Experimental flexibility is a luxury often taken for granted in typical laboratory science. Adapting a protocol in response to unfolding information as the experiment proceeds is a key element in discovery and is standard practice. In spaceflight, this luxury is not as accessible; crew time is highly regulated, orbital resources are limited, and protocols are tightly controlled once in place. In addition, the up-mass available to each researcher is limited, which restricts the choices of experimental treatments and orbital fixation options. These constraints often translate into hard choices that limit the scientific return on investment, including sacrificing the number of replicates to enable additional treatment and fixation scenarios. As a consequence, options that allow maximization of the data obtained from a single sample are highly valued.
Biological experiments are typically required to be segregated based on the analyses to be performed upon return from spaceflight. These segregated conditions are usually either molecular or morphological, due to the separate fixatives required for each application (e.g., Stutte et al., 2006). Investigations focused on transcriptome analyses, for instance, would not be likely to commit a portion of their limited up-mass on samples and associated fixation hardware for morphological applications, and would miss the opportunity to use both morphological and molecular perspectives. These limitations are keenly felt in scenarios where the transcriptome analyses reveal spaceflight-specific patterns of gene expression central to morphological patterns of cell growth and development. This scenario was encountered in a recent spaceflight experiment, in which telemetric imaging revealed interesting spaceflight-associated root morphologies (Paul et al., 2012). All material was fixed on orbit in RNAlater (Ambion, Grand Island, New York, USA) to facilitate transcriptome analyses on returned samples.
Although it is possible to partition an experiment into molecular and morphological components, using samples that correspond directly to those used in molecular applications would be of extreme value. A single fixation protocol better enables the full use of replicates to support a robust power analysis; all plant material allowed by the up-mass constraints can be dedicated to a single application per orbital treatment. Additionally, the plant material apportioned for microscopy can be taken directly from the pool of material whose balance represents a molecular replicate; thus, any morphologic change seen in these plants would directly reflect the effects of the altered gene expressions from those same plants. Terrestrial field studies could also benefit from this single fixative approach for similar reasons.
Despite its use as a molecular fixative, RNAlater is not commonly used as a morphologic fixative. RNAlater fixation produces unclear images with high background staining (Cadoret et al., 2013) and reduces immunofluorescent signals (Zaitoun et al., 2010; Paul and Ferl, 2011). This intracellular occlusion eliminates the possibility of clear subcellular localizations. However, the external morphology appears to be preserved in a fresh-like state, which suggests that RNAlater may be acceptable for use in scanning electron microscopy (SEM). Two cases in the literature report the use of RNAlater in a morphological investigation. Behrendt et al. (2011) reported that coral covered in crustose coralline algae was fixed in RNAlater, postfixed in 2% glutaraldehyde, rinsed in sodium cacodylate, and postfixed again in OsO4 prior to SEM investigations. Hofer et al. (2012) used SEM in their study on floral organs of Clermontia species using RNAlater as their main fixative with a hexamethyldisilazane drying step. In the current study, both of the above techniques resulted in significant tissue damage in Arabidopsis.
An alternative to fixation in SEM sample preparation is freezing. Cryo-SEM does not require prior fixation, but instead preserves tissues in a fresh-like state for imaging (Koroleva et al., 2010; Sarsby et al., 2012; Yu et al., 2013). However, few laboratories have access to such equipment, and the equipment and operation can often be expensive. Therefore, this study investigated the possibility of emulating cryo-SEM on a standard SEM, a “cryo-analog,” where tissue would first be returned to a fresh-like state for freezing prior to imaging.
METHODS AND RESULTS
For process development, test samples of 9–12-d-old A. thaliana var. columbia (Col-0) or Wassilewskija (WS) (The Arabidopsis Information Resource [TAIR], Columbus, Ohio, USA) were harvested into RNAlater and fixed overnight at ambient room temperature. Fixed plants, along with freshly harvested individuals of the same age, were washed in distilled water with gentle rocking for four 1-h increments. This washing step was designed to desalinate RNAlater-fixed samples, as the salt crystals were found to damage tissue during subsequent steps (Fig. 1A, B). Desalination also returned the plants to a fresh-like state (Fig. 1C, D) suitable for flash-freezing. Tissues were dissected immediately following the final wash into 5–10-mm lengths to fit on the SEM sample stage (“stub”).
The first two protocols tested provided valuable lessons. The first cryoanalog protocol was flash freezing in liquid nitrogen (LN2). Tissues were severely damaged, primarily due to rapid crystallization of water into ice, physically compromising cellular integrity. The second protocol tested was high-pressure freezing (HPF; Bal-Tec HPM100; Leica Microsystems, Wetzlar, Germany), which resulted in vitrified ice and a greater freezing depth. The noncrystalline structure prevented physical damage, but the roots had a tendency to stick to the planchettes and break (Fig. 1E, F).
The third protocol tested was based on the lessons learned and resulted in minimal tissue damage. The risk of water–ice crystal formation and planchette adhesion was minimized using a variety of methods. Samples were run through a methanol dehydration gradient, starting at 10% and increasing at 10% intervals, for 10 min at each step. An optional wash of 0.1% tannic acid in methanol was used to enhance electron density (Mizuhira et al., 1972; Roholl et al., 1981). Arabidopsis tissues were hand-plunged using forceps into LN2-cooled liquid ethane, which vitrified any remaining water in each sample. These rapid hand-performed plunges provided more control and less damage than automated plunge devices. Samples were immediately transferred to LN2-cooled individual cryovials, containing either 2% OsO4 in MeOH or 0.5% glutaraldehyde in 0.1% tannic acid. The headspace in each vial was filled with LN2 to prevent samples being lost or damaged due to static electric effects. Cryovials were placed into freeze substitution (Leica EM AFS2; Leica Microsystems) to increase the temperature to 10°C over the course of 4–10 d. Cryovial caps were left slightly loose to vent nitrogen gas. Once the temperature reached 10°C, cryovials were removed, opened, and tissues were transferred to tissue baskets (SPI microporous specimen capsules, 120–200 µm pore size; Structure Probe Inc., West Chester, Pennsylvania, USA) while submerged in chilled (4°C) methanol. This transfer took place at ambient room temperature; the chilled methanol submersion is enough to maintain the integrity of the tissues. Individual baskets were labeled using pencil on a small piece of paper approximately 2 mm2, as external labels were removed during the drying phase. Drying occurred via liquid CO2 critical point drying (CPD; Bal-Tec CPD 030; Leica Microsystems). Eleven quick-cycled washes replaced most of the loading methanol with CO2, followed by three 10-min infiltrations. One final exchange was performed and left overnight. Two additional 10-min infiltrations were performed, followed by CPD with controlled venting. Samples were in solution until the end of CPD. Dry samples in baskets were transferred to a desiccant-filled (Drierite; W A. Hammond Drierite Company, Xenia, Ohio, USA) sealable container until being mounted on stubs using carbon adhesive and returned to a stub rack within a different desiccant-filled sealable container. Mounted samples were sputter-coated for 90 s with a gold-palladium alloy (Desk V HP; Denton Vacuum, Moorestown, New Jersey, USA) and imaged on a scanning electron microscope (Hitachi S-4000; Hitachi, Tokyo, Japan).
Initial observations were focused on roots, resolution of cell files, and retention of root hairs throughout the processing of the tissue (Fig. 2). Later tests observed trichomes and more durable tissues, such as adaxial leaf surfaces and seed coats (Fig. 2A–C). The fresh-frozen tissues, used as protocol development control material, retained finer details, including closed stomata and normal trichome morphology (Fig. 2D, E). RNAlater-fixed test materials yielded similar images as the fresh-frozen plants, with minimal additional tissue damage (Fig. 2F–K).
Spaceflight materials and the corresponding ground controls, both fixed in RNAlater, were processed using the protocol discussed above and detailed in Appendix 1. Tissues yielded images of acceptable quality, with minimal cell wrinkling and high retention of fine details (Fig. 2J, K). The focus of these samples was root structure, morphology, and root hair retention because of interesting molecular and genetic observations from a recent spaceflight experiment (Paul et al., 2012; Paul et al., unpublished).
Cost of the developed protocol is minimal, while the accessibility is high. Assuming the microscopy core to be used has standard SEM capability, the additional cost of this protocol is merely the cost of consumables. These components are readily available and likely already obtained by the laboratory interested in using this protocol. Another advantage of this protocol is a near retention of tissue morphology at a fresh harvest state. Not only are finer elements of tissues retained (Fig. 2D–I), but other tissues, such as the adaxial leaf epidermis, seed coat, and floral stalk (Fig. 2A–C), produce results comparable to those processed through dedicated morphologic means (e.g., Bhargava et al., 2013). The amount of time needed for this process is slightly longer than other standard SEM processing times, with an additional 4 h for the desalination washes and several days required for freeze substitution. Hand-plunged samples may be processed in batches to increase throughput, being limited to the number of samples able to be held with a pair of forceps without compromising structural integrity. Examples of other SEM protocols may be seen in Liu et al. (2008), Koizumi et al. (2010), Springer et al. (2010), and Yi et al. (2010). As a result of this protocol, samples fixed with RNAlater retained many fine features including root hairs and trichomes, with a majority of cells inflated and displaying fresh-frozen–like morphology, regardless of whether they were harvested during spaceflight or on the ground.
The protocol presented here provides distinct advantages for both spaceflight and ground-based experimental applications. Arabidopsis plants harvested on the International Space Station by astronauts were successfully retrieved from RNAlater and processed to produce acceptable quality images without the use of a dedicated morphologic fixative (Fig. 2J, K). Field studies may also benefit from the use of this protocol. As in spaceflight, the amount of materials able to be carried and used during an expedition is limited. With the option of using one stable, safe chemical fixative for all desired studies to be conducted upon return, scientists in the field may be able to collect more data with fewer trips. This protocol may also assist laboratories with libraries of samples in RNAlater by expanding the available techniques for study.
Alternative imaging techniques outside the scope of this investigation provide alternatives to this protocol. True cryo-SEM would likely be an acceptable alternative (examples may be seen in Koroleva et al., 2010; Sarsby et al., 2012; and Yu et al., 2013), though desalination of specimens prior to tissue processing would still be beneficial. A second option is a cryo-holder, seen in Tang et al. (2012), which may prove to be an acceptable alternative step in this protocol if the SEM being used can handle the LN2 vapors. A third alternative is scanning ion conductance microscopy (SICM; Marti et al., 1988; Hansma et al., 1989; Ushiki et al., 2012). Due to the high ionic content of RNAlater and the necessity for an ionic submersion liquid for SICM, it is possible that this instrument could generate comparable images without any tissue processing. This possibility was not tested due to lack of access to the necessary equipment.
The protocol reported here will likely require optimization for each desired tissue to be tested. Arabidopsis is a very fragile tissue, with limited flat surfaces and small organ sizes. Other plants that may be more robust may tolerate a rougher technique, while more fragile tissues (i.e., floral tissues) may need a more gentle approach. Some potential additional steps to enhance tissue integrity include a postfixation step or reversing CPD and sputter-coating. A postfixation step (as seen in Behrendt et al., 2011), which would take place after the desalination, could improve tissue integrity during the cryo-analog plunge, especially if fixatives for electron microscopy, such as Karnovsky's or OsO4, are used. Sputter-coating prior to CPD (as seen in Yi et al., 2010) may also improve the physical tissue integrity, though both mounting prior to CPD and pulling a vacuum during the coating process may prove difficult.
The use of this protocol provides more morphological opportunities for samples that would have previously been wasted due to improper fixation. It is designed to be a supplement to rather than a replacement of SEM standard practices. Although this process needs to be adapted to each sample being studied, it enhances flexibility in experiments when the main objective is not imaging.
 The authors wish to thank Kim Backer-Kelley and Byung-Ho Kang for additional support in experimental design and execution. The authors also wish to thank Agata Zupanska, Fiona Denison, and Lawrence Rasmussen for editorial input on this manuscript. This work was supported by the National Aeronautics and Space Administration (NASA; grants NNX07AH270 and NNX09AL96G to R.J.F. and A-L.P).