Influences of thermal acclimation and acute temperature change on the motility of epithelial wound-healing cells (keratocytes) of tropical,temperate and Antarctic fish

Despite extensive research into thermal relationships of ectothermic animals, there remain large gaps in our understanding of how temperature affects the ability of ectotherms to cope with pathogens through both specific and nonspecific immune responses. Temperature affects both specific and nonspecific immunity, so these thermal effects are apt to be of considerable importance to ectotherms that encounter variations in environmental temperature (see Le Morvan et al.,1998 for a review). Low temperatures, consistent with normal seasonal fluctuations, correlate with increased pathogenesis in teleost fish(Baudouy et al., 1980; Bly and Clem,1992; Le Morvan et al.,1998). Most infectious diseases occur over wide ranges of body temperatures and the outcome of the pathogenesis is likely to depend on the differential effects of change in temperature on the efficacy of the immune response and the growth rate of the pathogen. In the case of primary antibody production, both delayed response (Rijkers et al., 1980) and suppression(Le Morvan et al., 1998) by cold temperatures have been reported. This temperature sensitivity results from the inability of T helper cells to undergo rapidly homeoviscous adaptation, leading to changes in membrane physical state that affect membrane-bound receptors and cell surface carbohydrates(Le Morvan et al., 1996). In addition to temperature mediation of the humoral response, cold temperatures also suppress lymphocyte proliferation(Ellsaesser and Clem, 1986; Bly and Clem, 1991) and specific toxicity (Verlhac et al.,1990). The effect of temperature on specific immunity is also species specific (Bly and Clem,1992).

Less is known about the effects of temperature on nonspecific immunity,although available data indicate these effects to be complex and of potentially great importance to organisms. Low temperatures appear to increase the efficacy of certain nonspecific immune responses, including cytotoxic cell lytic activity (Le Morvan et al.,1995), phagocyte respiratory burst(Dexiang and Ainsworth, 1991; Collazos et al., 1994) and macrophage response to activating factor and burst activity(Le Morvan et al., 1997). Thus, there is evidence that nonspecific immunity may respond to cold stress relatively quickly and protect the organism from further damage by a pathogen until cold-temperature suppression of specific immunity is overcome.

A potentially important contributor to the nonspecific immune responses is the wound healing function of cells termed keratocytes, which are terminally differentiated epithelial cells found in teleosts and amphibians. The motility of these cells was first described in goldfish (Carassius auratus) by Goodrich (1924). Subsequent research, including in vivo studies using Xenopus laevis,showed that keratocytes migrate across wounded areas to provide a barrier to infection (Radice, 1980a,b; Euteneuer and Schliwa, 1984). Keratocytes may crawl together as a sheet of cells or move individually. The cells extend large lamellipodia, and motility is driven by actin polymerization (Theriot and Mitchison,1991; Small et al.,1995). As is true for other types of actin-based cell motility,locomotion is thought to occur through the coordination of several distinct steps. First, the cell extends its leading edge forward, and this protrusion of the membrane is coupled with adhesion to the substrate. Next, traction force leads to a forward translocation of the cell body. Lastly, contact with the substrate at the rear of the cell is dissociated and the rear edge of the cell is released to follow the forward procession. Many of the cellular processes involved in keratocyte movement, such as actin polymerization and membrane fluidity, are strongly affected by the physical and chemical environment (Cossins et al.,1987; Hochachka and Somero,2002). Thus, it is to be expected that keratocyte motility and the wound healing capacity it supports would be highly sensitive to environmental temperature. The potential scale of these thermal effects is suggested by the finding that actin-based movement in mammalian fibroblast cells increased with a Q₁₀ of approximately 4 between 29°C and 39°C(Hartmann-Petersen et al.,2000).

Currently, little is known about thermal effects on keratocyte motility in ectotherms and how this thermal sensitivity varies among species evolutionarily adapted to different temperatures or among conspecifics acclimated to low and high temperatures. To examine these phenomena, we studied keratocyte motility in four species of fish: two eurythermal species -a temperate goby, Gillichthys mirabilis (10-35°C), and a desert pupfish, Cyprinodon salinus (10-40°C) - and two species with narrow environmental temperature ranges - an Antarctic notothenioid, Trematomus bernacchii (-1.86°C), and a tropical clownfish, Amphiprion percula (26-30°C). In order to examine the long-term effects of acclimation as well as acute effects of experimental temperature,the first two species were acclimated to multiple temperatures within their physiological thermal range, and the stenothermal T. bernacchii was maintained at 5°C, the highest temperature to which this species can be acclimated (see Somero and DeVries,1967; Hofmann et al.,2000). We describe variation in keratocyte morphology and quantify keratocyte motility in terms of cell speed and direction in relation to evolutionary, acclimatory and experimental temperatures. These data provide insights into thermal effects on nonspecific immunity and into the physiological determinants of organismal thermal tolerance limits.

Gillichthys mirabilis (Cooper) (Family: Gobiidae) are found in estuarine habitats in the Gulf of California and along the California coast as far north as Tamales Bay (Eschmeyer et al., 1983). G. mirabilis were collected in coastal waters near Santa Barbara, CA in 1997-1999 using fish traps. Cyprinodon salinus (Miller) (Family: Cyprinodontidae) is found in Salt Creek, Death Valley, California (Page and Burr,1991). Specimens were obtained from Dr G. Bernardi (University of California at Santa Cruz). Amphiprion percula (Lacepède)(Family: Pomacentridae) occur throughout the Western, Central and South Pacific Ocean, the Indian Ocean and the Red Sea (W. Capman, http:www.augsberg.edu/biology/aquaria). Specimens of A. percula were obtained from a California tropical fish store and were captive bred and raised. Trematomus bernacchii(Boulenger) (Family: Nototheniidae) occurs in coastal waters around Antarctica and the Antarctic Peninsula, South Shetland Islands, South Orkney Islands and Peter I Island from the shore to 695 m depth(Fischer and Hureau, 1985). T. bernacchii were collected in McMurdo Sound in December 1998 and January 1999 using hook and line or fish traps.

Stocks of field-collected G. mirabilis were maintained at Hopkins Marine Station in a continuous-flow, open-circulation tank containing ambient water (16°C) from Monterey Bay. Individuals were acclimated to 10°C,16°C and 25°C (±0.5°C) for a period of 4 weeks in closed recirculating tanks equipped with a flow-through heater/chiller. C. salinus were acclimated for 4 weeks at temperatures of 26±0.5°C or 35±0.5°C in closed recirculating tanks in seawater taken from Monterey Bay that was adjusted to 50%thou salinity. T. bernacchii were studied at the US McMurdo Station. Fish were maintained in a continuous-flow, open-circulation tank in ambient water (-1.86°C)pumped in from McMurdo Sound. T. bernacchii acclimated to 5°C were held for 3 weeks in a closed recirculating tank heated to 5±0.5°C. A. percula were maintained at 26±0.5°C for 4 weeks in seawater made from Instant Ocean© (MediaCybernetics,Silver Spring, MD, USA).

Keratocytes were cultured in L-15 medium made with 50% filtered seawater(0.2 μm filter) and supplemented with Hepes, 5% antibiotic/antimycotic(Sigma) and 10% fetal bovine serum (FBS)(Ream, 2002). Several scales were removed, using forceps, from 2-3 fish from a single acclimation for each temperature experiment. In the case of T. bernacchii, scales were removed from only one fish for each experiment because of limited specimen availability. The scales were washed in medium for 15 min and then plated between untreated glass cover slips with fresh medium. Cultures of keratocytes from G. mirabilis acclimated to 10°C, 16°C and 25°C were maintained at 12°C, 20°C and 25°C, respectively; keratocytes from C. salinus and A. percula were maintained at 25°C; and keratocytes from T. bernacchii acclimated to -1.86°C and 5°C were maintained at 0°C and 4°C, respectively. Keratocytes cultured from G. mirabilis matured in 12 h and were viable for 36 h after maturation; keratocytes from C. salinus and A. perculamatured in 6 h and were viable for 12 h after maturation; and keratocytes from T. bernacchii matured in 48 h and were viable for 36 h after maturation.

T. bernacchii can be supercooled to -6°C if they are previously warmed to 0°C to melt any ice crystals present in their body fluids (DeVries and Cheng,1992). Keratocytes cultured from T. bernacchii(-1.86°C) were subjected to supercooling experiments in order to determine whether or not individual cells in primary culture could withstand supercooling. Glass cover slips containing keratocytes were placed in seawater at 0°C and slowly cooled to -6.0°C. Keratocytes were then slowly returned to 5°C and visualized on an Olympus light microscope to determine if they were still capable of locomotion.

Cells were imaged on three different inverted light microscopes (located at the three different biological laboratories where the research was conducted)under 10×(2×), 20×(2×) or 40×(2×)magnification. The Nikon Diaphot microscope (Hopkins Marine Station) was equipped with a Peltier temperature-controlled stage, whereas the Olympus BH-2(McMurdo Station) and Nikon Diaphot-300 (Stanford University School of Medicine) microscopes were equipped with temperature-controlled stages attached to a circulating water bath. Cell cultures, cell behavior and cell speed remained consistent across imaging conditions. Temperature was measured at the center of the surface of the cover slip in the area being imaged using a thermocouple probe. Temperature was maintained within ±0.5°C of the designated experimental temperature. The temperature of the cover slip and surrounding medium was slowly brought from room temperature to experimental temperature, taking as little as 10 min or as long as 4 h, depending on the temperature differential. Once the experimental temperature was reached, the cells were held at that temperature for a minimum of 20 min before recording data. There was no discernible difference in cell speed or behavior between cells monitored after 20 min or 4 h atexperimental temperatures. Each cell was recorded for 15 min and video frames were captured every 15 s, resulting in a total of 61 frames per cell. Depending on the speed and path of the cell, many cells crawled out of the field of view. When this occurred, the time lapse was paused, the stage was moved in order to replace the cell in the field of view,and the time lapse was then continued. Each pause resulted in a temporal and spatial discontinuity that was accounted for in the analysis but which often reduced the total number of useful frames per cell. A total of 20 cells from a minimum of two cultures were imaged at each experimental temperature. No more than 12 cells per culture were imaged for a given temperature condition.

Movies of keratocytes imaged on the Nikon (Hopkins Marine Station) or Olympus microscopes were acquired using a Silicon Intensifier Target (SIT)camera (Nikon; Hopkins Marine Station) or chilled charge-coupled device (CCD)camera (Olympus) attached to a VCR and recorded in real time to VHS. Images were pulled from the tapes every 15 s using ImagePro Plus® data analysis software (MediaCybernetics). Time-lapse video was acquired digitally for keratocytes imaged on the Nikon (Stanford University Medical School)microscope using an intensified charge-coupled device (ICCD) camera (GenII Sys/CCD-c72) with the shutter driver and acquisition capabilities of MetaMorph® image analysis software (Universal Imaging Corporation™;Downingtown, PA, USA). Frames were captured every 15 s and compiled into time-lapse movies. Keratocytes were tracked by hand using a mouse and the`point-and-click' method utilized by both software packages, yielding discrete(x,y) coordinates for the keratocyte at every 15 s interval. The cell body and leading edge of the lamellipodium were tracked for >60 cells and no discernible difference was found between the two tracking methods. The cell body was chosen for simplicity and all cells used in this analysis were tracked with that method.

Keratocyte paths were characterized with respect to directional movement. Directional movement can be visualized by plotting the trajectory of each cell as (x,y) coordinates. Each trajectory begins at the origin and is rotated so that the first step taken by each cell is in the same direction(x=0). A matrix using non-overlapping intervals was used to determine all possible path lengths for this analysis. Additionally, the magnitude of the turns made by keratocytes can be examined by plotting the mean cos(θ), or turning angle, against the path length traveled by each cell for each pair of (x,y) coordinates. This was done using non-overlapping intervals for all possible path lengths. This method does not reveal directionality, because it takes into account turning magnitude only. Mathematical formulas used in these analyses can be found in Ream(2002).

Calculation of mean speed, median speed, standard deviation and standard error for each cell was performed using Microsoft Excel®. The statistical analysis package in Sigma Plot® version 5.0 was used to perform Student's t-tests to determine whether or not the increase in keratocyte speed between experimental temperatures was statistically significant. The Microsoft Excel® statistical software package was employed to run two-way analyses of variance (ANOVAs) between experimental temperatures within each acclimation to determine if thermal acclimation had an effect on speed.

Keratocytes exhibited wide variation in size, shape, pattern of shape change and lamellipodial ruffling, and this morphological variation differed among species. Fig. 1illustrates the range of morphologies that were observed. Keratocytes from all species underwent some degree of change in shape that was consistent with large-scale directional changes, which were frequently coupled to changes in the leading edge designation. The amount of time spent by a cell determining which edge of the cell became the new leading edge, along with the frequency of morphological changes involved, varied among species.

Keratocytes from G. mirabilis exhibited the greatest size variation. Within a single culture, cells of average size (≈35 μm wide along the long axis) could be found along with cells as large as ≈80 μm and as small as ≈15 μm in width. For the other species, size variation within a species was only approximately ±5 μm. Regardless of their size, G. mirabilis keratocytes displayed the characteristic `canoe shape' first described by Goodrich(1924), which denotes an elongated, elliptical cell body with a smooth-edged, narrow lamellipodium running along one side of the cell body and smoothly curving around each end(Fig. 1A). Some degree of ruffling was apparent over time and usually occurred at the leading edge of the cell, although the edges of the lamellipodium were generally smooth. Shape-changing events in G. mirabilis keratocytes had durations of 1-2 min, occurred in fewer than half of the cells and involved few lamellipodial rearrangements.

Keratocytes of C. salinus were ≈35 μm wide and varied little in size. The keratocytes were less canoe-shaped in appearance due to a lengthening in the lamellipodia between the cell body and the leading edge,which gave the cells a fatter appearance. Additionally, the lamellipodia of these cells ruffled >50% of the time. These ruffles were almost always in the first third of the lamellipodium at the leading edge of the cell and appeared as large, rapidly changing, phase-dense regions rather than small dark creases. Shape changes in C. salinus keratocytes occurred in slightly fewer than half of the cells, lasted approximately 2-4 min and involved more lamellipodial rearrangements than in cells of G. mirabilis and A. percula.

Keratocytes of A. percula were ≈35 μm wide along the long axis and had the stereotypical canoe shape. The leading edges of their lamellipodia were gently rounded, and most ruffling occurred at the ends of the lamellipodia rather than at the leading edge. Shape changing events in A. percula keratocytes had durations of 1-2 min, occurred in less than half of the cells undergoing shape changes and few had lamellipodial rearrangements.

T. bernacchii keratocytes also displayed the characteristic canoe shape and had minimal variation in size (≈35 μm wide along the long axis). The lamellipodia always appeared smooth and any ruffling was in the form of small creases in the lamellipodia at the edges of the cell body. The most notable feature of T. bernacchii keratocytes was their extremely slow rate of locomotion. Locomotion, lamellipodial ruffling or any manner of shape change was only detectable using time-lapse microscopy, in contrast to the cells of other species where dynamic events were often apparent to the eye. Shape changes occurred in a little over half of the cells, lasted approximately 4-5 min and involved a large number of lamellipodial rearrangements.

Keratocyte motility can be characterized in terms of a velocity comprising a magnitude (speed) and a complex directional component (Figs 2, 3). The directional component can be further broken down into persistence, or movement in a single direction, and turning behavior, namely the size of the angles made by a cell as it turns. Both components of velocity as well as the thermal range over which motility was observed, varied among species and, in some cases, with acclimation history.

The temperature ranges over which keratocyte motility occurred reflected the evolutionary adaptation temperatures of the species(Fig. 4). Keratocytes of T. bernacchii maintained at the normal McMurdo Sound temperature(-1.86°C) or acclimated to 5°C were functional over the full range of measurement temperatures (5-20°C). Keratocytes from T. bernacchii(-1.86°C) that were supercooled to -6.0°C for 10 min had normal motility when they were returned to 5°C. Keratocytes cultured from G. mirabilis acclimated to 10°C, 16°C and 25°C were able to function at 10-35°C. The keratocytes isolated from A. perculamaintained at 26°C were also functional at 10-35°C. Keratocytes from C. salinus acclimated to 26°C and 35°C were functional at 10-40°C and 5-40°C, respectively. In all cases, the thermal range of the keratocytes was at least as broad as the thermal range of the fishes'habitats, and, for the two stenothermal species (T. bernacchii and A. percula), the thermal range of keratocyte motility was much greater than the environmental temperature range of the fish.

Mean speed of keratocyte movement varied among species and as a function of experimental temperature (Fig. 4). Autocorrelation analysis of instantaneous speeds measured over 15 min varied from cell to cell and showed no strong correlation and no oscillation in speed for any species examined(Table 1), indicating that each instantaneous speed is discrete and is not affected by previous cell speed.

G. mirabilis keratocyte motility significantly increased in speed with increasing temperature from 5°C to 25°C (P<0.001 in all cases) with a Q₁₀ of ≈2.5(Fig. 4A). At 25°C, mean speed reached a maximum and then decreased at higher temperatures. Variation in instantaneous speed also increased with increasing measurement temperature between 5°C and 35°C. Although the mean speed of keratocytes from individuals held at all three acclimation temperatures exhibited the same response to experimental temperature (P=0.53), mean speed varied with acclimation temperature, with cells from the 10°C group most commonly having the slowest speeds (Fig. 4A). At low experimental temperatures of 5-15°C, keratocytes from 25°C-acclimated fish moved more rapidly than keratocytes from the two cold acclimations (16°C and 10°C). Conversely, cells from cold-acclimated fish (10°C) moved more slowly at warmer temperatures(20-35°C) than those from the two warmer acclimations (16°C and 25°C). The maximal speed of ≈45 μm min^-1 was reached by keratocytes from the 16°C acclimation. The mean speed for all the acclimation groups at 15°C, the experimental temperature that was closest to the habitat temperatures of the population used, was 11.4 μm min^-1.

C. salinus keratocyte speed increased with increasing temperature from 5°C to 35°C, with a Q₁₀ of 2.0, and decreased between 35°C and 40°C. The speeds are statistically different at each temperature (P<0.001 in all cases) except for cells of individuals acclimated to 35°C, which moved at the same rate at 25°C and 30°C. The variation in speed also increased with temperature. C. salinuskeratocytes from both acclimation temperatures exhibited the same response to experimental temperature (P=0.46), although mean speed varied between acclimation groups at several measurement temperatures. Keratocytes from both thermal acclimations did not vary in speed at low (10-15°C) and high(35-40°C) experimental temperatures but crawled at statistically different speeds at the middle experimental temperatures (20-30°C; Fig. 4B). At 20°C, the cold-acclimated cells moved faster, but at 25-30°C the warm-acclimated cells were the fastest moving. The mean speed for all the acclimation groups at 30°C, a common physiological temperature for C. salinus, was 13.8 μm min^-1 (Fig. 5). The maximal instantaneous speed was ≈45 μm min^-1 at 35°C. Both the mean speed at physiological temperature and the maximal speed were very similar to those of keratocytes from G. mirabilis.

Motility of keratocytes of A. percula significantly increased in speed as temperature increased from 10°C to 35°C(P<0.001), except between 15°C and 20°C, at which temperatures the keratocytes exhibited similar speeds (P=0.095; Fig. 4C). The overall Q₁₀ was ≈2.0. The maximal instantaneous speed measured was≈45 μm min^-1. Variation in instantaneous speed also increased with increasing temperature between 5°C and 15°C and between 20°C and 35°C. The mean speed at 30°C, a common physiological temperature,was 10.8 μm min^-1 (Fig. 5).

Keratocytes cultured from T. bernacchii held at -1.86°C also exhibited temperature-dependent rates of motility(Fig. 4D, inset). Keratocyte speeds are statistically different (P<0.05) between the experimental temperatures of 5°C, 10°C and 15°C; however,keratocytes moved at the same speed at 15°C and 20°C(P=0.068). The maximal instantaneous speed measured was 0.046 μm min^-1, a small fraction of the maximal velocity of the other species' keratocytes. The mean speed at 5°C, the experimental temperature closest to physiological (-1.86°C) was 0.008 μm min^-1. Thus,no temperature compensation of rate of motility was found for keratocytes of T. bernacchii (Fig. 5). Although the mean speed of keratocytes from T. bernacchii held at both acclimation temperatures exhibited the same response to experimental temperature (P=0.63), acclimation groups were statistically different (P<0.01) from each other at each experimental temperature. The strong acclimation effect on the speed of keratocytes from animals acclimated to 5°C is most obvious at the experimental temperatures of 15°C and 20°C, where speed was 10-20-fold greater than in the -1.86°C-acclimated cells. Variability in speeds was enormous, with speeds in the range of 0.065-1.05 μm min^-1. Keratocyte speeds were statistically different (P<0.01) at each experimental temperature.

Two types of analysis were used to quantitatively describe the directional component inherent in cell velocity. The first type of analysis plots the trajectory of each cell, beginning at the origin and oriented such that the first step taken by each cell is along the x-axis. Trajectories were compared at all experimental temperatures; however, general trends can be summarized by comparing trajectories at 10°C and 20°C(Fig. 6). This analysis made it possible to visualize path shape and to determine the degree of persistence in a single direction. There were cells taking very straight and persistent trajectories at all experimental temperatures and acclimations; however, the number of individual cells exhibiting this type of directional movement varied with experimental temperature and acclimation. The second type of analysis,plotting the cos (θ) of vectors tangential to the cell trajectory versus path length traveled between each pair of vectors, revealed turning magnitudes in highly persistent cells that were not obvious in the trajectory analysis (Fig. 7).

Keratocytes exhibited a range of turning behaviors that varied with both experimental and acclimation temperatures. At lower experimental temperatures,cells from all species traveled in persistently straight trajectories,regardless of acclimation history. Keratocytes from T. bernacchii had the greatest persistence at all experimental temperatures of any species examined, and persistence was not affected by experimental temperature. For all non-Antarctic species, as experimental temperature increased, persistence in a single direction decreased and turning associated with random walk behavior increased. Interestingly, persistent circular trajectories appear with increasing experimental temperature in the case of keratocytes cultured from G. mirabilis and C. salinus. G. mirabilis keratocytes took circular paths only at experimental temperatures of ≥20°C. Additionally, keratocytes from individuals acclimated to 10°C were more likely to have circular trajectories than those acclimated to 16°C, while fish acclimated to 25°C had the greatest number of cells with circular trajectories. C. salinus keratocytes were less persistent at lower experimental temperatures, and at higher experimental temperatures they exhibited turns of larger radii compared with G. mirabiliskeratocytes, with few C. salinus cells making completely circular paths at experimental temperatures below 35°C. Furthermore, trajectories varied with acclimation temperature; keratocytes from cold-acclimated fish(26°C) were less persistent with more random walks at experimental temperatures below 20°C and showed a greater incidence of curved and circular paths at temperatures greater than 20°C than cells from the warm-acclimated group.

Analysis of the cos (θ) versus path length reveals that the turning angles between two adjacent video time-lapse intervals (path lengths≤20 μm) are less sharp than the turning angles between non-adjacent intervals spaced further apart, which shows that the turning angle increases with increased path length [average cos (θ)=0.1 to 0.9, depending on path length]. Thermal acclimation affected the magnitude of directional changes taken by cells of all species examined. Keratocytes from G. mirabilis and C. salinus acclimated to 10°C and 26°C,respectively, displayed greater turning angles than the cells from warmer acclimations. G. mirabilis cells acclimated to 16°C took the straightest paths at 20°C, and thermal acclimation had the strongest effect on turning magnitude at 20°C. The turning magnitudes of keratocytes cultured from C. salinus exhibited greater turning magnitude at higher temperatures, and cells acclimated to 35°C took the straightest paths at 10°C. A. percula keratocytes made larger angle turns than the other non-Antarctic species, and their turning behavior was statistically the same (P>0.05) at every experimental temperature. Thermal acclimation affected turning behavior in T. bernacchiikeratocytes at path lengths greater than 2 μm, with -1.86°C-acclimated cells making larger turns at low temperatures and 5°C-acclimated cells making larger turns at high temperatures.

This is the first comparative description of the effects of temperature on fish keratocyte motility. A primary conclusion from this study is that the speed and persistence of movement of teleost keratocytes are highly variable and are complex functions of evolutionary, acclimatory and measurement temperatures. The effect of measurement temperature on rate of motility yielded Q₁₀ values of ≈2 for most species and acclimation groups, which suggests that the biochemical processes that are responsible for actin-based motility have relatively consistent thermal sensitivities across species. However, the absolute rates of motility differed significantly among species. Keratocytes of the two eurythermal species, G. mirabilis and C. salinus, were capable of crawling the fastest, with maximal instantaneous rates of ≈40-45 μm min^-1. It is possible that there is some limit to keratocyte motility indicated by this maximal speed that might be reflected in an underlying cellular process governing motility rates.

For all species except the Antarctic notothenioid T. bernacchii,rates of movement reflected compensation for temperature(Fig. 5). Thus, at common habitat temperatures, the keratocytes of G. mirabilis, C. salinus and A. percula had mean speeds of 11.4 μm min^-1 (15°C),13.8 μm min^-1 (30°C) and 10.8 μm min^-1(30°C), respectively. Although we did not directly examine the process of wound healing in this study, we conjecture that conservation of a certain range of motility rates is advantageous for ensuring that the role of keratocytes in the wound-healing process is retained across adaptation temperatures. Therefore, it is likely that keratocytes moving at 11-14 μm min^-1 are capable of adequate migration to be effective in wound closure for prevention of infection. Slower rates may not be as effective and more rapid locomotion might be too costly, metabolically, and not yield a sufficient increase in protective value.

Although temperature compensation of keratocyte motility was observed in comparisons among the non-Antarctic species, none of the species studied showed acclimatory temperature compensation of motility. The lack of acclimatory temperature compensation suggests that seasonal variation in wound closure behavior may exist. However, until it is known how pathogen activity varies with temperature, the potential importance, if any, of this seasonal change cannot be established.

The very low speeds recorded for keratocytes of T.bernacchii are paradoxical in the context of the proposed importance of conservation of wound-healing behavior. Keratocytes of the Antarctic fish moved at rates that were less than one percent those of the other species at their physiological temperatures. We do not interpret these extremely low rates of movement to be an indication of a `failure' to cold-adapt, because Antarctic notothenioids have a pronounced level of cold-adapted metabolic and enzymatic activity in certain tissues(Somero et al., 1968; Kawall et al., 2002) and there is no a priori basis for assuming that cold adaptation is precluded in any particular cell type. Thus, the biological significance, if any, of the extremely slow rates of movement of the keratocytes of T. bernacchiiis not apparent. Likewise, the significance of the increased mobility of keratocytes from 5°C-acclimated T. bernacchii is unclear. Because 5°C is several degrees above the highest temperatures that this species experiences (and likely has experienced for 10-15 million years), the effects of acclimation to 5°C could represent a pathological condition.

The range of temperatures over which keratocyte motility was observed and the thermal optima of keratocyte speed reflected both evolutionary and acclimation temperatures (Fig. 4). The ability of keratocytes of T. bernacchii to survive 10 min of supercooling at -6°C and display apparently normal rates of movement at 5°C after the bout of supercooling supports the observation of DeVries and Cheng (1992)that T. bernacchii can withstand these very low temperatures as long as ice crystals are absent in the internal fluids and medium. The ability of keratocytes of T. bernacchii to sustain motility at a temperature of 20°C, which is approximately 15°C above the upper lethal temperature of this species (Somero and DeVries,1967; Hofmann et al.,2000), indicates that the mechanisms underlying acute thermal death may involve complex physiological functions (for example, failure of synaptic transmission) rather than the survival of individual cells(Somero, 1996; Hochachka and Somero,2002).

In only one instance did thermal acclimation affect the functional thermal range of cultured keratocytes: cells cultured from C. salinusacclimated at 26°C did not survive at 5°C, whereas cells from fish acclimated to the warmer 35°C did. This suggests a potential thermal stress for C. salinus acclimated at 26°C, which was responsible for decreasing the range of cellular thermal tolerance.

Effects of experimental temperature were noted on the persistence and directionality of keratocyte movement, with persistently straight trajectories giving way to random walks and persistent circular trajectories as experimental temperature increased. Thermal acclimation also affected turning behavior, with cells from cold-acclimated specimens generally making turns of greater magnitude. Although G. mirabilis, C. salinus (35°C acclimation) and A. percula keratocytes appear very persistent at 5°C and 10°C in the path length versus displacement analysis,the cos (θ) analysis reveals that they are making large turns and must be moving in an oscillatory, yet persistent, manner. One of the most noteworthy results is that thermal acclimation does not affect how the keratocytes' mean speeds change with experimental temperature but does influence how the experimental temperature affects directional movement, in terms of persistence and turning angle, in G. mirabilis and C. salinus. Speed memory and oscillation are also affected by temperature acclimation.

Collectively, these results indicate that there is more than one temperature-sensitive mechanism governing cell motility. The rate-limiting process(es) responsible for speed is distinct from the mechanism(s) underlying directionality and persistence, and each mechanism is affected differently by temperature. Manipulation of temperature may allow a decoupling of these mechanisms and permit their individual study, allowing a detailed analysis of the effects of temperature on motility within and between species to be made. For example, the polymerization and depolymerization of cytoskeletal tubulin subunits in Antarctic fish occur at a reduced rate compared with those in mammals (Detrich et al., 2000). Similar differences in orthologous biochemical systems related to keratocyte motility may be present in species from different thermal niches. Examining the thermal responses of such processes as actin polymerization, membrane fluidity or binding/release of adhesion proteins may lead to insights into the mechanisms that underlie the thermal sensitivities of keratocyte speed and directional behavior. However, we have previously shown that there are only small, conservative changes in the sequences of α-actins from vertebrates adapted to habitat temperatures ranging over 42°C, andα-actins of Antarctic notothenioids exhibit no apparent adaptation at the level of primary structure (Ream,2002). Therefore, temperature-dependent variation in motility is probably not attributable to intrinsic variation between actin homologs.

One fundamental question that remains for future analysis is the interplay that may exist between speed and directional behavior in wound healing. It is possible that the variation in trajectory in response to temperature acclimation and experimental temperature has a functional significance, much as we conjectured for the conservation of speed per se, because memory and oscillation are present in a significant number of cells only at physiological temperatures.

In this study, we attempted to isolate external variables and focus solely on thermal effects; however, in situ wound healing occurs in the presence of numerous signaling molecules that are known to influence migration in other cell types. Obviously, temperature will affect not only the rate of diffusion of these molecules but also the transduction of chemical signals. Further examination of keratocyte motility during wound healing in situ may provide information on the interplay of chemical signals and/or an electrical gradient (established across the wound) on the temperature-dependent speed and directionality of these cells. Our initial descriptions of thermal effects on keratocyte function thus open up a number of avenues of study in cell biology and evolutionary physiology that can be addressed using this promising experimental system.

We acknowledge support from National Science Foundation grants IBN-0133184 to G.N.S. and OPP-9812707 to Dr Donal Manahan in support of the studies in Antarctica conducted during the McMurdo Antarctic Biology Course. We thank Dr Manahan and the staff of the Crary Laboratory for their efforts on behalf of this research. J.A.T. was supported by a David and Lucile Packard Fellowship in Science and Engineering.