Download Abstract

Imaging Sub-Cellular Structures with Ultrahigh Resolution Radiology
P. Y. Tseng*, W. L. Tsai*, Y. Hwu *, G. C. Yin†, M. T. Tang†, K. S. Liang†, Fu-Rong
Chen†,‡ , Wenbing Yun§, H. –I. Yeh¶, J. H. Je||, G. Margaritondo**
*
Institute of Physics, Academia Sinica, Nankang, Taipei, Taiwan
†National
‡Engineering
Synchrotron Radiation Research Center, Hsinchu, Taiwan
and System Science, National Tsing Hua University, Hsinchu, Taiwan
§Xradia
Inc., 4075A Sprig Drive, Concord, CA, 94520, USA
¶Mackay
||Department
Memorial Hospital, Taipei, Taiwan
of Materials Science and Engineering,
Pohang University of Science and Technology, Pohang, Korea
**
Faculté des Sciences de Base, Ecole Polytechnique Fédérale,
CH-1015 Lausanne, Switzerland
ABSTRACT
We present the first successful test of microscopy of intracellular structures with hard-x-rays.
This new approach yields better lateral resolution than optical microscopy and provides
3-dimensional and topographic images of thick specimens. The key factors in our success are
the use of a coherent synchrotron source with a high definition phase zone plate x-ray
microscope and the exploitation of contrast mechanisms based on the refractive index. In
particular, using the zone plate to magnify the projected x-rays we produced high-quality
images of human tumor cells with 60 nm resolution. We also present the results of a phase
retrieval method based on a wave propagation algorithm from a series of defocused images, to
improve high-resolution cell radiographs taken with unmonochromatized hard-x-rays.
INTRODUCTION
Resolution and contrast, two key characteristics of every microscopy, can be improved in the
x-ray spectral region by taking advantage of the steadily improving source brightness and
coherence and by using phase contrast. We present experimental evidence that with this
strategy hard x-rays can now image thick biological samples down to the sub-cellular level.
Coherence plays a multiple role in this approach. First, it enables the optical elements to better
perform focusing or magnification. It also solves the fundamental problem of limited contrast
caused by the weak absorption of x-rays by using a non-conventional image formation
mechanism: contrast based on the differences in the refractive index. Such a mechanism was
successfully tested in recent years (1, 2) – and we show here that it leads to subcellular
resolution using highly penetrating hard x-rays.
Absorption microadiology with hard-x-ray (wavelength ~1Å) did achieve quite high lateral
resolution (3). However, no results have been presented so far on sub-cellular level imaging
with nanometer scale resolution.
Cells can be observed in microscopy using soft-x-rays with wavelengths in the so-called “water
window” (284-543 eV) that maximizes the contrast between carbon-containing areas and water
(4, 5). With advanced x-ray optics, subcellular structures could be observed in hydrated states
(6-8). However, x-rays in this spectral range cannot penetrate specimens much thicker than
about 10 µm - approximately a single cell - preventing applications in cell biology at the tissue
level.
Higher-energy x-rays – the object of our present tests - can image thicker samples in their
natural state with the use of refractive index differences (also know as phase contrast). The
high resolution imaging in 3-dimensions of large biological samples permits biologists to
explore the relations between structures and biological functions.
In preliminary tests, we recently succeeded in using hard-x-ray to image cells in a fresh and
unstained state (9). The thickness reached several mm in some cases. We also demonstrated
that with a reasonably simple fixation process tomography reconstruction can be performed
with limited loss of lateral resolution -- starting from a large number of projection images of a
thick sample
The question then is: how far can we push the resolution of hard-x-ray images of biological
samples? Recently, using high definition x-ray zone plates microradiology reached 60 nm
resolution with first order diffraction and 30nm with third order diffraction (10, 11). We
applied this approach to image human tumor cells and found that even with coherence-based
mechanisms the natural contrast within biology samples is still limited and must be enhanced
when ultrahigh lateral resolution is required. Staining becomes essential in this case to obtain
high quality subcellular images. Furthermore, to study for example proteins and cellular
organelles it is necessary to label the specific target.
Immunocytochemistry is the most commonly used method for specific labeling in visible-light
microscopy, fluorescence microscopy, laser confocal microscopy and transmission electron
microscopy. However, labeling strategies developed for such microscopy techniques do not
automatically match the requirements of x-ray microscopy. We therefore developed methods
of specific cell labeling for x-ray microscopy based on conventional immunocytochemistry. By
combining the high penetration and high resolution of x-ray microscopy with the specificity of
immunocytochemistry, we could investigate specific protein structures and 2-dimensional or
3-dimentional distributions -- achieving nanometer-level resolution.
Our tests went beyond mere labeling. A novel phase retrieval technique based on wave
propagation algorithm enabled us to obtain high quality images at the subcellular level. Phase
effects also enabled the zone plate microscope with a phase retrieval device (a phase ring
similar to the device used in optical phased contrast microscopes) to obtain high resolution
microradiology images with strongly enhanced contrast.
These combined improvements produced the first subcellular micrographs of cells with
resolution better than optical microscopy and much simpler sample preparation methods.
Substantial further improvements can be foreseen. Efforts are underway to achieve more
versatile functional labeling and radiological functional imaging in the nanometer scale. The
current 30nm lateral resolution with the x-ray zone plate microscope is still far from the
theoretical diffraction limit and even from the practical instrumentation limits. With better
microfabrication, the lateral resolution of the x-ray phase zone plate can be further improved.
Higher brightness x-ray sources can improve the time resolution reducing the motion blur for
live specimens and increasing the throughput.
BACKGROUND
A major factor in our successful tests was contrast based on the refractive index, previously
used (1) to image unstained cells. This mechanism enhances the visibility of the edges between
regions with different refractive index. As a result, fine details can be observed at the cellular
and sub-cellular level without any staining or other contrast-enhancing cell preparation.
This is an interesting complementary approach to the standard optical microscopy of living
cells. In fact, x-rays microradiology can examine thicker samples in a more “natural”
environment.
A key result of the present work is the demonstration that sub-cellular details within a sample
as thick as 5 mm can be clearly identified at a selected depth from the sample surface without
sacrificing the lateral resolution. Furthermore, we obtained radiographic images from
hard-x-ray with 30-60 nm resolution -- well beyond the diffraction limit of optical microscopy.
Similar nanoscale resolution using hard-x-rays was only demonstrated previously with
sophisticated phase retrieval method with diffraction x-ray microscopy (12).
Edge enhancement based on the refractive index cannot be easily observed with conventional
x-ray sources due to their large size and angular spread (1). In fact, the enhancement is due to
small deviations of the x-rays propagating through the edge, a region with high
refractive-index gradient. A large source size and angular spread wash out the effect. To reduce
this problem, one can use an aperture to limit the effective source size or increase the
source-sample distance. However, this greatly reduces the portion of the emitted x-rays that is
actually used for radiological imaging.
Synchrotron sources eliminate the problem because of their natural coherence (1, 13). The
x-rays are emitted from a small source area and are strongly collimated. Edge enhancement
thus becomes easily visible and – as we demonstrate here – micro-radiology of live cell details
becomes feasible. Our results show indeed that fine details can be imaged with high lateral and
time resolution in a variety of biology systems -- including leaf skin cells, human tumor cells,
mouse neurons and rabbit bone cells.
MATERIALS AND METHODS
Instrumentation
The tests were performed with a custom-designed Nano-TXM system made by Xradia, Inc. in
operation at the beamline 01B of the National Synchrotron Radiation Research Center
(NSRRC), Taiwan (10, 11). By using the third order light from a Fresnel zone plate objective
lens, this system achieved 30nm of spatial resolution, the world record for hard x-ray
microscopy techniques.
BL01B is a 5 Tesla-superconducting wavelength shifter (SWLS) beamline equipped with a
double crystal monochromator (DCM) and a toroidal focusing mirror that provides
monochromatic photon beams with energies from 5 keV to 20 keV. The energy resolving
power (E/E) of the DCM is 1000 and the average photon flux is of the order of 3x1011
photon/sec/200mA at the sample position (14).
The x-rays emitted from SWLS are focused by a focusing mirror in order to reach higher
photon intensity, and then monochromatized by the DCM.
After passing through the
condenser lens, the monochromatic x-rays illuminate the object at its focal point. The
transmitted x-rays are then magnified by Fresnel zone plate objective and projected on the
imaging detector. The focal length of the current version of the zone plate is ~3cm and its
diameter is 85μm with 50 nm outermost zone width. Estimated with the Rayleigh criterion, its
optimal resolution is 60 nm. The image detector consists of a CCD camera with 20X optical
objective lens coupled to a scintillator. The magnification of the x-ray imaging system by the
zone plate is 45X, yielding a combined magnification of 900X. The microscope also has a
phase contrast mode. This imaging mode requires a special condenser and a phase ring made
out of gold. The phase contrast imaging mode is very useful for imaging features with low
absorption contrast, such as organic materials in a biological specimen.
The cell samples used for ultrahigh resolution microradiography were labelled by
nickel-compound staining to increase the absorption contrast. The photon energy was fixed at
8 keV to simultaneously optimize the focusing efficiency of the zone plate and the x-ray
penetration. The typical exposure time for acquiring a single image with 900X magnification
was 8 minutes.
The ultrahigh resolution tests at NSRRC’s Nano-TXM were accompanied by tests performed
with a microradiography system of high imaging speed whose details are described in (15, 16).
Unmonochromatized x-ray beams from the beamline 7B2 of the Pohang Light Source (PLS)
and the beamline 1B2 of the Taiwan Light Source (TLS, NSRRC) were used to illuminate the
sample. The transmitted x-rays were captured by a cleaved CdWO4 single-crystal scintillator
and converted to a visible image. This image was then magnified by an optical lens before
being captured and stored by a CCD camera.
The samples were mounted on a translation/rotation stage for precise positioning. For
biological samples with intrinsically limited x-ray absorption, a single image (typically
1600×1200 pixel, horizontal field of view (FOV) 500 µm) could be taken within 100 ms. The
corresponding radiation dose did not produce any detectable damage.
The sample can be translated along the direction of the incoming synchrotron x-rays, varying
the sample to detector distance from <1mm to 1.2 m. For the phase retrieval based on the wave
propagation algorithm, a series of defocused images can be taken at different sample detector
distances and these images are manually aligned before the iteration process to generate phase
and intensity maps.
Cell culture and immunocytochemistry
For the ultrahigh resolution (~60nm) imaging, HeLa cells were cultured on plastic film with
thickness of less than 100 µm to increase the contrast and specimen penetration. To localize the
vimentin, the intermediate filaments inside the cytoplasm, the cells were fixed by 4 %
paraformaldehyde and then blocked by 3 % bovine serum album in phosphate buffer saline
(PBS) plus 0.1 % Triton X 100 to eliminate the non-specific binding and by 0.1 % H202 to
neutralized the endogenous peroxidase. The cells were incubated in 100X diluted primary
antibody vimentin (obtained from Dako®) for 2 hours at room temperature. After rinsing by
PBS, the cells were incubated in 200X secondary antibody with peroxidase conjugation
(obtained from Chemicon®) for 1 hour. The 3,3’-diaminobentidin (DAB) kit with nickel
enhancement (VectorLab) was used as a chromogen. For x-ray imaging, the cells were
dehydrated by gradient ethanol 30 %, 50 %, 70 %, 95 %, 100 %. Parallel tests of laser confocal
fluorescence imaging to compare the vimentin pattern were performed by incubation of a Cy3
fluorophore conjugated secondary antibody. A non-specific staining procedure was also
applied to the cells. In this case, the cells were post-fixed by 1 % osmium tetroxide for 1 hour,
then stained by 2 % uranyl acetate for 1 hour.
The natural contrast is sufficient in many cases to see detailed cell structures. For these tests,
leaf cell samples were obtained from Hippeastrum epidermis. Tissue specimens were obtained
by peeling off the bottom skin of the leaf and then sandwiched between two moisturized
Kapton foils to preserve water. The thickness of the leaf skin was approximately 30-50 µm and
that of the Kapton foil approximately 50µm. Such a procedure was not required to observe
cells: it was possible to see individual cells even in the leaf without peeling. The peeling and
sandwiching procedure, however, eliminated a problem: the superposition of many cell layers
results in complicated images. An alternate solution for this problem is tomography. We
demonstrated the tomography reconstruction of plant cell by using thick samples (>500µm)
taken from the skin part of an aloe leaf.
A fibroblast cell line (HIG-82) was prepared from the fibroblast of rabbit joints. The cells were
cultured in 35 mm culture dishes. The medium was Ham’s F-12 medium (2mM L-glutamine
and Earle’s BSS adjusted to contain 1.5 g/L sodium bicarbonate, 0.1mM non-essential amino
acids, and 1.0 mM sodium pyruvate 90%; fetal calf serum, 10%). Dishes with ~ 80% cell
confluence were prepared for microradiology tests after fixation.
The collagen used for sealing was obtained from 3 rat tails and dissolved in 0.1% acetate at the
concentration of 2 mg/ml. The fibroblast cells were observed with the culture dish water-tight
sealed and vertically placed in the path of the x-ray beam.
For preparation of mouse aorta, adult C57/BL6 mice were anesthetized with ether inhalation
and were perfusion-fixed via direct intracardiac injection, initially with heparinized PBS (10
units/ml) followed by phosphate-buffered 2% paraformaldehyde (pH 7.4) for 10 minutes (17).
The thoracic aortae were dissected and cut into transverse rings. Thereafter the aortae were
embedded in resin using standard preparation procedures for thin-section electron microscopy
(18).
The above-described preparation procedures are much less complicated than the standard
methods for optical microscopy and other microscopy techniques – and resulted in specimens
very close to their natural state. Specifically, no staining or other contrast-enhancing procedure
was used. The cultured cells were fixed to preserve their morphology since the x-ray beam
propagates horizontally and therefore the specimens had to be placed vertically. The specimens
were much thicker than for optical microscopy and an opaque medium could be used.
RESULTS AND DISCUSSION
Ultrahigh resolution x-ray microscopy and staining
We used DAB with nickel enhancement as choromogen to identify the distribution of vimentin.
DAB yields a brown stain. When nickel chloride is added to the substrate solution, a gray-black
stain appears. Vimentin is an intermediate filament protein (58 kiloDaltons) that is generally
found in a variety of cells of mesenchymal origin. Vimentin filaments help supporting cellular
membranes. Vimentin networks also help keeping the nucleus and other organelles in fixed
places within the cell. Figure 1 shows the vimentin network revealed by fluorescence (Fig. 1a)
and visible light phase contrast microscopy (Fig. 1b) (FOV = 170μm). In this image, the cell
margin with sparse vimentin filaments can be identified. However, the resolution was
unsatisfactory: the individual vimentin bundles were blurred and hard to distinguish from each
other.
High resolution x-ray microscopy using unmonochromatic synchrotron x-rays can achieve
similar resolution (~1µm) as in Fig.1. For example, in Fig. 2 the staining of the sub-cellular
structure can be clearly seen in each cell (FOV = 600μm). The addition of nickel provides
good absorption contrast. Because the vimentin is located inside the cytoplasm rather than in
the nucleus, the nucleus is not affected by the vimentin staining and therefore has lower x-ray
absorption. However, the cytoplasm contains dense vimentin that causes staining especially at
the peri-nucleus area and produces sufficient x-ray absorption making it look darker than the
nucleus. This phenomenon could also be observed with an optical microscope and confirms the
dense meshwork structure of vimentin in the peri-nucleus area. When examined with a laser
confocal fluorescence microscope, the vimentin bundles could be clearly identified at the
lamellipodial structure of cell edges. However, the vimentin bundles formed dense meshwork
supporting the nucleus and could not be clearly identified. Figure 1A is an example of such
image obtained with a laser confocal microscope where the vimentin meshwork near the
nucleus produced only a blurred image (area marked by a rectangle).
With an ultrahigh resolution zone plate x-ray microscope, the image quality was substantially
improved. In Fig. 3, for example, the vimentin inside the cell and outside the nucleus manifests
itself as linear arrays throughout the cytoplasm (FOV =30μm). The texture of vimentin bundles
near the nucleus could be clearly identified. Vimentin also appears as a meshwork structure at
the peri-nucleus area. There were some vimentin bundles in the image area corresponding to
the nucleus; these features should be the projections of vimentin bundles located before or after
the nucleus and overlapped in the 2D radiography images. The nucleus and the cell boundary
were clearly delineated by the vimentin meshwork. Although most vimentin bundles formed
meshworks inside the cytoplasm, especially close to the nucleus, sparse vimentin filaments
could be observed.
Higher magnification view of vimentin staining at the cell margins revealed the individual
vimentin bundles in the cells filopodia (indicated by arrows in Fig. 4, scale bar=1 µm). Note
that in this case, the width of the smallest vimentin bundles detected is less than 100nm,
showing that the 60nm resolution is maintained in this image. This level of performance
clearly exceeds what optical microscopy can offer. Because of the linear relation between
absorption and image contrast, the light, gray staining pattern indicated sparse, loose filament
bundles, and the heavy, dark staining pattern indicated dense, overlapping filament bundles.
Vimentin and other cytoskeletal elements not only provide a support of cell structure but also
play an important role in determining the dynamic response to shear forces. The distribution of
vimentin bundles and the pattern of vimentin meshwork can be related to the cell responses to
exogenous forces (19). Therefore, a high resolution x-ray microscope can also study the
structure and dynamic properties of these intermediate filaments or other cytoskeletons.
We also stained these cells by osmium tetroxide and uranyl acetate, a common method to
enhance contrast for TEM. In this case, the cell nuclei showed strong absorption contrast in
x-ray microscopy because uranyl acetate stained the nucleotide acid. Osmium tetroxide also
provided limited contrast for delineating the cell margins (Fig. 5A, the cell margins were
marked by a red line, FOV = 600μm). These cells were also examined by SEM (scanning
electron microscopy), but only the cell margins could be detected (Fig. 5B) and the nucleus
could not be identified. These tests indicated that this standard staining procedure for TEM can
be also applied to x-ray microscopy. The results are superior to those of SEM, a standard
method for high resolution cell imaging: under the same sample preparation without specific
enhancing methods SEM only shows the cell surface and does not reveal structures inside the
cells. This was perhaps one of the reasons comparatively limits the SEM applications in cell
biology and histology.
To study the expressions, functions, distribution, and locations of molecules, proteins and
sub-cellular organelles, specific labelling of the target is required. As we already mentioned,
immunocytochemistry is widely used to specifically label the target molecules in both
visible-light and transmission electron microscopes. A variety of immunocytochemical agents
such as fluorescent probes, immuno-gold particles, DAB and other chromogens are used for
specific proteins. DAB chromogen with nickel enhancement is a common immunolabelling
method for visible light microscopy and TEM. The nickel particles provide good contrast for
both cases. We found that DAB with nickel enhancement also provides good contrast for x-ray
microscopy. When examining the cells with DAB and nickel labelling with x-ray microscopy,
the fine structure of vimentin meshwork could be observed with much better resolution than
visible-light microscopy. DAB with nickel enhancement provided a uniform staining along the
filament bundles. This property makes DAB a better staining method to study the
protein-protein interaction than other standard methods for TEM immunolabeling, such as
immunogold particle staining with silver enhancement. Nano-gold conjugated secondary
antibodies were enhanced with silver to form aggregates approximately 50 nm in diameter.
This produced a discontinuous, aggregated, speckled staining pattern that would distort the fine
details in ultrahigh resolution imaging (20). Also note that DAB staining is not limited to Ni
enhancement. Simple modifications of the staining procedure result in silver-gold or uranyl
enhancement (21). By adjusting the X-ray photon energy to match different metal absorption
edges, the effect of dual or triple staining can be achieved. We can therefore conclude that
DAB is so far the most effective staining method for ultrahigh resolution x-ray microscopy.
These results show in general that with this simple method we can specifically label the
proteins of interest to study their structure and distribution. Although immunocytochemistry is
used for specific labelling of proteins in visible-light microscopy, the corresponding spatial
resolution is limited at least for some proteins and protein complexes. In order to obtain more
information about the precise protein location and protein-protein interaction, images with
better resolution are required.
TEM provides superior resolution, but the processes of dehydration, embedding and preparing
ultra thin sections are time-consuming. Besides, TEMs are restricted to examining very thin
specimens (typically less than l μm) and provide very low contrast. Although 3D images can be
obtained from TEM by serial sections and algorithm reconstruction, the z-axis resolution of
these images is limited. Hard-X-ray microscopy can penetrate thicker specimens (more than
300μm, with nano-TXM and 10 mm with ~1µm resolution), compared to soft X ray
microscopy (up to 10μm) and TEM (less than 1μm). This high penetration power makes it
possible to examine an entire cell and multi-cellular structure without serial sectioning. This
eliminates the consequent possibility of losing information or obtaining distorted information.
The image in this case is the 2D projection of the 3D structure of a cell. However, 3D
tomographic reconstructions of cells can be foreseen.
High speed high resolution microradiology of live cells with refractive index imaging
Without using advanced nanostructured zone plates, the x-ray imaging resolution is limited to
~0.5 µm. This is mainly the result of low detector resolution. Since the typical size of a cell is
a few µm, this resolution level is sufficient in many cases to see subcellular structures. For
plant cells, due to the differences in composition between the cell wall and the cytoplasm, it is
not difficult to see the outline of the cell using hard-x-rays. In a previous study (9), we took
advantage of this characteristic and successfully performed tomography reconstruction to
reveal the full 3D structure of the leaf skin cells including stoma cells.
With these cells we also demonstrated real-time imaging of live specimens. A careful
examination of sequences of images like those in Fig. 6 reveals features beyond the mere
opening and closing. These include sub-cellular features and the induced changes (such as
compression and displacement) of the surrounding cells. The live detection of such minute
movements is quite remarkable if one considers the complete absence of staining.
This approach provides only limited information of the sub-cellular structures but has the
marked advantage of reaching very high time resolution. Thanks to the high brightness of the
synchrotron x-rays, a snapshot can be obtained in ~1ms with spatial resolution ~1µm. This
time resolution, combined with the high penetration of x-rays, makes it possible to examine
specimens in their living environment in air and moisture and to observe their natural
movements. In fact, without time resolution the spatial resolution is degraded by motion
blurring – which is quite limited in the images of Fig. 6.
Plant cells are rather easy to image with our approach. Figure 2 already demonstrated that
animal cells can be observed as well. We also verified in a previous work on cultured and fixed
neuron cells from a mouse brain (9) that not only neuron bodies can be detected but also the
interconnecting axons. Note that in this study the neuron cell specimens were not specifically
optimized for x-ray imaging. In fact, the collagen used for sealing was not specially designed
for microradiology and therefore led to weak contrast. The fixation collagen were also quite
thick, ~300µm, much more than normally required for optical microscopy.
Figure 7 shows cultured fibroblast cells from rabbit joints. The cells shown in Fig. 7A were
fixed with paraformaldehyde before mounting the specimen in a vertical position. The fixed
cells were tightly packed and all with a fibril or spindle shape. In contrast, Fig. 7B shows the
same type of cells imaged without being fixed: the cells are not close packed and
morphological modifications can be seen. In this case, the cells were still alive before the x-ray
microscopy examination. Nevertheless, the image quality is still quite reasonable, indicating
that our approach can be used to image mammalian cells in their natural state.
Our results thus show that refractive index imaging based on coherent x-rays can produce clear
images at the cell level without staining. It is quite important to compare this approach to other
techniques and assess its advantages, limitations and possible future improvements. The
resolution obtained so far is similar to optical microscopy and can be substantially improved.
Soft-x-ray microscopy has better resolution but has not been implemented for samples thicker
than a single layer of cells. Transmission electron microscopy and scanning electron
microscopy require specific sample preparation procedures and are limited as far as the sample
thickness is concerned.
Overall, the strongest advantage of our approach is the high
penetration of hard-x-rays, and the consequent possibility to examine systems in natural living
conditions. No microscopy can match refractive index microradiology in imaging thick
specimens with the resolution demonstrated above.
Our approach is not immune from problems. The possible radiation damage is less severe than
in absorption-contrast radiology but it cannot be automatically neglected. Although radiation
effects are not evident in our tests, this issue must be carefully explored to assess its impact.
The longer data taking time required for the tomography could aggravate the problem. On the
other hand, the current time per frame (1ms) (22) and the corresponding dose can be reduced
by 1-2 orders of magnitude with better detectors.
With the continuous improving of the lateral resolution, more radiation effect is expected to the
biology specimen and cyro-preservation (20) will become essential to the routine use of x-ray
microscopy. Development of better sample preservation during the measurement is inline with
the development of better sample preparation and is currently underway to facilitate this
technology to biomedical imaging. As to other future improvements, the full development of
tomographic reconstruction will even better exploit the key advantage of our approach in
exploring thick living samples up to several mm with minimal preparation. The technique
might be further enhanced by the future ultrabright x-ray laser and by the possibility to
implement new image detection strategies based on their ultrashort pulses.
Phase retrieval and contrast enhancement
As we already argued, phase contrast with coherent x-rays is a suitable solution when
absorption contrast is too limited without staining. However, the limited differences between
organelles and cytoplasm sometimes restrict even the phase contrast effectiveness. Methods to
separate the phase and absorption effects, such as DEI (diffraction enhanced imaging) or the
Zernick approach (implemented in the phase detection mode of the Nano-TXM and shown in
Fig. 4), are very helpful in extracting more information.
The same objective can be reached with numerical phase retrieval methods. One of the most
popular algorithms is the transport intensity equation (TIE) (23) that makes use of the relation
between the phase and the gradient of intensity. The non-interferometric TIE processing has
been applied for phase retrieval in optical microscopy with partially coherent illumination (24)
and in x-ray microscopy using a plane polychromatic x-ray wave (25). Iterative phase retrieval
methods based on wave propagation were further developed and were applied to phase
recovery for soft x-rays (26) and to retrieve the exit wave in HRTEM imaging (27, 28). Both
the smooth variation of phase and the sharp edges of the object can be obtained with the
propagation algorithm (27, 28).
We tested this phase retrieval approach on microradiographs taken with the high-resolution
microradiology system at NSRRC using non-monochromatic x-rays from a wavelength shifter.
We present here results obtained using the propagation algorithm (27) to retrieve the phase of
the exit wave for partially coherent synchrotron x-rays. The peak intensity for this beamline
occurs at the wavelength of ~0.08 nm. The minimum wavelength (defined as 1% of the peak
intensity) is ~0.2 nm.
Images of HeLa cells were taken at 25 different sample-detector distances, ranging from 1 mm
to 300 mm. The results of phase retrieval are shown in Fig. 8. The vimentin inside the
cytoplasm were stained by DAB with nickel enhancement. The motivation for staining in this
case was to obtain a clear difference between the amplitude and phase images as shown in Figs.
8A and 8B. In the amplitude image, the stained subcellular structure is clearly visible. The
comparison with the structures revealed by phase contrast shows the different nature of the
phase and absorption subcellular images. It is clear that with this approach, even without
staining, it is possible to detect subcellular structures based on phase contrast.
Thick samples
The ability of nondestructively examining thick samples is one of key advantages of x-ray
imaging, such as the widely deployed medical computed tomography in hospitals. Resolution
alone cannot justify the technically sophisticated use of x-rays for microscopy since higher
resolution has been achieved by other microscopies. Suitable sample preparation methods
were developed for many different types of samples and microimaging techniques.
Nevertheless, the high penetration depth gives x-rays a marked advantage. High penetration
combined with high spatial resolution provides extreme volume sensitivity. Figure 9 shows a
good example: fine vessels of a few µm diameter can be clearly observed in the abdomen of a
mosquito of size ≈400 µm. With a typical spatial resolution better than 1µm and a penetration
depth >1mm, the volume selectivity is > 109. This selectivity level sharply reduces the
difficulties in preparing the sample and finding the region of interest; furthermore, it makes it
possible to keep the samples in a much more natural environment.
Note that other
microscopies typically observe only very thin samples and therefore require sectioning to
reveal 3D structures. As we already mentioned, highly penetrating x-rays and tomographic
reconstruction eliminate the need for sectioning and associated information loss.
With suitable fixation procedures, we could even obtain high resolution tomography and 3D
reconstructed images of mammalian cells. Figure 10A shows an example: a piece of aorta
fixed in resin; the overall sample dimension, including the resin, is ~ 5mmx7mmx10mm.
Figure 10B shows one of the tomographically reconstructed slices. In this cross section of the
mouse aorta (FOV approximately 240µm×240µm), the detailed structure of the vessel wall is
clearly observed in spite of the fact that the sample is larger than the imaged area. The aorta
wall is layered by the waving elastic laminae, the innermost of which, the internal elastic
lamina (marked by black arrows), is isolated from the lumen (L) by a monolayer of endothelial
cells (white arrows). Figure 10C is the volume rendered 3D structure of this sample. The white
arrows show borders between neighbouring cells, whereas a black arrow emphasizes a nucleus
sticking out from the surface whose outline is well preserved in the reconstruction. Therefore,
imaging in this case goes well beyond the mere outline and shape of a cell and provides
sub-cellular details.
We also performed successful tests of tomographic reconstruction for plants cells – see the
already mentioned results on stoma cells. We typically used a standard filtered back projection
algorithm based on 200-1000 projections. Figure 10 shows another example of reconstruction:
the ~5mm thick freshly prepared skin of an aloe leaf. The image taking was quite fast: 10ms
per projection for 360 projections. The resolution of the reconstructed images were quite good,
for example, that the darker lines on the guard cells are those from the cellulose microfibrils (9),
were reconstructed without much blurring by the “local tomography” effects (29, 30) due to the
fact that the sample is larger than the FOV of each projection. The high imaging speed also
avoided the blurring due to sample movement and clearly shows that tomographic
reconstruction of wet cells is feasible for thick samples.
CONCLUSIONS
The limited absorption by biological tissues and the low lateral resolution so far prevented
x-rays from producing radiological images of individual cells and of subcellular structures. We
could overcome this limitation by taking advantage of the coherence of synchrotron x-rays and
of x-ray zone plates fabricated with state-of-the-art nanotechnology. We specifically presented
hard-x-ray images and tomographic reconstructions - for very thick samples - of different types
of living and non-living cells with a resolution reaching 60nm.
The results of high resolution and high speed microradiology were achieved by using coherent
x-rays from a synchrotron source with no monochromatization (since very limited longitudinal
coherence was required) (9, 13), x-rays detection with high lateral resolution and a detection
geometry that enhances the contrast due to variations in the refractive index. This approach
provides high image taking speed and high resolution (<1 µm) while simplifying the sample
preparation.
We also improved the detector resolution by magnification of the x-ray
transmitted through the sample with a high-definition zone plate. This enabled us to achieve
the best resolution of 60nm resolution while using x-rays to reveal subcellular features of cell
specimens in a rather natural environment.
The importance of these results for cell biology is evident: not only is this approach
complementary to other microscopies, but it could in some cases replace them – for example,
the TEM in the study of tissues. The results at the present stage already surpass those of optical
microscopy, notably as far as lateral resolution is concerned and the possibility to examine
opaque samples with thickness of a few mm.
Acknowledgement
This work was supported by the National Science Council (Taiwan), by the Academia Sinica
(Taiwan), by Mackay Memorial Hospital (MMH-E-94003), by the MOST (KOSEF) through
National Core Research Center for Systems Bio-Dynamics, by the Fonds National Suisse de la
Recherche Scientifique, and by the Ecole Polytechnique Fédérale de Lausanne. We thank Y. F.
Song of NSRRC and Fred XXX of Xradia for helps in the Nano-TXM studies.
References
1.
Hwu, Y., Tsai, W. L., Groso, A., Margaritondo, G. & Je, J. H. (2002) J. Appl. D: Appl.
Phys. 35, 105–120.
2.
Cloetens, P., Ludwig, W., Boller, E., Helfen, L., Salvo, L., Mache, R. & Schlenker, M.
(2002) Proc. SPIE. 4503, 82–91.
3.
Yun, W., Lai, B., Krasnoperova, A., Di Fabrizio, E., Cai, Z., Cerrina, F., Chen, Z.,
Gentili, M. & Gluskin, E. (1999) Rev. Sci. Instrum. 70, 3537–3541.
4.
Kirz, J., Ade, H., Anderson, E., Buckley, C., Chapman, H., Howells, M., Jacobsen, C.,
Ko, C. H., Lindaas, S., Sayre, D., et al.,. (1994) Nucl. Instrum. Methods. 87, 92–97.
5.
Jacobsen, C., Beetz, T., Feser, M., Osanna, A., Stein, A. & Wirick, S. (2002) Surf. Rev.
Lett. 9, 185–191.
6.
Larabell, C.-A., Yager, D. & Meyer-Ilse, W. (2000) Proc. of the 6th X-Ray Microscopy
Conference, American Institute of Physics, College Park, MD. 507, 107–112.
7.
Ford, T., Meyer-Ilse, W. & Stead, A. D. (2000) Proc. of the 6th X-Ray Microscopy
Conference, American Institute of Physics, College Park, MD. 507, 119–122.
8.
Weiß, D., Schneider, G., Niemann, B., Guttmann, P., Rudolph, D. & Schmahl, G.
(2000) Proc. of the 6th X-Ray Microscopy Conference, American Institute of Physics,
College Park, MD. 507, 123–128.
9.
Hwu, Y., Tsai, W. L., Chang, H. M., Yeh, H. I., Hsu, P. C., Yang, Y. C., Su, Y. T., Tsai,
H. L., Chow, G. M., Ho, P. C., et al. (2004) Biophys. J. 87, 4180-4187
10.
Yin, G.-C., Tang, M. Y., Song, Y. F., Liang, K. S. & Yun, W. (2005) Himeji
Microscopies Conference.
11.
Yun, W. (2005) [online]. http://www.xradia.com.
12.
Miao, J.-W., T. Ishikawa, B., Johnson, E. H., Anderson, Lai, B. & Hodgson, K. O.
(2002) Phys. Rev. Lett. 89, 088303.
13.
Margaritondo, G. (2002) Elements of Synchrotron Light for Biology, Chemistry, and
Medical Research. (Oxford University Press, New York).
14.
Song, Y.-F., Chang, C. H., Liu, C. Y., Huang, L. J., Chang, S. H., Chuang, J. M., Chung,
S. C., Tseng, P. C., Lee, J. F., Tsang, K. L., et al. Nucl. Inst. Meth. (in press).
15.
Tsai, W.- L., Hsu, P. C., Hwu, Y., Je, J. H., Ping, Y., Moser, H. O., Groso, A. &
Margaritondo, G. (2003) Nuclear Instrum. Methods B, 199, 436-440.
16.
Baik, S., Kim, H. S., Jeong, M. H., Lee, C. L., Je, J. H., Hwu, Y. & Margaritondo, G.
(2004) Rev. Sci. Instrum.75, 4355–4358.
17.
Yeh, H.-I., Lu, C. S., Wu, Y. J., Chen, C. C., Hong, R. C., Ko, Y. S., Shiao, M. S.,
Severs, N. J. & Tsai. C. H. (2003) Arterioscler. Thromb. Vasc. Biol. 23, 1391–1397.
18.
Blackburn, J.-P., Peters, N. S., Yeh, H. I., Rothery, S., Green, C. R. &. Severs, N. J.
(1995) Arterioscler. Thromb. Vasc. Biol. 15, 1219–1228.
19.
Helmke B. -P., Goldman, R. D. & Davies, P. F. (2000) Circ Res. 86, 745-752.
20.
Meyer-Ilse W., Hamamoto, D., Nair, A., Lelievre, S. A., Denbeaux, G., Johnson, L.,
A.L. Pearson, A. L., D. Yager, D., M.A. Legros, M. A. & C.A. Larabell, C. A. (2001) J.
Microsc. 201, 395-403.
21.
Goto S., Nagahiro, S., Ushio, Y. & Hofer, W. (1992) J. Histochem. Cytochem. 40,
1423-1425.
22.
Hwu, Y., Tsai, W. L., Je, J. H., Seol, S. K., Kim, B., Groso, A., Margaritondo, G., Lee,
K. H. & Seong, J. K. (2004) Phys. Med. Biol. 49, 501–508.
23.
Barty, A., Nugent, K. A., Paganin, D. & Roberts, A. (1998) Opt. Lett. 23, 817.
24.
Barty, A., Nugent, K. A., Roberts, A. & Paganin, D. (2000) Opt. Commun. 175, 329–36
25.
Paganin, D., & Nugent, K. A. (1998) Phys. Rev. Lett. 80, 2586.
26.
Allen, L.-J., McBride, W. & Oxley, M. P. (2004) Opt. Commun. 233, 88-82.
27.
Hsieh, W. -K., Chen, F. R., Kai J. J. & Kirkland, A. I. (2004) Ultramicroscopy 98,
99-114.
28.
Allen, L. J., McBride, W., O’Leary, N. L. & Oxley, M. P. (2004) Ultramicroscopy 100,
91-104.
29.
Anastasio, M. -A., De Carlo, F. & Pan, X. (2003) Proc. SPIE. 5030, 120–126.
30.
Faridani, A., Finch, D., Ritman, E. L. & Smith, K. T. (1997) SIAM J. Appl. Math. 57,
1095–1127.
Figure Captions
Figure 1 -
Micrographs show the vimentin network revealed by A) laser confocal optical
microscopy image of a HeLa cell stained by cy3 where the stained vimentin
meshwork near the nucleus can only be identified by the blurred area (marked by a
rectangle) and B) visible light phase contrast microscopy. FOV = 170μm.
Figure 2 -
High resolution x-ray micrographs using unmonochromatic synchrotron x-rays of
human HeLa tumor cells. The staining of the subcellular structure can be clearly
seen in each cell (FOV = 600μm). The addition of nickel provides enhanced
absorption contrast.
Figure 3 -
Image taken by the ultrahigh resolution zone plate x-ray microscope of the same
sample of Fig.2 shows much improved resolution and quality. (FOV =30μm) For
example, the vimentin inside the cell and outside the nucleus manifests itself as
linear arrays throughout the cytoplasm. The texture of vimentin bundles nearby
the nucleus could be clearly identified.
Figure 4 -
Ultrahigh resolution x-ray micrographs showing the vimentin staining at the cell
margins revealed the individual vimentin bundles in the cells filopodia (indicated
by arrows).
This image was taken with the (Zernick) phase ring showing
enhanced phase effect. The image is stitched by 4 smaller images. Scale bar = 1
µm.
Figure 5 -
A) High resolution x-ray micrographs of HeLa cells with uranyl acetate and
osmium tetroxide staining. The staining provides strong contrast for cell nuclei
and little contrast for delineating the cell margins (cell margins were marked by
red line). FOV = 600μm. B) SEM micrograph of the same HeLa cells. Only the
cell margins could be detected. Nuclei could not be identified.
Figure 6 -
Snap-shots microradiographs of a Hippeastrum leaf taken with broadband
(unmonochromatized) x-rays showing the opening movement of a live stoma cell.
Figure 7 -
A) Radiograph of cultured fibroblast cells extracted from a rabbit bone. The cells
were fixed with collagen before mounting the specimen in a vertical position. The
(horizontal) FOV was 500 µm. B) Image of the same type of cells as in A); in this
case the cells were again vertically mounted but not fixed.
Figure 8 -
Microradiographs of HeLa cells processed with a wave propagation algorithm.
The difference in A) the amplitude and B) the phase maps shows the well
separated absorption and phase of the sub-cellular features.
Figure 9 -
Isosurface representation of tomographic reconstructed 3D model of the abdomen
of a mosquito. A) shows the inside of the abdomen with front surface of the
abdomen as seen in B) removed (by software) and revealed fine vessels of a few
µm diameter. C) shows volume rendered representation. The size of the abdomen
of this mosquito is ≈400 µm.
Figure 10 – A) X-ray micrographs of a mouse aorta sample fixed in resin; the overall sample
dimension, including the resin, is ~ 5mm×7mm×10mm, Scale bar: 400 µm. B) one
of the tomography reconstructed slices. The detailed structure of the vessel wall is
clearly seen even if the sample is larger than the imaged area. The aorta wall is
layered by the waving elastic laminae, the innermost of which, the internal elastic
lamina (marked by black arrows), is isolated from the lumen (L) by a monolayer of
endothelial cells (white arrows). Scale bar: 15 µm. C) The volume rendered 3D
structure of the same aorta sample. The white arrows point to borders between
neighbouring cells, while black arrow points to a cell nucleus which extrudes from
the surface and whose outline is well preserved in the 3D reconstruction. Scale bar:
50 µm
Figure 11 – Isosurface presentation of tomography reconstructed 3D model of an aloe leaf
with stoma. The sample is a ~5mm thick freshly prepared skin of an aloe leaf.
The image taking was quite fast: 10ms per projection using the “local
tomography” mode. A) - C) are the “virtual slicing” through the stoma which
show the cell structures. FOV: A) and B) ~250 µm and C) ~80µm.