Animals
Samples used in this study were collected from nineteen male dromedary camels aged 4-5 years old, and with a body weight range of 276~416 kg. Body weights for all groups were calculated at baseline date and every five days thereafter using the formula, live weight (Kg) = Shoulder height x chest girth x hump girth x 5068. The camels were supplied with alfalfa hay as feed and were ranch-housed outside Al Ain, United Arab Emirates during the hot months (April and May) of 2016, under careful veterinary supervision. After a short adaptive phase, the camels were divided into a control group (n=5), a WD group (n=8), and a rehydrated group (n=6). The control group had free access to food and water throughout the experimental period. The WD group was supplied with food ad libitum but without access to water for 20 days. The rehydrated group was WD for 20 days followed by an unlimited water supply for 3 days. Blood samples were collected during the experiment by jugular venipuncture. After the experiment, the camels were sacrificed in the local central abattoir for human consumption. The hypothalamus samples were harvested, frozen on dry ice and shipped frozen on dry ice to the University of Bristol under the auspices of a DEFRA Import Licence (TARP/2016/063) and stored at -80 °C. This study was approved by the Animal Ethics Committee of the United Arab Emirates University (approval ID: AE/15/38) and the University of Bristol Animal Welfare and Ethical Review Board.
Plasma ANG II assay
Blood samples were collected between 8:00 and 9:00 am (if not specified) from all groups into heparinized vacutainers, on ice, for hormone plasma measures. Blood from control and WD groups were collected on days 0, 5, 10, 15 and 20 of WD. For the rehydrated group, blood samples were taken on day 0, and at 0, 1, 5, 8, 12, 24, 48 and 72 hours (hrs) following the re-administration of water. Plasma ANG II concentration was determined by specific radioimmunoassay using T-4007 antibody from Peninsula Laboratories, Inc. (San Carlos, CA, USA) as described in a previous study69. The ANG II assay sensitivity was 0.39 pg mL-1 and the intra‐ and inter‐assay variations were 5.9% and 8.3%, respectively.
Genomic DNA sequencing
To clone the dromedary AVP and OXT genes, genomic DNAs were extracted from camel kidney samples using the DNeasy Blood & Tissue Kit (QIAGEN, 69504), then amplified by PCR (primer sequences available in Supplementary Data S5a) using the Phusion High-Fidelity Master Mix with GC Buffer (Thermo Fisher Scientific, F532L), extracted from the gel using the QIAquick Gel Extraction Kit (QIAGEN, 28704, 28706, 28506 and 28115), A-tailed and ligated into the pGEM®-T Easy Vector Systems (Promega, A1360), and transformed into DH5-alpha competent cells (Thermo Fisher Scientific, 18265017), following the manufacturers’ instructions. Vectors harbouring the inserts of interest were harvested using PureYieldTM Plasmid Miniprep System Kit (Promega, A1222) and sequenced by Eurofins Genomics using the Sanger dideoxy sequencing method. In order to predict the gene features (i.e., CDS, exons, introns), the obtained sequences were aligned to the published Camelus ferusAVP (NCBI: XM_032461957.1) and Camelus dromedariusOXT (NCBI: MF464533.1) genes using the Align Sequences Nucleotide BLAST online tool, and also with the dromedary reference genome Camdro2 (GCA_000803125.2)70. The sequenced dromedary camel genes with predicted structural features were released to GenBank (AVP accession number: OM963135, OXT accession number: OM963134).
Hypothalamic sample processing
The methodologies for identifying and dissecting the SON from the camel hypothalamus samples are illustrated in Figure S5. Utilizing the 3V and OX as landmarks, the ventral part of the hypothalamus sample containing the SON was dissected (Figure S5A). The rostral and caudal orientations of the SON part were confirmed by recognizing the formation of the OX from OT (Figure S5B) before being mounted to the cryostat sample holder. The brain was sliced into 16 μm thick coronal sections along the rostral-caudal axis using a cryostat set at -20 °C (Leica, CM3050 S). The sections were mounted on Superfrost® Plus slides (Thermo Fisher Scientific, J1800AMNZ) and MCNS of the SON were identified by staining with toluidine blue. This region appeared by eye as a light brown region due to the highly vascularized feature of SON, which was lining the dorsal surface of OX and then lengthening dorsolaterally (Figure S5C). Once identified, sections were collected in a slide box in the cryostat chamber, stained with toluidine blue every 10th slide to trace the journey of the SON. More caudally, a second subregion of MCNs of the SON with was identified between the OX and the 3V (Figure S5D). The two subregions of SON were denominated as the rostral SON and the caudal SON. The sections were stored in slide boxes at -80 °C.
RNA in situ hybridization (RNAscope)
RNAscope probes for dromedary AVP (GenBank: OM963135) and OXT (GenBank: OM963134) mRNA were designed by Advanced Cell Diagnostics (ACD) (probe sequences available in Supplementary Data S5b). The mRNA distribution of AVP and OXT transcripts in the SON of the dromedary camel was analysed by RNAscope Multiplex Fluorescent Assay (ACD, 320851) following the manufacturer’s protocol. Briefly, frozen camel brain 16 μm thick sections mounted on slides were fixed in 4% (w v-1) paraformaldehyde (PFA) for 15 min on ice, and then immersed in a series of ethanol solutions with increasing concentration (50%, 70%, 100%, 100% v v-1), 5 min incubation for each concentration. Slides were air-dried at room temperature (RT) for 5 min before being subjected to the protease IV treatment for 30 min at RT in a humidity control tray. Brain sections were hybridized with probes in humidity control tray at 40°C for 2 hrs. Hybridization signals were amplified with RNAscope® Fluorescent Multiplex Detection Reagents (ACD, 320851). DAPI from the RNAscope kit was applied to the brain sections for 45 secs at RT. Brain sections were then coated with mounting medium FluoroshieldTM histology mounting medium (Sigma, F6182), coverslipped, and stored at -20 °C. Images were captured with a widefield microscope (Leica, DMI60000) or a confocal (Leica, SP5-II) fluorescent microscope and were analysed using ImageJ (bundled with 64-bit Java 1.8.0_112).
3D modelling of camel SON
A 3D reconstruction of the dromedary camel SON was built from RNAscope images. Software including ImageJ (bundled with 64-bit Java 1.8.0_112) and MATLAB (matrix laboratory) were used to process the images. Regarding the landmark structures of hypothalamus (3V and OX), the images were aligned manually using the trakEM2 (blank) function of ImageJ. The brightness and contrast of each channel were also adjusted to increase the signal/noise ratio in ImageJ. The processed images were then firstly scaled down in XY by a factor of 2 using bicubic scaling in ImageJ and then interpolated to aid the visualisation of the 3D data in MATLAB (MATLAB R2018a, version 9.4.0). Briefly, each colour channel was loaded separately, and data was linearly interpolated along the z direction using built-in MATLAB interp3 function to interpolate from the acquired 12 z planes to the desired 96 planes. Data was then saved as individual tiff images, reconstructed to be RGB image, and processed using the 3Dscript plugin of ImageJ to build the 3D model.
RNA extraction
Hypothalamic samples were sliced into 100 μm thick coronal sections in a cryostat set at -20 °C (Leica, CM3050 S). The start of the SON was mapped by staining with toluidine blue. SONs were punched from 47~60 consecutive slices per sample using a 1mm micro punch (Fine Science Tools, 18035-01). The punches were dispensed into 1.5 ml tubes maintained on dry ice within the cryostat chamber. At the end of collection, 1 mL of Trizol (Thermo Fisher Scientific, 10296010) and samples were thoroughly mixed by vortexing and stored at -80 °C. Total RNA was extracted using a Direct-zol™ RNA MiniPrep kit (Zymo research, R2052) following the manufacturer’s instructions. A Nanodrop spectrophotometer (Thermo Fisher Scientific, ND-1000) was used to determine the RNA concentration.
RNAseq
RNA integrity number (RIN) was established for each sample by using the Agilent 2200 TapeStation system (Agilent Technologies, 2503). Control and WD samples (n=5 for each condition) had an average RIN value of 6.27 (range 5.2-7). cDNA libraries were prepared using the TruSeq® Stranded Total RNA Library Prep (Illumina, 15031048) following the manufacturer’s instructions. The ribosomal RNA was depleted from the total RNA sample using rRNA removal beads. Following purification, the depleted samples were broken into small fragments (75 bp) that were used for first strand cDNA synthesis and a subsequent second strand cDNA synthesis. The cDNA fragments then went through a single adenine addition and ligation of adapter indices. Library amplifications were performed by PCR to generate the final cDNA libraries. Sample with average cDNA library size close to 260 bp (following the manufacturer’s protocol) and higher final library concentration was used for RNAseq. Approximately 200 ng individual libraries were sequenced using the NextSeq500 High Output Version 2.5, 2 x 75 bp kit (Illumina, 15057931) following the manufacturer’s manual. The samples generated averagely 22.2 million mapped reads per sample (range 17.7-37 million). Data was processed using the Real-Time Analysis (RTA) software (version 2.4.6).
cDNA synthesis and qRT-PCR
Total RNA (138 ng) was reverse transcribed using the GOScriptTM cDNA synthesis system (Promega, A276A). Primers of target genes (Supplementary Data S5c) were designed based on the reference sequences from National Centre for Biotechnology Information (NCBI)71, the genome assembly Camdro2 (GCA_000803125.2) and our sequencing data of AVP and OXT. All primers were synthesized by Sigma-Aldrich®. Intron-specific primers were designed to detect the heteronuclear RNA (hnRNA, pre-mRNA) for AVP and OXT72. The optimization and validation of primers was performed according to ABI protocols and relative standard curve method73. cDNA samples were used as templates for the qRT-PCR which was conducted in duplicates in 12 µL reaction volumes using PowerUpTM SYBR Green Master Mix (Thermo Fisher Scientific, 100029283) on an ABI StepOne-Plus Real-Time PCR System. For selecting reference gene of camel SON, the expression level of six commonly used reference genes (ACTB, B2M, GAPDH, HMBS, HPRT1 and PPIA) in rat hypothalamus74 were checked in our camel RNAseq data. The housekeeping gene HPRT1 was highly stable in expression under the experimental conditions, thus was selected as the reference gene for the qRT-PCR validation.
Statistics and reproducibility
For plasma ANG II measures over the 20 days of WD (n=14, data obtained from the WD treatment in both WD and rehydrated group) in comparison to the control (n=5), data was analysed using two-way repeated measures ANOVA with Šídák's multiple comparisons test in Graphpad Prism (version 9.1.0). Statistics available in Supplementary Data S6a. For plasma ANG II of the rehydrated group (n=6) over 72 hrs of rehydration in comparison to control and WD states, data was analysed using one-way mixed-effects model (restricted maximum likelihood) for repeated measures with Tukey’s multiple comparison test by using Graphpad Prism (version 9.1.0). Statistics available in Supplementary Data S6b.
RNAseq alignment and downstream data analysis were first performed in a Linux-based high-performance computer “Hydra” (PowerEdgeR820 12 core supercomputer; Dell, Round Rock, TX, USA). The paired end sequencing files (FASTQ) were first merged and then trimmed for adaptor sequences using BBDuk tool, followed up by a MultiQC quality check (version 1.9). A dromedary camel reference genome named Camdro2 (GCA_000803125.2) was indexed using Spliced Transcripts Alignment to a Reference (STAR) aligner (version 2.5.3a)75. The reads were aligned to the indexed genome. The resulting files were loaded into R (version 4.0.3)76. The mapped reads were counted by FeatureCounts77 where the number of aligned read pairs to each gene for each library were counted. Raw read counts were normalized using Median of ratios method78 inbuilt in DESeq279, differentially expressed genes between control and WD (n=5 for each condition) were identified using DESeq2. The inbuilt statistics in DESeq2 was Wald test with Benjamin-Hochberg adjustment79,80. Genes with padj≤0.05 were considered to have significant differential expression between groups. Gene annotations were retrieved by using an online tool g: Profiler (g:Convert, Orthology search)81.
Principle component analysis showing separations between control and WD conditions were plotted by inbuilt function of DESeq2 (ntop=500) based on the regularized log transformed read counts. For comparing different gene sets, unscaled venn diagrams were generated by using an online tool of Bioinformatics & Evolutionary Genomics (http://bioinformatics.psb.ugent.be/webtools/Venn/). Volcano plot showing statistical significance against LFC of genes was generated using ggplot2 package (version 3.3.3)82 in R. Lollipop chart was generated using ggplot2 to list DEGs by LFC ranking and to deliver information about statistical significance and expression abundance (baseMean: averaged normalized read counts across all samples) of the DEGs.
To compare the common DEGs between WD camel and rat, simple linear regressions and spearman correlation tests were performed using Graphpad Prism (version 9.1.0) on the LFC values of the genes in both species, and the absolute LFC and -log10padj values in each species. The absolute LFC values in the two species were compared by Wilcoxon matched-pairs signed-rank test (two-tailed) via Graphpad Prism (version 9.1.0). Statistics available in Supplementary Data S2f-h.
Gene ontology was performed using ClusterProfiler package (version 3.18.1)83 in R. For gene ontology of camel, we used the model organism human as references. First, the camel DEG Ensembl IDs were converted to the ortholog human Ensembl IDs by using g: Profiler (g:Orth, Orthology search)81, then converted to human Entrez IDs using AnnotationDbi package (version 1.52.0)84 and org.Hs.eg.db package (version 3.12.0)85, a genome wide annotation database for human, in R. Over-representation analysis86 was carried out based on the converted gene IDs and the human biological pathway and KEGG pathway annotation databases. Benjamin-Hochberg adjustment80 was applied for multiple comparison correction to reduce the false discovery rate. Pathways with padj≤0.05 were identified as significantly enriched pathways. Dot plots that visualizing the significantly enriched pathways ranked by padj and the associated DEGs with LFC and abundance (baseMean) in gene expression were plotted using ggplot2 in R.
The 2-ΔΔCT method was applied for the relative quantification of gene expression by qRT-PCR87. ΔCT is the difference in cycle threshold (CT) values of the gene of interest and the housekeeping gene and was used in statistical tests. ΔΔCT = ΔCT (treated sample) - ΔCT (control sample). 2-ΔΔCT was used for plotting. When comparing between control, WD and rehydrated camels (n=5 for each condition), qRT-PCR data was analyzed using Brown-Forsythe and Welch one-way ANOVA with Dunnett T3 post-hoc test in Graphpad Prism (version 9.1.0). Data was illustrated by box and whisker plots to show the dispersion of the dataset. Genes with padj≤0.05 were considered as significantly changed in expression. When comparing between control and WD camels (n=5 for each condition), qRT-PCR data was analyzed using two-way, unpaired t test with Welch correction in Graphpad Prism (version 9.1.0). Genes with p≤0.05 was considered as significantly changed in expression. Statistics available in Supplementary Data S6.