Two groups of five primiparous goats from divergent selection for extreme breeding values for the somatic cell score (one group from the high SCS line and the other from the low SCS line) were selected to have similar milk production (3.2 ± 0.5 kg/d) and provided by INRA. Basically, an experimental genetic evaluation was performed using data from 140,000 primiparous alpine goats in 3625 flock by year combination. The trait considered was the lactation mean of monthly somatic cell scores (LSCS) as detailed in Rupp et al. . Breeding values for LSCS were expressed in genetic standard deviation with inverted sign (positive index are favorable). Two groups of 5 and 6 bucks with extreme high and low breeding values for LSCS (−1.38 ± 0.68 vs 1.17±0.48, respectively) were selected to sire daughters at the INRA experimental facility of Bourges (UE0332, OSMOY, France). Out of 52 born kids, two groups of 5 daughters (sired by 3 and 4 high and low SCC bucks, respectively) were transferred two months before first kidding to “Centro Zootecnico Didattico Sperimentale dell’Università degli Studi di Milano-Italy” (http://www.veterinaria.unimi.it/Facolta/2586_ITA_HTML.html) for experimental infection.
The study was carried out when the goats were of first parity and at the peak of lactation (48 ± 2 days in milking). Goats were monitored for intra-mammary infections (particularly for S. aureus) from parturition to the day of challenge with bacteriological analysis and somatic cell scores measured on milk samples each week as described by Pisoni et al.  and Moroni et al.. Fore milk samples were collected on the three days immediately prior to the experimental infection and all animals were shown to be free of any mastitis pathogens and to have SCC below 250,000 cells/mL.
All experimental procedures were performed according to the Italian legislation, following approval by the ethics committee of University of Milan.
Staphylococcus aureus strain
S. aureus strain DV137, which was originally isolated from chronic case of caprine mastitis , was used for the experimental infections. S. aureus DV137 is positive for clumping factor, free coagulase, enterotoxins C and L, toxic shock syndrome toxin TSST-1 and leukocidin LukDE.
The S. aureus strain DV137 was grown from an individual colony by transfer to 10 mL of brain heart infusion broth (Becton-Dickinson Diagnostic Systems, Inc., Milan, Italy) and incubated for 6 h at 37°C. Thereafter, 1 mL of the culture was transferred to 99 mL of tryptic soy broth (TSB, Difco, Milan, Italy) and incubated overnight at 37°C. The concentration of the bacterium in this stock culture was maintained at 4°C overnight. Before infection the stock was diluted with sterile pyrogen-free Phosphate Buffered Saline (PBS, Invitrogen, Milan, Italy) to give a final concentration of 103 Colony Forming Unit/mL (CFU/mL) for the experimental infection.
Prior to intra-mammary challenge the goats were milked by hand and their udders emptied; the teat ends were carefully disinfected with chlorhexidine. The left udder half of each goat was infused with 1 mL (103 CFU/mL) inoculum of S. aureus. The right udder half was infused with 1 mL of sterile pyrogen-free PBS. Inoculation was administered intra-cisternally through the teat canal using a sterile blunt needle. Milk samples were collected from all goats from both udder halves just prior to the challenge (T0) and at 6 (T1), 12 (T2), 18 (T3), 24 (T4) and 30 (T5) hours after the experimental infection. At each time point each goat was recorded for: general demeanour, stance, position, food/water intake and vocalization, temperature and the udder examined for temperature, swelling, colour, pain, lumps, injuries to the teats/udder, milk letdown, milk colour and milk clots. A paired t-test (with threshold for statistical significance set to 0.05) was applied to body temperature of LSCS and HSCS to test if the differences observed at each time point were significant.
Bacteriology and milk somatic cells analysis
Twenty ml milk samples were taken under strictly hygienic conditions from each udder half into two sterile 10 ml tubes. One of the tubes was used for bacteriology, while the other tube of milk was used for SCC analysis. Subsequently, 200 mL of milk were collected from the left infected udders for isolation of cells and further analysis. A paired t-test (with threshold set to 0.05) was applied to milk SCC of right udders in LSCS and HSCS goats to test if the changes at each time point were significant.
Milk SCC were determined in 2-mL of milk, which was heated for 15 min up to 60°C then maintained at 40°C until analysed, in duplicate, on automated fluorescent microscopic somatic cell counter (Bentley Somacount 150, Bentley Instrument, Milan, Italy). For differential cell counts (DCCs), 10 ml of milk were centrifuged for 15 min at 400 x g, the cream layer and supernatant were discarded and the cells were washed twice in PBS. DCCs of isolated cell suspensions were estimated by means of esterase stain . Evaluation of the slides was carried out using light microscopy and oil immersion (100-fold magnification). One-hundred cells of each slide were counted meander-shaped and defined as lymphocytes, macrophages, PMN and epithelial cells. Cell identification was achieved using standard methods [47, 48].
Aseptically collected milk samples were diluted in sterile PBS and plated onto blood agar plates. The number of CFU was determined after 16 h of incubation at 37°C. Colonies displaying haemolysis were initially counted as S. aureus, and subsequently confirmed microscopically and biochemically by the presence of Gram-positive cocci that were both catalase- and coagulase-positive. S. aureus isolates were tested by RAPD-PCR  to confirm that they were the strain used in the experimental challenge.
Blood sample analysis
Ten ml of blood samples were collected into EDTA from the jugular vein just prior to the challenge and 6, 12, 18, 24, and 30 hours after inoculation.
Blood samples were analyzed using the ADVIA 120 (Siemens Healthcare) haematology system and the multispecies software provided by the manufacturer. The following parameters were calculated: erythrocyte count (RBC), haemoglobin concentration, haematocrit, mean corpuscular volume (MCV), mean corpuscular haemoglobin (MCH), mean corpuscular haemoglobin concentration (MCHC), leukocyte count (WBC), platelet count (Plt), mean platelet volume (MPV). Leukocyte different counts (percentage and number of neutrophils, lymphocytes, monocytes, eosinophils, basophils and large unstained cells or LUC), were also performed. A paired t-test (with threshold set to 0.05) was applied to blood sample parameters of LSCS and HSCS animals to test if differences at each time point were significant.
Isolation of white blood cells (WBC)
Ten ml of fresh blood were transferred into a 50 ml tube and 20 ml of 0.2% NaCl were added, mixed with 5 ml of NaCl 3.7% and then centrifuged at 1000 x g for 10 min at 4°C. The pellet was resuspended in 15 ml of PBS and the solution was centrifuged at 1000 x g for 10 min at 4°C. The supernatant was discarded and the pellet was resuspended in 1 ml of Trizol (Invitrogen, Milan, Italy).
Preparation of milk somatic cells
Fore milk was collected aseptically from both halves of the udder of each goat prior to the challenge and 6, 12, 18, 24, and 30 hours post-infection. Fifty mL of milk were transferred to falcon tubes and centrifuged at 750 x g at 4°C for 10 min. The fat layer and the supernatant were discarded and the cell pellet was re-suspended and washed in PBS pH 7.2. After a centrifugation at 450 x g for 10 min, the supernatant was discarded and the pellet was resuspended in 3 mL of Trizol (Invitrogen, Milan, Italy).
From WBC and somatic cells total RNA was extracted following the instructions of the supplier (Invitrogen, Milan, Italy), further purified using an RNeasy MinElute spin column (Qiagen, Milan, Italy) and eluted in RNase-free water. RNA was quantified using a NanoDrop spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA) and quality-checked using a Bioanalyser 2100 (Agilent, Santa Clara, CA). RNA samples with RNA Integrity Number (RIN) values between 7.0 and 10.0 were used for the microarray analysis.
The microarray design has been described previously , briefly all available bovine transcript sequence information was downloaded from Ensembl release 50, and Unigene and dbEST databases (Sept. 2008). A bioinformatic pipeline was created to align the sequences and select a unique set of minimally redundant bovine transcripts. This dataset was used to design 43,768 unique probes with a length of 35 nucleotides, each representing a single bovine transcript. The microarray probes were designed at the 3′ end of the bovine sequences. These probes were synthesised in duplicate, along with negative and quality controls, on a 90K feature custom array from CombiMatrix (Seattle, WA).
Array hybridization and statistical analysis
One μg RNA was amplified and labelled with Cy5-ULS using the RNA Amplification and Labelling Kit from CombiMatrix (ampULSe Cat. no. GEA-022; Kreatech Biotechnology, Amsterdam, The Netherlands). All procedures were carried out according to the manufacturer’s protocols. The purified labelled aRNA was quantified using a NanoDrop spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). Four μg of labelled RNA were fragmented to a uniform size and hybridized to the custom array following the Combimatrix CustomArray 90K Microarray Hybridization and Imaging Protocol Arrays were stripped and re-hybridised using the CustomArray Stripping Kit for 90K (CombiMatrix Cat. No. 610049) following the protocols of the manufacturer. Each array was used up to 4 times with no deterioration in signal or increase in background.
To verify the powerful of this bovine custom array for the caprine gene expression study, a series of technical replicates were carried out showing a signal intensities correlation higher than 97% (data not shown).
The hybridised arrays were scanned with a GenePix 4000B microarray scanner (Axon, Toronto, CA) and the images (TIF format) were exported to the CombiMatrix Microarray Imager Software, to perform quality checks of the hybridizations and the spots on the slide. Data were extracted and loaded into the R software using the Limma analysis package from Bioconductor. A design matrix was created using Limma functions to describe the experimental samples and replicates. The raw intensities were processed using quantile normalization and data were then transformed in log2 to be used for the statistical analysis.
Limma performs a linear regression analysis on the hybridizations, using a group-means parameterization approach to compare the different conditions and performs a false discovery rate adjustment with Benjamini-Hochberg correction for multiple testing . False discovery rates of 1% and 5% were accepted for SC and WBC respectively and DE genes were selected using an adjusted P-value cut off equal to 0.01 and 0.05.
The microarray data files have been deposited in NCBI’s Gene expression Omnibus (GEO; http://www.ncbi.nlm.nih.gov/geo/) at the identifier number GPL14856 and the experiment at the identifier number GSE33894.
Assignment of affected genes to pathways, networks and biological functions
Each gene symbol of the affected genes identified with R was mapped to its corresponding gene object in the Ingenuity Pathways Knowledge Base. Feeding the lists of affected genes as input to the IPA library identified associated canonical pathways, biological functions and networks which were used to investigate the biological context.
The IPA library items were ranked based on significance of association with the input list of genes. For the canonical pathways this significance was determined based on two parameters: (i) ratio of the number of genes from the input data set that map to the canonical pathway divided by the total number of genes of that pathway and (ii) p-values calculated using Fischer’s exact test determining the probability that the association is explained by chance alone.
For the biological functions and networks the significance was linked to the p-value only, calculated by right-tailed Fisher’s exact test. The p-values for the network analysis take into account the number of affected genes in the network and the size of the network.
Data from all time points were loaded into R software, the signal intensities were processed and normalized using standard Limma procedures, then all the expression data were put together.
MetaMA package was used to perform a moderated t-test with a Benjamini Hochberg (BH) correction at a 0.01 % threshold (to take into account the multiple testing problem) to each time point compared with T0. Once obtained, the p-values from study-specific analysis we combined them into one p-value in sense of sum of logs.
The Venn diagram was built using a modified version of R script “Venn”.
Real time PCR validation
For the optimization of all real-time assays, ten milk RNA samples at T0 were pooled and normalized to a final concentration of 200 ng/μl and the same procedure was used for samples from T4 and T5. RNA from 10 WBC samples at T0 were pooled and normalized to a final concentration of 100 ng/μl and the same procedure was applied to T5 samples. Pooled samples were serial diluted and used for the set-up of the standard curve (5 points in triplicates); in each assay a negative control was also included.
Samples were analysed one by one: 1 μg of milk and 0.5 μg of blood samples were individually reverse-transcribed using the Superscript II RT-PCR System (Invitrogen Life technologies) following the manufacturers instructions. Each sample was tested in triplicates.
Eleven genes from the milk samples and four from blood samples showing high FC and p-values different between post (T4, T5) and pre-infection (T0), were selected for quantitative real-time PCR validation of microarray results.
Caprine specific rtPCR primers were designed using goat sequence where this was available. When goat sequence was not available transcript-specific primers were designed for the conserved regions of the bovine sequence using Primer Express software (version 3.0) running standard settings.
ACTB (actin β), GAPD (glyceraldehyde 3-phosphate dehydrogenase), HMBS (Hydroxymethylbilane synthase), RPL13A (Ribosomal protein L13a), and YWHAZ (Tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein, zeta polypeptide) were selected from literature  and tested in order to choose a reference gene common to milk and blood analyses according to the variation in the array data. Control cDNA dilution series were created for each gene to establish a standard curve; all real-time reactions were performed in triplicate.
The real-time reaction mixture, in a final volume of 10 μl, included 2 μl dilution 1:10 of the cDNA as template, 1X Power SYBR Green Master Mix (Applied Biosystems, Foster City, CA, USA) and 0.5 μM of each primer forward and reverse (Additional files 6 and Additional file 7) except for blood-gene egf-like module containing, mucin-like, hormone receptor-like 1, where final concentration was 0.05 μM for primer forward and 0.9 μM for primer reverse. The real-time PCR reaction set up was made in 384 optical well plates with a Freedom Evo Robot (Tecan) and carried out on an ABI 7900HT Fast Real-Time PCR System (Applied Biosystem) with a standard programme (50°C*2’/95°C*10’/40 cycles 95°C*15” and 60°C*1’). Data were analyzed with the GeneAmp 7900HT sequence detection system software (PerkinElmer Corp., Foster City, USA).The log input amount of the standard curve was plotted versus the output Ct values and the log input amount of each sample was calculated according to the formula (Ct - b)/m, where b is the Y-intercept, and m is the slope. The log input amount was converted to input amount according to the formula 10^(log input amount), and triplicate input amounts were averaged for each sample.
Data were imported and processed into R using the ddCt analysis package from Bioconductor, time point T0 and RPL13A were set as reference sample and reference gene respectively and the 2-ΔΔCt algorithm was applied to find the relative level expression .