A first AFLP-Based Genetic Linkage Map for Brine Shrimp Artemia franciscana and Its Application in Mapping the Sex Locus

We report on the construction of sex-specific linkage maps, the identification of sex-linked markers and the genome size estimation for the brine shrimp Artemia franciscana. Overall, from the analysis of 433 AFLP markers segregating in a 112 full-sib family we identified 21 male and 22 female linkage groups (2n = 42), covering 1,041 and 1,313 cM respectively. Fifteen putatively homologous linkage groups, including the sex linkage groups, were identified between the female and male linkage map. Eight sex-linked AFLP marker alleles were inherited from the female parent, supporting the hypothesis of a WZ–ZZ sex-determining system. The haploid Artemia genome size was estimated to 0.93 Gb by flow cytometry. The produced Artemia linkage maps provide the basis for further fine mapping and exploring of the sex-determining region and are a possible marker resource for mapping genomic loci underlying phenotypic differences among Artemia species.


Introduction
Artemia, known as brine shrimp, is a genus of small planktonic crustaceans found worldwide in natural salt lakes and salterns [1]. Their larvae (nauplii) are the most commonly used live food in aquaculture activities, specifically for larval growth of more than 85% of the marine species reared in aquaculture [2,3]. Adult Artemia survive extreme salinities, while their encysted embryos (cysts), produced under stressful conditions, have a unique tolerance for high doses of UV and ionizing radiation, anoxia, thermal extremes and desiccation-hydration cycles [4][5][6]. Cysts remain viable for years and produce nauplii within 24 h after hydration.
Genetic linkage maps are invaluable in forward genetic analyses for the identification of the genomic loci responsible for phenotypic differences. From this perspective, Artemia offers a number of major advantages for time-effective generating of experimental mapping populations and for mapping natural allelic variation. They have a short generation time (2-4 weeks), offspring production of several hundred individuals per brood, storability of cysts for years, easy breeding in large numbers and levels of genetic variability that are among the highest within crustaceans [16,41,52]. In addition, we expect that forward genetic approaches in Artemia are not only restricted to Artemiaspecific traits, but are also valuable for mapping traits such as sex, Vibrio pathogen resistance and growth rate, segregating in commercially important crustaceans. We believe therefore, that Artemia could be a useful model species for other crustaceans.
In the present study, we report on a first AFLP-based linkage map of A. franciscana. We additionally present eight sex-linked markers that disclose the linkage group corresponding to the W chromosome and confirm female heterogamety in A. franciscana. Finally, we report on the estimation of the A. franciscana genome size by flow cytometry.

Mapping population
Cyst material of the A. franciscana strains from San Francisco Bay, USA (SFB; ARC1364) and Vinh Chau, Vietnam (VC; ARC1349) was obtained from the Laboratory of Aquaculture & Artemia Reference Center cyst bank (http://www.aquaculture. ugent.be). The SFB strain was first introduced in Vinh Chau, Vietnam in 1982, eventually resulting in the new VC strain in the late 1980s [53]. First, cysts from both strains were hatched separately in aerated seawater (28uC, salinity 35 g.l 21 ). The instar I nauplii of each strain were then harvested and reared for a week in aerated seawater with added sea salt (Instant OceanH, 28uC, final salinity 70 g.l 21 ) and fed with Tetraselmis suecica, a marine unicellular green alga. The Artemia were subsequently transferred to individual Falcon tubes and kept there under the same conditions for seven days until sexual maturation. A controlled cross between VC (R) and SFB (=) was then made, resulting in F 1 progeny that was collected over a sieve every two days and grown until maturity under the same conditions as the parental generation. Adult F 1 progeny was rinsed with sterile distilled water and the phenotypic sex of each F 1 offspring individual was determined visually. For gut evacuation before DNA extraction, the offspring and parents were starved during 24 h, followed by removal of the brood pouch in females. Parents and progeny were stored individually at 220uC.

DNA extraction
DNA was extracted from parents and their 112 F 1 offspring according to a modified CTAB-method for shrimp tissue [54]. Briefly: to each sample, ground in liquid N 2 , 150 ml of CTAB buffer was added. After homogenization, 750 ml of extra CTAB buffer was added and the mix was left at 25uC for 30 min. PCA solution was added (600 ml; 25:24:1 phenol/chloroform/isoamylalcohol). After centrifugation, 800 ml of the upper aqueous phase was added to 600 ml of CA solution (24:1 chloroform/isoamylalcohol) and the mix was homogenized. To 700 ml of the upper aqueous phase, 630 ml of isopropanol was added. The mix was incubated for 1 h at 270uC. After centrifugation, the pellet was washed with 600 ml of ethanol 70%, air-dried in a 60uC oven and resuspended in 20 ml of sterile distilled water. DNA quality and concentration were assessed on a 1% agarose gel.

Segregation analysis and linkage mapping
AFLP analysis with fluorescent dye detection was performed on a LI-COR long read-IR 2 4200 (LI-COR Biosciences) as described by Vuylsteke et al. [55]. Sixty-five EcoRI+3/MseI+3 primer combinations (PCs) listed in Table S1 were used. AFLP analysis of parents and 112 offspring was done on two separate 64-lane gels per PC.
The degree of polymorphism between the two parental strains was estimated based on AFLP fragments amplified by four PCs (E112M212, E112M213, E112M233 and E112M234).
AFLP markers were scored using the specific image analysis software AFLP-QuantarPro (http://www.keygene-products.com) as described in Vuylsteke et al. [55]. Each AFLP marker was identified by (1) a code referring to the corresponding PC (Table S1), followed by (2) the molecular size of the fragment in nucleotides as estimated by AFLP-QuantarPro, and (3) a tag referring to the type of marker. Parental AFLP markers segregating 1:1 in the F 1 progeny are heterozygous in either the female (female marker, tagged as ''F'') or the male parent (male marker, tagged as ''M'') and homozygous absent in the other parent. AFLP markers heterozygous in one of the parents and homozygous present in the other were not included in the linkage analysis, because heterozygotes could not be reliably discriminated from individuals homozygous for the ''band present'' allele. No tag was used for biparental markers, which are heterozygous in both parents and thus, segregate 1:2:1 in the F 1 progeny. Parental and biparental AFLP markers were scored co-dominantly based on relative fragment intensities resulting in more genetic information compared to dominant (present/absent) scoring and hence, speeding up the mapping process [55]. However, biparental markers were scored dominantly when the heterozygotes could not reliably be discriminated from the individuals homozygous for the ''band present'' allele.
Linkage and segregation analyses were performed using the software package Joinmap 4 [56]. The mapping population type was set to CP (i.e. a population resulting from a cross between two heterogeneously heterozygous and homozygous diploid parents, linkage phases originally unknown). The segregation type was encoded according to Joinmap 4 recommendations [56]. A logarithm of the odds (LOD) threshold range between 2.0 and 14.0 was initially used to group parental markers. Only linkage groups containing at least three markers were considered for map  (Table S1), followed by (2)  construction. Segregation distortion of markers was tested by using a x 2 -test as implemented in Joinmap 4. Graphical presentation of linkage groups was done with the software MapChart [57].

Artemia genome size estimation by flow cytometry
The haploid genome size (GS) of Artemia was assessed against the rainbow trout (haploid GS 2.4-3.0 pg or 2.3522.93 Gb [58]) and the chicken genome (haploid GS 1.07 pg or 1.05 Gb [59]), both used as internal standards.
The consistency of the used method was assessed by calibrating rainbow trout nuclei (2 ml of freshly drawn heparinized Oncorynchus mykiss blood) against chicken erythrocyte nuclei (2 ml of 10x diluted BioSureHCEN singlet, Gallus gallus domesticus, Rhode Island Red female).
Each of the four Artemia individuals (i.e. four full-sib males from the VC (R) x SFB (=) cross) were chopped together with internal standard material using a razor in 1 ml of Galbraiths buffer as described in Dolezel and Bartos [60]. Cell suspensions were filtered through a 30 mm mesh, put on ice and nuclei were costained in the dark for 2 min with 50 ml of fluorescent DNA stain Propidium Iodide (Sigma-Aldrich PI solution in water 1 mg/ml). The use of PI staining on A. franciscana (GC% 32) [61], O. mykiss (GC% 42) [58] and G. domesticus (GC% 47) [62] was chosen to avoid a GC content-linked bias, as occurs with DAPI staining [60]. At least 5,000 nuclei were analyzed for each co-stained sample, using a Modular Flow cytometer and cell sorter (MoFlo Legacy, Cytomation) with a 488 nm Argon laser and PI emission bandpass filter of 580/30 nm. Instrument calibration was performed using Flow-check Fluorospheres (Beckman Coulter) and internal standards. Fluorescence of the nuclei was recorded linearly with the software Summit v4.3. For each co-stained sample, fluorescence histograms were generated and mean fluorescence values were calculated with the flow cytometry data analysis software Cyflogic 1.2.1. The haploid GS of for each Artemia sample was calculated according to the following formula [13]: is the mean fluorescence of the sample and F is is the mean fluorescence of the internal standard.

Segregation analysis and linkage mapping
A total of 65 AFLP PCs resulted in a total of 531 markers, of which 433 were parental (239 female, 194 male) and 98 markers were biparental. Based on only four primer combinations (PCs) yielding 180 AFLP fragments, 36% of the fragments segregated between both parents.
First, a parental map including only parental markers was constructed. Summary statistics for the parental maps are listed in Table 1. The grouping of parental markers at a LOD score ranging from 5.0 to 6.0 resulted in a number of linkage groups corresponding with the haploid chromosome number (n = 21). The female map, containing 225 markers (Figure 1), resulted in 22 ''female'' linkage groups (LG) spanning 1,312.9 cM; the male map, containing 181 markers (Figure 2), resulted in 21 ''male'' LG spanning 1,041.3 cM. Twenty-eight percent of the analyzed parental markers showed significant (p,0.05; x 2 test) segregation distortion. Male markers were more often distorted than female markers (31% resp. 25%). Some larger genomic regions did not contain any markers (e.g. 32.5 cM in LG Female_6, Figure 1; 38.0 cM in LG Male_2, Figure 2), despite the low median intermarker distances of 3.9 and 3.1 cM for the female and the male linkage map (Table 1).
Next, an integrated map was created including the 98 biparental markers and 406 previously mapped parental markers ( Figure 3). By including biparental markers, groups consistent with  (Table S1), followed by (2) the molecular size of the fragment in nucleotides and (3)  linkage groups of the parental map were obtained at a LOD threshold ranging between 6 and 10. Sixty-nine percent of the biparental marker loci showed significant (p,0.05; x 2 test) segregation distortion. These loci were still included in map construction and evaluated for quality afterwards, since significant segregation distortion is inherent to relatively small experimental mapping population sizes of ,100 individuals. Forty-nine biparental markers (50%) could be mapped in the female as well as in the male map, identifying 15 homologous linkage groups including the sex linkage groups (Figure 3, Figure 4).

Mapping of the sex locus
Staelens et al. [27] described segregation patterns of sex-linked AFLP markers that unequivocally differentiate the WZ-ZZ and XX-XY sex-determination system. We observed eight AFLP markers, spanning a region of 0.2 cM on LG Female_1 (markers in green, Figure 4) segregating according to pattern 1 and a single marker (E112M122M167.3F) according to pattern 2. Both segregation patterns are expected under the assumption of female heterogamety. None of the markers segregated according to patterns 6, 7 and 8, expected under the assumption of male heterogamety. The male linkage group Male_10 was identified as homologous to Female_1 (Figure 4). In conclusion, the observed segregation patterns of sex-linked AFLP markers strongly favour female over male heterogamety in Artemia.

A. franciscana genome size estimation by flow cytometry
Using trout blood as the internal standard, the haploid female chicken genome size (GS) determined by flow cytometry was 1.05 Gb (1.07 pg) as previously reported for female chicken [59]. We preferred rainbow trout nuclei as the internal standard in the assessment of the Artemia GS because their fluorescence values did not overlap with those of Artemia, as was the case with fluorescence values obtained from chicken nuclei. Using rainbow trout nuclei as the internal standard, the A. franciscana haploid genome size was estimated to 0.9360.09 Gb (0.9760.09 pg; n = 4). Fluorescence histograms for each sample and for chicken are shown in Figure 5. Fluorescence peaks were relatively broad due to cell debris from the previously frozen Artemia individuals, but average DNA content estimates were consistent throughout the different samples, shown by the small standard error.

Discussion
We present the first sex-specific AFLP linkage maps and sexlinked markers as well as a consistent genome size (GS) estimation for the brine shrimp A. franciscana.
The linkage analysis of 433 parental AFLP markers segregating in a 112 full-sib family identified 21 male and 22 female linkage groups, corresponding very well with the haploid chromosome number in A. franciscana (2n = 42) [8]. Most likely, the markers in small linkage groups (LG) such as Female_20 (Figure 1) would join one of the other 21 LG by adding more markers to the female map. More female than male markers were generated, suggesting that maternal A. franciscana strain Vinh Chau (VC) has more unique alleles compared to the paternal strain San Francisco Bay (SFB). This seems a logical consequence of the SFB origins of VC. The level of polymorphism between the two A. franciscana parental strains was estimated at 36%, which is in the range of 9-50% estimated previously by Kappas et al. [53]. Given their high marker density, the produced genetic maps are adequate for the anchoring of Artemia genome sequences to facilitate the future construction of physical maps for each of the 21 chromosomes. Figure 3. Artemia franciscana homologous autosomal male and female linkage groups. Fourteen homologous autosomal linkage group pairs. Each AFLP marker was identified by (1) a code referring to the corresponding PC (Table S1), followed by (2)   This will be especially useful, considering the numerous reports of repetitive sequences in Artemia [63]. Artemia linkage maps will also allow future linkage studies in Artemia for important crustacean traits such as resistance to Vibrio, the most common bacterial pathogen in worldwide marine fish and shellfish aquaculture.
Fifteen homologous linkage groups, including the LG representing the sex chromosomes, were identified between the female and male linkage maps by including biparental markers in the linkage analysis. This study identified eight sex-linked AFLP marker alleles mapping to one locus and inherited from the female parent, suggesting A. franciscana adopts a genetic WZ-ZZ sexdetermining system. Artemia sex-linked markers will enable the study of nauplii sex ratios and their dynamic in natural Artemia populations. They will also enable the further fine-mapping of the sex-determining locus and the subsequent identification of the primary sex-determining gene(s). Furthermore, based on sequence homology with Artemia, sex-determining genes might be identified in commercially valuable crustaceans, enabling PCR-based allelespecific assay development in the framework of the development of mono-sex cultures in shrimp [64].
The clustering of eight sex-linked markers in a 0.2 cM region suggests reduced recombination, which is often found in sex-linked regions [65]. Genes from a region that stopped recombining in the early evolution of sex chromosomes have a high sequence divergence, allowing an estimate of when the W and Z chromosomes first stopped recombining and thus, the age of the sex chromosome system [65].
The estimated Artemia GS in this study (0.93 Gb) is smaller than earlier estimates: 2.93 Gb by Feulgen densitometry [9] and 1.47 Gb by DNA reassociation kinetics [10]. ''A. salina'' used to be a general name for all Artemia species, presently confounding the identity of the investigated Artemia in many studies [1]. Because the Artemia DNA content measured by Feulgen densitometry on ''A. salina'' is almost a twofold of that measured by DNA reassociation kinetics, Feulgen densitometry might have been performed on a tetraploid A. parthenogenetica, as suggested by Vaughn [10]. Also, the absolute A. franciscana karyotype size varies between 60.68 mm and 139.26 mm [8], showing that significant intra-specific variation in DNA content could explain the high Feulgen densitometry values as well.
Vaughn [10] calculated the Artemia haploid GS by DNA reassociation kinetics, based on an A. franciscana GC content of 42%. More recent measurements however, show an A. franciscana (SFB) GC content of 32% determined by CsCl centrifugation and confirmed by direct chemical analysis and renewed thermal denaturation [61]. An estimated GC content lowered by 1% results in a 0.018% lower haploid DNA content estimated by DNA reassociation kinetics [66]. Hence, based on a GC content of 32%, the corrected A. franciscana DNA content estimated by Vaughn [10] is 1.23 Gb, approximating more closely the 0.93 Gb Artemia GS estimated in this study.
Currently, out of the 50,000 known Crustacea species, the GS of 278 crustaceans has been determined, covering a 400-fold-wide genome size range between Cyclops kolensis, a Cyclopoid copepod (0.14 pg) and Ampelisca macrocephala, an Arctic Amphipod (64.62 pg) [67,68]. In comparison, A. franciscana has a relatively small genome of 0.97 pg. This makes it a potential new model crustacean for which genome sequencing is currently feasible, unlike for crustaceans with a much larger genome size. To date, the only publicly accessible sequenced crustacean genome is the branchiopod D. pulex, with an average genome size of 0.23 pg [69].
Ultimately, the further development of genomic resources for Artemia such as the whole genome sequence, will add a completely new dimension to Artemia research and its use as live food in aquaculture. Moreover, knowledge of the A. franciscana sexdetermining system will facilitate future evolutionary studies of sex chromosomes in sexually dimorphic (WZ female/ZZ male) and parthenogenetic Artemia. Considering the presence of sexual and asexual reproduction strategies, the Artemia genus shows promise as a model system for the study of asexuality, its evolution and its evolutionary purpose. Finally, since Artemia is considered a potential crustacean model species, increasing knowledge about Artemia genetics and genomics in general and sex-related genetics in particular, are expected to be valuable to crustacean aquaculture, presently lacking in molecular breeding strategies despite their contribution of 23% to the total aquaculture production value [70].

Supporting Information
Table S1 List of the 65 primer combinations used for AFLP analysis 1 E: EcoRI primer with three selective bases; M: MseI primer with three selective bases (1, 2, 3, 4 correspond to A, C, G, T). (DOC)