likely explanation is that in vitro and extracted amyloid fibers are in general purified systems in which the cross-β core structures can self-assemble laterally in a rather compact and regular way, unlike the in situ cross-β core structures, which are associated with other components forming a complex aggregation scaffold. The scaffold of the AA components may prevent lateral direct association between amyloid fibrils, thus leading to the absence or weakening of a specific amyloid signal around 10 Å. The scattering features from the cross-β core structures are thus superimposed against those of the other components, and are consequently less visible than the signal of pure cross-β core structures. For instance, one can assume a contribution from AA components such as SAP, which are known to have a mostly β-sheet secondary structure. This assumption is reinforced by previous studies on tissue-extracted AA amyloids, which self-assemble again after disruption of the macrofibril scaffold (18).

Moreover, the fine-structural examination of fibrils using ultrastructural techniques has suggested that in situ AA proteins are made of 10- to 30-Å-diameter helical filaments that are aggregated into a microfibril-like structure, able to contribute to the diffuse 10 Å signal.

Another likely explanation, which is not in contradiction to the previous one, consists in assuming that the precursor proteins could also preserve their native secondary folds and associate longitudinally, giving rise to the cross-β signal without the necessity of interactions for stabilizing their lateral structure. This is supported by in vitro studies of extended β-structured peptides, which show that long fibers can be formed without any lateral packing (10), giving rise to a very weak 10-Å feature. Indeed, it is likely that the absence of a well-defined protofibrillar structure leads to an absence of this 10-Å reflection.

All these assumptions are in agreement with electron microscopy images of different samples of various amyloids, which showed that sections from AA-containing samples produced diffuse images. Those images are difficult to interpret and cannot provide the number of protofilaments in the fiber section, whereas this was easily done for other types of ex vivo amyloid samples (28).

It can be summarized from our x-ray patterns that only the association during the axial growth seems similar to that observed for in vitro preparations, unlike the lateral packing. This indicates the specificity of the interactions between some amino acids that initiate the self-assembling into cross-β on large coherent domains as long as ∼85 Å.

Morphology-related parameters from microdiffraction: implications for fibrillogenesis in a tissue

It follows from this study that fiber orientation can be measured with a 2-μm precision at the x-ray beam size resolution. Our results show that in the whole glomerulus, fibers are preferentially oriented in the direction parallel to the axis that joins the afferent and efferent arterioles to the proximal convoluted tubule (Fig. 1 b, dashed arrow).This direction is indicated on the diffraction pattern with the dashed arrow joining the two intensity maxima in Fig. 2 c, although the orientation is in fact continuous over 360°. This anisotropic and uniform orientation is somewhat surprising given the heterogeneity of the glomerulus. Indeed, to obtain in vitro equivalent quality of orientation, various sample constraints are necessary, such as shear constraints, mechanical stretching, or intense magnetic field application.

The common average orientation of the fibrils in a given area could have been detected with birefringence analyses (30). However, two main points should be reported here. First, birefringence studies are always carried out after Congo red staining, and it is the dye intercalation that likely follows fibril orientation that is detected. This might introduce artifacts that prevent a possible correlation of the measured birefringence signal with structural information. In our diffraction study, the measured parameter is the direct fiber orientation without any extrinsic staining. This may allow us to measure well-characterized parameter values such as the orientation angle and its width and the number of oriented molecules. Moreover, these parameters can be directly correlated to an exact position in a structural tissue unit such as the glomerulus in this case, with a resolution defined by the operator as the mapping step size. On another hand, observations have been made about fibril orientation at the electron microscopy level, showing a correlation between fibril shape and the surrounding cells or connective tissues (31). This is in agreement with the correlation established between fibril average orientation and glomerulus morphology.

An explanation for the origin of the observed common preferential orientation could be due to a physicochemically driven growth process of the amyloid deposit, but it could also result from a specific interaction of the fibers with given subglomerulus components, the surrounding tissue applying an effect similar to that of mechanical constraints.

To date, data about in vivo fiber growth are very limited and are based only on the assembly of in vitro amyloidlike fibrils using amyloidogenic peptides. Their kinetics is consistent with a nucleation process. Accordingly, monitoring the accumulation of amyloid in animal models is consistent with the nucleation phenomenon that follows a sigmoidal shape. The first scenario would imply starting from a unique fiber acting as a seed, which would afterward impose its orientation during the whole-deposit growth. The existence of a unique seed seems not realistic, however; several seeds probably appear simultaneously in the glomerulus. The uniform orientation, which is present in the whole glomerulus, therefore suggests another scheme for fiber deposition.

Our suggestion is an alternative scenario where fiber growth is governed by tissue-specific physical or chemical factors and takes place when an adequate precursor concentration is present. The interaction with subglomerulus components is much more likely. Indeed, the direction of the average orientation of the fibers on the section corresponds to the axis of the glomerulus, which joins the afferent and efferent arterioles to the proximal convoluted tubule. Looking at a histological image, one can see that the capillary tuft of the glomerulus is formed by a series of sections of the capillary that are mainly directed along the axis of the glomerulus and that form loops between two sections. Therefore, the amyloid fibers would be deposited along the glomerular basement membrane (with the fiber axis lying on the membrane), and an anisotropic angular distribution of the fibers would be expected.

Another hypothesis may involve a growth steered by molecules other than the amyloid protein, and which may be involved in the fiber polymerization process, leading to their intrinsic alignment. The tissue can then be seen as a composite material of normally folded proteins reinforced by a well-organized distribution of fibrous molecules soaked into the amorphous matrix; this would lead to the exceptional resistance of the deposits to various treatments such as proteolysis. To understand fibrillogenesis more fully, it would be useful to complete this work with a detailed analysis of the in vivo driving conditions of these structures monitoring the 4.7-Å reflection, following the process at various stages of the pathology with animal models, and to track down the zones where the deposits are initiated.

X-ray microdiffraction as a structural imaging tool in amyloidosis

Amyloidosis is typically identified by the birefringence of Congo red staining of a biopsy, whereas its typing relies on immunohistochemical tests. Congo red has been shown to react with any highly aggregated material, regardless of its composition (32).

Although the staining process is easy and inexpensive, the chemical mechanisms are poorly understood, and results can vary widely depending on the protocol or the reagent quality. Moreover, the Congo red dye was shown to be nonspecific, as it induces circular dichroism even with native structures, as demonstrated on a variety of native proteins (33). This study showed that the Congo red must be used with caution as a diagnostic tool for the presence of amyloid fibrils.

Our results presented here emphasize the high capabilities of x-ray diffraction as an imaging tool for ex vivo samples with no post treatment after renal tissue resection. This technique is quite sensitive to the presence of amyloid fibers inside a tissue, as well as to fiber orientation. It is also the only approach we know of that unambiguously gives access to the molecular fingerprint of amyloid fibers, able to overcome the weaknesses of other techniques, such as FTIR-microscopy. The FTIR technique has already been used for Aβ-peptide secondary-structure distribution (34) inside a tissue section, but it is not sensitive enough to distinguish cross-β structures from extended-β ones.

Experimental x-ray maps could give access to even more information than the fiber distribution and orientation parameters, such as size and possibly basic structural characteristics. One could expect to have more variability in the structures, depending on the precursor protein, the amyloid-laden tissue, or the local biological conditions. In vitro studies show that a given peptide can give rise to a large variety of cross-β fibers according to the experimental conditions, which might be of a biological significance. Given the potential importance of the environmental effects on the formation of amyloid aggregates, in situ diffraction studies of deposits formed in vivo may also provide information about this precise environment state.

A further interesting aspect of our work concerns these additional structural parameters becoming available to characterize amyloidoses. The current classification of amyloidoses is based on fibril precursor proteins, which are normally found in the plasma and usually without any biochemical connection. However, differences at the supramolecular level of the protofibril assembly are known to exist ex vivo (28). Our study opens the possibility of access to new parameters that could characterize amyloid deposits, from the atomic or molecular level to the microscopic level: crystallite size from coherence length, possible lateral crystallinity parameters, quality of organization, and orientation. We shall then have means to characterize the various amyloidoses using additional parameters, which become significant in situ, whereas they might be influenced by the sample preparation when measured in vitro.

Conclusion

We succeeded in recording in situ fiber x-ray microdiffraction patterns from an amyloid-laden human kidney tissue specimen, opening new perspectives for future investigation tools in the field of amyloidosis.

Our data provide evidence that the in situ fibers in amyloid deposits are partly folded into cross-β sheets as in fibers formed in vitro. Moreover, we prove that despite its weak intensity, this structural signature is directly observable in tissue sections, opening new opportunities for histological examination.

New pieces of information were added to the existing data from studies carried out on in vitro systems. First, the lateral organization of the cross-β core objects is disrupted, probably because of other components that are associated with the amyloid molecules, unlike the longitudinal cross-β ordering, which seems to be the same as for in vitro systems. Second, the amyloid fibers are more or less naturally aligned locally within the beam footprint (2 × 2 × 50 μm³). Third, the mapping mode at a few micrometers resolution showed that the alignment direction of the amyloid fibers is roughly the same over the whole glomerulus section. The orientation of the fibers, which is usually used in vitro to improve the quality of the diffraction pictures and thus facilitate the structural interpretation, is observed here as intrinsic to the tissue. This structural feature is certainly significant, because it can be specific to the tissue type or to the precursor proteins and their fibrillation process.

The detailed analysis of these various parameters in diverse controlled conditions will be very important for advancing our understanding of the biological processes involved in the mechanisms of the amyloid fiber formation and the exact conditions of their deposition. To our knowledge, these effects are not accessible by any other method in a condition so close to in vivo and with molecular resolution. Designing well-controlled experiments that will allow us to obtain similar information in vitro would be very helpful in improving the understanding of fibrillogenesis conditions and in following the effects of possible treatments on a tissue.