Ice accommodation in plant tissues pinpointed by cryo-microscopy in polarised light CURRENT STATUS:

Background Freezing resistant plant organs are capable to manage ice formation, ice propagation, and ice down variable Insights in ice management strategies are essential for the fundamental understanding of plant freezing and frost survival. However, knowledge about ice management is scarce. Ice crystal localization inside plant tissues is challenging and is mainly based on optical appearance of ice in terms of colour and shape, investigated by microscopic methods. Notwithstanding, there are major uncertainties regarding the reliability and accuracy of ice identification and localisation. Surface light reflections, which can originate from water or resin, even at non-freezing temperatures, can have a similar appearance as ice. We applied the principle of birefringence, which is a property of ice but not of liquid water, in reflected-light microscopy to localise ice crystals in frozen plant tissues in an unambiguous manner. Results In reflected-light microscopy, water was clearly visible, while ice was more difficult to identify. With the presented polarised cryo-microscopic system, water, including surface light reflections, became invisible, whereas ice crystals showed a bright and shiny appearance. Based on this, we were able to detect loci where ice crystals are accommodated in frozen and viable plant tissues. In Buxus sempervirens leaves, large ice needles occupied and expanded the space between the adaxial and abaxial leaf tissues. In Galanthus nivalis leaves, air-filled cavities became filled up with ice. Buds of Picea abies managed ice in a cavity at the bud basis and between bud scales. By observing the shape and attachment point of the ice crystals, it was possible to identify tissue fractions that segregate intracellular water towards the growing ice crystals. Conclusion Cryo-microscopy in reflected-polarised-light allowed a robust identification of ice crystals in frozen plant tissue. It distinguishes itself, compared with other methods, by its ease of ice identification, time and cost efficiency and the possibility for high throughput. Profound knowledge about ice management strategies, within the whole range of freezing resistance capacities in the plant kingdom, might be the link to applied science for creating arrangements to avoid future frost damage to crops.

and attachment point of the ice crystals, it was possible to identify tissue fractions that segregate intracellular water towards the growing ice crystals.
Conclusion Cryo-microscopy in reflected-polarised-light allowed a robust identification of ice crystals in frozen plant tissue. It distinguishes itself, compared with other methods, by its ease of ice identification, time and cost efficiency and the possibility for high throughput. Profound knowledge about ice management strategies, within the whole range of freezing resistance capacities in the plant kingdom, might be the link to applied science for creating arrangements to avoid future frost damage to crops. Background 3 Low-temperature resistance of plants is highly diverse, ranging from chilling susceptible species in tropical biomes to extreme freezing resistant species of the boreal zones that even survive immersion in liquid nitrogen in the dormant state. Stress caused by low temperature limits global plant distribution, threatens crop survival and by this can lead to tremendous economic damage [1]. A challenging factor for frost survival is ice management, that is where it forms, how it propagates and how it is accommodated within the plant body [2]. Investigation of ice management has largely been neglected and is compared with other aspects of freezing resistance chronically understudied [2].
Presumably, this is due to difficulties to visualise ice in plant tissues as we lack simple and unambiguous detection techniques. Nevertheless, knowledge of ice accommodation in plant tissues will be essential for the fundamental understanding of frost survival of plants within the whole range of freezing resistance capacities (< 0 °C to -196 °C). Additionally, an improved knowledge might help to create improved arrangements for avoiding frost damage to crops in the future [2].
Precise temperature measurements can detect ice formation, as the phase transition from liquid water to ice releases measurable heat, the so-called freezing exotherm. Based on this principle the established and noninvasive methods are DTA and IDTA. Differential Thermal Analysis compares the sample temperature with a reference temperature and allows to identify freezing, but cannot detect where ice has nucleated first and where and how it spreads [1]. In contrast, IDTA allows tracking of 4 the spatiotemporal ice propagation. This enables to detect the point of ice nucleation, the propagation pattern and its speed [6,7]. As the release of heat is not quantified, both methods only provide a qualitative statement about the intensity of freezing [40]. Although the resolution of the techniques improved rapidly, the localisation of ice inside of samples can be a challenging task and implicate shortcomings due to dissection [20].
Remaining liquid water can be quantified by NMR spectroscopy [21,22], while NMR micro-imaging or magnetic resonance imaging (MRI) localises liquid water [1,19,20,23]. Both are noninvasive techniques. In contrast to IDTA, MRI visualises the spatio-temporal pattern of unfrozen water, which shows supercooled and non-frozen areas. With MRI resolutions of 20-100 µm can be obtained, whereas it was noted that monitoring of rapid phenomena while maintaining high resolutions is difficult [20,41]. MRI in florets and buds is established [20,42], but for leaves there is limited research available [43]. MRI is a powerful tool for investigating freezing behaviour of plants, but there have been only a few high-resolution studies because of the limitation of machine time and expense [20].
Ice crystals can be visualised and located inside of plant tissues by cryo-SEM and/or light microscopy.
Microscopic methods require sectioning at a certain point in time; this entails the potential risk of ice crystal formation during incision. Further, temporal monitoring of ice formation in a specimen is not possible [20]. However, microscopy is a standard tool for detection of ice in plant tissues. Light microscopy was mainly applied to identify extracellular ice whereas cryo-SEM in addition can detect intracellular ice [36]. Cryo-SEM is also a valuable tool for monitoring freezing responses of plant tissues or cells [38]. As fixation requires very low temperatures (-160 °C) [32], direct observation of naturally formed ice is not possible. Only indirect observation of naturally formed ice is possible as it can be distinguished whether ice crystals have formed before or after the fixation process [38]. By observation under vacuum conditions around − 110 °C, there is a risk of ice crystal formation and deformation [20]. In former studies, sectioning was performed by fracturing under vacuum conditions which produced irregular surfaces and structures difficult to interpret [44]. However, techniques to gain flat surfaces were developed by which ice can be detected unambiguously [44]. Nonetheless, a trained eye is crucially important. Overall, the cryo-SEM procedure seems to be elaborate and expensive technical equipment is required.
To directly detect and observe ice in plant tissues, optical light microscopy techniques have been widely used and applied since a long time [30]. Continuous improvements of resolution and particularly image quality gained more detailed insights. Transmitted-light microscopy in temperature-controlled conditions [45][46][47][48] and frozen specimens, which are incised and microscopically observed in reflected light [20,29,49], are the most common approaches. For transmitted-light microscopy specimens (organisms to cross-sections) are mounted on microscopic slides in liquid water. Indeed, freezing of this water may simulate extracellular ice, which allows monitoring of cellular responses to the presence of ice. Undoubtedly, the amount of frozen water and the location of ice masses inside the tissue are unrealistic and can cause misleading interpretations.
To directly localise ice crystals, plant organs can be cut and inspected in reflected-light microscopy.
The identification of ice crystals is based on the optical appearance in terms of colour and shape.
Unfortunately, this can be a challenging task, as it does not work equally well for all plant tissues. In certain tissues, differentiation between ice crystals, liquid water or resin can be impossible as surface light reflections interfere and cause a similar appearance. Although light microscopic methods are easy and comparably cheap, ambiguousness of ice detection might have triggered the deployment of the much more complex methods presented before.
To improve visibility of ice in plants, a prerequisite is a reliable distinction between the solid and liquid aggregation state of H 2 0. The optical property "birefringence" is a feature of ice but not of liquid water. While water in the liquid state is known to be optically isotropic, ice crystals are optically anisotropic [50]. Birefringence can be assessed by polarised light microscopy. Polarised transmittedlight microscopy is a standard tool in mineralogy for material characterisation in thin sections or powders [51].
In this study, we tested whether ice can unambiguously be distinguished from liquid water using polarised light in reflected light microscopy. To rule out disruptive effects during preparation of thin cross sections and the necessity of a liquid sample carrier for transmitted-light microscopy, we used 6 reflected-light microscopy to inspect frozen plant organs immediately after cutting. Inspections were performed in structural different plant organs from three freezing resistant species: in leaves of a perennial species (Buxus sempervirens), in leaves of an herbaceous species (Galanthus nivalis) and in overwintering buds of conifers (Picea abies).

Distinction between water and ice in reflected light
To verify whether liquid water can be reliably distinguished from frozen water, experiments were performed with water in different aggregation states at different temperatures (Fig. 1). By parallelorientated polarisation filters or without polarisers a droplet of water could clearly be detected in reflected-light microscopy (Fig. 1a). In the centre of the droplet, a strong light reflection occurred.
Under crossed polarisers, the droplet of water, including the light reflection, disappeared (Fig. 1b).
With parallel polarisers or without polarisers ice crystals were difficult to identify (Fig. 1c). However, under crossed polarisers, ice crystals showed a glowing appearance and could be identified much more explicitly (Fig. 1d).

Detection of ice in plant tissues
Identification of ice in plant tissues using cryo-microscopy in reflected-polarised-light (CM rpl ) is demonstrated for three species in Fig. 2: In unfrozen leaves of Buxus sempervirens a void between the adaxial leaf side (mainly parenchyma cells and vascular tissue) and the abaxial leaf side (spongy tissue) was found (Fig. 2a). When frozen down to -7 °C, large ice masses could clearly be identified by their bright shiny appearance in the CM rpl -image. Ice masses occupied and expanded the space between the two mesophyll layers, which in the frozen state became completely separated (Fig. 2b). The ice masses consisted of parallel oriented ice needles, originating from the adaxial side, where the vascular tissue is located.
Presumably, they grew from where cellular water is segregated into the apoplast during freeze dehydration. In the CM rpl -image further bright spots in the vascular tissue and in close proximity to the lower epidermis could be discerned (white arrows in Fig. 2a,b). Since these spots appear similar in the unfrozen leaf at +20 °C, they very likely originated from birefringent tissue components other 7 than ice.
Unfrozen leaves of Galanthus nivalis showed huge air filled cavities (mean cross sectional diameter = 363 ± 146 (SD) µm) (Fig. 2c). When frozen, these cavities filled up with ice masses (Fig. 2d) which consisted of separate ice crystals. The ice crystals appeared rather irregularly shaped, but seemed attached to the inner surfaces of cell walls bordering the cavities, which produced a hollow-cylindrical or cylindrical appearance.
Buds of Picea abies have a complex architecture (Fig. 2e) [24]. The chlorophyll containing preformed new shoot and needles and the apical meristem are further referred to as "bud". The bud is separated from the stem by the so-called crown, below which an air-filled cavity is formed. Multiple layers of bud scales surround the bud. When frozen, in the CM rpl -image no ice was found in the supercooled bud, but ice could be clearly identified in various surrounding spaces. Ice formed in the stem below and accumulated in the cavity below the crown and in spaces between the bud scales (Fig. 2f). The compact ice mass formed in the cavity below the crown adheres to the crown and consisted of parallel oriented needle-like columns of ice. These ice masses originated from water segregated across the crown during so-called extra-organ (translocated) ice formation. In the bud scales, bright birefringent structures were detected already in the unfrozen state. Nevertheless, in the CM rpl -image at -10 °C rather irregularly-formed ice crystals between the bud scales could explicitly be identified.
However, due to other birefringent components in the scales the extent of these ice masses was difficult to judge.

Discussion
With the experimental setup we developed, a robust differentiation between water and ice using polarisation filters in reflected-light microscopy was possible. If light passes through a polariser this yields plane polarised light, which can be extinguished by a second polarisation filter (analyser) when the filters are crossed. However, this only holds if there is isotropic material between the polariser and the analyser. If there is birefringent material in between, this will result in a wave inclined at some angle compared with the original plane wave [50], which will be not fully extinguished by the analyser. By this, birefringent objects become visible under crossed polarisers. 8 With CM rpl visibility of ice was considerably improved and liquid water could be made invisible. Water surface reflections can be deceptive in reflected-light microscopy as their distinction from ice is often hardly possible. With CM rpl , they fully disappeared under crossed polarisers ruling out potential uncertainties. Similar to water, resins from pines can cause deceptive surface light reflections in reflected-light microscopy. As resins did not show birefringence, surface light reflections from resins could also be extinguished by crossed polarisation filters (data not shown). The only obstacle were birefringent tissue components. However, their potential misleading interpretation can be ruled out by investigation of unfrozen control samples, which is recommended.

Advantages of CM rpl for detection of ice in plant tissues
By the application of CM rpl , it is possible to gain detailed new insights into ice accommodation and segregation strategies in plant tissues. The system allows a rapid inspection of a high number of samples at low costs compared with other methods. Besides the reliable determination of ice crystals in frozen plant tissues, CM rpl offers additional valuable information about ice growth. The resolution is high enough to identify the shape of the apoplastic ice crystals, which can be valid information to understand how they are formed. Additionally, in all three examples (Fig. 2) attachment to cell walls and orientation of ice crystals was clearly visualised. This information provides valuable hints about the ice formation process itself. In particular, the precise localisation of ice crystals indicates at which cell wall surfaces water is segregated into the apoplast.
Additional preliminary results showed that lethal intracellular ice formation could be identified in mesophyll cells of Trachycarpus fortunei. Hence, with the CM rpl , investigations in respect to the cause of frost damage are possible.

Conclusion
To understand how plant species with different freezing tolerance limits accommodate ice masses in their organs, without becoming damaged, will be essential for the fundamental understanding of frost survival. Unambiguous ice identification at a high resolution with CM rpl allows profound insights into ice accommodation and growth of ice crystals. CM rpl is a valuable tool in addition to the existent pool 9 of methods for assessing ice management in plant tissues. The various methods available provide different information. Hence, combining the results of the various methods will provide the most comprehensive picture for a better understanding of plant freezing processes. The methods differ largely in the time requirement and expense. Also spatial resolution differs greatly (freeze-fracture and freeze-substitution EM > cryo-SEM > light microscopic methods > MRI > IRVT (IDTA) > the unaided eye > DTA [1]). The required level of detail has to be chosen by choosing the appropiate method. CM rpl has major benefits in ease of ice identification, less time requirement, inspection of a high number of samples and cost efficiency.

Material And Methods
The experiments were performed with a reflected-light microscopic system under temperaturecontrolled conditions. All manipulation steps were exclusively executed in a temperature-controlled environment to prevent condensation effects (Fig. 3).

Microscopic system
A customisable BXFM-F microscope unit containing the following components (all parts from Olympus, Tokyo, Japan) was used: Reflected light illuminator U-KMAS mounted on BXFM-ILHS, single port tube with lens U-TLU and a video camera mount adapter U-CMAD3. Illumination was provided by a cold light source (KL1600LED, Schott, Mainz, Germany). For magnification, the microscope was equipped with the following objective lenses: MPLFLN5X, MPLFLN10X, LMPLFLN20X (Olympus, Tokyo, Japan).
Images were prepared with a digital camera (UC90), which can be fully controlled by the image analysis software cellSense Standard (Olympus, Tokyo, Japan). For the investigation of optical properties (e.g. birefringence) a polariser (U-PO3) and a rotatable analyser (U-AN360-3); (both from Olympus, Tokyo, Japan) were used.

Controlled freezing treatment
The microscopic system except for the camera unit was placed inside of a laboratory chest freezer (ProfiLine Taurus PLTA0987, National Lab, Moelln, Germany) (Fig. 3) which is fully temperature controlled. For thermal insulation the lid of the laboratory chest freezer was substituted by a customised acrylic glass lid. The lid was equipped with thermally insulated gloves, which allowed manipulations inside the cold compartment during the freezing treatment. Control technology enabled to regulate the temperature inside the laboratory chest freezer [52]; hence simulation of temperature profiles with defined cooling/warming rates was possible. Cooling rates were limited to a maximum of 3 K/h to avoid artificial supercooling [39]. Target freezing temperatures were set species specifically to sublethal temperatures. Input variables for the desired temperature treatment were set with a control software (programmed in Lab View by O. Buchner). With this technical equipment, a precise temperature control between ambient temperature and -21 °C was experimentally tested; however the cooling capacity of the laboratory chest freezer would even be -86 °C. As initiating the thaw procedure by removing the acrylic glass lid from the cooled laboratory chest freezer caused intense condensation on the microscope; this was eliminated by rewarming without removing the lid and increasing the temperature with multiple heating pads (total heating power = 49.8 Watt) attached to the microscopic unit. Controlled rewarming ensured that the temperature of the microscope was always higher than the air inside of the cooling compartment, which completely ruled

Sample preparation
For sectioning of the unfrozen as well the frozen leaves under the microscope a special sample holder for specimens was designed (Fig. 4). Before the start of the freezing experiment, leaves were clamped cautiously between the sponge rubbers of the holder and leaf petioles were inserted into 0.5

Consent for publication
All authors consent for the publication. The authors assume no liability regarding the introduced methods. Its application takes place at the user's own risk.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.  (e, f) at room temperature and in a frozen state. Ice crystal identification was facilitated by the crossed arrangement of the polarisation filters. In leaves of B. sempervirens large ice crystals formed, which expanded the preexistent cavity between the adaxial and abaxial leaf tissues, which was formed after the first freeze event. The bright spots (white arrows) are very likely due to other birefringent tissue components, as they are visible at room temperature and in the frozen state (b). In leaves of G. nivalis ice occurred prominently in 20 the air-filled cavities (d). In buds of P. abies ice occurred in the cavity (cav) below the crown (cr) and between the bud scales (bs) during translocated ice formation (f). White bars indicate 500 µm.