1. Introduction

Pollen grains are male gametophytes of higher plants that comprise the plant genetic information as well as water-soluble proteins and glycoproteins. These proteins control pollen-stigma recognition, fertilization and also play a role in plant defense mechanisms [1], however, these glycoproteins also stimulate IgE antibodies when in contact with the mucosa of sensitive individuals. This stimulation results in type I allergic symptoms typical of hayfever, e.g., rhino conjunctivitis and asthma [2].

The outer wall of a pollen grain consists of two layers: a porous inner layer, the intine that is composed predominantly of cellulose, hemicellulose and pectins, and an outer layer, the exine, that is formed by various organic and inorganic substances. Amongst these is sporopollenin, a highly resistant biopolymer [3] that is responsible for the high chemical robustness of the exine.

It has been shown that the allergens are rapidly released from pollen grains upon hydration [4, 5]. In accordance with this finding, attempts to localize the birch major allergen, Bet v1, protein on ultrathin sections of birch pollen grains by immunogold labeling with monoclonal antibodies were successful only after strictly anhydrous fixations of the pollen grains [6–8]. The allergens were localized only in the cytoplasm of the pollen grains, while the intine, exine and the surface of the pollen grains remained unlabeled. However, when dry birch pollen grains are rehydrated for a short time, Bet v1 proteins are found in the wall and on the surface of the pollen grains after only 1 minute of hydration, while the density of labeling in the cytoplasm decreased. The allergens in the pollen wall were often detected in clusters, which suggest the presence of preferred pathways of allergens through the pollen wall [5, 8].

Existing techniques for imaging pollen grains, such as Scanning Electron Microscopy (SEM) [9] or Transmission Electron Microscopy (TEM) [10], provide either few internal details (SEM) or require the pollen grain to be sliced into thin segments for imaging (microtoming; TEM) because of multiple scattering. Such slicing, however, may lead to undesirable damage of fragile internal structures. One of the advantages of X-ray microscopy [11] is that multiple scattering is negligible. As a result, the internal structures of few-micrometer-sized biological single particles, including pollen grains, can be visualized without sectioning the specimen. Another advantage of using x-rays is that sample staining is not required for feature contrast. We demonstrate here that Fresnel Coherent X-ray Diffractive Imaging (FCDI) [12] is indeed capable of examining the interior structure of a single, intact, pollen grain without sectioning. We show an example image from a single projection of a Convallaria majalis pollen grain, and indicate that this technique may be generalized to three dimensions.

Coherent X-ray Diffractive Imaging (CXDI) is a relatively novel imaging technique capable of investigating comparatively thick specimens at potentially very high resolutions [13, 14]. In this technique, a coherent x-ray beam illuminates a sample that alters the well-defined magnitude and phase distribution of this wave as it transits. The resulting wave, termed the exit surface wave (ESW), propagates downstream where it is measured by a two-dimensional detector. This diffraction pattern is usually recorded in the far-field and can be inverted through an iterative procedure to reconstruct the complete ESW, which in many cases can be treated as a projection image of the sample (for a review of CXDI methods see [15]). A number of applications to biological specimens have been shown in recent years [16–22] (for a review see [23]).

One limitation of CXDI using plane-wave illumination of the sample is that it requires an object of finite extent for the inversion procedure to succeed. Larger or extended objects, such as pollen grains, are not accessible with the traditional implementation of this technique. In contrast, FCDI employs a well-defined, strong phase curvature, as shown in Fig. 1 . This allows the observation of sub-regions of a specimen and permits a more robust reconstruction, which additionally converges in fewer iterations compared to plane-wave CXDI measurements [24]. The ability to observe sub-regions of a specimen is also provided by a scanning variant of CXDI, known as ptychography [25–27] though at the expense of requiring many overlapping illumination positions, which is not necessary in FCDI [12, 28, 29]. FCDI has been used to image whole biological specimens in the past [28], and we exploit its unique properties to image a sub-region of a pollen grain using only one illumination position on the sample.

 

Fig. 1 Schematic of the FCDI experiment. Coherent x-rays are focused using a zone plate to a focal plane upstream of the sample. The diverging beam (with phase curvature) is incident on a sub-region of the pollen grain. The radiation diffracted by the pollen grain propagates to the detector where it is measured. The radiation within the footprint of the direct beam forms an in-line hologram, while that outside is forms a coherent diffraction pattern. In our experiment the following geometrical conditions were used: z1 = 11.8 mm, z2 = 737.5 mm, z3 = 1.05 mm.

Download Full Size | PDF

2. Materials and methods

2.1 Sample preparation

Pollen samples were obtained from commercially purchased Convallaria majalis. Flower heads were dissected and the stamens were suspended in water. The samples were briefly subjected to a rotary shaker and residual portions of the stamen were removed using forceps. The suspension was then briefly centrifuged at 200 rpm to remove excess water. The samples were again suspended by the brief application of a rotary shaker. For the FCDI experiment, a small volume (2 μl) of the suspension was pipetted onto a glass cover slip and the pollen grains were transferred onto 20 nm thick silicon nitride membranes using crystal mounting loops from Hampton Research. The unstained samples were then allowed to air dry and transferred to the beamline.

2.2 FCDI Experimental geometry

The imaging experiment was performed at the 2-ID-B undulator beamline [30] at the Advanced Photon Source, Argonne National Laboratory, USA. The measurements were conducted in vacuum using a dedicated FCDI instrument [31] installed at the beamline. A 160 µm diameter zone plate with a 50 nm outermost zone was used to focus the 1820 eV monochromatic (ΔE/E≈1.25 × 10⁻³) x-rays to a focal spot size of ~60 nm with a focal length of z₁ = 11.8 mm and a flux of ~1 × 10⁸ ph/s. The sample was placed z₃ = 1.05 mm downstream of the focus, giving a spot size of approximately 14 µm at the sample. A 15 µm order sorting aperture was placed at the focus to isolate the first diffraction order of the zone plate. The detector, a charge-coupled device camera (PI-MTE, 2048 × 2048 pixels each 13.5 × 13.5 µm²), was positioned at z₂ = 737.5 mm downstream of the focus. We illuminated only a portion of the entire pollen grain near one extreme of its extent. This explicitly exploits one of the key advantages of the FCDI method: its ability to image sub-regions of extended objects. The complete geometry is outlined in Fig. 1, and allows the calculation of the Fresnel number, here N_F ≈22 across the largest dimension of the object, which is comfortably higher than the empirical ideal of N_F = 5 [24].The high numerical aperture of the zone plate enables the specimen to be imaged at a spatial resolution of ~60 nm; the higher-angle diffraction data permits imaging at higher resolution [32].

2.3 Data assembly

The data were collected in a series of exposures to improve the measurement statistics. However, the measurement may still suffer from the effects of sample movement with respect to the focused beam. To mitigate this problem, 600 short (2.5 s) exposures, with the sample in the beam, were recorded and correlated with an arbitrary frame. Those frames with a correlation coefficient of higher than 0.992 were used to create a data set, the mean of 305 such individual frames, for the reconstruction. The “white-field” images (frames taken without the sample in the beam) and “black-field” images (frames taken with the beam off) do not suffer from these effects and a mean for each was calculated using 832 and 66 frames, respectively. The mean images of the diffraction data and white-field data are shown in Fig. 2 . The sample data in Figs. 2(a) and 2(b) shows both the holographic region (a) which yields a readily interpretable, lower resolution image of the object and the higher angle coherent diffraction data (b) which is exploited in an iterative reconstruction to access higher resolution information about the sample. The high degree of correlation between the frames used implies that radiation damage to the sample, at least at the resolutions we are sensitive to here, is not present in these measurements.

 

Fig. 2 Measured FCDI data. (a) In-line holographic region of the data. (b) The same data shown on a truncated and logarithmic intensity scale to highlight the coherent scattering from the pollen grain. (c) White field data, measured with the sample removed from the beam. (d) Detail of coherent scattering (also on a truncated, logarithmic scale).

Download Full Size | PDF

2.4 Data reconstruction

The reconstruction was carried out in two steps. First, the white-field was constructed by attributing an ideal, parabolic phase curvature to the measured beam intensities. Then, the sample data were reconstructed for these beam parameters using 50 iterations of the FCDI algorithm [33], which utilized a shrink-wrap [34] support-size reduction every ten iterations. This refines the support by taking the thus-far reconstructed image, convolving it with a five by five (5 × 5) pixel square and defining the new support by selecting the region that was above 5% of the maximum value of the resultant, convolved image. The reconstruction was repeated 20 times, each initialized with random phase seeds. An average ESW (corrected for the structure of the beam) was calculated from these reconstructions and is shown in Fig. 3(a) .

 

Fig. 3 (a) Reconstructed (amplitude) image of the pollen grain region with color bar. We interpret the higher density features in the body of the reconstruction (highlighted with white arrows “i”) as consistent with projections of surface protrusions. The pore (approximately 1 μm diameter) is indicated with a double-headed white arrow (“ii”). The intine/exine wall separation is also indicated with a white double-headed arrow (“iii”). The estimated separation in the FCDI reconstruction (1.0 μm) is broadly consistent with that measured in the EM image shown in (b; 1.4 μm). (b) An electron micrograph of the external wall of a pollen grain taken from the same batch as in (a), with the intine/exine wall separation and surface protrusions indicated with red arrows. The difference in contrast between the X-ray and EM images is attributed to the different preparation techniques used in each case, in particular the staining, dehydration steps and sectioning required for the TEM sample.

Download Full Size | PDF

2.5 Transmission Electron Microscopy

Pollen grains isolated directly from the plant stamen, from the same sample batch used in the x-ray experiment, were rinsed with phosphate buffered saline and fixed with 2.5% glutaraldehyde in cacodylate buffer (pH 7.2) for 30 minutes. After fixation they were rinsed 5 times (5 min. each) with cacodylate buffer and stained with 1% osmium tetroxide for 40 minutes, then rinsed with water (4x), stained with 0.5% uranyl acetate for 40 minutes, and rinsed again with water (4x). All fixation and staining steps were performed on ice. The samples were then dehydrated in a series of ethanol and infiltrated with EPON (Carl Roth GmbH) and acetone in a series of steps (EPON/acetone 1:3 2hrs, 1:1 overnight, 3:1 3hrs, pure EPON 3x 30 min.). The resin blocks were polymerized by incubation in an oven at 60°C for 48 hours. Sections were cut 60 nm thick, placed on copper palladium slot grids coated with 1% Formvar (Sigma-Aldrich), and stained with uranyl acetate and lead citrate. Images were taken on a Philips CM120 Biotwin microscope.

3. Results and discussion

Figure 3(a) presents a reconstructed FCDI projection image through the pollen grain. We can identify features that give us confidence in the fidelity of the reconstruction. The most marked of these are the extine and intine walls of the pollen grain, seen at the edge of the grain. The distance between the extine and intine walls is consistent with that obtained by TEM as presented in Fig. 3(b) (average distance of 1.0 μm from the FCDI reconstruction and 1.4 μm from the TEM image). This variation is within expectations of the variation between individuals anticipated within this biological system. Several high-density features are also consistent with the scale of surface protrusions that are clearly visible in the EM image. In addition, the FCDI reconstruction shows a so-called 'aperture', the opening through the pollen wall into the grain cytoplasm, located near the tip of the grain (labeled with an arrow). These structures are common in pollen grains and are characterized by a reduction in density at a point in the pollen wall.

The holographic region of the data yields a lateral resolution of approximately 60 nm (defined by the focus of the zone-plate). The final, iteratively reconstructed image shown contains details finer than could be observed using the holographic region alone. This procedure can, in general, yield an improvement in resolution that is only limited by the wavelength and maximum angle at which the diffracted signal is detected.

4. Conclusions

In conclusion, we have shown a reconstructed image through part of a pollen grain using FCDI. We have observed features consistent with our expectations from electron microscopy and have demonstrated that FCDI is a viable method to image the internal structure of pollen grains.

Such measurements generalize readily to three dimensions through tomographic collection and processing of multiple angular projections [35]. This is important for the investigation of biological specimens, where two-dimensional projection information alone may be insufficient to answer many questions. Future experiments, utilizing higher flux and cryo- preservation, would allow intact pollen grains to be imaged either without treatment, or under a variety of hydration conditions. Furthermore, nanoscale imaging by FCDI can be applicable to other biological specimens of similar dimensions.

Three-dimensional imaging of micrometer-sized biological structures is of increasing importance in modern cell biology, providing insights into the function of bio-machinery at sub-cellular dimensions. Combining FCDI with tomographic and cryo-preservation methodology has the potential to be the tool to provide precisely this information.