Developmental Biology 415 (2016) 391–405

Contents lists available at ScienceDirect

Developmental Biology

http://d0012-16

n Corrnn CorE-m

jlynch421 Cu

Changes

journal homepage: www.elsevier.com/locate/developmentalbiology

Resource

Dissection of the complex genetic basis of craniofacial anomalies usinghaploid genetics and interspecies hybrids in Nasonia wasps

John H. Werren a,n, Lorna B. Cohen b, Juergen Gadau c, Rita Ponce a,1, Emmanuelle Baudry a,d,Jeremy A. Lynch b,nn

a Department of Biology, University of Rochester, Rochester, NY 14627, United Statesb Department of Biological Sciences, University of Illinois at Chicago, Chicago, IL 60607, United Statesc School of Life Sciences, Arizona State University, Tempe, AZ 85285, United Statesd Ecologie Systematique Evolution, Univ. Paris-Sud, CNRS, AgroParisTech, Universite Paris-Saclay, 91400 Orsay, France

a r t i c l e i n f o

Article history:Received 3 July 2015Received in revised form13 November 2015Accepted 21 December 2015Available online 23 December 2015

Keywords:NasoniaCraniofacialEpistasisQTLHybridCleft paletteIntrogression

x.doi.org/10.1016/j.ydbio.2015.12.02206/& 2015 The Authors. Published by Elsevier

esponding author.responding author.ail addresses: jack.werren@rochester.edu (J.H.@uic.edu (J.A. Lynch).rrent address: CE3C – Centre for Ecology, E(Ce3C), Faculdade de Ciências da Universida

a b s t r a c t

The animal head is a complex structure where numerous sensory, structural and alimentary structuresare concentrated and integrated, and its ontogeny requires precise and delicate interactions amonggenes, cells, and tissues. Thus, it is perhaps unsurprising that craniofacial abnormalities are among themost common birth defects in people, or that these defects have a complex genetic basis involving in-teractions among multiple loci. Developmental processes that depend on such epistatic interactionsbecome exponentially more difficult to study in diploid organisms as the number of genes involvedincreases. Here, we present hybrid haploid males of the wasp species pair Nasonia vitripennis and Nasoniagiraulti, which have distinct male head morphologies, as a genetic model of craniofacial developmentthat possesses the genetic advantages of haploidy, along with many powerful genomic tools. Viable,fertile hybrids can be made between the species, and quantitative trail loci related to shape differenceshave been identified. In addition, a subset of hybrid males show head abnormalities, including clefting atthe midline and asymmetries. Crucially, epistatic interactions among multiple loci underlie several de-velopmental differences and defects observed in the F2 hybrid males. Furthermore, we demonstrate anintrogression of a chromosomal region from N. giraulti into N. vitripennis that shows an abnormality inrelative eye size, which maps to a region containing a major QTL for this trait. Therefore, the geneticsources of head morphology can, in principle, be identified by positional cloning. Thus, Nasonia is wellpositioned to be a uniquely powerful model invertebrate system with which to probe both developmentand complex genetics of craniofacial patterning and defects.& 2015 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND

license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

The interpretation of the apparent correlation between formand function in living beings was one of the intellectual ad-vancements that led to formulation of Darwin's theory of evolu-tion, and the concept of homology (Appel, 1987). The problems ofhow and why forms change in evolution is still timely, and is oneof the pillars of the young but maturing field evolutionary devel-opmental biology (Moczek et al., 2015). In regard to evolution,specific forms that are more apt to perform particular functionsshould be favored, and increase in frequency over time. The

Inc. This is an open access article u

Werren),

volution and Environmentalde de Lisboa, Portugal.

developmental basis of changes in form within natural popula-tions are still relatively poorly understood despite the in depthknowledge of developmental mechanisms in few model species.Questions that have not been adequately answered include: Howare shape and size regulated at the cellular level during develop-ment, and how are these features encoded in the genome? How issymmetry maintained between independently developing halvesof bilateral structures, and how is the fusion of multiple tissuesinto a functioning organ accomplished?

A key innovation in biological formwas the cephalization of theearly ancestors of the Bilateria. The head is the major structurethrough which bilateral animals perceive and interact with theirenvironments, and is thus a crucible for interactions betweenform, function, evolution and development. The concentration of alarge number of sensory organs of different types and embry-ological origins in a relatively small space poses significant de-velopmental challenges, requiring exquisite communication acrossa complex set of structures, each of which is crucial for the survival

nder the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

J.H. Werren et al. / Developmental Biology 415 (2016) 391–405392

of the organism (Young et al., 2000).Consistent with the complexity of head development and rapid

evolution of head size and shape in Homo sapiens, cranialanomalies are some of the most common birth defects occurringin humans (Stanier and Moore, 2004). A major class of craniofacialdefects is oral clefts primarily occurring in the lip and/or palate(CL/P), which occurs in up to 1:300 live births (Wyszynski et al.,1996) and is known to have a complex genetic inheritance patterninvolving interactions among several loci (Carter et al., 1982;McKusick, 1994; Prescott et al., 2001; Shields et al., 1981; Wilkieand Morriss-Kay, 2001).

The advent of advanced sequencing techniques has led to theapplication of large scale approaches, such as genome wide asso-ciation studies (GWAS) to human craniofacial development((Dixon et al., 2011; Twigg and Wilkie, 2015). These studies haveidentified very strong candidate protein coding genes (Leslie et al.,2015; Wolf et al., 2015) and even regulatory regions (Attanasioet al., 2013; Fakhouri et al., 2014) affecting craniofacial develop-ment. However, given the large number of genes and epistaticinteractions that appear to underlie craniofacial development anddisease, the handful of genes with demonstrated roles in humansare likely a tip of a very large iceberg (Hallgrimsson et al., 2014).

Insects can be valuable models systems when attempting tounderstand complex developmental processes such as craniofacialdevelopment. Given that cephalization is a synapomorphy of theBilateria, a common evolutionary origin unites the vertebrate andinvertebrate head, and indeed many of the molecules involved inhead patterning and morphogenesis (e.g., Otx/otd genes, BMP,WNT, hedgehog signaling) are highly conserved between verte-brate models (Greene and Pisano, 2010; Hide et al., 2002; Zhanget al., 2002) and Drosophila (Royet and Finkelstein, 1997; Shyamalaand Bhat, 2002; Stultz et al., 2006; Won et al., 2015). Insects havethe added advantage of being easier to maintain as well as havingfaster generation times than vertebrate model systems, andtherefore can be used to reveal candidate genes involved in cra-niofacial abnormalities and development.

A major contribution to the genetic complexity of craniofacialdefects is epistasis-the phenotypic effect of non-additive interac-tions among alleles at different loci (Lidral and Moreno, 2005).Epistasis has been shown to play an important role in the de-generative craniofacial development of cavefish (Gross et al.,2014), and skull shape trait complexes in crosses between mousestrains (Wolf et al., 2005).

Epistatic interactions among alleles can be difficult to study dueto the complexity of the genetics involved. For example, as thenumber of interacting genes increases, there is an exponentiallyincreasing rarity of progeny homozygous for all of the requiredalleles. To illustrate, in crosses between two strains differing in atarget phenotype in a typical diploid organism, the proportion ofF2 offspring (i.e. F1� F1 parents) revealing a recessive epistaticinteraction between two autosomal loci is 1/16, for 3 loci is 1/64,and for 4 loci is 1/256. Backcrossing F1 progeny to either parentalstrain will not reveal such epistatic interactions among the pro-geny. The problem is even greater when screening for recessiveepistatic interactions within populations (e.g., GWAS), as suchapproaches are dependent upon allele frequencies among the in-teracting loci.

In contrast, recessive epistatic interactions are much morereadily revealed in organisms with haploid adults, such as thehymenoptera (ants, bees and wasps). These organisms have hap-lodiploid sex determination, where haploid males develop fromunfertilized eggs, and diploid females derive from fertilized eggs.Thus, hymenopteran F2 haploid males will reveal recessive epi-static interactions among 2, 3, and 4 loci at frequencies of 1/4, 1/8,and 1/16 respectively (compare to the respective values of 1/16, 1/64, and 1/256 in diploids). Thus, the statistical power of detecting

epistatic interactions is greatly enhanced in such organisms,should they be laboratory tractable.

The parasitoid wasp genus Nasonia is emerging as a powerfulgenetic model system, particularly for complex traits, and devel-opmental and evolutionary genetics (Beukeboom and Desplan,2003; Lynch, 2015; Werren and Loehlin, 2009). These insects havea short generation time (�two weeks), large family sizes, are ea-sily reared in the laboratory, can be kept under refrigeration forlong periods (allowing storage of many strains). They have ex-tensive genomic and transcriptome resources (Werren and Loeh-lin, 2009; Werren et al., 2010), and genetic tools such as systemicRNA interference (Lynch and Desplan, 2006) and visible and mo-lecular markers to facilitate mapping and cloning of phenotypictraits (Desjardins et al., 2013; Niehuis et al., 2013). A major ad-vantage of Nasonia is the ability to perform crosses among closelyrelated interfertile species (Breeuwer and Werren, 1995). Thisfeature permits the mapping and positional cloning of quantitativetrait loci (QTL) involved in species differences in development andmorphology (Loehlin et al., 2010a; Loehlin and Werren, 2012).

Males of Nasonia species differ in cranial shape, with the largestdifferences being between Nasonia giraulti and N. vitripennis((Darling and Werren, 1990), Fig. 1). These two species are sepa-rated by approximately 1 million years of independent evolution,and are diverged by �2 to 3% at the nucleotide level (Werren et al.,2010). In this manuscript, we describe a range of head shapes aswell as cranial abnormalities, including abnormal cranial midlinefurrowing, dorsal–ventral asymmetries, and lateral asymmetriesthat occur in F2 interspecies hybrid males. Analysis of gene in-teractions involved in cranial development and cranial abnorm-alities in hybrid males is greatly facilitated by the ability to moreeasily detect epistatic interactions in haploids, because at any locusthe male is hemizygous for genes from one or the other species,thus removing complications caused by dominance interactionsbetween alleles. This major advantage is enhanced by the powerfulgenetic toolbox available in Nasonia (Lynch, 2015; Werren andLoehlin, 2009).

Since F2 males show considerable variation in head shape, weare able to map major quantitative trait loci (QTL) and epistaticinteractions in cranial development. These QTL can be further in-vestigated by introgressing genetic regions containing cranial QTLfrom one species into the genetic background of the other.

Here we describe the potential of the Nasonia system for ge-netic and molecular analyses of craniofacial development, takingadvantage of differences in male head shape between closely re-lated species. We describe the basic system, results of a QTL ana-lysis for head shape revealing major QTL and epistatic interactions,cranial abnormality syndromes in hybrids, the introgression of ahead shape locus from N. giraulti into N. vitrpennis, and futuredirections using the tools available in Nasonia to reveal gene in-teractions involved in cranial development.

2. Materials and methods

2.1. Genetic crosses for Quantitative Trait Locus (QTL) analysis

Crossing methods to generate F2 haploid hybrid males forquantitative trait analysis are described in detail in (Gadau et al.,1999, 2002). The same F2 mapping population used in those studiesfor analysis of hybrid incompatibility loci and wing morphology QTLare used here for mapping of QTL for head morphology. Basically, across was initiated between two Wolbachia free (Werren, 1997) andhighly inbred strains of N. vitripennis (AsymCx) and N. giraulti(R16A). The N. giraulti strain R16A is derived from an introgressionof N. giraulti nuclear genome in a N. vitripennis cytoplasm to avoidnuclear-mitochondrial incompatibilities (Breeuwer and Werren,

Fig. 1. Head shape differences between males of the parental species. N. vitripennis (left) and N. giraulti (right) and, with head measurements shown on an N. giraulti image(below), (HL) Length from top of head to bottom of head, (AOL) Length from top of head to center of antennal sockets, (OIO) Interocular distance through ocelli, (MHW)Maximumwidth of the head, (AIO) Interoccular distance through center of antennal sockets, (MIO) Interocular distance at maximimum head width, (FE) Line fromwhere eyebegins to center of mandible, (FEP) Farthest point of cheek perpendicular to line FE.

J.H. Werren et al. / Developmental Biology 415 (2016) 391–405 393

1995). Fifteen hybrid F1 virgin females were used from the inter-species cross. These were provided with fly hosts for parasitization,and by setting the females as virgins, they produce all male pro-geny. Because males are haploid, this facilitates the detection of QTLfor morphology and epistatic interactions among QTL. These fe-males produced 178 males that were used for the mapping and QTLanalysis as described in (Gadau et al., 1999, 2002).

2.2. Segmental Introgression Lines (SILs)

As a consequence of other studies, SILs have been generatedthat contain chromosomal regions of N. giraulti in an N. vitripennisgenetic background. Many of these have been characterized usingthe genotyping microarray (Desjardins et al. 2013) or sets of PCRbased molecular amplifications to confirm the regions in-trogressed and absence of other N. giraulti regions from otherchromosomes in the lines.

One introgression line involving the proximal region of chro-mosome 2 has been examined for an associated QTL for headshape. The line (INT_2C7#49-19) was produced as follows: Aninitial cross of N. vitripennis females to N. giraulti males generatedF1 hybrid females, who were then backcrossed to vitripennis malesto produce F2 females. The N. giraulti proximal region on Chr2 wasthen introgressed into a N. vitripennis background for eight gen-erations while selecting for females heterozygous for two mole-cular markers, s3 and s13, which flank the centromeric region.Primers were designed to amplify two regions that are poly-morphic for an indel between the two species; the markers s3 ands13 are approximately 10 cM apart on the molecular linkage mapin cluster 2.035 and 2.048 respectively (Desjardins et al., 2013).Individual females were first mated and allowed to parasitizehosts, and then DNA was extracted and genotyped using poly-merase chain reaction and agarose gel electrophoresis; hetero-zygous females were selected for the next generation.

The homozygous INT_2C7##49-19 line was then created fromsingle male recombinant between the two flanking markers asfollows: One hundred F11 males were mated singly to AsymCx

virgins, then genotyped for the 2 flanking indels to identify re-combinants. Six of the hundred were recombinant in the region ofinterest. A recombinant male was then mated to N. vitripennis(AsymCx) females. A subset of the resulting female progeny wasprovided hosts as virgins, and resulting males were mated toheterozygous females from the previous generation (their“aunts”). Individuals were genotyped with the molecular markerto establish a line homozygous for marker s3 (INT_2C7##49-19).

2.3. Measurements

To control for orientation, heads were positioned on a micro-scope slide with double-sided tape such that the dorsal ocellarridge and the ventral clypeus were visible and in the same focalplane. Photographs were then taken using a Zeis M5A microscopewith 50� magnification with phototube attachment and shotusing Kodak Ektachrome 160T Tungsten balanced slide film. Aslide micrometer was photographed under the same conditions touse for millimeter calibration of measurements. Slides were thenscanned using ScanMaker 35T plus and the ScanWizard program(Microtek). TIFF files of heads (a calibration micrometer photos)were then measured using Adobe Photoshop, and then calibratedusing measurements from the micrometer photographs. The spe-cific measurements made are shown in Fig. 1. The initial mea-surements were made in 1998.

A second batch of parental heads were measured in 2015, be-cause the full set of head dimensions taken for hybrid heads hadnot been made for parentals in the original study. For these secondparental measurement series, wasps are first fixed in 5% for-maldehyde for at least 4 h, soaked in 30% H2O2 for 8–10 days,stained with Congo Red dye and imaged on an Andor RevolutionWD spinning disc confocal system. Images were captured viaAndor iQ3 software and samples were illuminated with 561 nmand 405 nm diode laser. Z-stacks were acquired using a 10� ob-jective at 3.58 μm increments, and were processed, colorized andmeasured in Imaris7.1.1.

A subset of measurements were compared between the

Table 1Relevant measurements of the male heads of the parental species N. vitripennis andN. giraulti, and hybrid F2 males.

Measurement N. vitripennis N. giraulti Hybrids

Mean SD Mean SD Mean SD

*MHW/HL 1.419 0.060 1.384 0.052 1.418 0.055*MEYE/HL 0.450 0.080 0.498 0.034 0.540 0.047*OIO/HL 0.853 0.037 0.827 0.032 0.821 0.082AOL/HL 0.673 0.024 0.654 0.047 0.690 0.062***OIO/AIO 0.893 0.035 0.933 0.027 0.912 0.09***OIO/MIO 0.882 0.024 0.932 0.021 0.936 0.08***FEP/FE 0.137 0.013 0.228 0.024 0.176 0.027

MHW¼Maximum head width, HL¼head length, MEYE¼Eye width at maximumhead width (¼MHW-MIO, see Fig. 1), OIO¼ Interocular distance through ocelli,AIO¼ Interocular distance through center of antennal sockets, AOL¼Length fromtop of head to center of antennal sockets MIO¼ Interocular distance at maximumhead width, FE¼Line from where eye begins to center of mandible, FEP¼Farthestpoint of cheek perpendicular to line FE. Statistical comparisons were made onlybetween the parental species. Triple asterisks represent measurements differingsignificantly between the two species at po0.001, double asterisk representspo0.01 and single asterisks represent po0.05. 16 individuals from each parentalspecies were measured.

J.H. Werren et al. / Developmental Biology 415 (2016) 391–405394

methods, and found to be mostly comparable except that theMHW/HL values overlapped between the two species in the sec-ond analysis. Data in Table 1 uses this second set of measurements,and the first set of measurements are presented in Table S1 forcomparison. Significance of the difference between the specieswas tested by using one tailed t-tests in Microsoft excel. Pearsonproduct moment correlations on head phenotypes were per-formed using the function available at http://www.r-project.org/(Team, 2014).

2.4. Genotyping and linkage analysis

The methods for RAPD genotyping and linkage map generationfor the QTL analysis are the same as in (Gadau et al., 1999, 2002),and uses the same F2 hybrid mapping population as in thosestudies. The linkage map was based on the segregation of 91 RAPDmarkers (Randomly Amplified Polymorphic DNA) in 178 haploidmales derived from 15 F1 females. Linkage group designationsfrom Gadau et al. (Gadau et al., 1999, 2002) are changed to cor-respond to the five chromosomes of Nasonia used in currentgenome wide linkage maps (Desjardins et al., 2013), and additionalmolecular and visible markers have been placed onto the map asdescribed below.

Additional crosses were conducted to place a set of visible andmolecular markers onto the linkage map (see Figs. 3 and 4). Forinsulin pathway genes, degenerate primers were used to amplifyfive genes (InR, PI3K, Pten, S6K, and mTor) which were then se-quenced in N. vitripennis and N. girualti (V and G, respectively), andallele restriction site polymorphisms were identified (or speciesspecific primers designed). The genes were then mapped onto theNasonia marker map using 50 F2 hybrid males from the mappingpopulation. Three visible mutants from N. vitripennis (eye colormutants red833 (R-locus), or123, rdh5) were placed onto the RAPDmap as described in (Gadau et al., 1999). In addition, two majorwing size QTL (ws1 and ws2) were placed on the linkage map(Gadau et al., 2002 ) and their locations have subsequently beenconfirmed by finer scale segmental introgression studies (Loehlinet al., 2010d; Loehlin and Werren, 2012). Together, these markersallowed us to anchor and orient the RAPD linkage map to thecurrent genome-wide molecular map for Nasonia. Linkage groupdesignations (Gadau et al., 1999, 2002) are changed to correspondto the five chromosomes of Nasonia used in current genome widelinkage maps (Desjardins et al., 2013).

2.5. QTL analysis

Having haploids as a mapping population for QTL analysis hasmultiple advantages; (1) The effect of an allele is directly mea-surable because there are no dominance interactions among al-leles of the same locus. (2) Epistatic interactions between nuclearloci are easier to analyze because for any two-loci interactions onlyfour genotypes are possible. (3) Linkage phase can be determinedin each individual even if dominant markers like RAPDs (Ran-domly Amplified Polymorphic DNA) are used.

MapQTL 4.0 (van Ooijen et al., 2000) was used to identify QTLfor all traits. First a standard interval mapping was done to identifythe major QTL. Then mqm mapping (multiple-QTL-model), im-plemented in MapQTL, was used to fit more than one QTL at atime. The mqm-mapping procedure uses markers closest to theQTL as cofactors to take over the role of the QTL. The genome wideLOD threshold for a significant QTL at the 5% and 1% false positiverate was determined for each trait individually using a standardpermutation test for interval mapping (Churchill and Doerge,1994) implemented in MapQTL 4.0.

EPISTAT (Chase et al., 1997) was used to reveal conditional QTL,i.e. QTL which only have a significant effect on the trait if a particularallele at a second unlinked locus is present. First an automatedsearch option was used to find all conditional QTL for all traits ex-ceeding a predetermined threshold (the default settings of theprogram were used, i.e. 5.0 LLR (Log Likelihood Ratio) for the nullthreshold and 6.0 LLR for the additive threshold, minimal group size10, see Epistat Tutorial). Then we discarded all interactions betweenmarkers on the same linkage group, considering only possible epi-static interactions between linkage groups. The initial p-value waschosen for particular LLR based on values provided in the EPISTATUsers Manual. These were then corrected for multiple comparisonsusing the formula p-corrected¼p/(1�(1�p)90) under the assump-tion of 100 independent comparisons (Chase et al., 1997; Lark et al.,1995). Individual markers on the map are not independent com-parisons due to linkage. The choice of 90 independent comparisonswas made by considering 3 regions per chromosome (two distal and1 proximal) for the 5 linkage groups and comparisons only betweenchromosomes for epistatic interactions.

Statistical comparisons using Fisher Exact Test (FET) or ChiSquare were performed at the web portal http://www.socscistatistics.com/tests/fisher/Default2.aspx.

3. Results

3.1. Species differences in head shape

Females of the two species (N. giraulti and N. vitripennis) arevirtually indistinguishable in terms of head shape and othermorphological features (Darling and Werren, 1990). In contrast,males differ in several features of head shape, but are fairly con-sistent within species (see Fig. 1, Table 1, (Darling and Werren,1990)). Landmarks on the Nasonia head include the two antennalsockets (antennae removed in the heads shown), clypeus in frontof the antennal sockets and before the mouthparts (not visible),two compound eyes on either side of the head, and three simpleeyes (ocelli) on the back of the head. A depression occurs behindthe antennal sockets, into which the antennae can be partiallywithdrawn. We refer to this as the antennal cup, which is vaguelyheart shaped in normal heads of both species. Antennae wereremoved in the figures to provide easier visualization of headshape.

To investigate the genetics of head shape, we evaluated par-ental and F2 hybrid males for various head shape parameters(Fig. 1). These include three measures of relative head shape along

J.H. Werren et al. / Developmental Biology 415 (2016) 391–405 395

the dorsal ventral and lateral axes (a) the ratio of head width atthe maximum (which falls between the antennae and ocelli) tohead length (MHW/HL), (b) length of the two eyes at the max-imum head width relative to head length (MEYE/HL), whichmeasures the contribution of the eyes to the ratio of head width tohead length, and (c) interocular distance at the ocellus relative tohead length (OIO/HL). A second ratio investigates relative headshape along the dorsal ventral midline by comparing the length ofa line from the antennal midline to the ocellus relative the headlength along the same line (AOL/HL). Heads with relatively largerregions between the antennae and clypeus will show smaller ra-tios for AOL/HL. The relative change in width from dorsal to ventralportions of the head is captured by (d) OIO/AIO (interocular at theantennal sockets relative to interocular at the ocellus) and (e) OIO/MIO (interocular at the ocellus relative to interocular at themaximum head width line). The smaller the OIO/AIO, the more ahead increased relative width between the eyes along the dorsal–ventral axis.

The most notable differences between male heads of N. giraultiand N. vitripennis are in size of the lateral cheek (FEP/FE) and as-pects of the head width/length ratio. N. giraulti has significantlylarger lateral cheeks than do N. vitripennis, and distributions in thisfeature (normalized to head size) are non-overlapping betweenmales of the two species (Fig. 1, Table 1). In terms of overall headshape, N. vitripennis heads are more rectangular (or ovoid)whereas N. giraulti heads are more square (Fig. 1, Table 1). Themaximum head width to head length (MHW/HL) ratios are sig-nificantly different (at the po0.05 level). Interestingly, compo-nents of this ratio (Eye width at maximum head width to headlength (MEYE/HL) and interocular distance at ocelli to head length(OIO/HL) were both highly significantly different between thespecies, indicating that both the eye field and medial head capsulefield differ independently in size between the two species.

The differences in cranial morphology between males of thetwo species almost certainly reflect the action of sexual selection,although it is currently unclear what aspects of mate competitionor mate choice have selected for these morphological differences.Nevertheless, they provide a useful tool for studying the geneticsof cranial morphology.

3.2. F2 hybrid male head shape

A comparison between measurements of males from bothparental species and F2 hybrid males showed that some values ofthe F2 males were intermediate between the two parental phe-notypes for head shape measurements (e.g. AOL/HL, FEP/FE, Ta-ble 1). For other ratios, the F2 mean is much closer to that of one ofthe parentals (e.g. MHW/HL, OIO/MIO), and some mean hybridhead shape ratios are actually larger or smaller than that of eitherparental species (e.g MEYE/HL, OIO/AIO). These patterns can re-flect epistatic interactions, transgressive phenotypes, and linkageof cranial loci to hybrid incompatibility loci (Gadau et al., 1999;Werren et al., 2010), which results in biased recovery of certainmarkers among adults. It is important to note that hybrid femaleshave normal heads that are indistinguishable from either of theparental species, and that male head shape characteristics arefairly consistent within species.

Correlations among traits in F2 are shown in Table 2. As ex-pected, many shape values that have at least one measurement incommon (i.e., (MHW/HL vs. OIO/HL, or OIO/MIO vs. OIO/AIO) arecorrelated in both parental species and in hybrids. More interest-ingly, there are weak to moderate correlations between cheek sizeand other head shape values in N. vitripennis, while cheek sizeseems to be regulated independently from other shape parametersin N. giraulti (Table 2). Strikingly, correlation between cheek sizeand shape values involving HL in F2 hybrid males are quite strong

(more so than in even N. vitripennis). In contrast, values involvingOIO are more weakly correlated with cheek size in the hybridscompared to N. vitripennis. The developmental and genetic basis ofthese correlated traits will be an interesting topic for futureresearch.

3.3. F2 hybrid male head developmental defects

Wild-type male heads within the two species show consistenthead shape parameters and visible developmental defects andasymmetries are extremely rare. Similarly, hybrid females havenormal heads that are indistinguishable from either of the par-ental species and again lack obvious defects in the vast majority ofcases. However, a significant proportion of F2 hybrid males (ap-proximately 25 percent) show various cranial abnormalities. Geneinteractions causing these abnormalities are apparently recessive,since they are expressed in haploid F2 males but are not visible inF1 hybrid females or F2 hybrid females produced by backcrossingto either parental strain. This feature provides an advantage forgenetic analysis because genetic regions involved in the pheno-types can be introgressed into a control genetic backgroundthrough females, and then gene interaction can be revealed bysetting females as virgins and examining their haploid re-combinant sons (since virgin females produce haploid eggs thatdevelop into males in this species).

Common abnormalities include cranial midline furrowing(clefting, Fig. 2C), lateral asymmetry (LA, Fig. 2D) and dorsal-ventralasymmetry (DVA, Fig. 2E). Other abnormalities include an abnormal“cranial pit” behind the antennal sockets (CP, Fig. 2F) and a “swollenhead” syndrome (SH, Fig. 2I). Morphometric and QTL analyses havebeen performed to characterize the phenotypes and reveal the ge-netic architecture of the DVA, LA, and clefting phenotypes.

4. Morphometric and QTL analysis of F2 hybrid male headshape differences and defects

QTL analysis was performed to gain an understanding of thegenetic architecture of the head shape differences of the twospecies and abnormalities found in the F2 hybrid males. In addi-tion to the identification of QTL of large effect (Table 3), QTL un-derlying epistatic interactions were also identified (Table 4). Theepistatic interaction revealed additional QTL in five of the eighthead phenotypes examined (those 7 listed in Table 1, plus cleft-ing), which include 26 newly found QTL not revealed in the majorQTL analysis. These interactions can be typed based on whetherthey occur between two major QTL (mQTL-mQTL), between amajor QTL and second conditional locus that only shows sig-nificant effects on the phenotype in interaction with the majorlocus (mQTL-cQTL), or between two loci that are not detected toaffect the phenotype additively, but do show an effect in interac-tion (epiQTL-epiQTL).

In general, the results indicate a rich web of interacting lociaffecting head morphology differences between the species.Linkage maps that indicate location of the major and epistatic QTLfor each trait are shown in Fig. 3 A–D for head measurementsnormalized to head length (MHW/HL, AOL/HL, OIO/HL, and MEYE/HL), whereas Fig. 4A–D maps QTL for dorsal ventral symmetrymeasurements (OIO/MIO and OIO/AIO) and cheek and cleftingQTL. In the following descriptions the terms G and V indicate an N.giraulti or an N. vitripennis allele, respectively, at a given locus.

4.1. Head width to length ratios

The basic head width to head length measurement (MHW/HL)revealed a single major QTL in the proximal region of Chr4

Table 2Significant correlation between phenotypes. Pearson's product moment correlations between phenotypes (above diagonal) and significant values (below diagonal) areshown for the 7 phenotypes for (a) N. giraulti, (b) N. vitripennis, and (c) F2 hybrids (þpo0.10, npo0.05, nnpo0.02, nnnpo0.005).

Measurement FEP/FE MEYE/HL N. giraulti AOL/HL OIO/AIO OIO/MIO

OIO/HL MHW/HL

FEP/FE �0.167 �0.3 �0.363 �0.251 �0.011 0.079MEYE/HL 0.535 0.620nn 0.843nnn �0.004 0.084 0.480þ

OIO/HL 0.259 0.01 0.839nnn 0.444þ 0.597nn 0.550n

MHW/HL 0.167 0 0 0.167 0.138 0.255AOL/HL 0.348 0.99 0.085 0.536 0.396 0.328OIO/AIO 0.968 0.757 0.015 0.61 0.129 0.708nnn

OIO/MIO 0.77 0.06 0.027 0.341 0.215 0.002N. vitripennisMeasurement FEP/FE MEYE/HL OIO/HL MHW/HL AOL/HL OIO/AIO OIO/MIOFEP/FE 0.472þ �0.566n 0.493þ 0.037 �0.641nnn �0.556n

MEYE/HL 0.065 �0.813nnn 0.945nnn �0.052 �0.649nn �0.476þ

OIO/HL 0.022 0 �0.696nnn 0.082 0.754nnn 0.732nnn

MHW/HL 0.052 0 0.003 �0.004 �0.705nnn �0.561n

AOL/HL 0.892 0.848 0.763 0.987 0.091 0.078OIO/AIO 0.007 0.006 0.001 0.002 0.738 0.853nnn

OIO/MIO 0.025 0.062 0.001 0.024 0.774 0F2 HybridsMeasurement FEP/FE MEYE/HL OIO/HL MHW/HL AOL/HL OIO/AIO OIO/MIOFEP/FE 0.025 �0.305nnn �0.244nnn �0.187nn �0.150þ �0.131þ

MEYE/HL 0.782 �0.344nnn 0.448nnn �0.151n �0.171n �0.043OIO/HL 0 0 0.438nnn 0.065 0.464nnn 0.610nnn

MHW/HL 0.001 0 0 �0.004 �0.085 �0.108AOL/HL 0.018 0.048 0.419 0.935 0.126þ �0.056OIO/AIO 0.061 0.025 0 0.262 0.097 0.632nnn

OIO/MIO 0.082 0.59 0 0.17 0.451 0

J.H. Werren et al. / Developmental Biology 415 (2016) 391–405396

(Fig. 3A). As expected, the G allele is associated with a smallerratio. This QTL accounts for accounts for �14% of F2 male phe-notypic variance, and the mean difference between the N. giraultiand N. vitripennis alleles accounts for a significant portion of thespecies difference (36%¼(1.43�1.39)/(1.48–1.37), Table 1). Thesame general regionwhere a major QTL for male-specific wing sizewas found and identified as involving cis-regulatory region arounddoublesex, a conserved sex determining gene (Loehlin et al.,2010a). However, that specific locus has not been shown to beinvolved in head shape.

In addition, an epistatic interaction occurs between markerregions on Chr3 and the distal end of Chr4 (Fig. 3A, Table 4). Thepresence of G alleles at these two loci induces a smaller meanMHW/HL ratio that is virtually the same as the N. giraulti parentalphenotype (1.37), whereas the other 3 genotypes are more similarto N. vitripennis (1.43, 1.45, 1.42 versus 1.47). Hence, in combinationthese three loci may account for the majority of the head ratiodifference between males of these two species. Testing this pro-position will involve genetic isolation of the respective QTL inreciprocal controlled genetic backgrounds (N. vitripennis and N.giraulti).

Antennal ocular height to head length ratio (AOL/HL) measuresthe relative contribution of the head length above the antennalsockets to full head length. Here we find a single major QTL(Fig. 3B), one on Chr1which accounts for �12% of phenotypicvariance and a web of three pairwise interactions between Chr2, 3,4, and 5. The proximity of markers involved on Chr2 and Chr4(Fig. 3B) suggest a complex interaction among four QTL. Suchcomplex epistasis can make identification of genes involved inmorphological differences between species challenging. But giventhat regions involved have been found, we can now reduce thecomplexity by selecting on interactions pairs in a controlled ge-netic background, which will reduce the complexity for positionalcloning.

OIO/HL measures the interocular distance at the central ocellusrelative to head length. In contrast to the previous phenotype, thisanalysis detected 4 major QTL (Chr2,3,4, and 5), but no epistatic

interactions (Fig. 3C, Table 3). In combination, these regions ac-count for 37.7 percent of the phenotypic variation among F2 males.For three of the mQTL, the G allele yields larger ratios, whereas theG allele for the region on Chr3 actually gives a smaller OIO/HL. Tworegions are shared with the previous width to length measure(MHW/HL). These involve the mQTL on Chr4, as well as the epiQTLfor MHW/HL on Chr3 which maps in the same vicinity at the OIO/HL major QTL. The GG genotype associated with this epiQTL in-duces a smaller MHW/HL ratio, as does the major QTL for the OIO/HL. Thus, these could reflect a single locus inducing similar effects,in one case epistatically and the other as a major additive effect.

MEYE/HL measures the contribution of eye width relative tohead length. Four major QTL were found on Chromosomes 2, 3, 4,and 5, and in each case the G allele induces a smaller MEYE ratio.Two regions are shared with OIO/HL, involving the mQTL on Chr2and Chr4 (Fig. 3D, Table 3). The fact that three other mQTL are notshared between them suggests that eye width and interocularwidth at the ocellar region are under different genetic controlmechanisms. We should note that the same proximal region onChr4 is involved in MHW/HL, MEYE/HL, and OIO/HL (Fig. 3A–C). Inthese cases the G allele results in a smaller ratio for the first twophenotypes and a larger for the last. Genetic dissection of thisregion could reveal whether they share a similar genetic under-pinning. To this end, we have already identified a segmental in-trogression line (SIL) that captures some head shape differences(see detailed description below). This line will be a starting pointfor positional cloning approaches that have already been suc-cessful in Nasonia (Loehlin and Werren, 2012; Lynch, 2015).

4.2. Dorsal–ventral comparisons

Two Dorsal–ventral measurements were analyzed, OIO/MIO(which measures InterOcular distance at the Ocelli relative to theMaximum InterOcular) and OIO/AIO, which measures the samedorsal region relative to InterOcular width at the Antennal sock-ets). Although they are similar phenotypes, the genetic archi-tecture for these two traits appears to be quite different (Fig. 4A-

Fig. 2. Hybrid Head Abnormalities. (A) Parental female N. vitripennis, (B) parental male N. giraulti, (C–I) Assortment of various F2 hybrid phenotypes. (C–C′′′) Midline furrowphenotype in increasing severity, (D–D′′′) laterally asymmetric phenotype in increasing severity, (E–E′′′) dorsal–ventral asymmetric phenotypes. (F–I) Other hybrid phe-notypes, (F) note severe medial indentation and eye shape, (G) note ocelli placement, (H) note antennal placement, (I) “swollen head” syndrome.

J.H. Werren et al. / Developmental Biology 415 (2016) 391–405 397

B). Analysis of OIO/MIO revealed three mQTL (Chr3, 4, and 5,Fig. 4A, Table 3), whereas the analysis of OIO/AIO detected a singlemQTL on Chr3 (mapping very close to the OIO/MIO mQTL on thatchromosome, Fig. 4B, Table 3), and a complex set of epistatic in-teractions involving 7-8 loci on four different chromosomes(Fig. 4B, Table 4). Complex four-way interactions are implied for amarker region on Chr1 with the proximal region on Chr4, Chr3 anda distal region on Chr4. By genetic isolation of components of

these interactions, the relationships can be defined more precisely.The dramatically different genetic architecture for these twophenotypes is consistent with complexity of head development.

It is noteworthy that the proximal region around chr4 containsa large number of mQTL and epiQTL affecting head shape (Cheek,AIO/HL, OIO/MIO, MEYE/HL), warranting further genetic dissectionof this region.

Table 3Significant major QTL using MapQTL. Shown are the marker, chromosome and LOD score, as well as percent variation explained by the QTL and mean values for the N. giraulti(G) and N. vitripennis (V) alleles. Range and mean and standard deviation among F2 is presented under the phenotype name. The direction of difference is also shown. AllLOD scores shown are significant at the po0.05 (*) or po0.01 (**) levels when corrected for multiple comparisons using a standard permutation test for interval mapping.

Trait mqm-mapping (marker (chr) – LOD % phenotypic variance explained Mean G allele Mean V allele Direction

MHW/HL 213-0.47 (4) – 5.20** 13.9 1.39 1.43 GoV1.25–1.6; 1.4270.05MEYE/HL P1-1.48 (2) – 5.38** 10.4 0.515 0.553 GoV0.38–0.65; 213-0.47 (4) – 5.16** 10.0 0.519 0.549 GoV0.5470.05 P1-0.76 (5) – 5.39** 13.3 0.551 0.517 G4VOIO/HL 323-0.65 (2) – 7.47** 12.1 0.873 0.814 G4V0.62–1.00; N15-0.89 (3) – 7.77** 12.5 0.820 0.867 GoV0.8470.06 307-0.25 (4) – 3.92** 6.1 0.859 0.827 G4V

407-0.7 (5) – 4.07** 7.0 0.824 0.863 GoVAOL/HL I30-1.11f (1) – 3.74** 11.9 0.672 0.713 GoV0.55–0.87; 0.6970.06OIO/AIO N16-0.89 (3) – 3.74** 9.8 0.901 0.971 GoV0.73–1.32; 1.4270.07OIO/MIO 28-1.41 (3) – 8.28** 15.3 0.934 0.975 GoV0.77–1.08; 91-1.22f (4) – 8.88** 16.5 0.975 0.934 G4V0.9670.05 297-1.2 (5) – 2.12* 4.7 0.944 0.966 GoVCheek (FEP/FE) 6-0.5 (1) – 3.96** 7.4 0.188 0.174 G4V0.06–0.27; D16-0.86 (3) – 10.0** 20.8 0.193 0.167 G4V0.1770.03 320-2.1f (4) – 2.50* 4.7 0.186 0.175 G4VCleft None0–1; 33.7%

J.H. Werren et al. / Developmental Biology 415 (2016) 391–405398

4.3. Clefting

Approximately 33% of F2 hybrids show abnormal midline fur-rowing. Expression of the trait varies from a short furrow adjacentto the central ocellus (simple eye) on the back of the head, to afurrow running across the head that fuses with the antennal cup(Fig. 2C). In normal crania, the ocelli “rest” above the surface of thecrania in association with an ocellar ridge (Fig. 2A–B).

The cleft phenotype was subjectively scored into two

Table 4Epistatic Interactions. Pairwise interactions are shown for different phenotypes, categvitripennis. Also shown is the log likelihood ratio (LLR) if an epistatic model is comparecomparisons (þpo0.10, npo0.05, nnpo0.02, nnnpo0.005 – see Section 2). Mean valvalues among all F2 males. Percent cleft for each genotype and all F2 is shown for this

Phenotype Type interaction QTL1 (chr) QTL2

MHW/HL epiQTL-epiQTL 83-026 (4) D16-01.25–1.6; 1.4270.05MEYE/HL None0.38–0.65; 0.5470.05OIO/HL None0.62–1.00; 0.8470.06AOL/HL epiQTL-epiQTL P4-146 (3) 18-1.10.55–0.87; 0.6970.06

epiQTL-epiQTL 330-1.1 (4) 209-0epiQTL-epiQTL 213-0.86 (4) 28-0.

OIO/AIO epiQTL-epiQTL D7-105 (4) O20-00.73–1.32; 1.4270.07

epiQTL-epiQTL P4-1.85 (4) O20-0epiQTL-epiQTL N9-0.65 (4) 30-1.1epiQTL-epiQTL I18-0.92 (3) 320-1epiQTL-epiQTL 28-1.02 (4) 28-1.3

OIO/MIO none0.77–1.08; 0.9670.05Cheek(FEP/FE) cQTL-mQTL 213-0.86 (4) D16-00.06–0.27; 0.1770.03

cQTL-mQTL 213-0.86 (4) 209-0mQTL-mQTL 209-0.98 (3) 30-1.2epiQTL-epiQTL 356-0.5 (3) 209-0epiQTL-epiQTL 76-1.1 (4) N16-0

Cleft epiQTL-epiQTL O20-0.95 (4) 34-0.4105nnn 0.0 37.5 38.10–1; 33.7%

epiQTL-epiQTL W7-0.97 (2) 297-1

categories, 0 (no clefting) or 1 (clefting ranging from moderateclefting restricted to the ocellar region, to clefts spanning ventrallyto the antennal sockets, see Fig. 2C). No major QTL for cleftingwere initially detected, which could reflect the fact that binarytraits violate the assumption of normality of the QTL program, orthat such abnormal development traits require epistatic interac-tions for their effect. A search for epistasis reveals two very strongpairwise interactions (Table 3, Fig. 4). Both interactions involve aregion on Chr5 with marker on either Chr4 or Chr2.

orized by type of interaction (see text); G indicates N. giraulti and V indicates N.d with an additive model (Chase et al. 1997) and the adjusted p-value for multipleues are presented for the 4 genotypes present, as well as the range, mean7SD ofphenotype.

(chr) LLR GG GV VG VV

86 (3) 7.1nn 1.37 1.43 1.45 1.42

2f (2) 8.9nnn 0.67 0.70 0.74 0.69

.68 (5) 6.2n 0.72 0.68 0.67 0.7089 (2) 5.4þ 0.74 0.69 0.68 0.69.59 (3) 11.7nnn 0.95 0.95 .089 0.97

.59 (3) 11.4nnn 0.94 0.95 0.88 0.971 (1) 7.0nn 0.95 0.93 0.90 0.9585 (2) 9.9nnn 0.95 0.88 0.94 0.95(2) 7.3nn 0.96 0.95 0.87 0.93

.92 (1) 9.3nnn 0.18 0.18 0.19 0.16

.98 (3) 7.7nn 0.19 0.18 0.19 0.163 (4) 7.2nn 0.19 0.19 0.18 0.16.87 (4) 8.1nnn 0.18 0.19 0.18 0.15.28 (2) 9.7nnn 0.16 0.19 0.18 0.1798 (5)

32.1

.21 (5) 37.4nnn 9.1 27.3 53.3 25.0

Fig. 3. QTL mapping of cranial shape measurements normalized to head length (a) MHW/HL, (b) AOL/HL, (c) OIO/HL, and (d) MEYE/HL. Potential QTL are marked with blackboxes, potential interacting loci are indicated with double headed arrows.

J.H. Werren et al. / Developmental Biology 415 (2016) 391–405 399

The individual effects of each locus on clefting percent are notsignificant (p¼0.85 for marker O20-0.95 (Chr4), p¼0.21 for 34-0.98 (Chr5), p¼0.33 for marker 297-1.21, and p¼1.00 for markerw7-0.97, FET). However, pairwise interactions reveal significanteffects on clefting percent. Marker 34-0.98 (Chr5) reveals a verystrong epistatic interaction with marker O20-0.95 (distal on Chr4).Individuals with GG genotype at these two markers show sig-nificantly reduced rates of clefting compared to the other threegenotypes (0% versus 34.3%, p¼0.017, FET). The second region ofChr5 (297-1.2) interacts with Chr2 marker region w7-0.97. In thiscase, VG individuals show elevated clefting compared to the othergenotypes. Although the marker regions are 34.5 cM apart on thelinkage map, it is possible that the two Chr5 regions represent thesame underlying locus.

Under the assumption of a single epiQTL on Chr5, we furtherinvestigated the possibility of a three-way interaction betweenthis Chr5 epiQTL and the two epiQTL on Chr2 and Chr4. To do thiswe used the marker (6-0.63) which falls between the two is pre-viously described epiQTL on Chr5 (34-0.98 and 297-1.2). Like theother clefting epiQTL, markers, 6-0.63 shows no effect on clefting

in isolation (p�1.00 FET, Table S2). However, a strong interactionamong marker regions on clefting percent is revealed when region6-0.63 is examined with the markers on Chr4 and Chr2 (Fig. 5,Table S2). The presence of the G allele at marker 6-0.63 (Chr5)strongly supresses clefting when at least one of the loci on Chr2 orChr4 also carries a G allele (0% clefting, Fig. 5, Table S2). In contrast,when individuals are G at the locus but V at both of the Chr2 and4 loci, clefting occurs at high rates (�33% clefting, Fig. 5, Table S2),indicative of strong non-additive effects. In contrast, if an in-dividual is V at the Chr5 epiQTL, the genotypes at the other twoloci do not significantly change the frequency of clefting (�25–39% clefting, Fig. 5). The pattern indicates a strong, but complex,epistatic interaction of the Chr5 region with those on Chr2 andChr4, with the G allele on Chr5 modulating epistatic interactions ofthe other two loci. Identifying the genes involved in this interac-tion will be an interesting topic of future research, which couldprovide insights into the complex genetic basis of abnormal cra-niofacial development.

Among all of the cranial defects identified here, clefting is theone that is most easily analogized to developmental abnormalities

Fig. 4. QTL mapping of cranial abnormalities and cheek size traits (a) OIO/MIO, (b) OIO/AIO, (c) Cleft, (d) Cheek (FEP/FE). Potential QTL are marked with black boxes, potentialinteracting loci are indicated with double headed arrows.

J.H. Werren et al. / Developmental Biology 415 (2016) 391–405400

common to humans. In addition to the similarity of the grossphenotype, the complex pattern of inheritance of these defects issimilar to that found for cleft lip/palette in humans. Thus, identi-fying the genes involved in clefting by positional cloning of theseepiQTL and evaluating their molecular interactions could havemedical relevance to craniofacial abnormalities in humans, andcould enhance our understanding of how developmental andmolecular processes can coordinate the growth of tissues of dis-tinct developmental origin, to mediate their fusion and ensuretheir integration into a functional, complex organ.

4.4. Cheek size comparisons

Cheek size (FEP/FE) shows a complex web of major QTL andepistatic interactions (Fig. 4D, Tables 3 and 4). Three major QTLregions are identified, on Chr1, Chr3 and Chr4. The major QTL onChr3 (marker D16–0.86) has a large additive effect, accountingalone for 20.8% of the phenotype. In addition there is a weaker

mQTL on Chr4 (marker 320-2.1f, 4.7% of variance) which interactsepistatically with marker 209-0.98 on Chr3. As this marker is only8 cM from the large effect mQTL on Chr3, it is possible this re-presents an epistatic interaction between this mQTL and the oneon Chr4. The same region on Chr3 (marker 209-0.89–D16-0.86)interacts epistatically with 213-0.86 (Chr4), which in turn interactsepistatically to marker D16-0.92 that occurs on Chr1 7.2 cM fromthe major QTL on that chromosome. Therefore, the pattern sug-gests direct and indirect interactions among the three major QTL,mediated in part by a cQTL on Chr4 (213-0.86). Additional epistaticinteractions occur on Chr2, Chr3, and Chr4 (see Fig. 2C). All theinteractions described above are significant even after correctingfor multiple comparisons.

Future research will focus on determining the interactionsmore precisely and identifying the genes involved. Towards thisend, segmental introgressions lines (SILs) containing the majorand epistatic QTL from one species introgressed into the other canbe used for finer scale mapping and positional cloning. Positional

Fig. 5. Three-way interaction on clefting probability. Strong epistatic interactionsare revealed between the epiQTL on Chr5 and two others on Chr 4 (blue) and Chr2(red). An individual that has the vitripennis (V) allele on Chr5 shows relatively highclefting percent and its genotype at the other two loci have only modest effects onclefting percent. However when giraulti (G) at the Chr5 epiQTL, then cleftingprobability is high only when V at both the other loci, and very low (zero percent)otherwise. A significantly higher clefting percent occurs when V at the other twoloci (FET p¼0.006), but this percent is not significantly different (ns) to the com-bined clefting when V at the Chr5 locus (p¼0.16). The result indicates that the Chr5locus is a strong modifier of non-additive interactions at the other loci.

J.H. Werren et al. / Developmental Biology 415 (2016) 391–405 401

cloning of QTL has been previously demonstrated in N. vitripennis(Loehlin et al., 2010d; Loehlin and Werren, 2012; Niehuis et al.,2013), by using the advantages of male haploid genetics. Given thelarge effect on the mQTL on Chr3, this will likely be a primarytarget of analysis.

4.5. Lateral Asymmetry (LA)

It is apparent from visual observation that some F2 hybridmales show extreme forms of lateral asymmetry not seen in par-ental males (see Fig. 2D-E). Asymmetries are apparent in cheeksize, in interocular distance from the midline, and in eye shape.We have only begun to characterize this feature. As an initial lookat eye asymmetry, we measured eye height (distance from theinner to outer edge of the eye at the maximal head width line) forthe left eye (LEW) and right eye (REW). There are a number ofstatistical methods for measuring lateral asymmetry (Klingenberget al., 2002; Palmer and Strobeck, 1986; Van Dongen, 2007). Threemeasures of asymmetry were examined, difference (LEW-REW),absolute value of the difference, and normalized mean difference((LEW�REW)/(LEWþREW)/2).

Approximately 15% of F2 hybrid males had normalized asym-metry of greater than 20% in eye width and 5% of males had agreater than 25% difference in eye width. The genetic and devel-opmental basis of lateral asymmetry is a topic of wide interest(Dongen, 2006; Klingenberg, 2010; Palmer and Strobeck, 1997). Amajor debate concerns whether lateral asymmetry (either fluctu-ating asymmetry or antisymmetry) in hybrids is due to specificgenotypic interactions or to a general destabilization of develop-ment in hybrids (Clarke, 1997; Moller and Thornhill, 1997; Palmerand Strobeck, 1997). Genetic dissection of this trait could revealthe extent to which these asymmetries are due to interactionsamong specific loci or to decanalization of the trait in hybridsthrough diffuse epistatic interactions and general disruption ofgene regulation in hybrids. It should be noted that no clear lateralasymmetries have been observed in other body regions (e.g.,wings, legs, abdomen), indicating that the asymmetries are likelynot due to some global disruption of developmental robustness.

5. Isolation of a head shape locus using a Segmental In-trogression Line (SIL)

Segmental introgressions are lines in which a chromosomalregion from one species in backcrossed into the genetic back-ground of another. These have been used effectively for char-acterization of phenotypic differences between Nasonia species(Desjardins et al., 2010; Hoedjes et al., 2014), as well as positionalcloning of QTL involved in development (Loehlin et al., 2010d;Loehlin and Werren, 2012). Here we describe head shape differ-ences found in a SIL where the pericentric region of Chr2 from theN. giraulti genome has been introgressed into a N. vitripennisbackground. Mothers of the SIL shown here are homozygous forthe N. giraulti region, resulting in males that are hemizygously N.giraulti at this locus. These males have significantly wider headsand eyes that appear to bulge out (Fig. 6B).

Male heads of this introgression are strongly statistically sig-nificantly different from wild-type N. vitripennis males in threecritical measurement ratios: MEYE/HL, OIO/HL, and AOL/HL, andmarginally significantly larger in cheek size (Fig. 6). This indicatesthat the SIL heads have much wider eyes, without significantlyincreasing overall head width. In fact, the interocular distance atthe ocelli as well as the ocellus to antenna midline length is sig-nificantly smaller than wild type (Fig. 6C), suggesting a trade-offwhere a smaller head capsule accounts for larger eyes that keepsoverall head size consistent. A similar trade off between intero-cular cuticle and eye size was seen between closely related Dro-sophila species (Arif et al., 2013). Altogether, large eyes in combi-nation with a decreased interocular region creates a “bulging eye”appearance. The presence of larger cheeks in the introgressionindicates either a single gene affects head shape and cheek size inthe centromeric region, or that two or more genes affecting headtraits are tightly linked in this region.

6. Discussion

In this manuscript we have introduced the species pair of N.vitripennis and N. giraulti as a novel, and uniquely powerful modelfor understanding the genetic and developmental basis of cra-niofacial defects and morphological diversity. There are distinct,non-overlapping aspects contributing to male head shape, both inseveral measures of the ratio of head length to head width, and inthe size of the lateral cheeks. As demonstrated by the largenumber of distinct QTL found to be associated with the differentaspects of head height/width ratio, differences in multiple genesmust underlie the expression of this species difference. Many ofthe QTL identified for this trait are modified by epistatic interac-tions. While a single major QTL explains a very large portion of the“cheekiness” trait in N. giraulti, multiple genes acting within andoutside the context of epistasis contribute to this character. Fur-ther genetic and molecular analysis of the head shape differenceswill contribute to the growing number of studies examining theevolutionary genetics of shape and form. In particular, our analyseswill complement similar approaches in understanding the pat-terning differences between closely (Arif et al., 2013; Posnien et al.,2012) and distantly (Carr et al., 2005) related flies with subtle andextreme, respectively, head shape differences.

More relevant to the topic at hand, the haploid male offspringof hybrid females display numerous diverse, but well-defined,cranial defects. These include clefts along the bilateral midline ofthe head, asymmetry along the DV axis, and lateral asymmetry.None of the traits are seen in the parental species or in hybridfemales. These anomalies are largely governed by epistatic inter-actions among 3 or more loci. Interestingly, it is well known thathuman craniofacial abnormalities are governed by complex

Fig. 6. A segmental introgression captures a head shape locus. (A) A fully wild-type N. vitripennis male head compared to (B) a SIL male that has the N. giraulti pericentricregion of Chr2 introgressed into a N. vitripennis background. (C) Key measurements show that the SIL differs significantly from wild type in MEYE/HL, OIO/HL, AOL/HL andFEP/FE ratios. Triple asterisks represent measurements differing significantly between the two species at po0.001, double asterisks represent po0.01, and single asterisksrepresent po0.05. See Table 1 and Fig. 1 for details on the meaning of abbreviations.

J.H. Werren et al. / Developmental Biology 415 (2016) 391–405402

epistatic interactions, which make understanding the underlyingorigin and etiology of the Nasonia defects of considerable interest.

Insect models of head development, and associated abnorm-alities, can be valuable in understanding craniofacial abnormalitiesseen in humans for many reasons. First, cephalization is an an-cestral feature of bilaterally symmetric animals so there is acommon evolutionary basis (homology) for comparison. In addi-tion, the fusion and integration of multiple tissue types and lat-erally symmetric imaginal discs to form a coherent functionalstructure in insects can be analogized to the formation of thevertebrate cranial case and integration of the associated sensoryand alimentary structures. Following from this, the range of de-fects we see in the hybrids can also be analogized to those seen inhuman craniofacial abnormalities. In particular, facial clefts are themost common birth defect in human beings, and result from failedinteractions among components along the facial midline. Theclefting phenotype we observe in a large proportion of F2 hybridmales appears to be analogous to the vertebrate anomalies such ascleft lip and palette syndromes, and possibly could rely onhomologous molecules and/or cellular mechanisms.

A relative handful of genes have been identified that contributesignificantly to CL/P in humans and mammalian model systems.Examples of these include the transcription factors IRF6 (Kondoet al., 2002; Leslie et al., 2013), GRH3 (Peyrard-Janvid et al., 2014),DLX6 (Wolf et al., 2014), FOXE (Ludwig et al., 2014), and PAX7(Leslie et al., 2015); the guidance signaling molecule Netrin1 (Le-slie et al., 2015); the metalloprotease ADAMTS20 (Wolf et al.,2015); and the FGF signaling receptor FGFR2 (Leslie et al., 2015).Almost all of the genes so far identified (with the notable excep-tion of IRF6 (Nehyba et al., 2009)) have clear orthologs in Nasonia(JAL personal observation), holding out the possibility of a con-served molecular basis between CL/P and the facial clefting we seein Nasonia hybrids.

Another human craniofacial anomaly that may be enlightened byNasonia is craniosynostosis, which is a premature fusion of cranialsutures, often causing facial malformations and brain problems(Wilkie, 1997). While many of the identified mutations and syn-dromes show autosomal dominant inheritance or haploinsufficiency

(El Ghouzzi et al., 1997; Howard et al., 1997) (which we do notobserve in Nasonia), recessive inheritance consistent with a complexgenetic basis (Melville et al., 2010) and epistasis (Sharma et al.,2013) are also observed. Since our understanding of the develop-mental basis for most of the cranial abnormalities we observe is stilllacking, we cannot exclude the possibility of premature fusion ordifferentiation of cuticular tissue underlying one or more of thephenotypes in Nasonia hybrids.

Due to challenges in fine-scale resolution of the RAPD map toNasonia linkage map generated using genome data (Desjardinset al., 2013), it is not currently useful to speculate whetherhomologs of the known CL/P or craniosynostosis genes are thecausative agents of some of the phenotypes seen in the Nasoniahybrids (i.e. correspond to Nasonia craniofacial QTL). The mainvalue of this system is to discover novel gene interactions that arevery difficult to detect even in GWAS studies (Hallgrimsson et al.,2014). Future goals will be to identify the genes involved by po-sitional cloning, which has been successful for other genes in-volved in developmental differences between the species (Loehlinet al., 2010d, Loehlin and Werren, 2012).

7. The Nasonia toolbox for functional genetic and genomicanalysis

Nasonia is an emerging model genetic and genomic system forseveral reasons (Lynch, 2015; Werren and Loehlin, 2009). First, thegenomic resources required for modern genetics are present in thespecies complex, and allow the plausible goal of identifying themutations that are correlated with species differences and ab-normal head capsule development, especially given the success inidentifying the bases for interspecies differences in wing shape(Loehlin et al., 2010d; Loehlin and Werren, 2012). Nasonia is alsoamenable to functional analyses, for example using RNAi for geneknockdown (Lynch and Desplan, 2006), and additional tools (Ro-senberg et al., 2014) for disrupting or altering gene function areunder development. Tools for interpreting the effects of functionalapproaches such as multiplex fluorescent in situ hybridization

J.H. Werren et al. / Developmental Biology 415 (2016) 391–405 403

(Ozuak et al., 2014), immunohistochemistry (Lynch and Desplan,2010), and live imaging (Buchta et al., 2013) are readily applicableto this species.

Another unique and distinguishing feature of the Nasonia sys-tem is the ability to produce viable hybrid haploid males, greatlyfacilitating genetic analyses, and allowing the detection andcharacterization of complex gene interactions. Such analyses aremore laborious and often impractical in other species primarilydue to the exponentially increasing number of progeny needed todetect homozygous epistatic interactions in diploids as the num-ber of genes involved increases. The combination of the well de-veloped developmental genetics tools, along with the awesomepower of haplo-diploid genetics make the promise of Nasoniaunparalleled among emerging model systems.

Segmental Introgression Lines (SILs) are particularly useful foraccelerating discovery of genes involved in species differences.SILs are lines in which a chromosomal region from one species inbackcrossed into the genetic background of another. These havebeen used effectively for characterization of phenotypic differ-ences between Nasonia species (Desjardins et al., 2010; Hoedjeset al., 2014; Loehlin et al., 2010a), as well as to jumpstart positionalcloning of QTL involved in development (Loehlin et al., 2010d;Loehlin and Werren, 2012).

A battery of SILs have been created with segments from N.giraulti in a N. vitripennis genetic background ((Desjardins et al.,2013), Werren unpublished), which cover �60% of the geneticmap of Nasonia, and additional ones are being generated routinely.Above we have described one example of a SIL which contains oneor more QTL affecting head shape. The battery of SILs can bescreened from other QTL for head shape, particularly those in re-gions hypothesized to contain QTL. In addition, crosses betweenSILs can be screened for epistatic effects on head shape, includingthose which could uncover epiQTL involved in head abnormalities.Therefore, the introgression and SIL method has promise for de-tecting QTL and epistatic interactions useful for cloning genes in-volved in head shape.

8. Future directions

The QTL analyses presented here were performed over 15 yearsago using state-of-the-art genotyping tools (RAPDs) available atthat time. We recognize that application of genome wide geno-typing methods (such as multiplex shotgun genotyping (Andol-fatto et al., 2011) or whole genome resequencing), coupled withRNA sequencing and RNA interference knockdowns, will provide amuch richer understanding of head shape regulation (and mis-regulation) in hybrids. Our goal here is to indicate with these olderanalyses, the rich potential of the system as an insect model forcranial development and mis-regulation due to epistatic interac-tions. However the basic crossing and phenotypic analysis strate-gies used here will be maintained, and with time we believe thevast majority of the genetic changes underlying the morphologicaldifferentiation and developmental abnormalities can be identifiedto single gene resolution, as shown for other morphological phe-notypes in Nasonia.

Much work remains to understand how the defects and mor-phological differences between species arise in the course of de-velopment. Head morphology is established in the eye-antennaldisc during larval and pupal stages. Numerous signaling pathways(e.g., BMP, EGF, Hedgehog, Notch, WNT (Royet and Finkelstein,1997; Shyamala and Bhat, 2002; Weasner and Kumar, 2013; Wonet al., 2015)) and transcription factors (e.g., orthodenticle (Finkel-stein and Perrimon, 1991), homothorax, defective proventriculus(Kiritooshi, 2014 #179), odd paired (Lee et al., 2007) and the hoxgenes labial and deformed (Stultz et al., 2012)) have been shown to

be crucial for proper patterning and proportioning of the Droso-phila head discs. It is tempting to look for these candidate genes inour QTL analysis. For example, The Hox cluster occurs in thepericentric region of chromosome 4, roughly corresponding theregion containing mapped gene Nv-s6k and the visible markeror123 (Fig. 3). Major QTL for a number of head shape phenotypesmap to this region, including for MHW/HL, OIO/HL, MEYE/HL, ANDOIO/MIO. However, many other genes occur in the region as well,including Nv-doublesex, the sex-determining locus which has beenshown to affect wing size differences between Nasonia species(Loehlin and Werren, 2012), and Nv-s6k, a known growth regulatorin the insulin signaling pathway. Therefore, it would be prematureto conclude that Hox genes are involved in these phenotypes, al-though they remain viable candidates. Once a set of candidategenes are narrowed by finer-scale mapping, taking a candidategene approach using larval RNAi, in situ hybridization, and im-munohistochemistry will help to improve our understanding ofthe genetic bases of patterning differences between the two spe-cies, and the origin of developmental abnormalities seen in thehybrid males.

Research points to a crucial role of the peripodial membrane inregulating the morphogenesis of the head capsule (Agnes et al.,1999; Gibson and Schubiger, 2001; Lee et al., 2007; Martin-Blancoet al., 2000; Zeitlinger and Bohmann, 1999). This process will alsobe illuminated by the use of live imaging of the growth and fusionof the discs. Protocols for the in vitro propagation, imaging, andanalysis of Drosophila imaginal discs have been developed (Aldazet al., 2010), and should be readily applicable to Nasonia discs.Comparison of the development of the parental and hybrid maleswill be quite informative in understanding the cellular basis of thedevelopmental abnormalities.

Another issue that needs resolved is the relationship betweenthe morphology of the head capsule and that of the internalstructures, such as the brain, glands and musculature. The adap-tive value, if any, of the novel head shape of the N. giraulti males isnot yet known. The shape itself may be adaptive, perhaps by fa-cilitating intra-host mating, a behavior much more prevalent in N.giraulti compared to the other Nasonia species (Drapeau andWerren, 1999).

Alternatively, or additionally, the head shape differences in N.giraulti may represent accommodations to changes in morphologyof internal structures, such as the brain, glands, or musculature. Ineither case, it is likely that external and internal morphologies co-evolve to maintain compatibility, since imaginal discs generallycan develop correct morphology autonomously and ectopically inthe absence of the normal internal head anlagen (Gibson andSchubiger, 2001). It will be interesting to explore whether some ofthe craniofacial defects we see are due to developmental in-compatibilities between internal and external structures.

While parallels between craniofacial defects with those ofvertebrates/ humans can justifiably be drawn, the disrupted de-velopmental process underlying the morphological abnormalitiescan also be models for basic processes at the cell and tissue level.For example, if the midline furrow stems from the failure in fusionof the epithelia of the two lateral eye antennal discs, this couldserve as a general model for epithelial interactions. While tradi-tional model organisms provide powerful means to dissect suchprocesses, the typical approaches in these organisms focus onmutations of large effect that act in isolation. Such mutations areunlikely to play important roles in natural populations, makingmodels such as Nasonia, where complex, subtle, interactionsamong multiple loci can be studied in depth, increasinglyimportant.

J.H. Werren et al. / Developmental Biology 415 (2016) 391–405404

Acknowledgments

We thank Celina Kennedy for head pictures and measurements,Rachel Edwards for assistance in figure preparation, Kevin Chaseand Gordon Lark for assistance with EPISTAT, R. Page for colla-borations on RAPD mapping in Nasonia, and the HumboldtFoundation for a Feodor Lynen Grant to J. Gadau. We also thankAisha Siebert and Zichao Yan for assistance, and Nico Posnien forsuggesting the use of Congo Red for cuticle imaging. This work wassupported by grants to JH Werren (NSF DEB-9707665, NSF DEB1257053, and NIH R01 GM70026) and by a University of Illinois atChicago Campus Research Board Pilot Grant (LYNCH 2013) to JAL.

Appendix A. Supplementary material

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.ydbio.2015.12.022.

References

Agnes, F., Suzanne, M., Noselli, S., 1999. The Drosophila JNK pathway controls themorphogenesis of imaginal discs during metamorphosis. Development 126,5453–5462.

Aldaz, S., Escudero, L.M., Freeman, M., 2010. Live imaging of Drosophila imaginaldisc development. Proc. Natl. Acad. Sci. USA 107, 14217–14222.

Andolfatto, P., Davison, D., Erezyilmaz, D., Hu, T.T., Mast, J., Sunayama-Morita, T.,Stern, D.L., 2011. Multiplexed shotgun genotyping for rapid and efficient geneticmapping. Genome Res. 21, 610–617.

Appel, T.A., 1987. The Cuvier-Geoffroy Debate: French Biology in the Decades beforeDarwin. Oxford University Press, New York.

Arif, S., Hilbrant, M., Hopfen, C., Almudi, I., Nunes, M.D., Posnien, N., Kuncheria, L.,Tanaka, K., Mitteroecker, P., Schlotterer, C., McGregor, A.P., 2013. Genetic anddevelopmental analysis of differences in eye and face morphology betweenDrosophila simulans and Drosophila mauritiana. Evol. Dev. 15, 257–267.

Attanasio, C., Nord, A.S., Zhu, Y., Blow, M.J., Li, Z., Liberton, D.K., Morrison, H., Plaj-zer-Frick, I., Holt, A., Hosseini, R., Phouanenavong, S., Akiyama, J.A., Shoukry, M.,Afzal, V., Rubin, E.M., FitzPatrick, D.R., Ren, B., Hallgrimsson, B., Pennacchio, L.A.,Visel, A., 2013. Fine tuning of craniofacial morphology by distant-acting en-hancers. Science 342, 1241006.

Beukeboom, L., Desplan, C., 2003. Nasonia. Curr. Biol. 13, R860.Breeuwer, J.A.J., Werren, J.H., 1995. Hybrid breakdown between two haplodiploid

species-the role of nuclear and cytoplasmic genes. Evolution 49, 705–717.Buchta, T., Ozuak, O., Stappert, D., Roth, S., Lynch, J.A., 2013. Patterning the dorsal-

ventral axis of the wasp Nasonia vitripennis. Dev. Biol. 381, 189–202.Carr, M., Hurley, I., Fowler, K., Pomiankowski, A., Smith, H.K., 2005. Expression of

defective proventriculus during head capsule development is conserved inDrosophila and stalk-eyed flies (Diopsidae). Dev. Genes Evol. 215, 402–409.

Carter, C.O., Evans, K., Coffey, R., Roberts, J.A., Buck, A., Roberts, M.F., 1982. A familystudy of isolated cleft palate. J. Med. Genet. 19, 329–331.

Chase, K., Adler, F.R., Lark, K.G., 1997. Epistat: A computer program for identifyingand testing interactions between pairs of quantitative trait loci. Theor. Appl.Genet. 94, 724–730.

Churchill, G.A., Doerge, R.W., 1994. Empirical threshold values for quantitative traitmapping. Genetics 138, 963–971.

Clarke, G.M., 1997. The genetic and molecular basis of developmental stability: theLucilia story. Trends Ecol. Evol. 12, 89–91.

Darling, D.C., Werren, J.H., 1990. Biosystematics of Nasonia (Hymenoptera, Pter-omalidae)-2 New Species Reared from Birds Nests in North-America. Ann.Entomol. Soc. Am. 83, 352–370.

Desjardins, C.A., Gadau, J., Lopez, J.A., Niehuis, O., Avery, A.R., Loehlin, D.W., Ri-chards, S., Colbourne, J.K., Werren, J.H., 2013. Fine-scale mapping Nasoniagenome chromosomes using a high-density genotyping microarray. G3 3,205–215.

Desjardins, C.A., Perfectti, F., Bartos, J.D., Enders, L.S., Werren, J.H., 2010. The geneticbasis of interspecies host preference differences in the model parasitoid Na-sonia. Heredity 104, 270–277.

Dixon, M.J., Marazita, M.L., Beaty, T.H., Murray, J.C., 2011. Cleft lip and palate: un-derstanding genetic and environmental influences. Nat. Rev. Genet. 12,167–178.

Dongen, S.V., 2006. Fluctuating asymmetry and developmental instability in evo-lutionary biology: past, present and future. J. Evol. Biol. 19, 1727–1743.

Drapeau, M.D., Werren, J.H., 1999. Differences in mating behaviour and sex ratiobetween three sibling species of Nasonia. Evol. Ecol. Res. 1, 223–234.

El Ghouzzi, V., Lajeunie, E., Le Merrer, M., Cormier, V., Renier, D., Munnich, A.,Bonaventure, J., 1997. TWIST mutations disrupting the b-HLH domain arespecific to Saethre-Chotzen syndrome. Am. J. Hum. Genet. 61, A332–A332.

Fakhouri, W.D., Rahimov, F., Attanasio, C., Kouwenhoven, E.N., Ferreira De Lima, R.L.,

Felix, T.M., Nitschke, L., Huver, D., Barrons, J., Kousa, Y.A., Leslie, E., Pennacchio,L.A., Van Bokhoven, H., Visel, A., Zhou, H., Murray, J.C., Schutte, B.C., 2014. Anetiologic regulatory mutation in IRF6 with loss- and gain-of-function effects.Hum. Mol. Genet. 23, 2711–2720.

Finkelstein, R., Perrimon, N., 1991. The molecular-genetics of head development inDrosophila-Melanogaster. Development 112, 899–912.

Gadau, J., Page, R.E., Werren, J.H., 1999. Mapping of hybrid incompatibility loci inNasonia. Genetics 153, 1731–1741.

Gadau, J., Page, R.E., Werren, J.H., 2002. The genetic basis of the interspecific dif-ferences in wing size in Nasonia (Hymenoptera; Pteromalidae): Major quanti-tative trait loci and epistasis. Genetics 161, 673–684.

Gibson, M.C., Schubiger, G., 2001. Drosophila peripodial cells, more than meets theeye? Bioessays 23, 691–697.

Greene, R.M., Pisano, M.M., 2010. Palate morphogenesis: current understanding andfuture directions. Birth Defects Res. C Embryo Today 90, 133–154.

Gross, J.B., Krutzler, A.J., Carlson, B.M., 2014. Complex craniofacial changes in blindcave-dwelling fish are mediated by genetically symmetric and asymmetric loci.Genetics 196, 1303–1319.

Hallgrimsson, B., Mio, W., Marcucio, R.S., Spritz, R., 2014. Let’s face it – complextraits are just not that simple. PLoS Genet. 10, e1004724.

Hide, T., Hatakeyama, J., Kimura-Yoshida, C., Tian, E., Takeda, N., Ushio, Y., Shiroishi,T., Aizawa, S., Matsuo, I., 2002. Genetic modifiers of otocephalic phenotypes inOtx2 heterozygous mutant mice. Development 129, 4347–4357.

Hoedjes, K.M., Smid, H.M., Vet, L.E.M., Werren, J.H., 2014. Introgression study re-veals two quantitative trait loci involved in interspecific variation in memoryretention among Nasonia wasp species. Heredity 113, 542–550.

Howard, T.D., Paznekas, W.A., Green, E.D., Chiang, L.C., Ma, N., Ortiz de Luna, R.I.,Garcia Delgado, C., Gonzalez-Ramos, M., Kline, A.D., Jabs, E.W., 1997. Mutationsin TWIST, a basic helix-loop-helix transcription factor, in Saethre-Chotzensyndrome. Nat. Genet. 15, 36–41.

Klingenberg, C.P., 2010. Evolution and development of shape: integrating quanti-tative approaches. Nat. Rev. Genet. 11, 623–635.

Klingenberg, C.P., Barluenga, M., Meyer, A., 2002. Shape analysis of symmetricstructures: quantifying variation among individuals and asymmetry. Evolution56, 1909–1920.

Kondo, S., Schutte, B.C., Richardson, R.J., Bjork, B.C., Knight, A.S., Watanabe, Y., Ho-ward, E., de Lima, R.L., Daack-Hirsch, S., Sander, A., McDonald-McGinn, D.M.,Zackai, E.H., Lammer, E.J., Aylsworth, A.S., Ardinger, H.H., Lidral, A.C., Pober, B.R.,Moreno, L., Arcos-Burgos, M., Valencia, C., Houdayer, C., Bahuau, M., Moretti-Ferreira, D., Richieri-Costa, A., Dixon, M.J., Murray, J.C., 2002. Mutations in IRF6cause Van der Woude and popliteal pterygium syndromes. Nat. Genet. 32,285–289.

Lark, K.G., Chase, K., Adler, F., Mansur, L.M., Orf, J.H., 1995. Interactions betweenquantitative trait loci in soybean in which trait variation at one locus is con-ditional upon a specific allele at another. Proc. Natl. Acad. Sci. USA 92,4656–4660.

Lee, H., Stultz, B.G., Hursh, D.A., 2007. The Zic family member, odd-paired, regulatesthe Drosophila BMP, decapentaplegic, during adult head development. Devel-opment 134, 1301–1310.

Leslie, E.J., Standley, J., Compton, J., Bale, S., Schutte, B.C., Murray, J.C., 2013. Com-parative analysis of IRF6 variants in families with Van der Woude syndromeand popliteal pterygium syndrome using public whole-exome databases.Genet. Med. 15, 338–344.

Leslie, E.J., Taub, M.A., Liu, H., Steinberg, K.M., Koboldt, D.C., Zhang, Q., Carlson, J.C.,Hetmanski, J.B., Wang, H., Larson, D.E., Fulton, R.S., Kousa, Y.A., Fakhouri, W.D.,Naji, A., Ruczinski, I., Begum, F., Parker, M.M., Busch, T., Standley, J., Rigdon, J.,Hecht, J.T., Scott, A.F., Wehby, G.L., Christensen, K., Czeizel, A.E., Deleyiannis, F.W., Schutte, B.C., Wilson, R.K., Cornell, R.A., Lidral, A.C., Weinstock, G.M., Beaty,T.H., Marazita, M.L., Murray, J.C., 2015. Identification of functional variants forcleft lip with or without cleft palate in or near PAX7, FGFR2, and NOG by tar-geted sequencing of GWAS loci. Am. J. Hum. Genet. 96, 397–411.

Lidral, A.C., Moreno, L.M., 2005. Progress toward discerning the genetics of cleft lip.Curr. Opin. Pediatr. 17, 731–739.

Loehlin, D.W., Enders, L.S., Werren, J.H., 2010a. Evolution of sex-specific wing shapeat the widerwing locus in four species of Nasonia. Heredity 104, 260–269.

Loehlin, D.W., Oliveira, D.C., Edwards, R., Giebel, J.D., Clark, M.E., Cattani, M.V., vande Zande, L., Verhulst, E.C., Beukeboom, L.W., Munoz-Torres, M., Werren, J.H.,2010d. Non-coding changes cause sex-specific wing size differences betweenclosely related species of Nasonia. PLoS Genet. 6, e1000821.

Loehlin, D.W., Werren, J.H., 2012. Evolution of shape by multiple regulatory changesto a growth gene. Science 335, 943–947.

Ludwig, K.U., Bohmer, A.C., Rubini, M., Mossey, P.A., Herms, S., Nowak, S., Reutter,H., Alblas, M.A., Lippke, B., Barth, S., Paredes-Zenteno, M., Munoz-Jimenez, S.G.,Ortiz-Lopez, R., Kreusch, T., Hemprich, A., Martini, M., Braumann, B., Jager, A.,Potzsch, B., Molloy, A., Peterlin, B., Hoffmann, P., Nothen, M.M., Rojas-Martinez,A., Knapp, M., Steegers-Theunissen, R.P., Mangold, E., 2014. Strong association ofvariants around FOXE1 and orofacial clefting. J. Dent. Res. 93, 376–381.

Lynch, J.A., 2015. The expanding genetic toolbox of the wasp Nasonia vitripennisand its relatives. Genetics 199, 897–904.

Lynch, J.A., Desplan, C., 2006. A method for parental RNA interference in the waspNasonia vitripennis. Nat. Protoc. 1, 486–494.

Lynch, J.A., Desplan, C., 2010. Novel modes of localization and function of nanos inthe wasp Nasonia. Development 137, 3813–3821.

Martin-Blanco, E., Pastor-Pareja, J.C., Garcia-Bellido, A., 2000. JNK and dec-apentaplegic signaling control adhesiveness and cytoskeleton dynamics duringthorax closure in Drosophila. Proc. Natl. Acad. Sci. USA 97, 7888–7893.

J.H. Werren et al. / Developmental Biology 415 (2016) 391–405 405

McKusick, V.A., 1994. Mendelian Inheritance in Man: A Catalog of Human Genesand Genetic Disorders, 11 ed. Johns Hopkins University Press, Baltimore.

Melville, H., Wang, Y., Taub, P.J., Jabs, E.W., 2010. Genetic basis of potential ther-apeutic strategies for craniosynostosis. Am. J. Med. Genet. A 152A, 3007–3015.

Moczek, A.P., Sears, K.E., Stollewerk, A., Wittkopp, P.J., Diggle, P., Dworkin, I., Ledon-Rettig, C., Matus, D.Q., Roth, S., Abouheif, E., Brown, F.D., Chiu, C.H., Cohen, C.S.,Tomaso, A.W., Gilbert, S.F., Hall, B., Love, A.C., Lyons, D.C., Sanger, T.J., Smith, J.,Specht, C., Vallejo-Marin, M., Extavour, C.G., 2015. The significance and scope ofevolutionary developmental biology: a vision for the 21st century. Evol. Dev. 17,198–219.

Moller, A.P., Thornhill, R., 1997. Developmental instability is heritable – reply. J.Evol. Biol. 10, 69–76.

Nehyba, J., Hrdlickova, R., Bose, H.R., 2009. Dynamic evolution of immune systemregulators: the history of the interferon regulatory factor family. Mol. Biol. Evol.26, 2539–2550.

Niehuis, O., Buellesbach, J., Gibson, J.D., Pothmann, D., Hanner, C., Mutti, N.S., Jud-son, A.K., Gadau, J., Ruther, J., Schmitt, T., 2013. Behavioural and genetic ana-lyses of Nasonia shed light on the evolution of sex pheromones. Nature 494,345–348.

Ozuak, O., Buchta, T., Roth, S., Lynch, J.A., 2014. Dorsoventral polarity of the Nasoniaembryo primarily relies on a BMP gradient formed without input from Toll.Curr. Biol. 24, 2393–2398.

Palmer, A.R., Strobeck, C., 1986. Fluctuating asymmetry-measurement, analysis,patterns. Annu. Rev. Ecol. Syst. 17, 391–421.

Palmer, A.R., Strobeck, C., 1997. Fluctuating asymmetry and developmental stabi-lity: heritability of observable variation vs. heritability of inferred cause. J. Evol.Biol. 10, 39–49.

Peyrard-Janvid, M., Leslie, E.J., Kousa, Y.A., Smith, T.L., Dunnwald, M., Magnusson,M., Lentz, B.A., Unneberg, P., Fransson, I., Koillinen, H.K., Rautio, J., Pegelow, M.,Karsten, A., Basel-Vanagaite, L., Gordon, W., Andersen, B., Svensson, T., Murray,J.C., Cornell, R.A., Kere, J., Schutte, B.C., 2014. Dominant mutations in GRHL3cause Van der Woude Syndrome and disrupt oral periderm development. Am. J.Hum. Genet. 94, 23–32.

Posnien, N., Hopfen, C., Hilbrant, M., Ramos-Womack, M., Murat, S., Schonauer, A.,Herbert, S.L., Nunes, M.D., Arif, S., Breuker, C.J., Schlotterer, C., Mitteroecker, P.,McGregor, A.P., 2012. Evolution of eye morphology and rhodopsin expression inthe Drosophila melanogaster species subgroup. PLoS One 7, e37346.

Prescott, N.J., Winter, R.M., Malcolm, S., 2001. Nonsyndromic cleft lip and palate:complex genetics and environmental effects. Ann. Hum. Genet. 65, 505–515.

Rosenberg, M.I., Brent, A.E., Payre, F., Desplan, C., 2014. Dual mode of embryonicdevelopment is highlighted by expression and function of Nasonia pair-rulegenes. Elife 3, e01440.

Royet, J., Finkelstein, R., 1997. Establishing primordia in the Drosophila eye-an-tennal imaginal disc: the roles of decapentaplegic, wingless and hedgehog.Development 124, 4793–4800.

Sharma, V.P., Fenwick, A.L., Brockop, M.S., McGowan, S.J., Goos, J.A., Hoogeboom, A.J., Brady, A.F., Jeelani, N.O., Lynch, S.A., Mulliken, J.B., Murray, D.J., Phipps, J.M.,Sweeney, E., Tomkins, S.E., Wilson, L.C., Bennett, S., Cornall, R.J., Broxholme, J.,Kanapin, A., Whole-Genome Sequences, C., Johnson, D., Wall, S.A., van der Spek,P.J., Mathijssen, I.M., Maxson, R.E., Twigg, S.R., Wilkie, A.O., 2013. Mutations inTCF12, encoding a basic helix-loop-helix partner of TWIST1, are a frequentcause of coronal craniosynostosis. Nat. Genet. 45, 304–307.

Shields, E.D., Bixler, D., Fogh-Andersen, P., 1981. Cleft palate: a genetic and epide-miologic investigation. Clin. Genet. 20, 13–24.

Shyamala, B.V., Bhat, K.M., 2002. A positive role for patched-smoothened signalingin promoting cell proliferation during normal head development in Drosophila.Development 129, 1839–1847.

Stanier, P., Moore, G.E., 2004. Genetics of cleft lip and palate: syndromic genescontribute to the incidence of non-syndromic clefts. Hum. Mol. Genet. 113 (1),R73–R81.

Stultz, B.G., Lee, H., Ramon, K., Hursh, D.A., 2006. Decapentaplegic head capsulemutations disrupt novel peripodial expression controlling the morphogenesisof the Drosophila ventral head. Dev. Biol. 296, 329–339.

Stultz, B.G., Park, S.Y., Mortin, M.A., Kennison, J.A., Hursh, D.A., 2012. Hox proteinscoordinate peripodial decapentaplegic expression to direct adult head mor-phogenesis in Drosophila. Dev. Biol. 369, 362–376.

Team, R.C., 2014. R: A language and environment for statistical computing. Vienna,Austria.

Twigg, S.R., Wilkie, A.O., 2015. New insights into craniofacial malformations. Hum.Mol. Genet. 24, R50–R59.

Van Dongen, S., 2007. What do we know about the heritability of developmentalinstability? Answers from a Bayesian model. Evolution 61, 1033–1042.

van Ooijen, J.W., Boer, M.P., Jansen, R.C., Maliepaard, C., 2000. MapQTL 4.0: Softwarefor the calculation of QTL positions on genetic maps.

Weasner, B.M., Kumar, J.P., 2013. Competition among gene regulatory networksimposes order within the eye-antennal disc of Drosophila. Development 140,205–215.

Werren, J.H., 1997. Biology of Wolbachia. Annu. Rev. Entomol. 42, 587–609.Werren, J.H., Loehlin, D.W., 2009. The parasitoid wasp Nasonia: an emerging model

system with haploid male genetics. Cold Spring Harb. Protoc. 4, 1–10.Werren, J.H., Richards, S., Desjardins, C.A., Niehuis, O., Gadau, J., Colbourne, J.K.,

Nasonia Genome Working, G., Werren, J.H., Richards, S., Desjardins, C.A., Nie-huis, O., Gadau, J., Colbourne, J.K., Beukeboom, L.W., Desplan, C., Elsik, C.G.,Grimmelikhuijzen, C.J., Kitts, P., Lynch, J.A., Murphy, T., Oliveira, D.C., Smith, C.D.,van de Zande, L., Worley, K.C., Zdobnov, E.M., Aerts, M., Albert, S., Anaya, V.H.,Anzola, J.M., Barchuk, A.R., Behura, S.K., Bera, A.N., Berenbaum, M.R., Bertossa, R.C., Bitondi, M.M., Bordenstein, S.R., Bork, P., Bornberg-Bauer, E., Brunain, M.,Cazzamali, G., Chaboub, L., Chacko, J., Chavez, D., Childers, C.P., Choi, J.H., Clark,M.E., Claudianos, C., Clinton, R.A., Cree, A.G., Cristino, A.S., Dang, P.M., Darby, A.C., de Graaf, D.C., Devreese, B., Dinh, H.H., Edwards, R., Elango, N., Elhaik, E.,Ermolaeva, O., Evans, J.D., Foret, S., Fowler, G.R., Gerlach, D., Gibson, J.D., Gilbert,D.G., Graur, D., Grunder, S., Hagen, D.E., Han, Y., Hauser, F., Hultmark, D., Hunter,H.Ct, Hurst, G.D., Jhangian, S.N., Jiang, H., Johnson, R.M., Jones, A.K., Junier, T.,Kadowaki, T., Kamping, A., Kapustin, Y., Kechavarzi, B., Kim, J., Kim, J., Kiryutin,B., Koevoets, T., Kovar, C.L., Kriventseva, E.V., Kucharski, R., Lee, H., Lee, S.L., Lees,K., Lewis, L.R., Loehlin, D.W., Logsdon Jr., J.M., Lopez, J.A., Lozado, R.J., Maglott,D., Maleszka, R., Mayampurath, A., Mazur, D.J., McClure, M.A., Moore, A.D.,Morgan, M.B., Muller, J., Munoz-Torres, M.C., Muzny, D.M., Nazareth, L.V.,Neupert, S., Nguyen, N.B., Nunes, F.M., Oakeshott, J.G., Okwuonu, G.O., Panne-bakker, B.A., Pejaver, V.R., Peng, Z., Pratt, S.C., Predel, R., Pu, L.L., Ranson, H.,Raychoudhury, R., Rechtsteiner, A., Reese, J.T., Reid, J.G., Riddle, M., Robertson,H.M., Romero-Severson, J., Rosenberg, M., Sackton, T.B., Sattelle, D.B., Schluns,H., Schmitt, T., Schneider, M., Schuler, A., Schurko, A.M., Shuker, D.M., Simoes, Z.L., Sinha, S., Smith, Z., Solovyev, V., Souvorov, A., Springauf, A., Stafflinger, E.,Stage, D.E., Stanke, M., Tanaka, Y., Telschow, A., Trent, C., Vattathil, S., Verhulst,E.C., Viljakainen, L., Wanner, K.W., Waterhouse, R.M., Whitfield, J.B., Wilkes, T.E.,Williamson, M., Willis, J.H., Wolschin, F., Wyder, S., Yamada, T., Yi, S.V., Zecher,C.N., Zhang, L., Gibbs, R.A., 2010. Functional and evolutionary insights from thegenomes of three parasitoid Nasonia species. Science 327; , pp. 343–348.

Wilkie, A.O., 1997. Craniosynostosis: genes and mechanisms. Hum. Mol. Genet. 6,1647–1656.

Wilkie, A.O., Morriss-Kay, G.M., 2001. Genetics of craniofacial development andmalformation. Nat. Rev. Genet. 2, 458–468.

Wolf, J.B., Leamy, L.J., Routman, E.J., Cheverud, J.M., 2005. Epistatic pleiotropy andthe genetic architecture of covariation within early and late-developing skulltrait complexes in mice. Genetics 171, 683–694.

Wolf, Z.T., Brand, H.A., Shaffer, J.R., Leslie, E.J., Arzi, B., Willet, C.E., Cox, T.C.,McHenry, T., Narayan, N., Feingold, E., Wang, X., Sliskovic, S., Karmi, N., Safra, N.,Sanchez, C., Deleyiannis, F.W., Murray, J.C., Wade, C.M., Marazita, M.L., Ban-nasch, D.L., 2015. Genome-wide association studies in dogs and humansidentify ADAMTS20 as a risk variant for cleft lip and palate. PLoS Genet. 11,e1005059.

Wolf, Z.T., Leslie, E.J., Arzi, B., Jayashankar, K., Karmi, N., Jia, Z., Rowland, D.J., Young,A., Safra, N., Sliskovic, S., Murray, J.C., Wade, C.M., Bannasch, D.L., 2014. A LINE-1insertion in DLX6 is responsible for cleft palate and mandibular abnormalitiesin a canine model of Pierre Robin sequence. PLoS Genet. 10, e1004257.

Won, J.H., Tsogtbartarr, O., Son, W., Singh, A., Choi, K.W., Cho, K.O., 2015. Cell type-specific responses to wingless, hedgehog and decapentaplegic are essential forpatterning early eye-antenna disc in Drosophila. PLoS One 10, e0121999.

Wyszynski, D.F., Beaty, T.H., Maestri, N.E., 1996. Genetics of nonsyndromic oralclefts revisited. Cleft Palate Craniofac. J. 33, 406–417.

Young, D.L., Schneider, R.A., Hu, D., Helms, J.A., 2000. Genetic and teratogenic ap-proaches to craniofacial development. Crit. Rev. Oral Biol. Med. 11, 304–317.

Zeitlinger, J., Bohmann, D., 1999. Thorax closure in Drosophila: involvement of Fosand the JNK pathway. Development 126, 3947–3956.

Zhang, Z., Song, Y., Zhao, X., Zhang, X., Fermin, C., Chen, Y., 2002. Rescue of cleftpalate in Msx1-deficient mice by transgenic Bmp4 reveals a network of BMPand Shh signaling in the regulation of mammalian palatogenesis. Development129, 4135–4146.