Hybrid breakdown and mitochondrial dysfunction in hybrids of Nasonia parasitoid wasps

Authors


  • C. K. Ellison and O. Niehuis contributed equally to this work.

Christopher K. Ellison, Marine Biology Research Division, Scripps Institution of Oceanography, University of California San Diego, La Jolla, CA 92093-0208, USA. Tel.: +1 858 344 4938; fax: +1 858 534 7313; e-mail: cellison@ucsd.edu

Abstract

Male F2 hybrids of the wasps Nasonia giraulti and Nasonia vitripennis suffer increased mortality during development. Previous studies suggested that the mitochondria may play an important role in this pattern of hybrid breakdown. The mitochondrial genome encodes 13 polypeptides, which are integral subunits of the oxidative phosphorylation enzyme complexes I, III, IV and V. We show that the mitochondrial ATP production rate and the efficacy of the enzyme complexes I, III and IV, but not that of the completely nuclear-encoded complex II, are reduced in F2 hybrid males of N. giraulti and N. vitripennis. We hypothesize that nuclear–mitochondrial protein interactions in the oxidative phosphorylation pathway are disrupted in these hybrids, reducing energy generation capacity and potentially reducing hybrid fitness. Our results suggest that dysfunctional cytonuclear interactions could represent an under-appreciated post-zygotic isolation mechanism that, due to elevated evolutionary rates of mitochondrial genes, evolves very early in the speciation process.

Introduction

Understanding how species come into being and what prevents them from fusing is one of the most demanding questions in evolutionary biology (Coyne & Orr, 2004). One facet of this problem, intrinsic post-zygotic reproductive isolation, may be a consequence of deleterious epistatic interactions between genes that have diverged via mutation, selection, and/or drift within isolated species or populations (Dobzhansky, 1934; Muller, 1939, 1940, 1942; Turelli & Orr, 1995). Considerable effort has been invested in the discovery of incompatibility genes, also frequently referred to as speciation genes (Orr, 2005), but relatively few have been identified so far (e.g. Wittbrodt et al., 1989; Schartl et al., 1994; Malitschek et al., 1995; Ting et al., 1998; Barbash et al., 2003; Presgraves et al., 2003; Brideau et al., 2006; Masly et al., 2006; see also Hutter, 1997; Orr et al., 2004; Orr, 2005; Mallet, 2006; Noor & Feder, 2006). Although a large number of pathways could be disrupted by genic incompatibilities, genetic factors involved in speciation processes might not be evenly weighted (Mishmar & Gershoni, 2007). Pathways, for example, that are vital and whose underlying interacting genes are subject to tight-and-fast coevolutionary adaptive processes might be of greater significance than others (Mishmar & Gershoni, 2007). One pathway that may fall into this category is the oxidative phosphorylation system (OXPHOS) (Burton et al., 2006; Mishmar & Gershoni, 2007).

Mitochondria perform a variety of cellular functions; notably, however, they are the principal source of cellular energy in the form of adenosine 5′-triphosphate (ATP). The metabolic production of ATP is mediated by the OXPHOS pathway, which consists of four large multimeric enzyme complexes (called complexes I–IV), plus one additional complex (complex V or ATPase) responsible for the synthesis of ATP (Scheffler, 2008). OXPHOS enzyme complexes I, III, IV and V are composed of both nuclear- and mitochondrial-encoded subunits, whereas complex II is completely nuclear-encoded (Hatefi, 1985; see also Blier et al., 2001; Rand et al., 2004). Interactions between the 13 mitochondrial-encoded subunits and the approximately 70 nuclear-encoded subunits of the OXPHOS enzyme complexes must be sufficiently functional to support cellular energetic demands. Since the mitochondrial genome generally evolves faster than the nuclear genome (Boore, 1999), incompatibilities between nuclear and mitochondrial OXPHOS genes are expected to evolve rapidly (Mishmar & Gershoni, 2007). In certain cases, reduced hybrid fitness can be completely attributed to such incompatibilities (Ellison & Burton, 2008). Due to the combined action of independent assortment and meiotic recombination, intraspecific coadaptation of genes encoding components of the OXPHOS system may be broken up in F2 hybrids. Hence, these hybrids harbour novel, untested combinations of nuclear-encoded OXPHOS genes with indeterminate function when paired with the uniparentally inherited mitochondrial genome.

Mounting evidence suggests that hybridization severely affects the efficacy of nuclear–mitochondrial interactions in the OXPHOS pathway, although the biochemical impact was only rarely tested directly (e.g. Kenyon & Moraes, 1997; Rawson & Burton, 2002). In the marine copepod Tigriopus californicus (Baker, 1912), the interaction of nuclear-encoded cytochrome c (CYC) with mitochondrial-encoded cytochrome c oxidase (COX) subunit I was studied extensively: in vitro activity of COX is the greatest when paired with CYC from the same population and shows a significant decline when paired with heterogeneous CYC (Rawson & Burton, 2002). Further studies demonstrated that this effect is, in fact, due to a single nucleotide substitution in CYC (Harrison & Burton, 2006). Studies of COX activity in hybrids of Drosophila simulans and Drosophila mauritania yielded similar results (Sackton et al., 2003) and strongly suggest that the reduced function of nuclear–mitochondrial mosaic enzyme complexes may also affect longevity of hybrids via the generation of reactive oxygen species (Rand et al., 2006). Recent evidence showed that intraspecific sequence variation in both the nuclear and mitochondrial subunits of COX in D. simulans have strong pleiotropic effects on mitochondrial metabolism and several life-history traits (Ballard et al., 2007; Katewa & Ballard, 2007).

Species of the genus Nasonia Ashmead, 1904, which barely differ in their size and morphology (Darling & Werren, 1990), are reproductively isolated from each other due to their infection with different Wolbachia endosymbionts (Breeuwer & Werren, 1990; Bordenstein et al., 2001); however, wasps cured of the Wolbachia infection readily hybridize (Breeuwer & Werren, 1990, 1995). Interspecific F1 hybrid females of Nasonia giraulti (Darling, 1990) and Nasonia vitripennis (Walker, 1836) are fully viable and fertile, whereas haploid F2 hybrid males, which are morphologically intermediate between the parental species (Weston et al., 1999; Gadau et al., 2002), suffer two to six times higher mortality during development than pure strain males of either species (Breeuwer & Werren, 1995). Niehuis et al. (2008) recently studied the genetic basis of this hybrid breakdown and found that the increased inviability of F2 hybrid males of N. giraulti and N. vitripennis is primarily caused by incompatibilities between the nuclear genome and a maternally inherited genetic factor in the cytoplasm, most likely the mitochondrial genome (Breeuwer & Werren, 1995; Niehuis et al., 2008). If true, the reported cytonuclear genic incompatibilities should manifest in a disruption of those mitochondrial functions where nuclear gene products interact with those of the mitochondrion or with the mitochondrial genome itself.

Hybrids of Nasonia provide a unique opportunity to study the disruption of mitochondrial function in detail. As in all Hymenoptera, Nasonia species have a haplodiploid sex determination system: females develop from fertilized eggs and are diploid, whereas males develop from unfertilized eggs and are, therefore, haploid (reviewed by van Wilgenburg et al., 2006; Beukeboom et al., 2007). Hence, there are no intralocus dominance interactions in males. Recombinant F2 hybrid males will inherit a single nuclear complement consisting of both maternally and paternally derived alleles, along with the mitochondrial genome and other cytoplasmic factors exclusively from the maternal lineage. Any changes observed in the integration of nuclear and mitochondrial loci in male hybrids can then unequivocally be attributed to the interaction of nuclear and cytoplasmic factors.

Here, we exploit the haplodiploid genetic system offered by Nasonia to: (1) study mitochondrial function in N. giraulti, N. vitripennis and reciprocal interspecific hybrids; and (2) to directly test the hypothesis of nuclear–mitochondrial incompatibility in these hybrids (Breeuwer & Werren, 1995; Niehuis et al., 2008). Mitochondrial function was measured both as in vitro ATP generation capacity as well as through individual enzyme activities of OXPHOS complexes I, II, III and IV. We hypothesize that coadaptation of nuclear and mitochondrial components of the OXPHOS pathway is disrupted by hybridization and that this may contribute to broader physiological hybrid failures.

Methods

Nasonia strains and cross experiments

We used the Wolbachia-free Nasonia strains AsymCX (=N. vitripennis) and RV2X(U) (=N. giraulti) for the cross experiments and comparative physiological measurements. Both strains are highly inbred (i.e. homozygote for every marker tested so far; Breeuwer & Werren, 1990) and have been used for past experiments mapping hybrid breakdown (e.g. Niehuis et al., 2008). All wasps were cultured in an incubator at 25 °C under permanent light on flesh fly pupae (Sarcophaga bullata Parker, 1916). The cross experiments were conducted at the School of Life Sciences (Tempe, AZ, USA) following the protocol given by Niehuis et al. (2008) and the F2 hybrid males were then shipped in their pupal stage to the Scripps Institution of Oceanography (La Jolla, CA, USA). Upon receipt, the Nasonia pupae were stored in culture tubes and maintained at 25 °C. Wasps were removed from 25 °C incubation temperature 24–48 h after they had eclosed and immediately prior to initiating biochemical assays.

Preparation of mitochondrial fraction

Preparations containing an enriched mitochondrial fraction were completed using individual animals based on a protocol modified from Ellison & Burton (2006). Nasonia individuals were removed from culture tubes and decapitated using a sterile razor. Bodies were placed in 1 mL of mitochondrial isolation buffer (10 mm MOPS, pH 7.5, 55 mm KCl, 500 μm EGTA) and kept on ice until mitochondrial preparations were begun.

All mitochondrial preparations were carried out on individual wasps and designed to produce a crude solution enriched in the subcellular mitochondrial. Briefly, digestion with trypsin was followed by manual homogenization in isolation buffer. A low-speed centrifugation step (600 g at 4 °C) pelleted nuclei and most cellular debris to the bottom of the tube and the supernatant containing most mitochondria was transferred to a fresh tube. A subsequent higher speed centrifugation (11 000 g at 4 °C) pelleted mitochondria (and some attendant proteins). The supernatant was removed and discarded, and the enriched mitochondrial fraction was resuspended in 60 μL of mitochondrial storage solution (for assays of enzyme complexes I, II, I + III, II + III and IV; mitochondrial storage solution: 5 mm HEPES, pH 7.5, 125 mm sucrose, 0.5 mm ATP, 40 μm ADP, 2.5 mm sodium succinate, 1 mm K2HPO4) or 60 μL of isolation buffer (for assays of mitochondrial ATP production).

Two 5-μL aliquots of each mitochondrial preparation were transferred to 0.2-mL tubes for subsequent protein quantification using the NanoOrange Protein Quantification Kit (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s specifications. The average of these values was used for subsequent calculations and no pair of readings differed by more than 10% of the mean value. Five microlitres of mitochondrial preparation was additionally used to briefly check membrane integrity of samples using a JC-1 assay (after Salvioli et al., 1997); samples without measurable fluorescent signal were discarded as inviable. Samples for estimating the mitochondrial ATP production rate were used immediately upon completion of the mitochondrial preparation protocol. Samples for enzyme activity assays were frozen at −80 °C until use; storage at this temperature fractures the mitochondrial membranes, allowing substrate full access to enzyme complexes for assays.

Measurement of OXPHOS enzyme complex activities

Enzyme activities were measured according to Ellison & Burton (2006) and modified from Trounce et al. (1996) (complexes I, II, I + III, and II + III) and Rawson & Burton (2002) (complex IV). Mitochondrial preparations were divided into three 15-μL aliquots for each assay so that each assay was performed in triplicate for a given individual (i.e. three measurements of enzyme complex I activity were taken for one individual and three measurements of enzyme complex II activity for another individual). All other parameters of measurement were performed precisely as described by Ellison & Burton (2006). Data were averaged and normalized to protein content. For each OXPHOS complex enzyme assay, we assayed 20 adult pure stain males (10 N. vitripennis and 10 N. giraulti) and 80 F2 hybrid males (40 with N. vitripennis cytoplasm and 40 with N. giraulti cytoplasm) (Table 1).

Table 1.   Summary of the results of the Shapiro–Wilk test for normal distribution of the sampled data, the Levene’s test for equality of variances between the four studied groups of males (see below) and the anova/Welsh’s test for differences in the mean values between these groups.
Enzyme activity/production rateGroupNShapiro–Wilk testLevene’s testanova/Welsh’s test
WPF3,96PF3,96/F3,37.958P
  1. The Welsh’s test was applied to the data of the ATP production rate as an alternative to the anova as the Levene’s test indicated a significant inhomogeneity of the variances between the groups. Significant comparisons are shown in bold. N. vit.♂♂ Nasonia vitripennis pure strain males, N. gir.♂♂ Nasonia giraulti pure strain males, gv[v] ♂♂ N. vitripennis × N. giraulti F2 hybrid males with N. vitripennis cytoplasm, vg[g] ♂♂ N. vitripennis × N. giraulti F2 hybrid males with N. giraulti cytoplasm.

OXPHOS complex IN. vit.♂♂100.9520.6872.0900.10728.1180.000
gv[v] ♂♂400.9540.100
vg[g] ♂♂400.9470.058
N. gir.♂♂100.9400.555
OXPHOS complex IIN. vit.♂♂100.9190.3480.5630.6410.5730.634
gv[v] ♂♂400.9570.131
vg[g] ♂♂400.9630.211
N. gir.♂♂100.9210.367
OXPHOS complex IIIN. vit.♂♂100.9390.5461.4400.23616.9090.000
gv[v] ♂♂400.9790.660
vg[g] ♂♂400.9460.057
N. gir.♂♂100.9390.543
OXPHOS complex IVN. vit.♂♂100.9690.8822.5620.05914.5570.000
gv[v] ♂♂400.9630.214
vg[g] ♂♂400.9580.138
N. gir.♂♂100.9170.333
In vitro ATPN. vit.♂♂100.9630.8145.5830.00136.3620.000
gv[v] ♂♂400.9720.404
vg[g] ♂♂400.9680.313
N. gir.♂♂100.9580.758

ATP production rate assays

The rate of mitochondrial ATP production in mitochondrial preparations was determined in an end-point assay following the protocol of Ellison & Burton (2006). Samples were divided into two 20-μL aliquots [hereafter denoted as (+) and (−)]. Five microlitres of substrate solution (1 mm ADP, 9 mm pyruvate, 4 mm malate in mitochondrial isolation buffer) was added to the (+) sample and 5 μL of mitochondrial isolation buffer was added to the (−) sample. Samples were then incubated at 25 °C for 5 min before ATP content was assayed using the CellTiter-Glo ATP Quantification Kit (Promega, Madison, WI, USA), a luminescence-based assay (luciferin–luciferase) of ATP concentration in solution. Samples were assayed in half-area 96-well plates. Twenty (+) and 20 (−) samples were assayed at a time and each set of 20 was independently paired with a series of ATP standards for quantification. ATP production rate was determined by correcting all wells for background luminescence, then subtracting the reading obtained in each (−) sample from the corresponding (+) sample. Finally, all values were normalized to protein content. Samples assayed consisted of 20 adult pure stain males (10 N. vitripennis and 10 N. giraulti) and 80 F2 hybrid males (40 with N. vitripennis cytoplasm and 40 with N. giraulti cytoplasm) (Table 1).

Statistical analyses

Oxidative phosphorylation system enzyme activities and ATP production rates were tested for normality within groups with the Shapiro–Wilk test as implemented in stats package of the R software version 2.7.1 (R Development Core Team 2008). Homogeneity of variances for enzyme activities and the ATP production rate between groups was assessed with the classical Levene’s test implemented in the lawstat package of the R software. Differences in enzyme activities were parametrically assessed with a one-way analysis of variance (anova) as implemented in the stats package. As the Levene’s test rejected equality of ATP production rate variances, we conducted a one-way analysis of means (a.k.a. Welch’s test; part of the stats package) to assess differences in the ATP production rate between groups. The Tamhane T2 post hoc test implemented in the spss software version 16 (SPSS Inc., Chicago, IL, USA) was applied to compare enzyme activities and the ATP production rate between all pairs of groups.

Results

The results of the Shapiro–Wilk tests (Table 1) were consistent with the assumption that the OXPHOS complex I–IV enzyme activities and the ATP production rates in the four studied groups of males (i.e. N. vitripennis, N. giraulti, N. vitripennis × N. giraulti F2 hybrids with vitripennis and giraulti cytoplasm, respectively) were normally distributed. The Levene’s test results (Table 1) further suggested equality of variances between groups for enzyme activities of the OXPHOS complexes I–IV, but not for the in vitro ATP production rate. The analysis of variances (anova) indicated significant differences between groups in the activity of OXPHOS enzyme complexes I, III and IV, but not in that of the completely nuclear-encoded complex II (Table 1). The Welch’s test (Table 1) further indicated significant differences in the ATP production rate between the four groups.

The T2 Tamhane post hoc test results (Table 2) showed that the enzyme activity of three OXPHOS complexes was significantly reduced in F2 hybrid males relative to the males of the parental strains (Fig. 1). No significant differences were found in OXPHOS enzyme activities between parental strains. OXPHOS complex I enzyme activity was significantly reduced in both hybrids with N. vitripennis cytoplasm and hybrids with N. giraulti cytoplasm. OXPHOS complex III activity was significantly reduced in hybrids with N. giraulti cytoplasm relative to N. vitripennis pure strain males (close to significantly reduced to N. giraulti males), but not in hybrids with N. vitripennis cytoplasm. By contrast, OXPHOS complex IV activity was significantly reduced in hybrids with N. vitripennis cytoplasm, but not in hybrids with N. giraulti cytoplasm. OXPHOS complex II activity, the only completely nuclear-encoded complex, was not significantly different from any parental strain in either hybrid cytoplasmic background. Mitochondrial ATP production rate in isolated mitochondria was significantly reduced in all hybrids relative to both parental strains (Table 2, Fig. 2).

Table 2.   Results of the Tamhane’s T2 post hoc test for assessing differences in OXPHOS complex I–IV enzyme activities and in vitro mitochondrial ATP production rates between Nasonia vitripennis × Nasonia giraulti F2 hybrid males and N. giraulti and N. vitripennis pure strain males.
ComparisonComplex IComplex IIComplex IIIComplex IVATP production
  1. Significant comparisons are shown in bold. N. vit.♂♂ N. vitripennis pure strain males, N. gir.♂♂ N. giraulti pure strain males, gv[v] ♂♂ N. vitripennis × N. giraulti F2 hybrid males with N. vitripennis cytoplasm, vg[g] ♂♂ N. vitripennis × N. giraulti F2 hybrid males with N. giraulti cytoplasm.

N. vit.♂♂/N. gir.♂♂0.9130.9970.8220.9911.000
gv[v] ♂♂/N. vit.♂♂0.0001.0000.1530.0160.000
gv[v] ♂♂/N. gir.♂♂0.0000.9940.9940.0010.000
vg[g] ♂♂/N. vit.♂♂0.0010.9910.0010.8630.000
vg[g] ♂♂/N. gir.♂♂0.0000.8280.0660.2620.000
vg[g] ♂♂/gv[v] ♂♂0.1600.9350.0000.0000.025
Figure 1.

 Box-and-whisker plots showing median (horizontal line), interquartile range (box) and maximum/minimum range (vertical line) of measured enzyme activities of OXPHOS complexes I, II, III and IV for Nasonia vitripennis and Nasonia giraulti parental strains and their F2 hybrids. Abbreviations: N. vit. = N. vitripennis pure strain males, N. gir. = N. giraulti pure strain males, gv[v] = N. vitripennis × N. giraulti F2 hybrid males with N. vitripennis cytoplasm, vg[g] = N. vitripennis × N. giraulti F2 hybrid males with N. giraulti cytoplasm.

Figure 2.

 Box-and-whisker plots showing median (horizontal line), interquartile range (box) and maximum/minimum range (vertical line) of measured in vitro mitochondrial ATP production rate for Nasonia giraulti and Nasonia vitripennis parental strains and their F2 hybrid males. Abbreviations: N. vit. = N. vitripennis pure strain males, N. gir. = N. giraulti pure strain males, gv[v] = N. vitripennis × N. giraulti F2 hybrid males with N. vitripennis cytoplasm, vg[g] = N. vitripennis × N. giraulti F2 hybrid males with N. giraulti cytoplasm.

Discussion

Hybridization between species is widely observed in nature, yet hybrid lineages rarely persist for long periods (Harrison, 1993). One explanation for this observation is that hybrids are less fit than individuals from the parental strains – a phenomenon known as hybrid breakdown – for reasons ranging from ecology (Schluter, 2000; Nosil et al., 2002) and development (Shapiro et al., 2004) to genetics (Ting et al., 1998; Presgraves et al., 2003; Brideau et al., 2006). Often, first-generation hybrids are equal in fitness to parental strains and hybrid breakdown is manifested only in the F2 or later generations (Wu & Palopoli, 1994; Breeuwer & Werren, 1995; Hutter, 1997; Burton et al., 2006). In such cases, epistatic interactions involving two or more loci must be assumed. This model of isolation, where hybrids suffer from deleterious interactions between genes that evolved independently in separate populations or species, is known as the Dobzhansky–Muller model (Dobzhansky, 1936; Muller, 1942; further developed by Orr, 1995; Orr & Turelli, 2001) and the interacting loci commonly referred to as Dobzhansky–Muller pairs. The model, however, is by no means limited to deleterious interactions involving only two loci.

Most studies of Dobzhansky–Muller interactions have focused on nuclear–nuclear interactions, but epistatic interactions between genes in nuclear and plastid genomes may be equally widespread and important for the speciation process (Turelli & Moyle, 2007; Bolnick et al., 2008). In the case of epistasis between nuclear and mitochondrial genes, all possible interactions between protein coding loci are located within the enzyme complexes of the mitochondrial OXPHOS pathway. The efficacy of these interactions is largely responsible for cellular ATP synthesis and is of paramount importance to the cell (Rand et al., 2004). Failure of mitochondrial function is known to have broad pleiotropic effects and has been correlated with a growing number of human disease pathologies (Shadel, 2004), apoptosis (Regula et al., 2003), aging (Dufour & Larsson, 2004) and sperm motility (Troiano et al., 1998).

The results of our experiments in the present study demonstrate that the ATP production rate of mitochondria isolated from N. vitripennis × N. giraulti F2 hybrid males is significantly reduced and has a wider variance relative to males of the parental lineages regardless of the origin of the cytoplasm (Tables 1 and 2, Fig. 2). This is consistent with the Dobzhansky–Muller model of genic incompatibilities predicting the appearance of deleterious epistatic interactions in F2 and later hybrids, but does not distinguish between effects of nuclear–nuclear vs. nuclear–mitochondrial interactions. At the enzymatic level, our measurements show that OXPHOS enzyme complex I activity is significantly reduced in both hybrids with N. vitripennis and hybrids with N. giraulti cytoplasm, whereas OXPHOS enzyme complex III activity is significantly reduced only in hybrids with N. giraulti cytoplasm (Table 2, Fig. 1). Conversely, OXPHOS enzyme complex IV activity is significantly reduced in hybrids bearing a N. vitripennis cytotype, but not in those with N. giraulti cytoplasm (Table 2, Fig. 1). In contrast to the ATP production rate, the variances of the OXPHOS complex enzyme activities do not differ significantly between the hybrids and pure stain individuals (Table 1, Fig. 1).

The symmetry of the OXPHOS enzyme complex I disruption in our experiments can be explained by both nuclear–nuclear and nuclear–mitochondrial incompatibilities, and we are thus unable to resolve nuclear–mitochondrial from nuclear–nuclear effects. However, the asymmetric reduction in efficiency in complexes III and IV must be based on nuclear–mitochondrial incompatibilities. In the absence of selection, F2 hybrid males from reciprocal crosses are not expected to differ systematically in their nuclear genetic composition. Any differences between reciprocal crosses can therefore be attributed directly: (1) to the maternally inherited cytoplasm; or (2) to interactions between nuclear and cytoplasmic elements. Asymmetrical biochemical responses to hybridization in complexes III and IV thus indicate that deleterious nuclear–mitochondrial interactions must exist in interspecies hybrids of Nasonia. Previous studies examining the nature of hybrid breakdown in Nasonia have found evidence for nuclear–cytoplasmic effects in hybrids of N. vitripennis and N. giraulti (Breeuwer & Werren, 1995; Gadau et al., 1999; Niehuis et al., 2008) and specifically identified four cytonuclear genic incompatibilities that cause increased mortality of F2 hybrid males (Niehuis et al., 2008). Three of the deleterious interactions were observed in F2 hybrid males with N. vitripennis cytoplasm, whereas one was found in F2 hybrid males with N. giraulti cytoplasm. These results are fully consistent with the deleterious effects of hybridization on the mitochondrial biochemistry that we observed in this study.

The close alignment of past mapping studies with the parameters of mitochondrial biochemistry measured in the present study strongly suggest that deleterious nuclear–mitochondrial epistasis disrupts the mitochondrial OXPHOS pathway of N. vitripennis × N. giraulti F2 hybrid males, leading to diminished mitochondrial ATP production capacity. Whether the reduced ATP production rate is responsible for the increased mortality of N. vitripennis × N. giraulti F2 hybrid males during larval development or metamorphosis, as reported by Niehuis et al. (2008), still remains to be shown as the current assay data were collected in adult wasps. Further quantitative genetic studies correlating the observed phenotype of complexes I–IV with their genotype in F2 hybrid males should enable mapping of the nuclear loci and, ultimately, the amino acid changes causally involved in the observed reduction in mitochondrial efficiency. These studies will be important to address the question of whether the cytonuclear incompatibilities are between coding genes of the OXPHOS pathway itself, as the OXPHOS complex-specific response to hybridization suggests, or whether other genic interactions have to be invoked.

An understanding of the basic genetic and physiological mechanisms underpinning biological isolation between species remains an important goal in evolutionary biology. This study, in combination with previous investigations, confirms the hypothesis of nuclear–mitochondrial incompatibility in Nasonia species hybrids and suggests that different lineages accumulate different mutations at OXPHOS coding loci, gradually diverging over time. Subsequent hybridization between such divergent lineages may then expose deleterious nuclear–mitochondrial epistatic interactions, disrupting mitochondrial biochemistry and, ultimately, hybrid fitness. Therefore, cytonuclear genic incompatibility could represent an under-appreciated post-zygotic isolation mechanism with the potential to evolve very early in the speciation process.

Acknowledgments

We thank J. D. Gibson, W. Willis and two anonymous reviewers for helpful comments and B. Allen, J. Overson, and A.K. Judson for their help in the laboratory. ON acknowledges the Alexander von Humboldt foundation for a Feodor Lynen stipend. This work was funded by a NIH marine biotechnology training grant to CKE and a NIH-NCRR grant (grant no. 1R21 RR024199-01) to JG.

Ancillary