Stickleback crosses and care
Japanese Pacific and Paxton Benthic sticklebacks were bred in the laboratory to generate age-matched clutches. All offspring were raised in identical laboratory conditions without parental care. For the population comparison, both populations were raised together in common garden tanks (Wark et al. 2011
). For genetic mapping, an in vitro
cross was made between a single, wild-caught Paxton Benthic female stickleback and a single, first-generation lab-raised Japanese Pacific male stickleback to generate an F1 family. Four F1 females were independently crossed to four F1 male siblings to generate four F2 families.
All sticklebacks were housed in 29-gallon aquarium tanks under summer lighting conditions (16 hr light, 8 hr dark) at approximately 15.5°. Tanks were filled with stickleback aquarium water (0.35% saltwater: 3.5g/l Instant Ocean salt, 0.4 ml/l NaHCO3). Water was oxygenated with an air stone and circulated through an external charcoal filter (AquaClear 20 Power Filter; Hagen, Montreal, Canada). Fish were fed live Artemia nauplii twice daily. All animal procedures were approved by the Fred Hutchinson Cancer Research Center Institutional Animal Care and Use Committee (protocol 1575).
Neuromast visualization and analysis
Eight Japanese Pacific sticklebacks, 8 Paxton Benthic sticklebacks, and 236 F2 hybrid sticklebacks were examined for lateral line morphology at approximately one year of age. Average standard lengths in cm ± SEM of the fish were: Japanese Pacific (5.0 ± 0.06), Paxton Benthic (5.0 ± 0.15), and F2s (4.72 ± 0.03). To count neuromasts, fish were stained with the fluorescent vital dye 2-(4-(dimethylamino)styrl)-N-ethylpyridinium iodide (DASPEI; Invitrogen/Molecular Probes, Carlsbad, CA). Live fish were placed in aerated 0.025% DASPEI in 30% tank water and 70% deionized water for 15 min. Fish were then deeply anesthetized in 0.016% MS-222 (tricaine methylsulfonate; Fisher Scientific, Pittsburgh, PA) for approximately 5 min, or until the fish were motionless and breathing very shallowly. Fish were gently submerged in a Petri dish containing 0.005% MS-222 and mounted on a Leica fluorescence dissecting scope with a FITC filter set (Leica Microsystems Inc., Bannockburn, IL). Neuromasts were counted in all 12 lines that compose the stickleback lateral line system (Wark and Peichel 2010
). Abbreviations for these lines are as follows: mandibular (MD), ethmoid (ET), supraorbital (SO), infraorbital (IO), oral (OR), preopercular (PO), otic (OT), anterior pit (AP), supratemporal (ST), main trunk line anterior (Ma), main trunk line posterior (Mp), and caudal fin (CF). Only neuromasts on the left side of the body were counted. Following staining and neuromast quantification, fish were returned to 0.016% MS-222 and killed. Fin tissue was extracted and placed in ethanol for subsequent DNA extraction. Bodies were placed in 10% buffered formalin.
To quantify neuromast patterning in the main trunk line, neuromasts in each body segment (myomere) were categorized according to the primary axis of patterning: dorso-ventral (vertical distribution) or anterior-posterior (horizontal distribution). Neuromast distribution could only be determined when sufficient neuromasts were present in a given body segment. In Ma, two neuromasts were required for classification because the dorsal-ventral midline was difficult to determine and neuromast position had to be compared relative to one another. In Mp, segments with single neuromasts could be categorized because the midline of each body segment could easily be observed, regardless of plating. A summary ratio of dorso-ventral patterning was calculated by dividing the number of segments with a vertical neuromast distribution by the number of segments that could be phenotyped.
DASPEI staining was not consistent across the body in all F2 hybrids, due to unequal stain penetration or high background. For each individual, any lines in which neuromasts could not be clearly visualized were excluded from the QTL data set. Furthermore, 32 F2 hybrids had weak or inconsistent staining on the majority of the body. These animals were used to make the linkage map and in the QTL analysis for skeletal traits, but they were excluded from the QTL analysis for neuromast number and pattern.
Groove depth in the supraorbital (SO) line was scored on a fluorescent stereomicroscope during DASPEI staining. Grooves were assigned a score based on qualitative observations of depth, ranging from 0 (no grooves detected) to 3.5 (deepest grooves observed).
Skeletal trait characterization
Lateral plates were visualized by staining all F2 hybrids with alizarin red (Fisher Scientific, Pittsburgh, PA), a calcium stain. Fish were removed from formalin and placed in dH20 overnight. Fish were placed in 0.008% alizarin red in 1% KOH for 24 hr and then de-stained in several washes of dH20. Plates were counted on the left side of the body. In addition to total plate number, the number of plates in the body regions corresponding to the anterior and posterior portions of the main trunk line were recorded. This boundary is defined as the position where the last plate in Ma contacts both the support structure for the second dorsal spine and the pelvic girdle. Animals were also assessed for the presence of a pelvic girdle and pelvic spines. Animals with a complete or partial pelvic girdle and pelvic spines were assigned a score of 1, and animals lacking any pelvic structures were assigned a score of 0.
Fluorescent images of neuromasts were captured using a Retiga camera (QImaging, Surrey, BC, Canada). The contrast of these images was adjusted uniformly using the automated “Levels” function in Adobe Photoshop. Alizarin red-stained animals were photographed on a Nikon SMZ1500 light stereomicroscope equipped with a Nikon Coolpix 4500 digital camera (Nikon, Melville, NY). Schematics of F2 hybrid phenotypes were created by overlaying DASPEI images and alizarin red images in Adobe Illustrator.
Statistical analyses were performed in SPSS 13.0 software (SPSS, Chicago, IL). Japanese Pacific and Paxton Benthic neuromast numbers were compared using multivariate analysis of variance (MANOVA). Overall differences among groups were tested with the Wilks lambda multivariate test. Epistatic interactions between linkage groups (LG) 4 and 21 were assessed by ANOVA.
Quantitative trait locus analysis
Genomic DNA was isolated from fin clips using phenol-chloroform extraction, followed by ethanol precipitation and resuspension in 50 μl TE (10 mM Tris, 1 mM EDTA). Both grandparents, 7 F1 parents, and 236 F2 hybrids were genotyped using 1536 genome-wide single nucleotide polymorphism (SNP) markers on a custom-built stickleback Golden Gate SNP array (Illumina, San Diego, CA; Jones et al. 2012a
). SNP genotypes were analyzed using GenomeStudio software (Illumina). There were 245 SNPs with fixed differences between the Paxton Benthic and Japanese Pacific grandparents (Table S1
); these were combined with five microsatellite markers on LG 21 (Table S2
) to create a linkage map with JoinMap 3.0 (Van Ooijen and Voorrips 2001
). The linkage map consisted of 22 linkage groups, including 2 linkage groups containing markers from chromosome 14 (labeled 14a and 14b). Three markers did not associate with any linkage groups, leaving 247 markers in the final map. Two F2 individuals had poor genotyping data and were excluded from the study, leaving 234 F2 hybrids in the QTL mapping analysis. All genotype and phenotype data for these 234 F2s are provided in File S1
Interval mapping was performed using MapQTL 4.0 (Van Ooijen et al. 2002
). Because we focused on identifying QTL segregating between the Paxton Benthic and Japanese Pacific populations, only markers with fixed differences between the grandparents were used for QTL mapping (i.e.
, all F1s were heterozygous), and data for all four F2 families were therefore combined in the analysis. Genome-wide likelihood of odds (LOD) significance thresholds were established for each trait using permutation testing (α
= 0.05, 1000 permutations). When permutation testing could not be used (pelvis, Ma pattern, and Mp pattern for plated segments only), we employed a conservative genome-wide LOD significance threshold of 4.2 (α
= 0.05), based on simulations for an F2 population (Van Ooijen 1999
). Only QTL that met genome-wide significance thresholds at a nearby marker are reported.