For: Bijlsma, R. and V. Loeschcke, eds.Stress, Adaptation, and Evolution. Birkh...user Verlag, Basel.

DISCLAIMER: This is a preliminary version of a publication and should not be regarded as the final form. Changes may have been or will be made during copy-editing and corrections of proof. If accuracy is critical, please contact the author or consult the final published version.


ECOLOGICAL AND EVOLUTIONARY PHYSIOLOGY OF HEAT-SHOCK PROTEINS AND THE STRESS RESPONSE IN DROSOPHILA: COMPLEMENTARY INSIGHTS FROM GENETIC ENGINEERING AND NATURAL VARIATION.

Martin E. Feder*+ and Robert A. Krebs*

Department of Organismal Biology & Anatomy,* The Commitee on Evolutionary Biology+, and The College,+ The University of Chicago, 1027 East 57th Street, Chicago IL USA 60637

Running head: Hsps and thermotolerance in Drosophila

Correspondence to: Dr. Martin E. Feder

Department of Organismal Biology & Anatomy

The University of Chicago

1027 East 57th Street

Chicago, IL 60637 USA

E-Mail: m-feder@uchicago.edu

Telephone: 312 702-8096

Fax: 312 702-0037

Evolutionary adaptational biology and genetic engineering as complementary approaches

The major issues concerning stress, adaptation, and evolution include how organisms vary in their tolerance of stress; the biochemical, physiological, morphological and behavioral mechanisms that underlie stress tolerance; and the evolutionary processes that create and sustain variation in stress tolerance. Investigations of these questions can proceed from at least two alternative research strategies (Fig. 1).



Fig. 1. The evolutionary adaptationist program and the molecular genetic program: two approaches for understanding stress, adaptation, and evolution. The former approach often proceeds by demonstrating that variation in performance and/or fitness is consistent with an hypothesized effect of a trait, but often cannot unambiguously attribute such effects to the gene or trait under study. By manipulating a gene of interest directly, the latter approach can often unambiguously demonstrate the consequences of that gene for performance and/or fitness, but may create genetic variation that will seldom if ever occur in nature. When combined, these two approaches may complement one another.

In the evolutionary adaptationist program, investigators either locate existing populations that differ in stress tolerance or create study populations through experimental selection on pre-existing variation, choose an underlying trait for analysis and hypothesize its effect on stress tolerance, and perform experiments or observations to test for the predicted covariation between the trait and stress tolerance. In the molecular genetic approach, by contrast, investigators choose a gene for analysis, create variation in that gene de novo, and observe any gain or loss of function that may result. Thus, the evolutionary approach focuses on genetic variants that may arise through natural or semi-natural genetic and evolutionary processes, while the molecular genetic approach may be free to create any desired variant. These alternative research strategies each have strengths and weaknesses. Our thesis, which we will illustrate with our work on the heat-shock proteins of Drosophila, is that these strategies are complementary; i.e., if deployed in combination, the resulting insights may exceed those yielded by either strategy alone. Neither strategy may be sufficient; both may be essential.

The complementarity of these research strategies arises from the source and magnitude of the variation they analyze: (1) The evolutionary approach can elucidate what variants have arisen and might arise through the natural processes that create and constrain genetic variation, whereas the genetic engineering approach can yield variants that might never occur naturally. Accordingly, the evolutionary approach can help define the "evolutionary feasibility" of engineered variants. (2) By creating greater variation than may be seen in nature, genetic engineering can facilitate experimental and comparative studies where natural variation is too limited for current analytical techniques. (3) The critical potential contribution of the genetic engineering approach is that it permits a single gene to be manipulated in isolation against a constant genetic background (Feder, 1996) . Organisms comprise multiple traits, and manipulating a single trait while all else is held constant is a near-impossible undertaking by conventional means. Even the most extraordinarily detailed and comprehensive research programs of natural genetic variants have conceded that they cannot exclude that their findings are due to linkage between a gene under study and some "mystery gene" (Powers et al., 1993) . Moreover, whereas in natural variation the suite of genetic changes responsible for a given phenotypic variant may be impossible to characterize and may differ in multiple evolutionary events (i.e., parallelism and convergence), the molecular genetic approach enables an investigator to specify which gene(s) will undergo manipulation.

Both the evolutionary adaptationist approach and the genetic engineering approach must surmount the same difficulty: that of providing an ecological context for the model under study. Ideally, practitioners of both approaches should examine organisms undergoing the same intensities, frequencies, and kinetics of stress that these organisms experience in the wild. Both specifying experimental conditions that correspond to nature and inferring the implications of laboratory study for fitness in nature can be problematic when an ecological context is absent. Evolutionary adaptationists often study organisms for which a rich background of natural history already exists, which can provide an adequate ecological context. Genetic engineers, by contrast, often choose model systems for their experimental tractability and convenience of laboratory culture. In many cases, these favored models have not been studied in the wild; basic ecological information can be lacking. Whatever the cause of inadequate ecological data, its provision is essential for meaningful experimentation and interpretation. In our case, the availability of an especially powerful system for controlled transgenic manipulation of Drosophila (Welte et al., 1993) led us to begin an evolutionary adaptationist analysis of thermal stress in the species, and then to alternate recursively between evolutionary and molecular genetic studies. Through this work, we hope to illustrate the contribution that each approach provides to understanding stress resistance in this species.

The stress response and heat-shock proteins

Since their discovery, the heat-shock response and heat-shock proteins have intrigued evolutionary and molecular biologists alike. In response to high temperature and nearly every other stress known, almost every organism studied expresses a characteristic suite of proteins, termed heat-shock proteins (Hsps) or stress proteins (Lindquist and Craig, 1988; Morimoto et al., 1994; Feder et al., 1995; Feder, 1996) . The Hsps are highly conserved and occur in several well-characterized families, most named by molecular weight (e.g., the Hsp70 family). These proteins are now known to be only part of a larger machinery that manages unfolded proteins within the cell (Hartl, 1996) . During initial synthesis, intracellular transport, and under the influence of cell stress, proteins may not be in their native conformation. In non-native proteins, residues that would ordinarily be internalized and isolated from one another in the native state can be exposed and interact, which can lead non-native proteins to aggregate. Such aggregations can be harmful if not lethal to the cell. Hsps and their constitutively-expressed cognates function as molecular chaperones; they recognize and bind to non-native proteins, thus impeding aggregation (Hartl, 1996) . Molecular chaperones then either release bound proteins, which can then fold, or target bound proteins for degradation and removal from the cell (Parsell and Lindquist, 1993) . At least one Hsp, Hsp104 of yeast, can remove proteins from aggregations (Parsell et al., 1994) . Because of their stress-inducible expression and their role in coping with stress-damaged proteins, the Hsps have long been considered important for stress tolerance. Unequivocal demonstrations of Hsps' necessity for stress tolerance have been few, however, due especially to the difficulty of distinguishing the effects of Hsps from those of the many other physiological and biochemical mechanisms that contribute to thermotolerance. Molecular genetic manipulations now provide these demonstrations in a growing number of eukaryotic systems. With respect to Hsp70 family members, for example, overexpression and introduction of exogenous Hsp70 improves the thermotolerance of various mammalian cells types in cell culture (Li et al., 1991; Liu et al., 1992; Lee et al., 1993; Heads et al., 1994; Mailhos et al., 1994; Li et al., 1995) , protects cells against ultraviolet radiation (Simon et al., 1995) , protects whole mammalian hearts against post-ischemic trauma (Marber et al., 1995; Plumier et al., 1995) , and improves the inducible thermotolerance of Drosophila cells in culture (Solomon et al., 1991) . Introduction of anti-Hsp70 antibodies disrupts transcription (Moreau et al., 1994) and the development of tolerance to ischemia (Nakata et al., 1993) and heat (Riabowol et al., 1988; Lee et al., 1993) .

The present study concerns the major Hsp of Drosophila melanogaster, Hsp70. In Drosophila cells in culture, Hsp70 is virtually absent from the unstressed cell and undergoes massive expression during and after heat shock, at its peak accounting for 1% of total cellular protein (Lindquist, 1980; Velazquez et al., 1980; Velazquez et al., 1983; Velazquez and Lindquist, 1984) . Although Hsp70 of Drosophila has not been characterized as a molecular chaperone, its counterparts in other species can clearly perform as molecular chaperones both in vitro and in the cell (Hartl, 1996) . Typically, 10 nearly identical genes encode Hsp70 in the wild-type genome (Ish-Horowicz et al., 1979) . Other family members in Drosophila include Hsp68 and a variety of constitutely-expressed cognates (Palter et al., 1986) . In transformed Drosophila cells in culture, thermotolerance is highly correlated with Hsp70 levels (Solomon et al., 1991) . These facts led us to investigate whether Hsp70 was a significant evolutionary adaptation to stress.

Natural thermal stress: a contribution of the evolutionary adaptationist approach

Understanding how Hsp70 contributes to thermotolerance in Drosophila requires information on whether Drosophila encounters thermal stress in nature, if such stress is sufficient to induce Hsp70 expression, and whether this induced expression varies genetically within and/or among Drosophila populations. The effects of temperature on maintenance, growth, and reproduction in Drosophila are well known (Fig 2).



Fig. 2. Effect of temperature on traits related to fitness in Drosophila melanogaster. Data are from (Parsons, 1978; David et al., 1983; Ashburner, 1989; Huey et al., 1992; Krebs and Loeschcke, 1994a; Feder et al., 1997a)
At temperatures above approximately 27¡C, various body functions fail in direct proportion to the magnitude and duration of heat shock. Parsons (1978) has cited 30¡C as the highest constant temperature at which D. melanogaster populations can persist in the wild (Parsons, 1978) . Individual Drosophila can tolerate higher temperatures, but only temporarily.

Do Drosophila encounter such temperatures in the wild? Surprisingly, despite the massive study of Drosophila, numerous investigations of the effects of temperature upon it, and its extensive characterization along climatic gradients, few studies report actual body temperatures of Drosophila melanogaster in the field (Feder, 1996) . Such direct measurements are necessary to document thermal stress in adults because their great mobility and small size potentially enable them to escape thermal stress by behavioral thermoregulation (Krebs and Bean, 1991; Toolson, 1997) . By contrast, the preadult developmental stages, eggs, larvae, and sometimes pupae, reside on or within necrotic fruit. For these stages, the thermal regime of the host fruit becomes that of the Drosophila within it. From studies of necrotic fruit and Drosophila outside the laboratory, the following points emerge:

If necrotic fruit (e.g., apples, peaches) is in the sun, it can readily attain temperatures that would be harmful if not lethal to any indwelling Drosophila. On warm summer days, temperatures of fruit containing larvae can be as warm as 44¡C, and frequently exceed 35¡C (Feder et al., 1997a) . Fruit temperatures largely are a function of size, as the greater thermal inertia and evaporative cooling capacity of very large host plants (e.g., columnar cacti) apparently results in more moderate temperatures (Krebs and Loeschcke, 1994b; Toolson, 1997) .

In D. melanogaster, behaviors that might mitigate such stress are either ineffective or non-existent. Adult females can avoid fruit that is stressfully warm at the time of oviposition (Fogleman, 1979; Schnebel and Grossfield, 1986) . The limited thermal inertia of small fruits, however, means that fruit temperatures at the time of oviposition are poor indicators of future thermal conditions. Although the physical-chemical-biological "signature" of past heating could be used as an indicator of a future thermal stress, ovipositing females do not avoid fruit that has previously been heated (Feder et al., 1997b) . Once deposited on an apple or peach, larvae are restricted to that environment. There larval Drosophila have little opportunity to thermoregulate behaviorally, as these fruits tend to be thermally homogeneous (Feder et al., 1997a) .

Larvae and pupae undergoing such natural thermal stress express Hsps in response. For example, Hsp70 levels in wild larvae equal or exceed the highest levels detected in wild-type strains undergoing heat shock in the laboratory (Feder et al., 1997a) .

Previous molecular genetic investigations of the heat-shock response in Drosophila had proceeded without any reference to temperatures that Drosophila naturally experience, prompting the criticism that Hsp expression in this species might have no bearing on fitness in nature and therefore be irrelevant as an adaptation to stress. The characterization of natural thermal stress in Drosophila creates an ecological context for the interpretation of molecular data and enhances the design of evolutionary and genetic engineering experiments to come. Elsewhere in this volume, Loeschcke reviews natural phenotypic and genetic variation among populations of Drosophila (Loeschcke, 1997) . Below we consider variation in Hsp70 expression and thermotolerance within Drosophila populations.

Controlled variation against a constant genetic background: a contribution of the molecular genetic approach

Much of our knowledge of the traits that contribute to stress tolerance has come from comparisons of individuals, populations, species, and/or higher taxa in contrasting environments, in which differences in traits are found to be consistent with differences in stress tolerance. Correlations of stress tolerance and underlying traits have readily led to reductionistic studies, which have elucidated numerous candidate mechanisms of stress tolerance. In addition to Hsps, these tolerance-related traits include fatty acid saturation, quantitative and qualitative changes in numerous enzymes, cell membrane stability, energy storage, alternative metabolic strategies, osmotic mediators of protein stability, and a host of others (Precht et al., 1973; Cossins and Bowler, 1987) . The multiplicity of candidate mechanisms of stress tolerance, however, creates a problem in moving in the opposite direction; i.e., inferring functional and/or evolutionary significance from an underlying trait. So many potential mechanisms of stress tolerance differ among individuals, populations, species, and/or higher taxa that attributing functional and/or adaptive significance to any one of them can be inherently equivocal. In many cases, investigators must content themselves with assuming that the trait they are studying (and not other correlated traits) is actually responsible for an observed functional or evolutionary difference, all else equal (Lewontin, 1978) . The key contribution of the genetic engineering approach is that it can hold other traits genetically constant while manipulating a gene of interest. With proper controls, such manipulation can establish that a gene of interest is necessary, sufficient, neither necessary nor sufficient, or both necessary and sufficient to yield a predicted phenotype. Similarly, some whole-organism traits with complex genetic bases (e.g., body size, reproductive status) are amenable to "phenotypic engineering" in studies of adaptation (Sinervo and Basolo, 1996) .

The execution of the genetic engineering approach with respect to Hsp70 in Drosophila is problematic, however, because the wild-type genome already includes 10 copies of the gene (Ish-Horowicz et al., 1979) . Thus, the effects of adding a single transgene or knocking out an existing copy might well be undetectable against this multi-copy background. Susan Lindquist and colleagues have contributed greatly to the resolution of this problem by developing a technique (Golic and Lindquist, 1989; Welte et al., 1993) wherein flies are transformed with a transgene construct bearing multiple copies of a gene of interest flanked by yeast recombination targets (Fig. 3).



Fig. 3. A scheme for controlled mutagenesis in Drosophila via site-specific recombination (Golic and Lindquist, 1989; Welte et al., 1993) . A transgene construct is created that includes the transgene of interest (in this case, 3 copies of the Drosophila hsp70 gene) and an eye-color marker, flanked by yeast FLP recombination target sequences (RT) and P element sequences (P). This construct is transformed into the germline of D. melanogaster, where it integrates at random into a chromosome (shown in black); transformants are made homozygous (Step 1). Male transformants are mated to females transformed with the FLP recombinase gene under control of the hsp70 promoter. Heat-shocking their offspring induces FLP recombinase, which sometimes catalyzes unequal recombination among sister chromatids (Step 2). Such recombination yields one chromatid with two copies of the sequence flanked by the recombination targets, and a second chromatid from which this sequence has been excised. Subsequent crosses then isolate the recombinants, eliminate the chromosome bearing FLP recombinase, and make the recombinants homozygous. These crosses result in two strains (Step 3), an extra-copy strain homozygous for two copies of the RT-flanked transgene construct (in this case, 12 copies of hsp70), and an excision strain with an identical site of transgene integration into the genome but lacking any copies of the transgene of interest. All strains bear 10 wild-type copies of hsp70. Strains can be identified by eye color.
Expression of the yeast Flp recombinase in transformed flies enables unequal homologous recombination such that two daughter strains eventually result: an extra-copy strain bearing two or more copies of the transgene construct, and an excision strain bearing only the construct's flanking elements, but not the transgene itself. In addition, both the extra-copy strain and the excision strain share the site of transgene integration; thus, the excision strain controls for positional or insertional mutagenesis (i.e., a phenotype due solely to disruption of the host genome rather than to the transgene). In rare cases, a difference in the size of the remaining construct in the extra-copy and excision strains can itself result in a phenotype due to unusual genetic circumstances. Examination of multiple pairs of extra-copy and excision strains can effectively rule out this possibility because the likelihood that two independent chromosomal interruptions would yield the same phenotype in relation to the size of the transgene insert is negligibly small (Welte et al., 1993) .

This entire procedure has now been used to create extra-copy strains of Drosophila bearing 12 hsp70 transgenes in addition to the 10 wild-type copies, and matched excision strains with only the wild-type copies (Welte et al., 1993) . Under diverse conditions and at multiple developmental stages, extra-copy Drosophila accumulate Hsp70 to higher levels than do their excision counterparts, and they generally express Hsp70 more rapidly as well (Figs. 4, 5) (Welte et al., 1993; Feder et al., 1996; Krebs and Feder, 1996) M. Tatar and J. Curtsinger, personal communication). Although



Fig. 4. Effect of age and hsp70 copy number on Hsp70 levels in pupae of two pairs of extra-copy and excision transformants. Hsp70 levels were determined for lysates of whole pupae with a Hsp70-specific ELISA and are expressed as a percentage of a standard lysate of Drosophila S2 cells (Feder et al., 1996; Feder et al., 1997a) . Means are plotted ± one standard error. II and III refer to the chromosomes bearing the transgenes in each pair of strains.


both the genes for other Hsps and the wild-type Hsp70s continue to express upon heat shock in the extra-copy Drosophila, the transgenic mRNA is specifically elevated in the extra-copy strains (Welte et al., 1993) .

Is the difference in hsp70 copy number between extra-copy and excision strains sufficient to affect tolerance of ecologically relevant heat stress? To answer this question, we compared thermotolerance of extra-copy and excision Drosophila with and without exposure to a Hsp70-inducing pretreatment, typically 36¡C. Differences in thermotolerance between pretreated and control (i.e., unpretreated) Drosophila are attributable to the entire suite of heat-inducible responses, whereas differences between extra-copy and excision strains are attributable specifically to hsp70 copy number. As has been previously reported for most ectotherms in which it has been studied, pretreatment dramatically improves the thermotolerance of both extra-copy and excision Drosophila, except in early embryos. The improvement of thermotolerance, however, is greater in extra-copy than in excision Drosophila in several circumstances (Welte et al., 1993; Feder et al., 1996) . In wandering 3rd-instar larvae, Hsp70 levels differ most greatly between extra-copy and excision strains at 1 hr after pretreatment. At this time, the thermotolerance of excision larvae has increased to approximately 150% of control levels, whereas the thermotolerance of extra-copy larvae has increased to approximately 350%. Pretreatment of 6-hr embryos also results in much greater hatching success in extra-copy strains than in excision strains.

Although candidate mechanisms for this Hsp70-enhanced thermotolerance are numerous, an appropriate target of investigation is the unfolding of proteins, as is consistent with the role of Hsps as molecular chaperones. We chose alcohol dehydrogenase (ADH) as a model protein for several reasons: ADH denatures at temperatures experienced by Drosophila in the field, ADH alleles vary in thermal stability, and this variation is correlated with climatic gradients (Sampsell and Sims, 1982; Sampsell and Barnette, 1985; Chambers, 1988) . Moreover, ADH has been the subject of numerous evolutionary investigations (see above references and elsewhere in this volume). To examine the effects of Hsp70 level on thermal inactivation and recovery of ADH, we exposed second-instar larvae to 40¡C for 15 minutes, which reduced ADH activity by 40% (Fig. 5).



Fig. 5. Effect of hsp70 copy number on reactivation of alcohol dehydrogenase in whole 2nd-instar larvae (Chromosome III transformants). Top: Groups of larvae were exposed to 40¡C for 15 min after either being exposed to 36¡C for 1 hour (pretreated) or being kept at 25¡C (unpretreated). Middle: Hsp70 levels in pretreated larvae of the extra-copy and excision strains. Data are plotted as in Fig. 4. Unpretreated larvae expressed no Hsp70 before or during heat shock. Bottom: Alcohol dehydrogenase (ADH) activity of whole larvae. ADH activity was determined from crude homogenates according to (McKechnie and Geer, 1984) , standardized by µg protein in the homogenate, and expressed as a percentage of the ADH activity determined before heat shock. Means are plotted ± one standard error.
If larvae of the extra-copy strain were pretreated at 36¡C before the 40¡C exposure, Hsp70 was present at high levels during the 40¡C heat shock. In these larvae, ADH activity recovered to near-initial levels during the 2 hours after heat shock. No such recovery occurred in extra-copy larvae that were not pretreated before heat shock, nor in excision larvae with or without pretreatment. The greater levels of Hsp70 in pretreated larvae of the extra-copy strain could have yielded this pattern by reactivating pre-existing ADH unfolded by the heat shock and/or by protecting de novo synthesis of ADH after the heat shock. Nonetheless, this is apparently the first example of Hsp-mediated restoration of thermally inactivated protein function in an intact multicellular eukaryote.

In summary, these data show that alteration in hsp70 copy number is sufficient to enhance inducible tolerance of temperatures that might well be encountered in the field by Drosophila larvae and pupae. Furthermore, the manner in which the experimental strains were created establishes the participation of Hsp70, rather than some linked trait, in the production of thermotolerance. Whether Hsp70 exerts these effects alone or combination with other chaperones (e.g., the Drosophila DnaJ homologue (Hartl, 1996) ), and whether Hsp70 itself promotes thermotolerance, protects some other thermotolerance factor against heat damage, signals the induction of other thermotolerance factors, and so on, remain for future investigation.

Molecular and evolutionary approaches intersect: Hsp70 as a trade-off

Diverse molecular investigations suggest that Hsp70 may sometimes harm cells in addition to protecting cells from stress and/or aiding cells' recovery from stress. Hsp70 is undetectable in Drosophila cells and tissues in the absence of stress (Velazquez et al., 1983) . A large array of transcriptional, translational, and post-translational regulatory mechanisms enforce both this absence of Hsp70 and its rapid degradation after recovery from stress is complete (Lindquist, 1993) . If these controls are overridden by experimentally expressing Hsp70 in the absence of stress, the growth of cells in culture slows until the Hsp70 is sequestered in intracellular granules (Feder et al., 1992) . Coleman et al. (1995) have suggested that these phenomena occur because massive Hsp expression depletes the cell of resources that are needed for other cellular processes (Coleman et al., 1995) . Alternatively, ectopic Hsp70 expression may be toxic to cells. For example, it may bind inappropriately to its cellular targets and/or fail to release them at appropriate times. Indeed, Dorner et al. (1988, 1992) have reported that overexpression of the mammalian Hsp70 family member Grp78 inhibits the secretion of certain proteins from the cell because it retains these proteins within the Grp78-dependent secretory pathway; proteins not dependent on Grp78 for secretion are unaffected (Dorner et al., 1988; Dorner et al., 1992) . These putative mechanisms are not mutually exclusive.

Evolutionary biology can supply a theoretical framework for interpreting such findings: that of a trade-off, in which a trait that increases fitness is linked to a trait that decreases fitness (Stearns and Kaiser, 1996) . Attempts to understand trade-offs, however, have often been frustrated by lack of a detailed understanding of the traits in question and how they actually interact to yield the trade-off (Stearns and Kaiser, 1996) . Hsp70, for which the positive and negative effects may share a common and increasingly well-understood mechanism (binding to unfolded proteins), may be a useful model for the study of trade-offs. This possibility manifested itself in our work with Drosophila when excision strains outperformed extra-copy strains in several tests of thermal tolerance (Krebs and Feder, 1996) . For example, in studies of 1st-instar larvae begun several hours after hatching, pretreated extra-copy larvae exhibited significantly lower survival to adulthood than excision larvae. This assay tests not only acute survival of a heat shock, but also the ability to grow, metamorphose, and eclose afterwards. Pretreatment and/or heat shock initially disrupted growth more severely in the extra-copy strain than in the excision strain, although extra-copy larvae made up this difference within 72 hours. Comparisons of thermotolerance in an independently derived pair of extra-copy and excision strains and in early 3rd-instar larvae of both pairs of strains revealed essentially the same pattern. Interestingly, even in the control treatment, which involved neither pretreatment nor heat shock, extra-copy larvae showed higher mortality than excision larvae.

Thermal injury of feeding and protection against it may have special ecological importance because intense larval feeding provides the nutrients and energy for metamorphosis and initial stages of reproduction. Our ongoing analyses of the pathogenesis of heat damage single out the gut as a target for these effects. One day after a 38.5¡C heat shock without pretreatment, cells die massively throughout the foregut, gastric caeca, midgut, and hindgut. Hsp70-specific immunofluorescence remains elevated during this time, unlike in other tissues. Whether these patterns differ between extra-copy and excision strains is not yet known. These strains differ in recovery of feeding after heat shock, however. A 38.5¡C heat shock dramatically depresses feeding for at least 16 hours afterwards (Fig. 6). Pretreatment alone has little effect on feeding at this time, but significantly ameliorates the effect of the 38.5¡C heat shock if administered beforehand (Fig. 6).



Fig. 6. Effect of heat shock, pretreatment, and hsp70 copy number on feeding rates of 3rd-instar larvae (Chromosome II transformants). Treatments: Control, sham transfer at constant 25¡; Pretreatment, 36¡C for 50 minutes and 25¡C for 1 hour; Heat shock, 38.5¡C for 50 minutes. Larvae were fed yeast paste for 16 hours after the end of treatment, and then transferred to a yeast slurry made with 1% FD&C Blue #1 dye. This dye does not transit the gut wall, and thus serves as a marker for ingestion (Edgecomb et al., 1994) . After 75 minutes in the dyed yeast, groups of larvae were homogenized and analyzed for dye-specific absorbance at 630 nm and total protein content. Dye ingestion per hour was standardized by µg protein in the homogenate. Means + standard error are plotted.
The improvement due to pretreatment is significantly greater in the excision strain than in the extra-copy strain, which expresses higher levels of Hsp70.

These findings prompted an experiment in which extra-copy and excision larvae underwent multiple pretreatments, during which Hsp70 levels were chronically elevated (but to the different levels characteristic of each strain). In the excision larvae, larva-to-adult survival was unrelated to the number of pretreatments (Fig. 7). In the extra-copy larvae, by contrast, larva-to-adult survival was inversely correlated with the number of heat shocks (Krebs and Feder, 1996) . Thus, engineered variation in hsp70 copy number can amplify aspects of a prospective trade-off that may be more difficult to recognize in natural variants.



Fig. 7. Effect of number of Hsp70-inducing treatments on survival to adulthood in extra-copy and excision larvae (Krebs and Feder, 1996) . For larvae given one or more treatments, the first pretreatment was 2-4 h after hatching, the second 24 h later, and the third another 24 h later. Each treatment exposed larvae to 36¡C for 1 hour. Means are plotted ± one standard error. Pretreatment number significantly reduced survival in the extra-copy strain but had no effect on the excision strain (P<0.001 for strain X pretreatment number interaction; analysis of variance on arcsine-square root transformed data).

Natural variation in Hsp70: a final contribution of the evolutionary approach

The preceding section suggests that stabilizing selection may maintain hsp70 copy number and Hsp70 expression at present levels in natural populations; i.e., the disadvantages of any increase in these traits would outweigh the corresponding advantages. Alternatively, the present hsp70 copy number and levels of Hsp70 expression could have become fixed long ago and bear no relationship to ongoing stabilizing selection. Transgenic experimentation cannot resolve this issue, which requires return to the evolutionary approach.

The conditions necessary for ongoing selection on Hsp70 in nature are that individuals vary in expression within populations, that this variation has a genetic basis, and that these genetic differences affect fitness. Natural variation in hsp70 copy number in D. melanogaster has never been examined thoroughly, although studies with laboratory stocks and cell lines suggest a considerable potential for natural variation (Mirault et al., 1979; Lis et al., 1981a; Lis et al., 1981b; Lis et al., 1981c) . To characterize variation in Hsp70 expression, we measured Hsp70 accumulation in 20 isofemale lines derived from a summer collection of D. melanogaster from a population in an Indiana Orchard (Krebs and Feder, 1997) . These lines varied approximately two-fold in Hsp70 expression in response to a standard heat shock (Fig. 8).



Figure 8. Natural inter-individual variation in Hsp70 expression and thermotolerance, as inferred from 20 isofemale lines derived from a single natural population (Krebs and Feder, 1997) . Hsp70 levels were determined and standardized as in Fig. 4. These levels are plotted against first-instar larva-to-adult survival after sequential exposure to 36¡C for 1 h, 25¡C for 1 h, and 41¡C for 1 h (left graph), and survival at constant 25¡C. Significance levels are for individual correlation coefficients.
Although comparable data are not yet available for other populations of Drosophila or most other species, diIorio et al. (1996) have demonstrated substantial genetic variation in the expression of two Hsp70 family members among hemiclones of Poeciliopsis (diIorio et al., 1996) . Next, we examined the potential relationship of this variation in Hsp70 to fitness by comparing the tolerance of the 20 isofemale lines to heat shock with their Hsp70 expression. Hsp70 level was positively correlated with tolerance of high temperatures, as assessed by larva-to-adult survival. McColl et al. (1996) provide additional evidence that Hsps can be subject to ongoing natural selection (McColl et al., 1996) . They selected Drosophila for resistance to high temperature stress, analyzed hsp68 and hsr-omega via denaturing gradient gel electrophoresis, and discovered that the experimental selection regime had brought about changes in both genes. While the nature and the functional consequences of these changes await characterization, clearly the genes for Hsps are not refractory to selection.

Unexpectedly, Hsp70 level and tolerance of high temperatures were each negatively correlated with larva-to-adult survival at 25¡C (Fig. 8). Thus, the isofemale lines derived from a natural population manifest the same inverse relationship between Hsp70 level upon heat shock and mortality in the absence of stress as do the genetically engineered strains. Here the evolutionary and molecular approaches corroborate one another. Both approaches raise the same concern, however: How can variation in a gene expressed only in the presence of stress affect survival in the absence of stress? Tests of several possible explanations are presently underway.

Conclusion

The complementarity of genetic and phenotypic analyses and of laboratory and field approaches within the evolutionary adaptationist approach is by no means a novel concept (Koehn, 1987; Feder and Watt, 1993; Powers et al., 1993; Partridge, 1994) , nor is the exploitation of transgenic variation in the analysis of evolutionary adaptation (Feder and Block, 1991) . The challenge is to incorporate all of these in a single research program. Often, the limiting factor will be the availability of a suitable model system (Feder and Watt, 1993) . In Drosophila, the incidence of natural thermal stress and the amenability to genetic engineering are key features lacking in many other candidate models.

The complementary contributions of the evolutionary adaptationist approach and the molecular genetic approach deserve re-emphasis. The field data establish that Drosophila naturally encounter thermal stress, to which they respond by expressing Hsps, and provide assurance that the transgenic variation has a counterpart in nature. Without these data, the analyses of the transgenic strains risk becoming an exercise in mechanistic pharmacology with no clear relevance to the function and fitness of organisms in nature. Experimentation with the transgenic strains unequivocally attributes variation in high temperature tolerance to Hsp70. Without such a comparison, hundreds of other traits known to affect thermotolerance must be accounted for before making a similar attribution. Genetic engineering can be an incredibly powerful addition to the tool kit of evolutionary adaptationists. However, realizing the full potential of transgenic manipulation as an analytical tool in adaptational biology will require its deployment in concert with the classical tools of evolutionary and functional biology.

Acknowledgements

We thank Susan Lindquist for continuing encouragement and insight. Evelyn Tin (Knox College) gathered data on alcohol dehydrogenase activity while supported by a summer research fellowship from the Howard Hughes Medical Institute. Figure 3 is modified from original artwork by Michael Welte. The Warner-Jenkinson Corp., St. Louis, MO, donated FD&C Blue #1 dye. Research was supported by grants from the National Science Foundation (IBN94-08216 and BIR94-19545) and the Louis Block Fund of the University of Chicago.

References

Ashburner, M. (1989) Drosophila: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

Chambers, G. K. (1988) The Drosophila alcohol dehydrogenase gene-enzyme system. Advances in Genetics 25, 39-108.

Coleman, J. S., Heckathorn, S. A. and Hallberg, R. L. (1995) Heat-shock proteins and thermotolerance: linking ecological and molecular perspectives. Trends in Ecology and Evolution 10, 305-306.

Cossins, A. R. and Bowler, K. (1987) Temperature Biology of Animals. Chapman and Hall, London.

David, J. R., Allemand, R., Van Herrewege, J. and Cohet, Y. (1983). Ecophysiology: abiotic factors. In The Genetics and Biology of Drosophila 3d (Edited by M. Ashburner, H.L. Carson & J.N. Thompson), pp. 105-170. Academic Press, Inc., London.

diIorio, P. J., Holsinger, K., Schultz, R. J. and Hightower, L. E. (1996) Quantitative evidence that both Hsc70 and Hsp70 contribute to thermal adaptation in hybrids of the livebearing fishes Poeciliopsis. Cell Stress & Chaperones 1, 139-147.

Dorner, A. J., Krane, M. G. and Kaufman, R. J. (1988) Reduction of endogenous GRP78 levels improves secretion of a heterologous protein in CHO cells. Mol Cell Biol 8, 4063-4070.

Dorner, A. J., Wasley, L. C. and Kaufman, R. J. (1992) Overexpression of GRP78 mitigates stress induction of glucose regulated proteins and blocks secretion of selective proteins in Chinese hamster ovary cells. EMBO J 11, 1563-1571.

Edgecomb, R. S., Harth, C. E. and Schneiderman, A. M. (1994) Regulation of feeding behavior in adult Drosophila melanogaster varies with feeding regime and nutritional state. Journal of Experimental Biology 197, 215-236.

Feder, J. H., Rossi, J. M., Solomon, J., Solomon, N. and Lindquist, S. (1992) The consequences of expressing hsp70 in Drosophila cells at normal temperatures. Genes & Development 6, 1402-13.

Feder, M. E. (1996). Ecological and evolutionary physiology of stress proteins and the stress response: the Drosophila melanogaster model. In Animals and Temperature. (Edited by I.A. Johnston & A.F. Bennett), pp. 79-102. Cambridge University Press, Cambridge.

Feder, M. E., Blair, N. and Figueras, H. (1997a) Natural thermal stress and heat-shock protein expression in Drosophila larvae and pupae. Functional Ecology, in press.

Feder, M. E., Blair, N. and Figueras, H. (1997b) Oviposition site selection and temperature: unresponsiveness of ovipositing Drosophila to cues of potential thermal stress on offspring. Animal Behaviour, submitted.

Feder, M. E. and Block, B. A. (1991) On the future of physiological ecology. Functional Ecology 5, 136-144.

Feder, M. E., Carta-o, N. V., Milos, L., Krebs, R. A. and Lindquist, S. L. (1996) Effect of engineering Hsp70 copy number on Hsp70 expression and tolerance of ecologically relevant heat shock in larvae and pupae of Drosophila melanogaster. Journal of Experimental Biology 199, 1837-1844.

Feder, M. E., Parsell, D. A. and Lindquist, S. L. (1995). The stress response and stress proteins. In Cell Biology of Trauma (Edited by J.J. Lemasters & C. Oliver), pp. 177-191. CRC Press, Boca Raton, FL.

Feder, M. E. and Watt, W. B. (1993). Functional biology of adaptation. In Genes in Ecology (Edited by R.J. Berry, T.J. Crawford & G.M. Hewitt), pp. 365-391. Blackwell Scientific Publications, Oxford.

Fogleman, J. (1979) Oviposition site preference for temperature in D. melanogaster. Behavioral Genetics 9, 407-412.

Golic, K. G. and Lindquist, S. (1989) The FLP recombinase of yeast catalyzes site-specific recombination in the Drosophila genome. Cell 59, 499-509.

Hartl, F. U. (1996) Molecular chaperones in cellular protein folding. Nature 381, 571-580.

Heads, R. J., Latchman, D. S. and Yellon, D. M. (1994) Stable high level expression of a transfected human HSP70 gene protects a heart-derived muscle cell line against thermal stress. Journal of Molecular & Cellular Cardiology 26, 695-699.

Huey, R. B., Crill, W. D., Kingsolver, J. G. and Weber, K. E. (1992) A method for rapid measurement of heat or cold resistance of small insects. Functional Ecology 6, 489-494.

Ish-Horowicz, D., Pinchin, S. M., Schedl, P., Artavanis, T. S. and Mirault, M. E. (1979) Genetic and molecular analysis of the 87A7 and 87C1 heat-inducible loci of D. melanogaster. Cell 18, 1351-1358.

Koehn, R. K. (1987). The importance of genetics to physiological ecology. In New Directions in Ecological Physiology (Edited by M.E. Feder, A.F. Bennett, W.W. Burggren & R.B. Huey), pp. 170-185. Cambridge University Press, Cambridge.

Krebs, R. A. and Bean, K. L. (1991) The mating behavior of Drosophila mojavensis on organ pipe and agria cactus. Psyche 98:101-109 98, 101-109.

Krebs, R. A. and Feder, M. E. (1996) Thermotolerance, Hsp70 and larval development: effects of transgenic manipulation of a heat-shock protein. American Naturalist, in review.

Krebs, R. A. and Feder, M. E. (1997) Natural variation in the expression of the heat-shock protein Hsp70 in a population of Drosophila melanogaster, and its correlation with tolerance of ecologically relevant thermal stress. Evolution, in press.

Krebs, R. A. and Loeschcke, V. (1994a) Costs and benefits of activation of the heat-shock response in Drosophila melanogaster. Functional Ecology 8, 730-737.

Krebs, R. A. and Loeschcke, V. (1994b). Response to environmental change: genetic variation and fitness in Drosophila buzzatii following temperature stress. In Conservation Genetics (Edited by V. Loeschcke, J. Tomiuk & S.K. Jain), pp. 309-321. Birkh...user, Basel.

Lee, Y. J., Kim, D., Hou, Z. Z., Curetty, L., Borrelli, M. J. and Corry, P. M. (1993) Alteration of heat sensitivity by introduction of hsp70 or anti-hsp70 in CHO cells. Journal of Thermal Biology 18, 229-236.

Lewontin, R. C. (1978) Adaptation. Scientific American 239, 156-169.

Li, G. C., Li, L. G., Liu, Y. K., Mak, J. Y., Chen, L. L. and Lee, W. M. (1991) Thermal response of rat fibroblasts stably transfected with the human 70-kDa heat shock protein-encoding gene. Proc Natl Acad Sci U S A 88, 1681-1685.

Li, L., Shen, G. and Li, G. C. (1995) Effects of expressing human Hsp70 and its deletion derivatives on heat killing and on RNA and protein synthesis. Experimental Cell Research 217, 460-468.

Lindquist, S. (1980) Translational efficiency of heat-induced messages in Drosophila melanogaster cells. J. Mol. Biol. 137, 151-158.

Lindquist, S. (1993). Autoregulation of the heat-shock response. In Translational Regulation of Gene Expression 2 (Edited by J. Ilan), pp. 279-320. Plenum Press, New York.

Lindquist, S. and Craig, E. A. (1988) The heat-shock proteins. Annu Rev Genet 22, 631-677.

Lis, J., Neckameyer, W., Mirault, M. E., Artavanis, T. S., Lall, P., Martin, G. and Schedl, P. (1981a) DNA sequences flanking the starts of the hsp 70 and alpha beta heat shock genes are homologous. Dev Biol 83, 291-300.

Lis, J. T., Ish, H. D. and Pinchin, S. M. (1981b) Genomic organization and transcription of the alpha beta heat shock DNA in Drosophila melanogaster. Nucleic Acids Res 9, 5297-310.

Lis, J. T., Neckameyer, W., Dubensky, R. and Costlow, N. (1981c) Cloning and characterization of nine heat-shock-induced mRNAs of Drosophila melanogaster. Gene 15, 67-80.

Liu, R. Y., Li, X., Li, L. and Li, G. C. (1992) Expression of human hsp70 in rat fibroblasts enhances cell survival and facilitates recovery from translational and transcriptional inhibition following heat shock. Cancer Res 52, 3667-3673.

Loeschcke, V. (1997). Thermal stress and resistance in Drosophila. In Stress, Adaptation, and Evolution (Edited by R. Bijlsma & V. Loeschcke), pp. . Birkhaeuser Verlag, Basel.

Mailhos, C., Howard, M. K. and Latchman, D. S. (1994) Heat shock proteins hsp90 and hsp70 protect neuronal cells from thermal stress but not from programmed cell death. Journal of Neurochemistry 63, 1787-1795.

Marber, M. S., Mestril, R., Chi, S. H., Sayen, M. R., Yellon, D. M. and Dillmann, W. H. (1995) Overexpression of the rat inducible 70-kD heat stress protein in a transgenic mouse increases the resistance of the heart to ischemic injury. Journal of Clinical Investigation 95, 1446-1456.

McColl, G., Hoffmann, A. A. and McKechnie, S. W. (1996) Response of two heat shock genes to selection for knockdown heat resistance in Drosophila melanogaster. Genetics 143, in press.

McKechnie, S. W. and Geer, B. W. (1984) Regulation of alcohol dehydrogenase in Drosophila melanogaster by dietary alcohol and carbohydrate. Insect Biochemistry 14, 231-242.

Mirault, M. E., Goldschmidt, C. M., Artavanis, T. S. and Schedl, P. (1979) Organization of the multiple genes for the 70,000-dalton heat-shock protein in Drosophila melanogaster. Proc Natl Acad Sci U S A 76, 5254-5258.

Moreau, N., Laine, M. C., Billoud, B. and Angelier, N. (1994) Transcription of amphibian lampbrush chromosomes is disturbed by microinjection of HSP70 monoclonal antibodies. Experimental Cell Research 211, 108-114.

Morimoto, R. I., Tissieres, A. and Georgopoulos, C. (eds) (1994) Heat Shock Proteins: Structure, Function and Regulation. Cold Spring Harbor, NY, Cold Spring Harbor Lab. Press.

Nakata, N., Kato, H. and Kogure, K. (1993) Inhibition of ischaemic tolerance in the gerbil hippocampus by quercetin and anti-heat shock protein-70 antibody. Neuroreport 4, 695-698.

Palter, K. B., Watanabe, M., Stinson, L., Mahowald, A. P. and Craig, E. A. (1986) Expression and localization of Drosophila melanogaster hsp70 cognate proteins. Mol Cell Biol 6, 1187-1203.

Parsell, D. A., Kowal, A. S., Singer, M. A. and Lindquist, S. (1994) Protein disaggregation mediated by heat-shock protein Hsp104. Nature 372, 475-478.

Parsell, D. A. and Lindquist, S. (1993) The function of heat-shock proteins in stress tolerance: degradation and reactivation of damaged proteins. Ann. Rev. Genet. 27, 437-496.

Parsons, P. (1978) Boundary conditions for Drosophila resource utilization in temperate regions, especially at low temperatures. American Naturalist 112, 1063-1074.

Partridge, L. (1994). Genetic and nongenetic approaches to questions about sexual selection. In Quantitative Genetic Studies of Behavioral Evolution (Edited by C.R.B. Boake), pp. 126-141. Univ. of Chicago Press, Chicago.

Plumier, J. C., Ross, B. M., Currie, R. W., Angelidis, C. E., Kazlaris, H., Kollias, G. and Pagoulatos, G. N. (1995) Transgenic mice expressing the human heat shock protein 70 have improved post-ischemic myocardial recovery. Journal of Clinical Investigation 95, 1854-1860.

Powers, D. A., Smith, M., Gonzalez-Villasenor, I., DiMichele, L., Crawford, D. L., Bernardi, G. and Lauerman, T. (1993). A multidisciplinary approach to the selectionist/neutralist controversy using the model teleost Fundulus heteroclitus. In Oxford Surveys in Evolutionary Biology (Edited by D. Futuyma & J. Antonovics), pp. 43-107, .

Precht, H., Christophersen, J., Hensel, H. and Larcher, W. (eds) (1973) Temperature and Life. Berlin, Springer-Verlag.

Riabowol, K. T., Mizzen, L. A. and Welch, W. J. (1988) Heat shock is lethal to fibroblasts microinjected with antibodies against hsp70. Science 242, 433-436.

Sampsell, B. M. and Barnette, V. C. (1985) Effects of environmental temperatures on alcohol dehydrogenase activity levels in Drosophila melanogaster. Biochem. Genet. 23, 53--59.

Sampsell, B. M. and Sims, S. (1982) Effect of Adh genotype and heat stress on alcohol tolerance in Drosophila melanogaster. Nature 296, 853-855.

Schnebel, E. M. and Grossfield, J. (1986) Oviposition temperature range in four Drosophila species triads from different ecological backgrounds. American Midland Naturalist 116, 25-35.

Simon, M. M., Reikerstorfer, A., Schwarz, A., Krone, C., Luger, T. A., Jaattela, M. and Schwarz, T. (1995) Heat shock protein 70 overexpression affects the response to ultraviolet light in murine fibroblasts. Evidence for increased cell viability and suppression of cytokine release. Journal of Clinical Investigation 95, 926-933.

Sinervo, B. and Basolo, A. (1996). Testing adaptation using phenotypic manipulations. In Adaptation (Edited by M.R. Rose & G.V. Lauder), pp. in press. Academic Press, New York and London.

Solomon, J. M., Rossi, J. M., Golic, K., McGarry, T. and Lindquist, S. (1991) Changes in Hsp70 alter thermotolerance and heat-shock regulation in Drosophila. The New Biologist 3, 1106-1120.

Stearns, S. C. and Kaiser, M. (1996) Effects on fitness components of P-element inserts in Drosophila melanogaster: analysis of tradeoffs. Evolution 50, 795-806.

Toolson, E. C. (1997). Effects of constant and fluctuating rearing temperatures on development and physiology in the cactiphilic Drosophila mojavensis (Diptera: Drosophilidae). In Stress, Adaptation, and Evolution (Edited by R. Bijlsma & V. Loeschcke), pp. . Birkhaeuser Verlag, Basel.

Velazquez, J. M., DiDomenico, B. J. and Lindquist, S. (1980) Intracellular localization of heat shock proteins in Drosophila. Cell 20, 679-689.

Velazquez, J. M. and Lindquist, S. (1984) Hsp70: nuclear concentration during environmental stress and cytoplasmic storage during recovery. Cell 36, 655-662.

Velazquez, J. M., Sonoda, S., Bugaisky, G. and Lindquist, S. (1983) Is the major Drosophila heat shock protein present in cells that have not been heat shocked? J Cell Biol 96, 286-290.

Welte, M. A., Tetrault, J. M., Dellavalle, R. P. and Lindquist, S. L. (1993) A new method for manipulating transgenes: engineering heat tolerance in a complex, multicellular organism. Current Biology 3, 842-853.