Kamilla Sofie Pedersen, Department of Ecology and Genetics, University of Aarhus, Ny Munkegade, Building 540, DK-8000 Aarhus C, Denmark. Tel.: +45 89423230; fax +45 89422722; e-mail: firstname.lastname@example.org
Induction of heat shock proteins (Hsp) is a well-known mechanism through which cells cope with stressful conditions. Hsp are induced by a variety of extrinsic stressors. However, recently intrinsic stressors (aging and inbreeding) have been shown to affect expression of Hsp. Increased homozygosity due to inbreeding may disrupt cellular homeostasis by causing increased expression of recessive deleterious mutations and breakdown of epistatic interactions. We investigated the effect of inbreeding and the rate of inbreeding on the expression of Hsp70, larval heat resistance and fecundity. In Drosophila melanogaster we found that inbred lines (F ≈ 0.67) had significantly up-regulated expression of Hsp70, and reduced heat resistance and fecundity as compared with outbred control lines. A significant negative correlation was observed between Hsp70 expression and resistance to an extreme heat stress in inbred lines. We interpreted this as an increased requirement for Hsp70 in the lines suffering most from inbreeding depression. Inbreeding depression for fecundity was reduced with a slower rate of inbreeding compared with a fast rate of inbreeding. Thus, the effectiveness of purging seems to be improved with a slower rate of inbreeding.
Molecular chaperones are major cell constituents in all organisms under benign conditions and they are essential to ensure proper folding and intracellular localization of newly synthesized polypeptides (Feder & Hofmann, 1999). The demand for molecular chaperones is larger under stressful conditions, since then the rate of damage to cell proteins or problems with proper folding increase markedly (Feder & Hofmann, 1999; Sørensen et al., 2003). Heat shock proteins (Hsp), a subset of the molecular chaperones, are part of this cellular defense. The intracellular signal for induction of Hsp is a sudden increase of abnormal polypeptides in the cytosol or nucleus (Ananthan et al., 1986; Sherman & Goldberg, 2001). Hsp function by promoting proper folding or refolding and by preventing potentially damaging interactions or protein aggregations, and aid in disassembling already formed protein aggregates (Gething & Sambrook, 1992; Feder & Hofmann, 1999).
The genes encoding Hsp are highly conserved between species and have similar functions in vertebrates and invertebrates (Gething & Sambrook, 1992). One of the most conserved Hsp is the inducible Hsp70, showing an amino acid similarity between Drosophila melanogaster and Homo sapiens of approximately 75% (Lindquist, 1986). Investigations on Hsp have mainly focused on Hsp70 family members, which in general have been studied in relation to extreme environmental stress, primarily heat stress. Hsp70 is the major stress inducible Hsp in Drosophila and is induced by a variety of extrinsic stressful conditions, e.g. heat shock, parasitism, alcohol, heavy metals, crowding and inflammation (for review see Sørensen et al., 2003). A common feature of this variety of stressors is that they result in increased levels of abnormal proteins in the cell (Feder & Hofmann, 1999). Recent evidence indicates that Hsp70 also has a function in coping with genetic stress, which, compared with environmental stress (extrinsic), is an internal (intrinsic) kind of stress due to e.g. aging, deleterious mutations, hybridization or inbreeding (Wheeler et al., 1999; Kristensen et al., 2002; Trotter et al., 2002). Both aging and inbreeding have been shown to cause an up-regulation of Hsp70 expression (Wheeler et al., 1999; Kristensen et al., 2002). Inbreeding increases homozygosity and thereby the expression of recessive deleterious alleles (Charlesworth & Charlesworth, 1999). This may result in polypeptides with abnormal conformations that prevent normal folding or the association of a polypeptide with other subunits or stabilizing cofactors (Sherman & Goldberg, 2001; Trotter et al., 2002). Furthermore, inbreeding may change epistatic interactions, which can disturb cellular homeostasis and thus affect protein regulation.
Kristensen et al. (2002) showed that inbred larvae of D. melanogaster and D. buzzatii had up-regulated expression levels of Hsp70 when compared with outbred control lines, under benign environmental conditions. It was suggested that this low but constant up-regulation of Hsp70 expression in inbred lines as compared with outbred controls was caused by an increased demand for chaperones in order to maintain cellular homeostasis. However, in the study by Kristensen et al. (2002) the effect of rate of inbreeding on inbreeding depression and the association between fitness and Hsp70 expression was not investigated. Studies on the potential association between Hsp70 expression level and fitness may reveal further information on mechanisms responsible for inbreeding depression, and emphasize a new role for the inducible Hsp70.
The aim of this study was to investigate the effect of inbreeding and the rate of inbreeding on the expression of Hsp70, as well as on two fitness components, larval heat resistance and fecundity. The experiment was designed to study these effects in replicate lines with equivalent expected inbreeding coefficients. Hsp70 quantification at a nonstressful temperature was chosen in an attempt to separate up-regulation as a consequence of heat stress (extrinsic stress) and a possible up-regulation caused by inbreeding (intrinsic stress).
Materials and methods
Population origin and inbreeding method
Populations of D. melanogaster from different geographical regions were mixed in order to ensure genetically diverse starting material. A mass population was set up in August 2002 by mixing 600–700 flies from each of four large pre-existing stocks from our laboratory (collected in Denmark, Australia and the Netherlands). The inbreeding procedure was started eight generations after foundation of the mass population. Lines with equivalent expected levels of inbreeding (F ≈ 0.67) were obtained by two different rates of inbreeding, through five generations of full-sib mating or by keeping the population sizes at two pairs for nine generations. For the full-sib procedure, offspring from each line from each consecutive generation were collected as virgins, and one full-sib pair was randomly chosen as parents for the next generation. For lines inbred by a slower rate of inbreeding, offspring from each line were collected as virgins and two males and two females were randomly selected as parents for the next generation. Some lines were lost through the procedure of inbreeding; hence, to ensure enough lines reached the expected inbreeding level excess lines were set up. Twenty independent lines were started to make up the full-sib mated lines. Within each line four vials with full-sib pairs were set up. One of them was used to establish the next generation. Fifteen independent lines were started to make the lines inbred by a slower rate of inbreeding (n = 4). Within each line two vials with two pairs were set up. One of them was used to establish the next generation. Ten outbred control lines founded by approximately 500 individuals were started when the inbreeding procedures were initiated. Flies were reared and maintained at standard laboratory conditions (25 ± 0.2 °C, 50% RH, 12/12 h light/dark cycle). Assuming the inbreeding level of the base population to be zero, the expected inbreeding levels (F) were calculated from the expression Ft = (1 + 2Ft−1 + Ft−2)/4 (Falconer & Mackay, 1996) for the fast-inbred lines and from the expression Ft = Ft−1 + (1 − 2Ft−1 + Ft−2)/2Ne (Crow & Kimura, 1970) for the lines inbred at a slower rate. For each inbreeding rate, 10 of the surviving lines were randomly selected and the population sizes flushed to minimally 500 individuals. Lines inbred at a fast rate reached the expected inbreeding level four generations before the lines inbred at a slower rate. Theoretically, the inbreeding coefficient increases minimally over generations when population sizes are high, so the inbreeding coefficients of the fast-inbred lines are assumed to be unchanged after the desired level of inbreeding was reached.
Adult flies (1 day old ± 24 h) were transferred to vials containing standard laboratory medium. On the third day (flies 3 days ± 24 h old), flies were allowed to oviposit for 12 h on standard laboratory medium. Third instar larvae were transferred to 2 mL Eppendorf tubes and exposed to 29 °C for 1 h and allowed recovery at 25 °C for 1 h, before being frozen at −80 °C. Exposure to 29 °C was chosen because this temperature revealed the largest difference between inbred and outbred control lines in the study by Kristensen et al. (2002). Hsp70 was quantified in five replicates of 10 larvae from each line by the ELISA technique using the monoclonal antibody 7.FB, which is specific for inducible Hsp70 in Drosophila (Welte et al., 1993), according to the protocol described by Sørensen et al. (1999). Two ELISA plates constituted each replicate; lines 1–5 from each breeding regime (outbred control and two rates of inbreeding) formed one ELISA plate and lines 6–10 formed another ELISA plate (one replicate = 2 ELISA plates). Variation between replicates (5 × 2 ELISA plates) was corrected by adjusting all replicates to replicate one.
From each line per breeding regime, young adult flies (12 h old ± 12 h) were transferred to bottles containing standard laboratory medium. On the third day (flies 3 days ± 12 h), flies were allowed to oviposit for 12 h. Third instar larvae were collected and transferred to vials with standard laboratory medium. Five replicates of 25 larvae per line were exposed to 38.5 °C for 1 h in a water bath. Afterwards the vials containing the larvae were transferred to 25 °C to complete their development. Emerging flies were counted and removed every day until eclosion stopped. All emerging flies (dead or alive) were included.
Young adult flies (12 h old ± 12 h) from each line per breeding regime were transferred to bottles containing standard laboratory medium. On the fourth day, flies (4 days ± 12 h) were separated by sex under CO2 anaesthesia and transferred to new vials with standard laboratory medium. One female and two males, randomly chosen, were transferred to each of 30 vials per line. Two days later flies (6 days ± 12 h) were transferred to new vials containing a plastic spoon with medium. Oviposition was allowed for 22 h. Subsequently the eggs were counted. All replicates were included in the analysis. Including replicates from which no offspring emerged means that fecundity in this study is a combination of the effect of inbreeding on sterility and fecundity.
To test for effects of breeding regime (inbreeding rate, fixed factor) and line (nested within breeding method, random effect) on Hsp70 expression, heat resistance and fecundity (dependent variables), full factorial nested anovas were used. Fecundity and Hsp70 expression data showed a non-normal distribution. However, anovas are quite robust to deviations from normality as long as sample sizes are large and equal (Zar, 1999). Homogeneity of variances between breeding methods was tested using Bartlett's test when the data was normally distributed and Levene's test of homogeneity of variance when the data was not normally distributed. For traits showing heterogeneity of variances, pairwise comparisons between breeding regimes were made using a Welch anova, allowing unequal variances. If homogeneity of variances was fulfilled pairwise comparisons were made using a nested anova. Sequential Bonferroni corrections were performed to correct for multiple testing (Rice, 1989). Correlation analyses of all traits were made by nonparametric Spearman rank tests, which do not require a normal distribution of data.
Breeding regime and line significantly affected Hsp70 expression (Fig. 1; breeding regime: F(2,27) = 4.18, P < 0.05; line: F(27,120) = 1.82, P < 0.05). Inbred lines showed up-regulated expression levels of Hsp70 as compared with outbred control lines. Levene's test revealed homogeneity of variances between breeding regimes (F(2,27) = 1.27, n.s.). Pairwise comparisons resulted in a significant effect of inbreeding, but expression levels of Hsp70 did not differ between the two inbreeding regimes (control–slower: F(1,18) = 7.02, P < 0.05; control–fast: F(1,18) = 6.82, P < 0.05; slower–fast: F(1,18) = 0.09, n.s.).
Breeding regime and line significantly affected heat resistance (Fig. 2; breeding regime: F(2,27) = 16.24, P < 0.001; line: F(27,120) = 3.47, P < 0.001). Inbred lines showed reduced heat resistance as compared with outbred control lines. Bartlett's test revealed heterogeneity of variances between breeding regimes (F(2,27) = 4.22, P < 0.05). Because of the heterogeneity of variances, pairwise comparisons between breeding regimes were made using a Welch anova allowing unequal variances (control–slower: F(1,18) = 48.79, P < 0.001, control–fast: F(1,18) = 83.32, P < 0.001, slower–fast: F(1,18) = 0.79, n.s.). No significant difference in heat resistance was observed between the two inbreeding regimes.
Breeding regime and line significantly affected fecundity (Fig. 3; breeding regime: F(2,27) = 21.29, P < 0.001; line: F(27,870) = 6.73, P < 0.001). Fecundity was significantly reduced in inbred lines as compared with outbred control lines. The data on fecundity did not fulfil the assumption of normality due to a number of flies laying no eggs. Levene's test, which does not assume normality, revealed heterogeneity of variances between breeding methods (F(2,27) = 5.16, P < 0.01). Pairwise comparisons were made using a Welch anova allowing unequal variances (control–slower: F(1,18) = 87.09, P < 0.001, control–fast: F(1,18) = 263.76, P < 0.001, slower–fast: F(1,18) = 33.77, P < 0.001). Fecundity showed a significant effect of the rate of inbreeding, with fecundity being reduced more in the lines inbred by a fast rate of inbreeding as compared with lines inbred by a slower rate of inbreeding.
Correlation between traits
Correlation analyses between all three traits (Hsp70 expression, heat resistance and fecundity) were made for each breeding regime, revealing no significant correlations between any of the traits (results not shown). However, larval heat resistance and larval Hsp70 expression showed a trend towards a negative correlation in both inbreeding regimes (slower: ρ = −0.56, P = 0.09; fast: ρ = −0.50, P = 0.14, Fig. 4). When the two inbreeding regimes were pooled, the correlation between heat resistance and Hsp70 expression was significantly negative (ρ = −0.52, P < 0.05). That is, inbred lines with high levels of Hsp70 expression have low resistance to heat stress.
Many studies have investigated inbreeding depression and evolution in small populations. However, direct experimental evidence for biochemical mechanisms involved in inbreeding depression is limited. We used D. melanogaster to investigate the complex effects of inbreeding and rate of inbreeding on the expression of Hsp70 and the efficiency of purging. Using a model species such as D. melanogaster makes it possible to test specific hypotheses in a precise manner using a highly replicated design. Drosophila has proved to be a reliable model species, because it has all the realistic genetic variables and similar genetic behaviour to most naturally outbreeding species (Frankham, 2000). Furthermore, Drosophila has at the cellular level many similarities with mammals (Lindquist, 1986). This makes it reasonable to assume that results obtained from studies on Drosophila will be qualitatively similar to studies on other outbreeding species.
Here we present results demonstrating that inbreeding affects expression of the molecular chaperone Hsp70. Inbred larvae showed on average higher expression levels of Hsp70 than outbred control lines, but we observed no significant difference in Hsp70 expression between the two regimes differing in rates of inbreeding. These results establish, in a highly replicated design, the up-regulation of Hsp70 as a consequence of inbreeding, first reported by Kristensen et al. (2002). Hsp70 is generally thought to be a Hsp that is exclusively induced by stressful environmental conditions (Gething & Sambrook, 1992; Feder & Hofmann, 1999). However, we find that inbreeding at benign temperatures also leads to up-regulation of Hsp70. Consequently, our results indicate that Hsp70, besides being a part of the emergency response to extreme environmental conditions, is induced by the deleterious effects of inbreeding. An increased level of abnormal proteins in inbred lines may trigger this up-regulation of Hsp70, or increased Hsp70 expression may reflect a more general cellular attempt to restore protein homeostasis. An alternative explanation could be that a change in the epistatic interactions because of inbreeding causes a break down of the homeostatic regulation of Hsp70 expression. But given the high level of replication in this study we do not find this last explanation feasible.
Both fitness traits showed a significant effect of inbreeding: reduced heat resistance and decreased fecundity. Furthermore, the slower rate of inbreeding resulted in less inbreeding depression for fecundity than the fast rate of inbreeding. This indicates that the effectiveness of purging is improved under a lower rate of inbreeding and that selection is capable of eliminating some of the deleterious alleles, as predicted from theory (Charlesworth & Charlesworth, 1999). Empirical studies support an increased efficiency of purging with lower rates of inbreeding, but the effect of purging has been shown to be minimal with fast inbreeding (Frankham et al., 2001; Day et al., 2003; England et al., 2003; Reed et al., 2003). We showed that rate of inbreeding had a significant effect on reproductive success, but no significant effect on heat resistance. The efficiency of selection in opposing inbreeding depression depends on both the rate of inbreeding and the selection coefficients operating at the loci under consideration (Hedrick, 1994; Wang et al., 1999). An explanation of the different efficiency of purging on different traits in our experiment could be that selection coefficients at loci controlling fecundity were relatively high, whereas little selection operated on loci controlling heat resistance. This different effect of rate of inbreeding on different fitness traits is in accordance with results obtained by Ehiobu et al. (1989). Despite an improved purging of deleterious alleles for reproductive success, the lines inbred at a slower rate still had lowered fitness when compared with the outbred control lines. Lethals and deleterious alleles with large effect will be purged relatively fast, but for mildly deleterious alleles selection will have minor impact compared with random drift; thus mildly deleterious alleles will often drift to fixation (Hedrick, 1994; Fu et al., 1998; Wang et al., 1999). Mildly deleterious alleles will certainly lower population fitness and affect the long-term persistence of populations undergoing both fast and slower inbreeding.
The Hsp70 expression, heat resistance and fecundity differed significantly between lines within breeding regimes. The impact of inbreeding depends on the genetic load of deleterious alleles. As most deleterious alleles are rare, sampling will cause the frequencies of such alleles to vary between lines. In addition, the effect of genetic drift increases with inbreeding, thereby further increasing the between line variation (Falconer & Mackay, 1996). This is consistent with the finding of lineage effects in other experimental studies (Reed et al., 2002; Kristensen et al., 2003).
In inbred lines (pooling the two inbreeding regimes) we observed a significant negative correlation between Hsp70 expression and resistance to extreme heat stress. This negative correlation indicates that lines that are more sensitive to temperature stress have higher up-regulation of Hsp70. One interpretation of this result is that lines with high expression levels of Hsp70 at nonstressful temperatures are less capable of surviving heat stress because they are more affected by inbreeding depression.
The inbreeding levels in the two inbreeding regimes were not estimated using molecular markers, but by pedigree analysis (see Materials and methods). Heterozygotes may on average survive the process of inbreeding better than homozygotes, thereby favouring offspring with lower than expected inbreeding levels, resulting in a lower than expected level of inbreeding (Rumball et al., 1994). Furthermore, the effective population size in the lines inbred at a slower rate may be lower than the census size, resulting in a faster rate of inbreeding and a higher level of inbreeding for these lines (Frankham et al., 1993). A highly replicated design, which we present here, will reduce these problems, and it is unlikely that such deviations in inbreeding level and rate of inbreeding are able to explain our results. Our results were obtained in a design using extreme rates of inbreeding, for both inbreeding regimes, as compared with what is normally observed in nature and animal breeding (Charlesworth & Charlesworth, 1987; Keller & Waller, 2002). In nature and in animal breeding, where rates of inbreeding generally are much lower than in the ‘slower inbred’ regime used here, purging of recessive deleterious alleles is expected to be more efficient than indicated in this study. Despite a relatively small difference between the two rates of inbreeding, we still demonstrated a significant difference in fecundity between the two inbreeding regimes. It is reasonable to assume that a larger difference between the two rates of inbreeding would have led to larger differences in the investigated parameters than was observed. Because of this our results can be interpreted as being conservative.
In conclusion, our results show that inbred D. melanogaster larvae up-regulate expression of Hsp70 as compared with outbred lines. The up-regulation of Hsp70 expression in inbred lines even at benign temperatures can be interpreted as a cellular defence against abnormal proteins or a cellular attempt to restore protein homeostasis. The reduced inbreeding depression for reproductive success in the lines inbred at a slower rate as compared with the fast-inbred lines indicates that a lower rate of inbreeding is capable of reducing inbreeding depression, although not capable of eliminating it.
We thank the Danish National Science Research Council for financial support by a frame and centre grant to VL. We are grateful to Richard Frankham, Kuke Bijlsma, Stuart Barker and Louise Toft Jensen for helpful comments on the manuscript, to Doth Andersen for excellent technical assistance and to S. Lindquist for providing the monoclonal antibody 7.FB.