Establishment of the mapping family
The L. vannamei family used for mapping was established at the National and Guangxi Shrimp Genetic Breeding Center (Guangxi Province, China). To develop the mapping family, a male shrimp from a family with a relatively strong nitrite tolerance and a female shrimp from a normal family were artificially inseminated, and their progeny were cultured in a pond. After cultivation for one year, a male shrimp and a female shrimp were randomly selected from the progeny and artificially inseminated. Their progeny (the F1 population) were used as the mapping family (LV-1).
Acute nitrite stress test
We randomly selected 157 shrimp from the mapping family (eight-months-old; average mass: 20.46 g). Selected shrimp were cultured in a 2 m × 4 m × 1 m indoor pool. The pool water remained aerated throughout the acclimation period and the experiment, with pH maintained at 8.2 ± 0.3; temperature maintained at 27.0 ± 0.5°C; salinity maintained at 30.1‰; and dissolved oxygen maintained at 7–8 mg/L. All shrimp were fed formulated pellets (Zhengda Corporation, China) daily at a ratio of 5% of average body weight. Shrimp were allowed to acclimate for 5 d before the acute nitrite stress test. Analytically-pure NaNO2 was dissolved in filtered seawater to prepare a concentrated nitrite stock solution; this stock solution was then added to the pool water to increase the nitrite concentration to 700 mg/L. This concentration was chosen because preliminary experiments (Additional file 1: Table S1) indicated that all experimental shrimp will die within 80 hours at nitrite concentration of 700 mg/L, and this time range was suitable for observation. The nitrite concentration in the pool was measured every 24 h using the standard method . Filtered seawater or nitrite stock solution was added to the pool as needed to maintain the nitrite concentration at 700 mg/L. Dead shrimp were collected every 1 h. Shrimp were considered dead when lying on the bottom of the pool out of balance and unreactive when touched with a wooden stick. Survival time was recorded to represent nitrite tolerance. Dead shrimp were collected immediately upon observation, frozen in liquid nitrogen, and transferred to an ultra-low temperature freezer (−80°C) until DNA extraction.
DNA extraction, SLAF library preparation, and sequencing
We extracted DNA from the tail muscles of the 157 selected F1 progeny and the two parents using Marine Animal DNA Extraction kits (Tiangen, China). The quality and quantity of each extracted DNA sample was measured using 1% agarose gel electrophoresis and a ND2000 spectrophotometer (NanoDrop, USA), respectively. We used restriction enzyme digestion prediction software (Biomarker Technologies Corporation, China) to predict appropriate endonucleases for genomic fragmentation based on the genome of L. vannamei (https://www.ncbi.nlm.nih.gov/genome/?term=Vannamei) . Based on this prediction, all genomic DNA samples were digested with the enzymes HaeIII and Hpy166II. The dual-index adapter was then ligated to the fragments using T4 ligase, and polymerase chain reactions (PCRs) were performed. Amplification products (314–414 bp, including the adapter) were purified and collected using gel extraction kits (Illumina, USA). These fragments were the re-amplified for SLAF sequencing using PCR. SLAF sequencing was performed using an Illumina HiSeq system (Illumina, USA), following the manufacturer's instructions. To evaluate the accuracy of library construction, we used the same procedures to construct SLAF sequence libraries for Oryza sativa L. japonica as a control. All library construction and sequencing were performed at Biomarker Technologies Corporation (Beijing, China).
SLAF-seq data analysis and genotyping
First, the raw sequencing data were assigned to samples based on the dual-index adapters. These adapter sequences were removed, and the reads were filtered to remove all reads containing the adapter sequence, as well as reads containing more than 10% unknown (N) bases. As the first few bp of each read were the residue left by the enzyme fragment, the sequencing quality in this area was low. Therefore, only bases 4–103 bp were analyzed (fragment length: 100 bp). The filtered clean reads were compared to the L. vannamei genome (https://www.ncbi.nlm.nih.gov/genome/?term=Vannamei)  using BWA software. Reads with >95% identity, and where both ends matched to the same location in the L. vannamei genome, were considered to be derived from the same SLAF marker. SNP-based polymorphic SLAF markers were identified by comparing reads derived from the same SLAF marker. These polymorphic SLAF markers were filtered to remove markers whose parental sequencing depth was less than 10×; markers with >5 SNPs; markers for which the proportion of genotypes covering all offspring was <70%; and markers with severe partial segregation (chi-squared test P-value < 0.05). The remaining polymorphic SLAF markers were encoded into eight segregation patterns: ab × cd; ef × eg; hk × hk; lm × ll; nn × np; aa × bb; ab × cc; and cc × ab. As the mapping population we used was an F1 population, the polymorphic SLAF markers with the aa × bb separation pattern were discarded, and the remaining polymorphic SLAF markers were used to construct the linkage map.
Genetic map construction and QTL analysis
The genetic map was constructed with iterative ordering and error correction, as implemented in HighMap . The polymorphic SLAF markers were imported into HighMap, and the linkage groups were determined using the pair-wise logarithm of the odds (LOD) test. The order of the markers in the linkage group was calculated using the enhanced Gibbs sampling, spatial sampling, and simulated annealing (GSS) algorithm . The genetic distances between markers were calculated in centiMorgans (cM) based on recombination values using the Kosambi function. QTL analysis was performed using R/qtl software package . The LOD threshold for each data set was established based on the permutation test (1,000 permutations; P < 0.05). QTLs with LOD values above this threshold were considered significant. The phenotypic variance explained (PVE) by the two QTLs was estimated using the following formula: 1 – 10–2L°D/n , where n was the sample size.
RNA-seq analyses were performed using shrimp from four families: the mapping family (LV-1) and three additional genetically-distinct families (LV-2, LV-3, and LV-4). Our previous analysis indicated that the 24-h median lethal concentration (LC50) of NaNO2 was 194.6 mg/L, 196.3 mg/L, 174.3 mg/L, and 250.8 mg/L for families LV-1, LV-2, LV-3, and LV-4, respectively (Additional file 2: Table S2).
We randomly selected 200 shrimp from each family. Each family group was subjected to the acute nitrite stress test (700 mg/L nitrite), as described above with minor modifications. Briefly, in each family group, we collected the 20 most nitrite-sensitive shrimp and the 20 most nitrite-tolerant shrimp. Nitrite sensitive shrimp were collected as soon as soon as they began to exhibit symptoms of toxicity (i.e., swimming out of balance). The hepatopancreases were removed from all collected shrimp, and pooled to generate nitrite-sensitive and nitrite-tolerant pooled samples for each family. After removal and pooling, hepatopancreas tissues were immediately frozen in liquid nitrogen and stored in an ultra-low temperature freezer (−80°C) until RNA extraction.
Total RNA was extracted from each pooled sample using TRIzol reagent (Invitrogen, USA), following the manufacturer’s instructions. Residual genomic DNA was removed with DNase I (Invitrogen, USA). RNA integrity was evaluated using the Agilent Bioanalyzer 2100 system (Agilent Technologies, USA), and RNA concentration was measured using a ND2000 spectrophotometer (NanoDrop, USA). Sequencing libraries were constructed using NEBNext UltraTM RNA Library Prep Kits for Illumina (NEB, USA), following the manufacturer’s instructions. The libraries were sequenced on an Illumina HiSeq 2500 platform and paired-end reads were generated. Library construction and sequencing were performed at Biomarker Technologies Corporation (Beijing, China).
Raw sequencing reads were cleaned using in-house Perl scripts to remove reads containing adapter or poly-N sequences, as well as low-quality (sequencing error rate > 0.1%) reads. Clean reads were mapped to the L. vannamei genome using Hisat2 2.1.0 (http://ccb.jhu.edu/software/hisat2/index.shtml) . Unique mapped reads with one or zero mismatches were used to calculate gene expression levels.
We used the DESeq2 package  to identify statistical differences in gene expression between samples. Genes were considered significantly differentially expressed when the false discovery rate (FDR) was <0.01 and fold change was ≥2.
The DEGs were annotated against the following databases: Nr (ftp://ftp.ncbi.nih.gov/blast/db/), GO (http://www.geneontology.org/) and KEGG (http://www.genome.jp/kegg/).
Candidate gene analysis
The DEG sequences were compared to the L. vannamei genomic region within each QTL using the NCBI blast tool (https://blast.ncbi.nlm.nih.gov/Blast.cgi). We considered a DEG putatively associated with nitrite tolerance only if the DEG was located within the QTL interval, and the expression profile of the DEG was consistent across all four families (LV-1, LV-2, LV-3, and LV-4). Preliminary analysis identified a single gene meeting these criteria: SLC26A6.
Verification of SLC26A6 expression using qRT-PCR
To validate our RNA-seq results, we used qRT-PCR to quantify the expression of the candidate gene (SLC26A6) in the nitrite-tolerant and nitrite-sensitive pooled samples from the four L. vannamei families (LV-1, LV-2, LV-3, and LV-4). qRT-PCRs and RNA-seq analyses were performed using the same samples. qRT-PCRs were performed with SYBR Premix Ex TaqTM II kits (TaKaRa, Japan), following the manufacturer's instructions. The primers used to detect SLC26A6 gene expression levels were RT-SLC-F and RT-SLC-R (Table 1). The qRT-PCR cycling program was as follows: preheating at 95°C for 30 s, followed by 40 cycles of 95°C for 5 s and 60°C for 30 s. We used L. vannamei 18S RNA as the internal reference gene; this gene was amplified using the primers 18s-F and 18s-R  (Table 1). Three parallel qRT-PCRs were performed for each sample. Relative gene expression levels were calculated using the 2-ΔΔCT method .
Effects of candidate gene silencing on nitrite tolerance
The dsRNA used to silence the candidate gene (dsRNA-SLC26A6) was synthesized with the T7 RiboMAX Express RNAi System kit (Promega, USA), following the manufacturer's instructions, using the primers T7-SLC26A6-F, T7-SLC26A6-R, SLC26A6-F and SLC26A6-R (Table 1). The primers T7-egfp-F, T7-egfp-R, egfp-F and egfp-R (Table 1) were used to synthesize dsRNA-egfp.
The dsRNA interference experiment was performed in 1000-liter glass saltwater tanks (30‰ salinity; 26–27°C; pH 7.5–8.1). We randomly selected 360 L. vannamei from family LV-1. The selected shrimp were divided into three groups (n = 40 shrimp per group; three replicates of each group): buffer control, negative control, and experimental. Each shrimp in the buffer control group was injected at the second abdominal segment with 20 μg 0.9% normal saline; each shrimp in the negative control group was injected with 20 μg dsRNA-egfp; and each shrimp in the experimental group was injected with 20 μg dsRNA-SLC26A6. Each shrimp was re-injected after 24 h. At 12 h after the second injection, all shrimp were subjected to the acute nitrite stress test, as described above, for 120 h. The time of death of each shrimp was recorded.
To evaluate the inhibitory effects of dsRNA on SLC26A6 gene expression, we randomly collected shrimp from the experimental group at 0 h, 36 h, 48 h, and 72 h after the first injection of dsRNA-SLC26A6 (three shrimp were collected at each time point). The hepatopancreases were removed from the sampled shrimp as described above, and pooled by time point. Total RNA was extracted from each pool using a TRIzol Reagent kit (Invitrogen, USA), following the manufacturer's instructions. qRT-PCRs was performed using SYBR Premix Ex TaqTM II kits (TaKaRa, Japan), following the manufacturer's instructions. The primers used to detect SLC26A6 expression (SLC26A6-F and SLC26A6-R; Table 1) were designed using Primer Premier v 6.2 (http:/ /www.premierbiosoft) based on the SLC26A6 gene sequence (NCBI accession: XM_027371736.1) . The qRT-PCR cycling program was as follows: preheating at 95°C for 30 s, followed by 40 cycles of 95°C for 5 s and 60°C for 30 s. We used L. vannamei 18S RNA as the internal reference gene; this gene was amplified using the primers 18s-F and 18s-R (Table 1). Three biological replicates and three technical replicates were performed for each group. Relative mRNA expression levels were calculated using the 2-ΔΔCT method .
Data were shown as mean ± standard deviation (SD), and analyzed using one-way ANOVAs in SPSS 13.0 . We considered P < 0.01 statistically significant.