Responses to Ice Formation and Reasons of Frost Injury in Potato Leaves

: Potato leaves are ice-tolerant but are frost-damaged at − 3 ◦ C. Freezing occurs in two steps, a ﬁrst non-destructive freezing event and a second independent lethal event. Localization of ice, and whether cells freeze-dehydrate after the ﬁrst freezing event remains unknown. The cause of frost damage during the second freezing event lacks experimental evidence. Cytological responses of mesophyll cells were examined during ice formation using cryo-microscopic techniques after high-pressure freeze-ﬁxation and freeze-substitution. CO 2 gas exchange on frozen leaves revealed functional responses, but also frost damage. After the ﬁrst freezing event, gas exchange was un-interrupted. Consequently, intercellular spaces are free of ice, and ice may be restricted to xylem vessels. The cellular shape of the mesophyll cells was unchanged, cells did not freeze-dehydrate but were supercooled. When thawed after the ﬁrst freezing event, leaves were initially photoinhibited but regained photosynthesis. During the second freezing event, cells froze intracellularly, and some palisade parenchyma cells remained intact for a prolonged time. Intracellular ice caused complete destruction of cells, and chloroplasts became invisible at the light microscopic level. When thawed after the second freezing, leaves were unable to regain photosynthesis. Consequently, freezing avoidance is the only viable strategy for potatoes to survive frost.


Introduction
Potato is the fourth most important food crop and is grown in more than 100 countries around the world [1]. Potatoes originate from high elevations in the Andes, an area that lies within the zonobiome of tropical rainforests and the zonobiome of tropical-subtropical monsoon forests and savannahs, and in the lowlands of southern Chile, which lie in the nemoral zonobiome with cold winters and deciduous forests [2]. Optimum growth conditions for potatoes are daily mean temperatures between 18 and 20 • C; further, a frost-free period of approx. 3 months is needed. However, due to climate change, the risk of devastating spring frosts is increasing [3] and will especially negatively affect potato yield in colder regions. Indeed, frost already is a chronic problem for potato production in temperate regions and at high elevations in the tropics [4,5]. At natural field conditions, ice nucleates in leaves of Solanum tuberosum between −0.5 and −3 • C [6,7], and leaves become frost-injured around −3 • C [8,9]. Potatoes are unable to frost-harden [10,11]. Through infrared differential thermal analysis (IDTA), freezing of potato leaves was shown to proceed in two separate steps [9]. During the first non-destructive freezing event (E1), ice spread extracellularly, first across the xylem but then appearing to capture the entire leaf blade, within 60 s. Potato leaves did not become frost-injured until 10-78 min later, when a second freezing event (E2) occurred, which produced a diffuse freezing pattern without a distinct starting point. This suggests that S. tuberosum has ice-tolerant leaves, which were originally only found in other Solanum species [8]. However, ice tolerance of leaves of S. tuberosum is limited and only possible at moderate temperatures higher Crops 2022, 2 379 than −3 • C [12] and short-term frost-exposure [9]. The non-destructive formation of ice must be extracellular. Tolerance of extracellular ice formation is thought to cause cellular freeze-dehydration. Still, the cellular responses of mesophyll cells to E1 at non-destructive temperatures, i.e., whether freeze-dehydration occurs, are unknown.
Potato leaves are lethally damaged by frost during the second freezing process. Without microscopic evidence, this was suggested to be caused by intracellular freezing [9]. Through intracellular freezing ice crystals form within the cell and destroy the protoplasts [13,14]. Intracellular freezing was described to cause the symptom of frost plasmolysis, i.e., loss of osmotic responsiveness, which is observed after thawing of frozen ice in susceptible tissues [14]. However, how intracellular ice formation proceeds is hardly understood, and various symptoms have been observed [14,15]. In these early studies, the measurement technologies employed were limited and without control of freezing and thawing rates; consequently, cellular responses were studied at unnaturally high freezing rates (2-4 K·min −1 ), which might have resulted in supercooling down to between −3 and −20 • C [14]. As experimental designs can dramatically affect the extent and kind of frost injuries, it must be questioned whether this behavior occurs in nature. How potato mesophyll cells are injured during the second freezing event has not been studied.
We hypothesized that: (1) After E1, mesophyll cells of potato leaves show freeze-dehydration; (2) After E1, although it is non-injurious, gas exchange activity stops as ice formation in the intercellular spaces between mesophyll cells blocks gas diffusion; (3) During E2, the mesophyll cells freeze intracellularly, which causes membrane rupture and irreversibly destroys the cells, resulting in lethality.
During E1, the first non-destructive, and E2, the second lethal freezing event in potato leaves, we aimed to assess functional and cytological responses of mesophyll cells to ice formation. In comparison to unfrozen leaves, we studied the specific cytological responses after E1 through the investigation of frozen, cryo-transferred high-pressure freeze-fixed and freeze-substituted tissue, from which semi-thin sections were prepared and investigated via light microscopy to study the possible occurrence of freezing cytorrhysis. Furthermore, we addressed the physiological performance after E1 and E2 via CO 2 gas exchange measurements. Cellular and structural responses during E2 were studied via cryo-light microscopy.

Plant Material
Leaves of Solanum tuberosum L. (var. Agria Bio) were studied. Plants were grown in the Botanical Gardens of the University in Innsbruck and the University of Salzburg. We used certified tubers, which were grown either in pots (d = 25 cm) embedded in the open field or in natural soil. The soil was sandy silt. The plants were not irrigated artificially and were thus dependent on precipitation at the growing sites. Only fully expanded leaves were sampled. Sampling time was always in the morning. Leaves were cut and kept watered until the start of the experiments.

Leaf Anatomy
Leaf anatomy was quantified on microscopic sections taken from mature leaves, which were sectioned in off-axis angles of 0 • , 45 • , and 90 • . Five stripes per off-axis angle were sectioned from 3 different leaves. Then, 3 randomly chosen stripes out of the 5 were used to cut 1 section. Sections were cut with a microtome (GSL1, Schenkung Dapples, Zürich, Switzerland) and had a thickness of 80 µm. Analyses were performed using a light microscope (Olympus BX50F, Olympus Corporation, Tokyo, Japan) coupled with a digital camera (Olympus DP25, Olympus Corporation, Tokyo, Japan). Measurements of cells and stomata were performed in the live view mode of the imaging software cell^D 3.1 (Olympus Soft Imaging Solutions GmbH, Muenster, Germany). The leaves of S. tuberosum are bifacial ( Figure 1) and amphistomatic. At the adaxial leaf side we found 100 ± 12 stomata·mm −2 ; at the abaxial leaf side, 231 ± 71 stomata·mm −2 . The mesophyll palisade parenchyma layer had a thickness of 72 ± 10 µm and the spongy parenchyma layer 116 ± 16 µm. The mean sectional cell area in the spongy parenchyma was 533 ± 252 µm 2 and in the palisade parenchyma 1170 ± 353 µm 2 . The area of intercellular spaces was found to be lower than 1%. The leaves had a SLA of 2.32 ± 0.59 (dm 2 ·g −1 DW).
were sectioned from 3 different leaves. Then, 3 randomly chosen stripes out of the 5 were used to cut 1 section. Sections were cut with a microtome (GSL1, Schenkung Dapples, Zürich, Switzerland) and had a thickness of 80 μm. Analyses were performed using a light microscope (Olympus BX50F, Olympus Corporation, Tokyo, Japan) coupled with a digital camera (Olympus DP25, Olympus Corporation, Tokyo, Japan). Measurements of cells and stomata were performed in the live view mode of the imaging software cell⌃D 3.1 (Olympus Soft Imaging Solutions GmbH, Muenster, Germany). The leaves of S. tuberosum are bifacial ( Figure 1) and amphistomatic. At the adaxial leaf side we found 100 ± 12 stomata•mm −2 ; at the abaxial leaf side, 231 ± 71 stomata•mm −2 . The mesophyll palisade parenchyma layer had a thickness of 72 ± 10 μm and the spongy parenchyma layer 116 ± 16 μm. The mean sectional cell area in the spongy parenchyma was 533 ± 252 μm 2 and in the palisade parenchyma 1170 ± 353 μm 2 . The area of intercellular spaces was found to be lower than 1%. The leaves had a SLA of 2.32 ± 0.59 (dm 2 •g −1 DW).

Freezing Treatments
All freezing treatments in the laboratory were conducted in either commercial or laboratory chest freezers. Cooling rates were limited to −3 K•h −1 ; such rates correspond to naturally occurring leaf-cooling rates at freezing temperatures [6,16].

Effects of Freezing Events on CO2 Gas Exchange
For gas exchange measurements, leaves from different individuals were used (n = 3). The petioles were immediately watered in a test tube, and on the underside of the leaf blade, customized temperature sensors [17] were attached. One sensor recorded the leaf temperature (wire diameter 0.05 mm; COCO-002, Omega Engineering Inc., Stamford, CT, USA), and a second sensor was placed in close proximity (2 mm) to the leaf to measure the reference temperature (TT-TI-40, same manufacturer). The leaves were placed in the standard cuvette of a gas exchange measurement system (GFS 3000, Walz, Effeltrich, Germany), and the entire assembly was transferred into an automatic freezing unit (AFU; [18]). The AFU can be precisely temperature-controlled based on software programmed in LabView 2012 (National Instruments, Austin, TX, USA). After temperature stabilization at +8 °C, the leaves were cooled (−3 K•h −1 ) to the specific target temperatures. For the detection of leaf-freezing exotherms, the leaf and reference temperatures were recorded at high frequency (1 to 4 Hz). A sudden increase in leaf temperature compared to the reference temperature (DTA, differential thermal analysis [19,20]) allowed detection of leaf ice

Freezing Treatments
All freezing treatments in the laboratory were conducted in either commercial or laboratory chest freezers. Cooling rates were limited to −3 K·h −1 ; such rates correspond to naturally occurring leaf-cooling rates at freezing temperatures [6,16].

Effects of Freezing Events on CO 2 Gas Exchange
For gas exchange measurements, leaves from different individuals were used (n = 3). The petioles were immediately watered in a test tube, and on the underside of the leaf blade, customized temperature sensors [17] were attached. One sensor recorded the leaf temperature (wire diameter 0.05 mm; COCO-002, Omega Engineering Inc., Stamford, CT, USA), and a second sensor was placed in close proximity (2 mm) to the leaf to measure the reference temperature (TT-TI-40, same manufacturer). The leaves were placed in the standard cuvette of a gas exchange measurement system (GFS 3000, Walz, Effeltrich, Germany), and the entire assembly was transferred into an automatic freezing unit (AFU; ref. [18]). The AFU can be precisely temperature-controlled based on software programmed in LabView 2012 (National Instruments, Austin, TX, USA). After temperature stabilization at +8 • C, the leaves were cooled (−3 K·h −1 ) to the specific target temperatures. For the detection of leaf-freezing exotherms, the leaf and reference temperatures were recorded at high frequency (1 to 4 Hz). A sudden increase in leaf temperature compared to the reference temperature (DTA, differential thermal analysis [19,20]) allowed detection of leaf ice formation. The leaves were illuminated at 1000 µmol photons·m −2 ·s −1 at temperatures above +4 • C and kept in darkness during cooling below +4 • C.
The effects of freezing processes on CO 2 gas exchange and on the viability of the leaves were investigated. Freezing of potato leaves proceeds in two separate steps, a first non-destructive freezing event (E1) and a second event (E2) which is lethal to the leaves [9]. We used two different experimental approaches: (1) After E1, the leaves were kept for several minutes in this state and were quickly thawed and irradiated again, before the occurrence of E2. (2) After E1, the leaves were cooled until E2 was registered. Only after completion of E2 the leaves were thawed and irradiated again.
Before and during the freezing treatment, and after thawing, the assimilation rate (A) and dark respiration rate (R D ), respectively, were recorded. Before and immediately after the experiment, the maximum quantum efficiency of photosystem II was determined through assessment of the chlorophyll fluorescence parameter F v /F m .

Semi-Thin Sections of High-Pressure Frozen Leaves
The effect of short-term, non-destructive freezing (E1) on the shape of the mesophyll cells was investigated on semi-thin sections of high-pressure frozen and freeze-substituted leaf samples. For preparation of freeze-fixed samples, leaves from 3 different individuals were detached and frozen (−3 K·h −1 ) to −2.5 • C. Once ice nucleated, circular areas were punched out from these leaves and immediately fixed in the frozen state by means of high-pressure freeze-fixation (HPF; Leica Empact, Leica Microsystems, Wetzlar, Germany). Controlled freezing was carried out in an Automatic Freezing Unit (AFU; see [18]). The leaves were placed on a polystyrene block and wrapped in cling film to thermally insulate the leaves. Fine wire thermocouple sensors (Type T, solder junction diameter < 0.2 mm, TT-TI-40, Omega Engineering Inc., Stamford, CT, USA) were inserted between the abaxial leaf side and the polystyrene block to monitor the leaf temperatures and to detect the occurrence of freezing exotherms ( Figure S1). The petioles were placed in Eppendorf tubes stuffed with cotton wool and filled up with water mixed with a solution of ice-nucleating active (INA) bacteria (Pseudomonas syringae; ref. [19]). The INA bacteria triggered ice nucleation at approx. −2.5 • C [21,22]. To ensure that all leaves froze at the same time, based on the method of Kuprian et al. [23], short glass-fiber-coated wires (GG-TI-28, Omega Engineering Inc., Stamford, CT, USA) were soaked in the INA solution, inserted into PVC tubes (to prevent the glass-fiber coating from drying), and then placed in two adjacent Eppendorf tubes. By doing this, once ice nucleated in one leaf, ice quickly spread to the other leaves, so that the leaves froze simultaneously. The whole arrangement was placed in the precooled AFU. After leaf temperatures had stabilized between +4 • C and 0 • C, the temperature was lowered until freezing of the leaves occurred at −2.5 • C. After an exposure time of 20 min at this temperature, leaf discs were collected for the high-pressure fixation (Leica EMPACT, Leica Microsystems, Wetzlar, Germany). Punching out leaf discs and placing them in gold-plated specimen holders (interior dimensions: 1200 × 200 µm, 16706897, Leica Microsystems, Vienna, Austria) was carried out in the AFU using thermally insulated gloves. Transfer of the prepared specimens to the pre-cooled HPF-device was performed without thawing, so that it was possible to fix the leaf tissue in the frozen state at approx. 12,000 K·s −1 and 2040 bar (for details, see [18]). The subsequent freezesubstitution and embedding in epoxy resin was carried out in accordance with a previously published protocol in a Leica EM AFS (Leica Microsystems, Wetzlar, Germany) [24]. Freezesubstitution was performed at −80 • C in acetone containing 2% OsO 4 and 0.05% uranyl acetate. The samples were epoxy-resin-embedded and heat-polymerized.
Semi-thin sections (0.6 µm) were prepared at a Reichert Ultracut S (Leica Microsystems, Wetzlar, Germany). The sections were transferred onto glass slides and stained with 0.3% Toluidine Blue O [25] and investigated with a Zeiss Axiovert 200 M microscope, and images were captured with a Zeiss AxioCam HRc camera (Carl Zeiss, Jena, Germany).

Cross-Sections Viewed by Light Microscope
To address cellular responses during leaf freezing, leaf cross-sections (n = 3) were monitored and captured using a Leica DM1000 light microscope (Leica Microsystems, Wetzlar, Germany) while being exposed to sub-zero temperatures. The freezing treatments were conducted inside a temperature-controlled chest freezer (GT 21012, Liebherr, Lienz, Austria), equipped with fans, to avoid temperature gradients, as described in detail by Kuprian et al. [26]. The microscope was equipped with a camera (Leica EC4, Leica Microsystems, Wetzlar, Germany), and the whole unit was placed inside the freezer. The software LAS EZ 3.0.0 (Leica Microsystems GmbH, Wetzlar, Germany) was used for live view and image capture. An acrylic glass lid with integrated thermally insulated gloves on top of the freezer enabled operation of the microscope throughout the freezing treatment. Cross-sections were prepared with a microtome (GSL1, Schenkung Dapples, Zürich, Switzerland) in an off-axis cutting angle of 90 • and a section thickness of 80 µm. To guarantee hydration and facilitate ice nucleation, a small ball of moist cotton wool inoculated with INA bacteria was placed on the microscope slide at the margin of the coverslip. From the cotton wool, a thin thread was elongated underneath the coverslip alongside the crosssection. A thermocouple was installed next to the cotton wool to detect freezing. Target temperature was −2.5 • C.

Results
A typical freezing response of a S. tuberosum leaf, with two temporally separated freezing events, shows E1, which was followed by E2 ( Figure 2). E1 was registered at −1.0 • C and E2 at −2.5 • C. E2 was detected 48 min after E1. While E1 is a short-time event releasing little heat, E2 lasts for several minutes and produces a significant amount of heat.
on top of the freezer enabled operation of the microscope throughout the fre ment.
Cross-sections were prepared with a microtome (GSL1, Schenkung Dapp Switzerland) in an off-axis cutting angle of 90° and a section thickness of 80 μ antee hydration and facilitate ice nucleation, a small ball of moist cotton woo with INA bacteria was placed on the microscope slide at the margin of the cove the cotton wool, a thin thread was elongated underneath the coverslip alongsid section. A thermocouple was installed next to the cotton wool to detect freez temperature was −2.5 °C.

Results
A typical freezing response of a S. tuberosum leaf, with two temporall freezing events, shows E1, which was followed by E2 ( Figure 2). E1 was regist °C and E2 at −2.5 °C. E2 was detected 48 min after E1. While E1 is a short releasing little heat, E2 lasts for several minutes and produces a significant amo

Cytological Aspects
Leaves of S. tuberosum were immediately freeze-fixed after E1. The shape of epidermal, palisade, and spongy parenchyma cells of unfrozen control cells ( Figure 3A) had the same appearance as the respective cells after the freezing event, which appeared unaffected by E1, i.e. the cells did not freeze-dehydrate ( Figure 3B,C).

Gas Exchange and Photosystem II
CO 2 gas exchange of S. tuberosum leaves was continuously recorded during controlled freezing treatments (Figure 4). Before the treatment, leaves had a Photosystem II efficiency (F v /F m ) > 0.8. In the example shown, the assimilation rate (A) was about +1.6 µmol CO 2 m −2 ·s −1 . During low-temperature exposure, leaves were darkened, and a constant dark respiration rate (R d ) of approximately −0.5 µmol CO 2 m −2 ·s −1 was recorded. When the first freezing exotherm (E1) was registered at −1.2 • C, the leaves were kept at this temperature for 24 min. Strikingly, R D was unaffected by E1. Thereafter, the leaf was rapidly thawed, and irradiation was switched on again. After thawing, the assimilation rate rapidly reached its initial value. F v /F m remained temporarily reduced to 0.35 but recovered to 0.74 within the next day. After E1, the leaves were always fully intact and did

Gas Exchange and Photosystem II
CO2 gas exchange of S. tuberosum leaves was continuously recorded during controlled freezing treatments (Figure 4). Before the treatment, leaves had a Photosystem II efficiency (Fv/Fm) > 0.8. In the example shown, the assimilation rate (A) was about +1.6 μmol CO2 m −2 •s −1 . During low-temperature exposure, leaves were darkened, and a constant dark respiration rate (Rd) of approximately −0.5 μmol CO2 m −2 •s −1 was recorded. When the first freezing exotherm (E1) was registered at −1.2 °C, the leaves were kept at this temperature for 24 min. Strikingly, RD was unaffected by E1. Thereafter, the leaf was rapidly thawed, and irradiation was switched on again. After thawing, the assimilation rate rapidly reached its initial value. Fv/Fm remained temporarily reduced to 0.35 but recovered to 0.74 within the next day. After E1, the leaves were always fully intact and did not show any symptoms of frost injury. Similar gas exchange responses were found on S. tuberosum leaves during and after exposure to −7 °C that had been kept free of ice in an unfrozen state for a prolonged time period (data not shown).

Gas Exchange and Photosystem II
CO2 gas exchange of S. tuberosum leaves was continuously recor trolled freezing treatments (Figure 4). Before the treatment, leaves had efficiency (Fv/Fm) > 0.8. In the example shown, the assimilation rate (A μmol CO2 m −2 •s −1 . During low-temperature exposure, leaves were dark stant dark respiration rate (Rd) of approximately −0.5 μmol CO2 m −2 •s When the first freezing exotherm (E1) was registered at −1.2 °C, the lea this temperature for 24 min. Strikingly, RD was unaffected by E1. Therea rapidly thawed, and irradiation was switched on again. After thawing rate rapidly reached its initial value. Fv/Fm remained temporarily reduc covered to 0.74 within the next day. After E1, the leaves were always fu not show any symptoms of frost injury. Similar gas exchange responses tuberosum leaves during and after exposure to −7 °C that had been kep unfrozen state for a prolonged time period (data not shown). . Below +4 °C, the leaves were not irradiated (yellow shading: pho flux density 1000 μmol photons•m −2 •s −1 ) and kept in darkness until thawing. A freezing event (E1) was registered, and after 24 min of being exposed to freezing without occurrence of the second freezing event (E2), the leaf was thawed. Blac change rate, blue lines: leaf temperature, arrow: freezing exotherm. The insert exotherm in the differential temperature plot (leaf − reference temperature). . Below +4 • C, the leaves were not irradiated (yellow shading: photosynthetic photon flux density 1000 µmol photons·m −2 ·s −1 ) and kept in darkness until thawing. At −1.2 • C, the first freezing event (E1) was registered, and after 24 min of being exposed to freezing temperatures, but without occurrence of the second freezing event (E2), the leaf was thawed. Black line: CO 2 gas exchange rate, blue lines: leaf temperature, arrow: freezing exotherm. The insert shows the freezing exotherm in the differential temperature plot (leaf−reference temperature).

Cytological Aspects
Leaf cross-sections of S. tuberosum were viewed with a light microscope during controlled freezing down to −2.9 • C ( Figure 5). In the unfrozen state, the shape of the cells was unaffected when exposed to zero ( Figure 5A) or the sub-zero temperature of −2.1 • C ( Figure 5B). After ice formation, injuries were detected, especially in the spongy parenchyma cells ( Figure 5C). Chloroplasts became invisible, likely due to fragmentation and destruction by intracellular ice formation. In lethally frozen cells, the entire cell lumen had a homogeneous, greenish, granular appearance. Cell walls of some spongy parenchyma cells were wrinkled. After thawing, at +5 • C, all spongy parenchyma cells appeared damaged, whereas in the palisade parenchyma, some cells seemed to have survived ( Figure 5D). The damaged cells appeared flaccid, and cell sap leaked out.

Cytological Aspects
Leaf cross-sections of S. tuberosum were viewed with a light microscope during controlled freezing down to −2.9 °C ( Figure 5). In the unfrozen state, the shape of the cells was unaffected when exposed to zero ( Figure 5A) or the sub-zero temperature of −2.1 °C (Figure 5B). After ice formation, injuries were detected, especially in the spongy parenchyma cells ( Figure 5C). Chloroplasts became invisible, likely due to fragmentation and destruction by intracellular ice formation. In lethally frozen cells, the entire cell lumen had a homogeneous, greenish, granular appearance. Cell walls of some spongy parenchyma cells were wrinkled. After thawing, at +5 °C, all spongy parenchyma cells appeared damaged, whereas in the palisade parenchyma, some cells seemed to have survived ( Figure 5D). The damaged cells appeared flaccid, and cell sap leaked out.

Gas Exchange and Photosystem II
The dark respiration rate of the investigated potato leaf ranged between −0.3 and −0.5 μmol CO2•m −2 •s −1 during controlled cooling in darkness and was unaffected by ice that formed at -1.0 °C during E1 (Figure 6). The second freezing process, E2, started at −2.5 °C, occurred 48 min after E1, and lasted approximately 35 min. Immediately after thawing, despite the onset of irradiation, no photosynthetic activity was measurable. An unchanged Rd was recorded. Fv/Fm, which was 0.79 before the freezing treatment, was 0.18, and the value successively decreased with the decay of cells. The day after, the leaf had lost its turgor and showed blackish discolorations, indicating necrosis.

Gas Exchange and Photosystem II
The dark respiration rate of the investigated potato leaf ranged between −0.3 and −0.5 µmol CO 2 ·m −2 ·s −1 during controlled cooling in darkness and was unaffected by ice that formed at -1.0 • C during E1 (Figure 6). The second freezing process, E2, started at −2.5 • C, occurred 48 min after E1, and lasted approximately 35 min. Immediately after thawing, despite the onset of irradiation, no photosynthetic activity was measurable. An unchanged R d was recorded. F v /F m , which was 0.79 before the freezing treatment, was 0.18, and the value successively decreased with the decay of cells. The day after, the leaf had lost its turgor and showed blackish discolorations, indicating necrosis.

Localization of Ice and Cellular Responses to the First Freezing Event
To test the consequences of the first freezing event, we measured gas exchange during non-destructive freezing and prepared semi-thin sections of freeze-fixed tissue frozen at a freezing rate, which would naturally occur. In frozen leaves, the dark respiration rate was unaffected, which indicates that gas exchange was not interrupted by ice. Non-injurious ice formation in intercellular spaces of the spongy tissue of evergreen leaves of Hedera helix and Helleborus niger was found to immediately block gas exchange, while in the ice-free mesophyll of Pinus mugo, even photosynthesis was possible [27]. Unaffected RD in potato leaves upon E1 indicates that intercellular spaces are free of ice at this stage of the freezing process. Records of IDTA during E1 showed a rapid ice propagation along the vascular system, followed by a weak signal all over the leaf [9]. Our results show that this weak signal is not caused by ice formation in the intercellular spaces of the mesophyll but rather originates from thermal diffusion.
In leaves freeze-fixed after E1 and in frozen leaf cross-sections inspected with a light microscope, virtually all mesophyll cells appeared to be in a similar shape as seen in unfrozen samples. This means that upon E1, mesophyll cells of potato do not freeze-dehydrate. Usually, once extracellular ice has formed in a plant tissue, a steep water potential gradient between the extracellular ice masses and the supercooled liquid cell water forces freeze-dehydration, and the gradient increases with decreasing freezing temperature [20,28,29]. Mesophyll cells of many freezing-tolerant species have been shown to freezedehydrate in equilibrium (Allium bulb scales [30]; Dendrosenecio keniodendron [28]; Hedera helix [29]; Sphagnum capillifolium [31]; Ranunculus glacialis [32]) or non-ideal equilibrium (Citrus [33]; Hordeum vulgare [29]; Pachysandra terminalis [34]). In a recent study, it was found that freeze-dehydration is governed by cell wall thickness to cell size ratio and the relative area of intercellular spaces [35]. Furthermore, mesophyll cells of freezing-sensitive species can have extracellular ice and show cell collapse [36], which contrasts with our observation on potato mesophyll cells.
Similar to the findings here, mesophyll cells of several species have been shown to avoid freeze-dehydration and to keep the cell water inside the cell and to supercool (Trach- Figure 6. CO 2 gas exchange of a Solanum tuberosum leaf recorded during a controlled freezing treatment (−2 K·h −1 ). Below +4 • C, the leaves were not irradiated (yellow shading: photosynthetic photon flux density 1000 µmol photons·m −2 ·s −1 ) and kept in darkness until thawing. At −1.0 • C, the first freezing event (E1) was registered. A total of 48 min after E1, a distinct second freezing event (E2) was measured at −2.5 • C, which lasted for 35 min. Black line: CO 2 gas exchange rate, blue lines: leaf temperature, arrows: freezing exotherms. Insert shows the freezing exotherms in the differential temperature plot (leaf−reference temperature).

Localization of Ice and Cellular Responses to the First Freezing Event
To test the consequences of the first freezing event, we measured gas exchange during non-destructive freezing and prepared semi-thin sections of freeze-fixed tissue frozen at a freezing rate, which would naturally occur. In frozen leaves, the dark respiration rate was unaffected, which indicates that gas exchange was not interrupted by ice. Non-injurious ice formation in intercellular spaces of the spongy tissue of evergreen leaves of Hedera helix and Helleborus niger was found to immediately block gas exchange, while in the ice-free mesophyll of Pinus mugo, even photosynthesis was possible [27]. Unaffected R D in potato leaves upon E1 indicates that intercellular spaces are free of ice at this stage of the freezing process. Records of IDTA during E1 showed a rapid ice propagation along the vascular system, followed by a weak signal all over the leaf [9]. Our results show that this weak signal is not caused by ice formation in the intercellular spaces of the mesophyll but rather originates from thermal diffusion.
In leaves freeze-fixed after E1 and in frozen leaf cross-sections inspected with a light microscope, virtually all mesophyll cells appeared to be in a similar shape as seen in unfrozen samples. This means that upon E1, mesophyll cells of potato do not freeze-dehydrate. Usually, once extracellular ice has formed in a plant tissue, a steep water potential gradient between the extracellular ice masses and the supercooled liquid cell water forces freezedehydration, and the gradient increases with decreasing freezing temperature [20,28,29]. Mesophyll cells of many freezing-tolerant species have been shown to freeze-dehydrate in equilibrium (Allium bulb scales [30]; Dendrosenecio keniodendron [28]; Hedera helix [29]; Sphagnum capillifolium [31]; Ranunculus glacialis [32]) or non-ideal equilibrium (Citrus [33]; Hordeum vulgare [29]; Pachysandra terminalis [34]). In a recent study, it was found that freeze-dehydration is governed by cell wall thickness to cell size ratio and the relative area of intercellular spaces [35]. Furthermore, mesophyll cells of freezing-sensitive species can have extracellular ice and show cell collapse [36], which contrasts with our observation on potato mesophyll cells.
Similar to the findings here, mesophyll cells of several species have been shown to avoid freeze-dehydration and to keep the cell water inside the cell and to supercool (Trachycarpus fortunei [37]; Sasa senanensis [38]; Pinus mugo [27]) despite the presence of ice. However, potato mesophyll cells have a limited supercooling capacity. The supercooling capacity of leaf tissues has been related to a specific leaf anatomy [20,35], and it has been shown to be fostered by narrow intercellular spaces and a densely package of mesophyll cells in usually sclerophyllous leaves [20]. Solanum tuberosum leaves are not sclerophyllous but have a small fraction of intercellular space (<1%) and would fulfill at least one anatomical trait that may aid supercooling. Intercellular spaces in herbaceous leaves with freeze-dehydrating mesophyll cells can be up to 35% [35]. Still, potato leaves have only a very low supercooling capacity. In a comparison of different Solanum species that differ in intercellular space size, no relationship to their freezing resistance was found [39]. Additionally, small-sized cells exhibited another important trait for supercooling, and this may be fulfilled in the mesophyll of S. tuberosum. The cell area of mesophyll cells of S. tuberosum is small. Compared to the freeze-dehydrating mesophyll cells of R. glacialis [32], spongy cells are a quarter the size, and palisade cells half the size.
Furthermore, it has been shown that stomatal density is increased in more ice-tolerant Solanum species [39], which somehow contradicts what others have found, i.e., that frost hardening of species of the same genus is coupled with decreased stomatal density (Citrus [40]; Secale sp. [41]).

The Second Freezing Event Is Intracellular
The second, thermally more pronounced freezing event occurs temporarily distinctly separated, up to 78 min later than E1, and coincides with frost damage. Freezing injuries to potato mesophyll cells are caused by intracellular ice formation that spontaneously occurs in supercooled cells. Until now, evidence for intracellular freezing was obtained by IDTA [9]. A second freezing exotherm, the so-called LTE, was indicative of intracellular freezing. The LTE was usually matched with frost damage. Our results show that after E2, mesophyll cells are damaged and unable to regain photosynthetic functions. This is clear, as in the cryo-light microscope, upon intracellular freezing, chloroplasts were seen to become invisible as chloroplasts were destroyed by ice formation. The methods employed in the current study additionally enabled visualizing the intracellular freezing processes during E2 at fully controlled freezing conditions. This is in contrast to several very early microscopic studies, where cooling of samples proceeded without control of freezing and thawing rates, cooling at unnaturally high freezing rates (2-4 K·min −1 ), resulting in various degrees of artificial supercooling down to even −20 • C [15,20,42]. In these earlier studies [15,20,42], symptoms of intracellular freezing were that chloroplasts became invisible, likely due to fragmentation and destruction by intracellular ice growth. In lethally freeze-damaged cells, the entire cell lumen took on a homogeneous, greenish, granular appearance. Cell walls of some spongy parenchyma cells were wrinkled, cells appeared turgor-less, and cell sap seemed to leak out, which is a clear indication of membrane damage. Interestingly, in the palisade parenchyma, the damage pattern was unclear, as in some palisade parenchyma cells, the chloroplasts appeared unaffected. However, in most of the cells, the chloroplasts must have been ruptured as they virtually diminished and the chlorophyll occupied the entire cell volume. This corresponds to some extent to earlier microscopic observations upon intracellular freezing [15,20,42] that were mechanical destruction of biomembranes resulting from the fast growth of ice crystals, symptoms of frost plasmolysis (loss of osmotic responsiveness) after thawing, and complete destruction of cellular components visible in the light microscope, such as chloroplasts. Despite very simple measurement technologies, the symptoms found partly agree with our observations. Images of intracellularly frozen Marchantia protoplasts were shown by Sakai and Larcher [20], being recognizable as an instantaneous darkening of these cells, though precluding visibility of the ongoing processes. Already at low temperatures above the freezing point, aggregation of chloroplasts of Marchantia polymorpha indicates stress-induced cellular reorganization [43]. However, such a darkening symptom was never observed in our experiments during intracellular freezing of potato mesophyll cells. Asahina [15] also sometimes observed flash freezing (instantaneous darkening) on parenchymatous cells during rapid cooling (3-4 K·min −1 ), but non-flash freezing could also occur. In non-flash freezing, several knife-like smooth ice crystals grew parallel to another within a few seconds, and the cell contents were entirely concentrated between these ice crystals as a small number of irregular masses [15].

Outlook
Leaves of S. tuberosum are unable to frost-harden. We were able to show that after the first freezing event, the structural and physiological functionality of mesophyll cells fully remains. Still, upon the stochastic event of the second freezing, ice ruptures cellular membranes and brings immediate frost damage. Consequently, avoidance of ice formation is the only effective strategy to prevent potatoes from frost damage.

Supplementary Materials:
The following material is available online at https://www.mdpi.com/ article/10.3390/crops2040026/s1, Figure S1: Controlled freezing of Solanum tuberosum leaves. (A) The petioles of detached leaves were placed in cotton-filled Eppendorf tubes. Then, an aqueous suspension containing a small amount of ice nucleation active bacteria was added. The Eppendorf tubes were connected to each other using glass-fiber-coated wire that had previously been soaked in the same suspension. To prevent the fiberglass coating from drying out, the wire was inserted into a narrow polyvinyl chloride tube. The leaves were fixed on a polystyrene block. Fine-wire thermocouple sensors were positioned on the lower surface of the leaves to record the leaf temperature and, thus, also the ice formation in the leaf tissue. Wrapping the leaves in cling film increased the thermal insulation of the leaves during the freezing treatment. (B) The same leaves 1 h after thawing. For freeze-fixation, leaf discs (arrows) were punched out from the frozen leaves for subsequent high-pressure freeze-fixation. The leaves were completely undamaged by the freezing treatment applied (−2.5 • C for 20 to 35 min). (C) Leaf temperatures of the three leaves recorded during the freezing treatment. By connecting the Eppendorf tubes via the glass-fiber-coated wire, once nucleated, the ice was able to spread from the first frozen leaf to the others within 2 min (arrows indicate the freezing exotherms), so that all leaves froze almost simultaneously.

Data Availability Statement:
The data that support the findings of this study are available from the corresponding author upon request.