Gene expression profiling of the olfactory tissues of sex-separated and sex-combined female and male mice

Olfactory experience can alter the molecular and cellular composition of chemosensory neurons within the olfactory sensory epithelia of mice. We sought to investigate the scope of cellular and molecular changes within a mouse’s olfactory system as a function of its exposure to complex and salient sets of odors: those emitted from members of the opposite sex. We housed mice either separated from members of the opposite sex (sex-separated) or together with members of the opposite sex (sex-combined) until six months of age, resulting in the generation of four cohorts of mice. From each mouse, the main olfactory epithelium (MOE), vomeronasal organ (VNO), and olfactory bulb (OB) were removed and RNA-extracted. A total of 36 RNA samples, representing three biological replicates per sex/condition/tissue combination, were analyzed for integrity and used to prepare RNA-seq libraries, which were subsequently analyzed via qPCR for the presence of tissue- or sex-specific markers. Libraries were paired-end sequenced to a depth of ~20 million fragments per replicate and the data were analyzed using the Tuxedo suite.


Background & Summary
Sensory activity plays an important role in guiding the development of the nervous system, in part through activity-dependent changes in gene expression [1][2][3] . In the olfactory system, activity mediates the formation and the refinement of connections between olfactory sensory neurons (OSNs) located in the MOE and postsynaptic neurons in the OB 4-8 , as well as the relative abundance of OSNs that express specific olfactory receptor (OR) genes [9][10][11][12][13][14][15] . The latter changes appear to occur via alterations in the turnover rates of specific OSNs, which are continually born and replaced throughout life 16,17 . Like OSNs, Vomeronasal sensory neurons (VSNs) also undergo turnover throughout life 17 , suggesting that the abundance of VSN subtypes may have a similar capacity for experience-dependent changes. Activitydependent changes to the representation of chemosensory neurons have been hypothesized to play a role in adapting an individual's olfactory system to the detection and/or discrimination of salient odors, which may vary from one olfactory environment to another 11 .
The datasets described here were generated to enable investigation of the scope of molecular and cellular changes that occur within the olfactory system as a function of mouse exposure to odors from the opposite sex for a prolonged time period. Mouse odors are complex mixtures of volatile and non-volatile chemicals derived from skin secretions and substances such as urine, tears, saliva, and feces that are known to differ substantially between males and females [18][19][20][21][22][23][24][25][26][27] and activate distinct subsets of OSNs and VSNs 18,20,21,[28][29][30][31][32][33][34][35] . Because male and female mice emit distinct odor profiles, we predicted that sexseparated males and females would have distinct olfactory experiences and would thus display differences in their profiles of olfactory sensory neuron subtypes and gene expression. In contrast, sex-combined Figure 1. Experimental design. From weaning (P21) until 6 months of age, mice experienced either a sexseparated environment, in which they were housed either 4 females/cage (SF mice; left) or 4 males/cage (SM mice; middle), or a sex-combined environment (CF and CM mice; right), in which they were housed 2 females + 2 males/cage. MOE, VNO, and OB tissues were dissected from each of 9 mice per sex/condition combination, resulting in a total of 108 tissue samples. RNA was extracted from each tissue sample and pooled in groups of 3, resulting in 36 RNA samples (3 biological replicates per sex/condition/tissue combination), and used to generate RNA-seq libraries.  -CF1  MOE-CF2  MOE-CF3  MOE-CM1  MOE-CM2  MOE-CM3  MOE-SF1  MOE-SF2  MOE-SF3  MOE-SM1  MOE-SM2  MOE-SM3 Ladder  Table 1. male and female mice would be expected to have more similar olfactory experiences and would thus display fewer differences in their profiles of OSN/VSN subtypes and gene expression. To generate the datasets described here, we housed male and female mice either separated from members of the opposite sex (sex-separated) or combined with members of the opposite sex (sexcombined) from the time of weaning until six months of age (Fig. 1). We then dissected the MOE, VNO, and OB tissues from a total of 36 mice (six mice per sex/condition/tissue combination) and extracted the RNA from each tissue. We generated a total of 108 RNA samples, which were combined in groups of 3 to generate a total of 36 pooled-RNA samples, with each sex/condition/tissue combination represented by three biological replicates. The integrity of each of the 36 pooled-RNA samples was analyzed and each sample was used for the generation of a stranded RNA-seq library. Libraries were analyzed by quantitative PCR (qPCR) for the presence or absence of tissue-and sex-specific markers and then pairedend sequenced to generate a total of approximately 20 million sequence pairs per library. Sequences were aligned to the mouse genome and gene expression was quantified using the Tuxedo suite 36 . Further analyses of the data, including assessment of the effects of sex separation on chemosensory neuron abundance and overall gene expression, have been published in a separate manuscript 37 .

Methods
These methods represent an expanded version of some of the methods described in our related work 37 . All procedures involving animals were carried out in accordance with NIH standards and approved by the University of Wyoming and Harvard University Institutional Animal Care and Use Committees (IACUC).

Preparation of olfactory tissues from sex-separated and sex-combined mice
C57Bl/6 mice were subjected to either sex-separated (SF and SM samples) or sex-combined (CF and CM samples) conditions, in which animals were housed four females per cage (SF), four males per cage (SM), or two females and two males per cage (CF and CM) from weaning (postnatal day 21) until 6 months of age ( Fig. 1). At the time of weaning, SF and SM cages were transferred to rooms containing only mice of the same sex to avoid exposure to opposite-sex odors from cages in the same room. Pups born in the sexcombined cages were euthanized within one day of birth to minimize exposure to pup odors. At 6 months of age, mice were sacrificed and MOE, VNO, and OB tissues were dissected as described 11 . Briefly, dissections were performed as follows: Using strong scissors, mice were decapitated and the bottom jaw was removed, along with and the skin, soft tissue, front teeth, and palate. Using fine scissors, the cranium was cut down the midline above the brain from the brainstem to the OB. Following removal of the dorsal and lateral cranial bones, including the bones surrounding the OB, the brain and OB were carefully lifted from the cranium and the OB was separated from the brain with a scalpel. The whole VNO was obtained by breaking the vomer bone with forceps and carefully lifting the vomer bone and attached VNO from the ventral nasal cavity. Finally the MOE was obtained by removing the dorsal and lateral bones surrounding the MOE and carefully lifting it from the dorsal nasal cavity. Immediately following dissection, each tissue was placed in a sterile microcentrifuge tube, flash-frozen on dry ice, and stored at −80°C until RNA extraction.  Technologies) following the manufacturer's protocol, resulting in a total of 108 RNA samples (Fig. 1). Trizol-purified RNA samples were quantified using a NanoDrop instrument (ThermoFisher Scientific). Equal quantities of three samples of the same sex/condition/tissue were combined and further purified using an RNeasy Plus Mini Kit (Qiagen) to generate 36 samples, representing three biological replicates per combination of sex/ condition/tissue. Integrity of the RNA was analyzed using a 2100 Bioanalyzer (Agilent) (Fig. 2; Table 1). Using the TruSeq Stranded Total RNA with Ribo-Zero Gold Kit (Illumina), each RNA sample was depleted of ribosomal RNA and used to prepare an RNA-seq library tagged with a unique barcode. Library identity and quality were confirmed via quantitative PCR (qPCR) analysis using primers specific for genes expressed in the MOE (Cnga2: TCTGTTGGTAGCCAGAGCCT and AGCCCTTGTTCTAG-GAAGCC), VNO (Vmn1r51: TGAGAACAGCAGGGTACACA and TGAATGCCATGACCAGTAGC), and male tissues (Utyl: GGTTCAGTGCACTTGCCTTT and TGATCCCTAGCTACTTGTCTGTTTT) (Fig. 3). Libraries were quantified using a Qubit instrument (ThermoFisher Scientific). Libraries were paired-end sequenced (2 × 50 bases) to a depth of~40 million reads/sample (~20 million paired-end fragments/replicate;    38 . For each sample, sequence pairs were aligned to the genome using Tophat2 39 , resulting in concordant alignments for~80% of the read pairs ( Table 2). Analyses of gene expression levels and differential expression were performed using Cufflinks and Cuffdiff, respectively 36 . The correlation of FPKM values between biological replicates was analyzed pairwise (Fig. 5). Significance testing for differential expression was performed on all genes with a minimum alignment count of 5 fragments.

Data Records
RNA-seq data files in FASTQ format were deposited at NCBI Sequence Read Archive (Data Citation 1). This accession contains a total of 144 FASTQ files resulting from paired-end sequencing for each of the     Validation of RNA-seq libraries Prior to sequencing, all libraries were analyzed by qPCR for the presence (or absence) of the following gene markers: Cnga2, a gene expressed in MOE but not VNO or OB tissues, Vmn1r51 (V1ra1), a gene expressed in VNO but not OB or MOE tissues, and Utyl, a gene expressed in male but not female tissues. This analysis revealed that Cnga2 expression was detected only in the MOE libraries, Vmn1r51 expression was detected only in the VNO libraries, and Utyl expression was detected only in the male libraries (Fig. 3).

Validation of sequencing data and alignments
FASTQ files obtained from Illumina sequencing were analyzed for quality using FASTQC (Andrews S. (2010). FastQC: a quality control tool for high throughput sequence data. Available online at: http://www. bioinformatics.babraham.ac.uk/projects/fastqc/). This analysis revealed that the raw sequence data was of high quality (Fig. 4). Alignment of the libraries resulted in an average of 90.1% of reads aligned to the mouse genome and 81.9% of pairs aligned concordantly ( Table 2). Analysis of the sequenced libraries using the CollectInsertSizeMetrics tool (http://broadinstitute.github.io/picard/) revealed a mean insert size for all libraries of 167 bp. (Table 2; Fig. 4). Following Cufflinks determination of gene expression values (FPKM) for each gene in each library, the pairwise correlation of FPKM values between biological replicates were analyzed (Fig. 5). This analysis revealed mean Pearson correlation coefficients (r) of 0.97, 0.96, and 0.97 for the MOE, VNO, and OB replicates, respectively (Fig. 5).