Insect fat body cell morphology and response to cold stress is modulated by acclimation

Insects employ a variety of strategies to survive at low temperatures, and while the majority of those studied (an estimated 70%) are freeze intolerant (Sinclair et al., 2003), the most cold-hardy species survive internal freezing (Worland et al., 2004; Sformo et al., 2010; Li, 2016). Currently, we lack mechanistic understanding about how the cells of some insects survive freezing, and whether survival links to protection or repair of cryoinjury (Pegg, 2007; Clark and Worland, 2008; Gosden, 2011; Štětina et al., 2018; Toxopeus and Sinclair, 2018). Uncovering these basic principles of insect overwintering physiology is imperative for predicting species' distributions and responses to climate change, and also has great potential for applied research; e.g. the efficacy of cold storage of insects (e.g. for biological control) is often limited by the detrimental effects of low temperatures on development, fecundity, performance and survival (Bloem et al., 2006; Colinet and Boivin, 2011). Manipulation of freeze tolerance is therefore crucial for production and management of quality arthropod stock populations and is the ultimate goal in cryopreservation of biological material (Leopold, 1991, 2007; Ayvaz et al., 2008; Ghazy et al., 2012).

Insect cold tolerance is considerably plastic (Ayrinhac et al., 2004; Kvist et al., 2013; Sgrò et al., 2016), and cold acclimation or dietary manipulation can even induce freeze tolerance in otherwise freeze-intolerant insects (Koštál et al., 2011a, 2012; McKinnon, 2015). The malt fly, Chymomyza costata (Zetterstedt 1838) (Diptera: Drosophilidae), is an emerging model for understanding freeze tolerance mechanisms and plasticity in insects (Koštál et al., 2003, 2011a, 2016; Rozsypal et al., 2018; Štětina et al., 2018). Cold tolerance in this species is broadly manipulable; active larvae tolerate freezing only at mild subzero temperatures (e.g. −5°C), but under appropriate rearing conditions the larvae can survive exposure to −196°C, making malt flies one of only a few animals known to withstand such extreme freezing in a hydrated state (Van Wyk et al., 1977; Ramløv and Westh, 1992; Moon et al., 1996; Suzuki et al., 2014). Survival of malt flies at −196°C is achieved by the combined effects of diapause and cold acclimation, but even dietary proline supplementation alone confers extreme freeze tolerance for some individuals (Koštál et al., 2011b). The mechanisms by which acclimation and dietary manipulation confer freeze tolerance are not well resolved.

Freezing presents multiple physiological challenges for the cell: rapid ice crystal formation can damage the cells mechanically, while growth of extracellular ice results in cellular dehydration stress linked to increased acidity, viscosity, toxic metabolic intermediate concentrations, macromolecular crowding and hyperosmotic shock (Neufeld and Leader, 1998; Minton, 2000; Storey and Storey, 2012; Pegg, 2015). These changes in the intracellular environment should severely threaten the chemical and conformational stability of proteins and other macromolecules (Hazel, 1995; Hochachka and Somero, 2002; Bhatnagar et al., 2007; Dias et al., 2010). As direct observation of such molecular damage is technically difficult, the effects of freezing are most often observed at the whole-organism level (via survival analysis). To bridge the gap between loss of molecular function and whole-organism survival, we can first look for evidence of cryoinjury at two technically feasible levels: cells and tissues. The insect midgut and fat body tissues appear to be particularly susceptible to freezing damage (Izumi et al., 2005; Teets et al., 2011). Freezing also damages the Malpighian tubules of freeze-tolerant insects (Neufeld and Leader, 1998; Marshall and Sinclair, 2011). Cytoskeletal proteins (e.g. actin and tubulin) are readily observable structural components of the cell that are modified by both cold stress (Belous, 1992; Russotti et al., 1997; Cottam et al., 2006; Kayukawa and Ishikawa, 2009) and cold acclimation (Michaud and Denlinger, 2004; Kim et al., 2006; Teets et al., 2012; Gerken et al., 2015; Des Marteaux et al., 2018), making them ideal indicators of freeze-attributed cellular damage.

We aimed to identify freeze-attributed changes to cell and tissue morphology that are associated with insect mortality. Using microscopy and immunostaining, we characterized the effects of lethal and sublethal cold stress on morphology of the fat body, Malpighian tubules and anterior midgut in malt fly larvae. Specifically, we compared cell morphology and cytoskeletal structure (F-actin and α-tubulin) of weakly freeze-tolerant (non-diapause) and extremely freeze-tolerant (acclimated, diapausing) malt fly larvae exposed to various cold stresses (from chilling at −5°C to extreme freezing at −196°C) and upon recovery from cold exposure. We also characterized the effects of dietary proline supplementation (which confers freeze tolerance to non-acclimated malt flies; Koštál et al., 2011a; Rozsypal et al., 2018) on cell morphology after freezing. We predicted that cell structure would be resistant to disruption by a freezing challenge and/or repaired during recovery in freeze-tolerant malt flies, but would be disrupted by a freezing challenge in freeze-intolerant malt flies, and that this protection or repair would correlate with survival.

A malt fly (Chymomyza costata) strain (Sapporo) originating from Sapporo, Hokkaido, Japan (Riihimaa and Kimura, 1988), was reared on an artificial diet (52 g malted barley, 28 g corn meal, 18 g dry yeast, 10 g agar powder, 1.7 g methyl hydroxybenzoate per 1.0 liter water) in mass colonies. Adults were housed in 200 ml glass jars containing 10 ml of diet. To collect eggs, a few drops of live yeast suspension were added to fresh apple pieces and kept overnight, after which they were added to flasks containing adults. Females oviposited on the apple for 3 days before the apple pieces were moved to 3×10 cm cylindrical glass vials containing 10 ml of diet and one drop of live yeast suspension. These vials contained developing larval populations of up to 200 individuals. Adult and larval colonies were housed in a Lovibond^® ET 650-8 incubator (Tintometer GmbH, Dortmund, Germany) at 18°C and a 16 h:8 h light:dark photoperiod.

Malt flies were reared from eggs to third instar larvae under different temperature and photoperiod or diet regimes to produce larvae of three different cold tolerance states: (1) weakly freeze tolerant ‘long-day’ (LD); (2) moderately freeze tolerant ‘long-day, proline-supplemented’ (LD Pro50); and (3) extremely freeze tolerant ‘short-day acclimated’ (SDA). LD and LD Pro50 malt flies were reared under warm, long-day (colony) conditions at 18°C and a 16 h:8 h light:dark photoperiod, and the latter group was supplemented with 50 mg proline per gram diet. SDA larvae were first reared for 6 weeks in a Lovibond^® ET 650-8 incubator at 18°C and a shorter day length (12 h:12 h light:dark, which initiates diapause; Koštál et al., 2016), transferred to an MIR 154 incubator (Sanyo Electric, Osaka, Japan) at 11°C and constant darkness for 1 week, and finally placed in a refrigerator at 4°C and constant darkness for 4 weeks (rearing and acclimation details in Fig. S1). Just over 70% of LD larvae survive to adulthood after supercooling at −5°C, while only 39% reach adulthood after freezing at −5°C (and none survive after exposure to −20°C) (Rozsypal et al., 2018). Almost half of LD Pro50 larvae reach adulthood even after freezing to as low as −40°C (Koštál et al., 2011a). For SDA larvae, just over 75% survive to adulthood after freezing to temperatures as low as −75°C, and approximately half reach adulthood after long-term exposure (6 months) to −196°C in liquid nitrogen (LN₂). SDA larval survival and successful emergence after LN₂ exposure requires pre-freezing to −30°C; this temperature marks the completion of glass transition for body liquids (which does not occur at pre-freezing to −20°C). Survival among treatments is listed in Table 1, from Rozsypal et al. (2018).

For cold exposure treatments, larvae were placed between two pieces of moistened cellulose in plastic tubes and immersed in a programmable Ministat 240 bath (Huber, Offenburg, Germany). Temperature inside the tubes was monitored using type T thermocouples and PicoLog software (Pico Technology, Cambridge, UK). Tubes were gradually cooled at 0.1°C min⁻¹ from 0 to −5°C (with and without freezing), −20°C and −30°C, held at the target temperature for 1 h, and then warmed at 0.6°C min⁻¹ to 5°C. Some larvae cooled to −20 or −30°C were immediately plunged into LN₂ for 1 h and then returned to −20 or −30°C, from which they were warmed. Fewer than 20% of SDA larvae survive LN₂ exposure if first cooled to −20°C, while nearly all survive LN₂ if first cooled to −30°C (this temperature likely allows larvae to complete the glass transition of body liquids) (Rozsypal et al., 2018). To separate the effects of temperature and freezing on cell morphology, larvae at −5°C were either supercooled (i.e. remained unfrozen; ‘−5°C SC’ treatment) or frozen by adding an ice crystal to the tube to promote inoculation (‘−5°C F’ treatment). Supercooling or freezing of the larvae was verified by the absence or presence of an exotherm (indicating heat released upon freezing), respectively. All cold exposure details are provided in Fig. S2A. To determine whether disruptions to cell morphology are repaired or persist upon recovery from sublethal freezing, some larvae were returned to colony rearing conditions and dissected after 3 days (LD −5°C F), 1 week (SDA −30°C) or at the prewandering stage (approximately 3 weeks; SDA −30°C).

We dissected larvae in chilled PBS (137 mmol l⁻¹ NaCl, 2.7 mmol l⁻¹ KCl, 10 mmol l⁻¹ Na₂HPO₄·2H₂O, 1.8 mmol l⁻¹ KH₂PO₄, pH 7.2) under a dissecting microscope either directly from rearing conditions (LD and LD Pro50 at 18°C, SDA at 4°C) or immediately after cold exposure regimes. Internal organs (the entire alimentary canal including the Malpighian tubules, and fat body) were extracted from each larva in a single mass and fixed overnight (17–18 h) at room temperature with 4% paraformaldehyde in PB (0.8 mol l⁻¹ Na₂HPO₄·7H₂O, 0.2 mol l⁻¹ NaH₂PO₄·2H₂O, pH 7.4). Tissue masses were rinsed 3×15 min in PB, permeabilized for 20 min in PBT (PB with 0.5% Triton X-100), blocked for 4 h in PBT with 5% normal goat serum (NGS), and immersed in 5% NGS containing a 1:12 dilution of α-tubulin primary antibody (anti-α-tubulin 12G10, Developmental Studies Hybridoma Bank, University of Iowa, USA) at 4°C overnight (17–18 h). After 3×15 min washing in PBT, tissues were then hybridized in 5% NGS containing a 1:200 dilution of secondary antibody (goat anti-mouse Alexa Fluor^® 647; Abcam, Cambridge, UK) for 24 h at 4°C. After 3×15 min rinsing in PBT, tissues were stained with a 1:100 dilution of Alexa Fluor^® 488 phalloidin (Life Technologies, Carlsbad, CA, USA) in PBT for 20 min and rinsed again for 3×15 min. The fat body, entire gut and Malpighian tubules were arranged on glass slides in ProLong™ Diamond antifade mountant with DAPI (Life Technologies) under a cover slip and sealed with clear nail polish. Slides were kept in the dark at 4°C until imaging.

We imaged the Malpighian tubules, anterior midgut and fat body (mid-section and surface) with a Confocal Olympus FluoView FV1000 microscope system (Olympus Europa) at 40× magnification in brightfield and excitation/emission wavelengths of 405/461 nm for DAPI (nuclei), 488/520 nm for phalloidin (F-actin) and 635/668 nm for α-tubulin antibody. We compared cell morphology among 15 treatments in total (Table S1). The number of larvae (biological replication) for each treatment was as follows: anterior midgut, n=9–12; Malpighian tubules, n=10–14; fat body mid-section, n=10–20; and fat body surface, n=9–20. We collected approximately two to four images (technical replicates) per biological replicate for each tissue. For details on biological and technical replication, see Figs S5–S10. Because we discovered that some freezing treatments cause F-actin aggregation (spotting) on the fat body surface, we compared the proportions of cells with many and/or large spots among LD treatments (excluding Pro50) and among SDA treatments (excluding LN₂) using ANOVA on arcsine-transformed data and determined pairwise differences using Tukey's honestly significant difference (HSD) tests. Differences in spotting between LD Pro50 treatments and between SDA LN₂ treatments were determined using Welch’s t-tests. All statistical analyses were performed in R (v3.4.4; https://www.r-project.org/) with α=0.05.

To characterize cell morphologies among different malt fly freeze tolerance variants and cold exposures (15 treatments), we analyzed a total of 2706 tissue images: 1052 fat body mid-sections (526 each for brightfield and combined fluorescence stains), 920 fat body surfaces (460 each for F-actin and α-tubulin), 432 Malpighian tubules and 302 anterior midguts (151 each of epithelia and muscle by α-tubulin and F-actin staining, respectively). We also assessed the extent of F-actin aggregation (spotting) for 6400 fat body cells.

Freezing caused a coalescence of malt fly fat body lipid droplets (represented by SDA larvae in Fig. 1, all treatments in Figs S4 and S5), and our preliminary work indicates that coalescence relates primarily to temperature rather than to time frozen (although time has some effect on LD larvae; Figs S2 and S3). Lipid coalescence was indeed modulated by temperature but also by freeze-tolerance status; coalescence did not occur at −5°C (with or without freezing), was substantial for LD larvae frozen at −20°C (a lethal temperature; Fig. 2), but was only apparent in SDA larvae if frozen to −30°C (a sublethal temperature). Coalescence became entire (formation of a single large lipid droplet displacing nucleus and cytoplasm to the periphery) upon recovery and persisted at least until the prewandering stage.

Freezing of the fat body was accompanied by disintegration of α-tubulin radial structure and (often) a displacement of the nucleus (represented by LD larvae in Fig. 2, all treatments represented in Fig. S5). This loss of tubulin radial structure was most prominent in lethally frozen LD larvae, whose fat bodies were irregular even compared with those of lethally frozen SDA larvae. In addition, the fine mesh structure of α-tubulin on the periphery of fat cells appeared to differ among control, lethally frozen and recovered larvae (described in more detail in Fig. S6); however, these differences were subtle. α-Tubulin disruption in the fat cell interior was not repaired by larvae that survived a freezing challenge (SDA larvae, Fig. 1). The fat body cells of LD larvae exposed to lethal freezing at –20°C exhibited some surface wrinkling (Fig. S7), but all SDA larvae exhibited substantial wrinkling of the fat body cell surfaces prior to freezing and even following recovery (Fig. 3; Figs S6 and S7).

Freezing was associated with the formation of apparent F-actin aggregates (spots) at the periphery of fat body cells (Fig. 4; Fig. S7). The extent of this aggregation differed by freeze tolerance status, and aggregates disappeared upon recovery (LD: F_4,64=26.4, P<0.001; SDA: F_5,54=30.2, P<0.001). In LD larvae, F-actin aggregates appeared in fat body cells when frozen at −5°C but were rare or absent if cells were merely supercooled to −5°C. Fat bodies of LD larvae lethally frozen to −20°C also exhibited F-actin aggregation; however, this was more variable among individuals. Cold acclimation lowered the temperature at which aggregates appeared: we observed aggregates in the SDA larval fat body only upon freezing to −30°C (none were apparent at −5 or −20°C).

Cell and tissue morphologies of SDA larvae exposed to LN₂ after pre-freezing to −20 and −30°C were very similar (Figs S4–S10). Fat body lipid droplet coalescence and loss of α-tubulin radial structure in both variants was severe, and both exhibited fat body F-actin aggregation. Proline supplementation produced variable morphological responses to freezing in malt fly larval tissues. For example, the extent of lipid droplet coalescence in LD Pro50 larvae exposed to −20°C ranged from none to severe (Fig. S5). Frozen LD Pro50 larvae also exhibited the highest proportion (87%) of fat body cells with F-actin aggregate spots (Fig. S7), and all Malpighian tubules exhibited lumenal F-actin spotting (Fig. 5). Supplementation with proline caused surface wrinkling in some fat body cells at 18°C (albeit to a lesser degree than in SDA larvae; Figs S4 and S7).

Anterior midgut and Malpighian tubule morphologies were generally unchanged among freeze tolerance variants and cold exposures (represented by various treatments in Fig. 6, all treatments in Figs S8–S10), with the exception of F-actin in the Malpighian tubule lumen (above). Sublethal freezing to −5°C (LD) and −30°C (SDA) caused lumen F-actin aggregation in approximately 14 and 50% of the Malpighian tubules, respectively, which disappeared completely upon recovery (F_9,94=8.46, P<0.001). F-actin aggregates were not apparent in the LD −20°C variant.

We expected to observe major morphological differences between unfrozen and frozen malt fly tissues and among different larval freeze tolerance variants; however, only the fat body generally exhibited obvious differences at the level of resolution investigated here. Furthermore, some of the apparent freezing injuries in fat body tissue were unrelated to mortality; it is unclear how these structural disruptions are seemingly without detriment. Below, we focus our discussion on the prominent fat body structural changes and their relationship to whole-animal mortality.

Coalescence of fat body lipid droplets upon freezing has been observed in other insects (Salt, 1962; Asahina, 1970; Lee et al., 1993) and, as in malt flies, this phenomenon appears to be primarily driven by temperature (Lee et al., 1993). The extent of lipid droplet coalescence was also modified by malt fly freeze-tolerance status: diapause and cold acclimation lowered the temperature of coalescence by 10°C. That this coalescence should become complete and persist in diapausing, cold-acclimated larvae upon recovery from sublethal freezing was unexpected. These larvae have good survival and emergence success (Rozsypal et al., 2018); therefore, droplet coalescence is clearly not a lethal form of freeze injury. Freeze-tolerant Eurosta solidaginis (Diptera) and Cephus cinctus (Hymenoptera) similarly survive such lipid coalescence in the fat body cells (Salt, 1962; Lee et al., 1993). It is therefore unclear whether protection of lipid droplet morphology to lower temperatures is somehow beneficial for diapausing, cold-acclimated malt fly larvae. Energy utilization problems stemming from such a substantial reduction in lipid droplet surface area could affect reproductive output or time to starvation, but these sublethal effects of freezing have yet to be determined for adults.

Disintegration of fat body α-tubulin radial structure upon freezing was less substantial for diapausing, cold-acclimated malt fly larvae (a variant that also defended lipid droplet structure to lower temperatures), but it is unclear whether failure of tubulin upon freezing caused coalescence of lipid droplets or vice versa, or whether these effects simply coincide. Still, α-tubulin disruption is clearly also a non-lethal consequence of cold exposure, as recovered larvae did not repair α-tubulin structure. Although the fine mesh structure of α-tubulin on the fat cell periphery showed subtle differences among control, lethally frozen and recovered larvae, it is unclear whether these differences have functional consequences for the cells.

Larval fat body cell surface wrinkling upon lethal freezing suggests freeze-attributed dehydration which could correspond with mortality (Sinclair and Renault, 2010; Rozsypal et al., 2018). However, extensive cell wrinkling prior to freezing and following recovery in diapausing, cold-acclimated malt flies suggests that this physiological state provides some degree of protective intracellular dehydration. Such dehydration could both lower the risk of intracellular freezing and subsequently prevent rapid fluctuations in cell volume and osmolality associated with freeze–thaw. Still, survival of these extremely freeze-tolerant larvae is essentially independent of the body's ice fraction across a broad range of sub-zero temperatures (Rozsypal et al., 2018). Dehydration could therefore instead concentrate intracellular proline or other cryoprotectants to thresholds that promote macromolecular protection (Koštál et al., 2011a; Rozsypal et al., 2018), or perhaps precondition cells against anticipated osmotic stress. Alternatively, fat body dehydration could be a non-protective consequence of dormancy (Danks, 2000) and/or the cold acclimation regime (i.e. an effect of age and loss of body water over time).

F-actin aggregation (spotting) at the periphery of fat body cells was modified by freeze tolerance status: diapause and cold acclimation lowered the temperature at which spots appeared by 25°C. We suspect that these apparent protein aggregates result from depolymerization of F-actin upon cold exposure (Kayukawa and Ishikawa, 2009), and that a reduction in the temperature at which actin aggregates form may indicate a general trend of protein stabilization at freezing temperatures. Disappearance of actin aggregates in the fat bodies of recovered larvae suggests that repair of the actin cytoskeleton may correlate with survival of a freezing challenge. Cold-acclimated malt flies indeed increase expression of genes involved in both protein stabilization (e.g. heat shock proteins) and degradation (e.g. ubiquitination and proteasome machinery; L.E.D.M. and V.K., unpublished data), and cytoskeletal protection appears to underlie cold acclimation in other insects (Kim et al., 2006; Kayukawa and Ishikawa, 2009) and plants (Örvar et al., 2000; Ouellet et al., 2001). However, it is impossible to determine whether lethally frozen malt flies perish owing to an inability to repair actin structure.

Unlike in the fat body, the general morphologies and structural (cytoskeletal) appearance of the malt fly anterior midgut and Malpighian tubules were mostly unchanged among treatments. In the anterior midgut, neither the epithelium (visualized by α-tubulin staining) nor muscle (visualized by F-actin staining) differed between freeze tolerance status or upon a freezing challenge (whether lethal or sublethal). Although damage to muscle cell ultrastructure is visible in lethally frozen Eurosta solidaginis gall flies (Collins et al., 1997), such injury in malt fly muscle may not have been detectible via phalloidin staining and the lower magnification used here. Freeze injury in the gut may also not be immediately apparent upon thawing. We did observe some (albeit variable) freezing-associated F-actin aggregation in the Malpighian tubule lumen, which, like in the fat body, was apparently repaired by larvae during recovery. General Malpighian tubule morphology and α-tubulin structure were otherwise unaffected. Investigation of other structural components in these tissues (e.g. cell junctions; MacMillan et al., 2017) either immediately or some hours after a freezing challenge may uncover other forms of cryoinjury.

Dietary proline improves survival of warm-acclimated malt flies at −20°C from zero to roughly 50% (Rozsypal et al., 2018); therefore, it is unclear why we observed more pronounced actin aggregation in the fat body and Malpighian tubules of these larvae compared with non-supplemented larvae frozen to the same temperature. Pre-freeze surface wrinkling in some fat body cells of proline-supplemented larvae could have resulted from the osmotic effects of proline, and may therefore confer some form of protective pre-dehydration (discussed above). Because cell morphologies were similar after exposure to LN₂ regardless of whether larvae were pre-frozen at −20°C (resulting in poor larval survival and no metamorphosis to adults) or −30°C (allowing for substantial larval survival and adult metamorphosis for ca. 40% of individuals), it is likely that the mechanisms conferring extreme freeze tolerance (e.g. glass transition of body liquids; Rozsypal et al., 2018) cause no detectable morphological changes in the structures and at the scale visualized presently.

Uncovering freeze-attributed damages that correlate with insect mortality remains a challenge, but the freeze-susceptible fat body and (to some degree) Malpighian tubules of malt fly larvae appear to be suitable indicators of cryoinjury. Our observations suggest that repair or removal of damaged (aggregated) cytoskeletal proteins is a promising avenue of further investigation for the mechanisms of acquired freeze tolerance, and we expect that macromolecule protection and repair may be a general feature of cold-acclimated insects. We therefore recommend similar morphological (and ultrastructural) assessments of the effects of acclimation and freezing on other macromolecules and cell features (e.g. cell junctions and mitochondria; Collins et al., 1997). Cellular pre-dehydration – which may protect cells from osmotic cryoinjury – is another potential mechanism of acquired freeze tolerance that warrants future consideration. We show that fat body lipid droplet coalescence is not an injury associated with freeze mortality, but we recommend rigorous assessment of any sublethal effects that this coalescence may cause in adults.

We would like to thank Jaroslava Korbelová for malt fly rearing and cold exposure, Irena Vacková for malt fly colony maintenance and Hana Sehadová for confocal microscopy assistance.