A total of 210 she buffaloes during their transition period (−30 to +30 days) were selected randomly from dairy farms and the veterinary referral hospital, Faculty of Veterinary Sciences and Animal Husbandry, Sher-e-Kashmir University of Agricultural Sciences & Technology of Jammu, R.S. Pura, UT of Jammu & Kashmir, India. Animal experiment was approved by the Committee for the Purpose of Control and Supervision of Experiments on Animal (CPCSEA), New Delhi (No.: 25/15/2018/CPCSEA). These buffaloes were categorized into three groups, on the basis of days of their transition period: Gp-I (−30 days), Gp-II (near parturition), Gp-III (+30 days). Buffaloes selected for this study had a body weight of 724.00 kg (±60), previous lactation length of 286 days (±7) previous lactation milk yield 1,800 to 2,058 kg (±125), body condition score of group-I; 3.04 (±0.04), group II; 2.914 (±0.038); and group III; 2.8 (±0.029) and with parity between 3rd-6th. The buffaloes were given stall feeding (containing 25 to 30 kg of dry fodder, salt lick and ad libitum water provided.

Blood samples were collected from each buffalo of all the groups on pre-decided days (a month pre-partum, close to partum and one-month post-partum) in ethylenediamine tetra-acetic acid (EDTA) contained vacutainer; VACUETTE (Cat. 455036, Greiner bio-one, Austria) for extraction of total ribonucleic acid (RNA) from peripheral blood mononuclear cells and in serum clot activator vacutainers; VACUETTE (Cat. XLGA-C5; Greiner Bio-One India Pvt. Ltd., Noida, India) for estimation of BHBA levels using enzyme linked immunosorbent assay (ELISA) kits (Immunotag; GBiosciences, St Louis, MO, USA). Blood samples collected in heparin containing vacutainers VACUETTE (Cat. 455051; Greiner bio-one, Austria) were used for estimation of oxidative biomarkers; glutathione peroxidase (GPX), LPO, SOD, and catalase. Buffaloes in mid-lactation period were considered a control group for estimation of differential gene expression of interest. Glucose concentration in whole blood (mg/dL) was estimated with Accu-check glucometer immediately with fresh blood.

Urine samples from the buffaloes were collected in sterile Uricol (Cat. PW016; Himedia, Mumbai, India). Then urine analysis strips (SD Urocolor, Standard diagnostics, INC, Kyonggi, Korea) were dipped in urine for seconds and compared with corresponding color chart on the bottle label and results were read immediately. Immediately after collection, Rothera’s test was performed with 1 to 2 gm of ammonium sulphate added in 5 mL of urine, 5% solution of sodium nitroprusside and concentrated ammonium hydroxide was layered over the sample to obtain purple ring within 2 to 3 min.

Total RNA was extracted from EDTA preserved blood by trizol method (Invitrogen Life Technologies, Carlsbad, CA, USA). Trizol reagent (Cat. 15596018; Thermo Fisher Scientific, Ahmedbad, India) which contains various components viz guanidine isothiocyanate, phenol and isoamyl which facilitate the extraction of RNA with high yield. EDTA contained blood was diluted in phosphate buffer solution (PBS) and then layered over the HiSep LSM-1077 (Cat. LS001; HiMedia, India). Subsequent centrifugation at 2,400 rpm for 24 min at 4°C was done in a refrigerated centrifuge (Centrifuge 5430R; Eppendorf India Ltd., Chennai, India). After separation of buffy coat in 1× PBS and its centrifugation at 17,800 rpm for 10 min thrice, a clear pellet was obtained at the bottom. TRIzol (Cat. 15596026; Thermo Fisher Scientific, India) was added after separation of supernatant and mixed properly with pellet. After a 30 to 40 min. at −20°C incubation period a TRIzol, chloroform (Cat. MB109; Himedia, India) and isoproponalol treatment was applied to the pellet. The pellet was subsequently washed three times with ethanol and collected by centrifuging at 7,500 rpm for 5 min. Following air drying of the pellet, 20 μL of diethyl pyrocarbonate (DEPC) treated water (Cat. R0601; Thermo Fisher Scientific, India) was added. Optical density (OD) of the final pellet was taken spectrophotometrically (Biospectrometer; Eppendorf India Ltd., India) to determine concentration and purity. Cut-off point of RNA concentration (μg/mL) recommended for cDNA synthesis was >1.0 μg/mL with ratio of 260/280 nm >1.5. Gel electrophoresis with 1% agarose of molecular grade containing 1 μg/mL of Ethidium bromide (Cat. H5041; Promega, New Delhi, India) was made with 1× TAE (Tris-acetate-EDTA) buffer (40× molecular grade, Cat. V4281; Promega, USA) treated in DEPC water (Cat. MB076; Himedia, India). Voltage 1 to 5 V/cm was applied across the gel until the bromophenol blue migrated to appropriate distance. Resultant band (28s band and 18s) in the gel was seen with gel documentation system (Vilber; Eppendorf India Ltd., India), under the illumination of UV light and confirmed the presence of RNA.

Total isolated RNAs were subjected to cDNA synthesis using Hi-cDNA synthesis kit (Cat. MBT076-100R; Himedia, India). A reaction mixture of 1 μg of RNA template, 2 μL of oligo-dtprimer (oligonucleotides that contained a segment of repeating deoxythymidines) (10 pmol) and 7 μL of nuclease free water in sterile polymerase chain reaction (PCR) tubes was made. This mixture was incubated for 5 min at 65°C in a thermal cycler (BioRad T100 Thermal cycler, Gurugram, India). Other components: 4 μL of RT buffer, 2 μL of 10× solution, 10 mMdNTP (Deoxynucleoside triphosphates) solution, 0.5 μL of ribo-nuclease inhibitor and 1 μL of reverse transcriptase enzyme were added in template RNA and primer mixture to make a final volume of 20 μL. This mixture was subjected to PCR conditions; one cycle of 42°C for 60 min, 70°C for 5 min and hold at 4°C in a thermal cycler. A negative control reaction mixture was made, containing all the kit components except RNA template.

Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as a house keeping gene for validation of cDNA; forward primer (AAGGCCATCACCATCTTCCA) and reverse primer (CCACTACATACTCAGCACCAGCAT). The reaction mixture totaled 20 μL containing 10 μL of PCR pre-mix (Thermo Scientific, India), 2 μL of cDNA, 0.3 to 0.5 μL of GAPDH forward and reverse primer (10 pm each) and nuclease free water to make the final volume in sterile PCR tubes. These reactions were subjected to PCR cyclic conditions and gel electrophoresis for the validation of cDNA products when compared with negative control (Table 1).

The analyses of gene expression levels and gene ontology (GO) annotation were carried out with GO-Elite software, using default options [19]. Microarray expression values were reported as log2 values. Differentially expressed genes were identified using a combination of a >2-fold change in expression with a statistical significance of p<0.05 (moderated t-test) using Benjamini-Hochberg correction method [20]. Differentially expressed genes were subjected to hierarchical clustering to identify gene clusters within the control and test groups [21]. Gene-set enrichment analysis and comparison was executed using the GO-Elite in AltAnalyze, where only terms with an false discovery rate adjusted enrichment p< 0.05 was considered for further evaluation.

To identify tissue/cell markers in both ketosis and healthy samples, we first filtered the genes with expression level of >90 percentile of all samples, and then we compared the abundantly expressed genes in both ketosis and non-ketosis samples to identify both common and uniquely expressed genes. A gene was defined as enriched in tissue ‘X’ if the average expression of the gene in tissue ‘X’ was at least 3 times greater than its average expression in all other 66 tissues. We defined a tissue-specific gene marker as the gene not only enriched in tissue ‘X’, but also expressed highest in tissue ‘X’ and when expression in tissue ‘X’ was at least 1.5 times higher than its expression in any other tissues. Further, Marker Finder algorithm in Alt Analyze was performed within each independent dataset to derive putative cell-population-specific markers for gene-set enrichment [22].

Real time PCR helps in the amplification of the targeted molecule in real time. It was conducted by quantitative real time PCR cycler- MyGo mini (MyGo 002). A reaction mixture with 10 μL Hi-SYBr master mix (Cat. MBT074-100R; Himedia, India), 0.5 μL of forward and reverse primer of gene of interest and nuclease free water up to 20 μL volume was made and subjected to qRT-PCR reaction condition (Table 2). For calculation of Cq values (quantification cycles), comparative CT method also referred to as the 2^−ΔΔCt method was used with internal gene control; GAPDH and buffaloes in mid-lactation as control group [23].

Various oxidative stress markers such as SOD, GPx, LPO, and catalase were evaluated from transition buffaloes (Group-I, II, and III). For evaluation, 5 mL of blood in heparin containing vacutainer was taken and centrifuged at 3,000 rpm for 10 min. After that, plasma was stored and left over red blood cells (RBC) lysate was used to determine the antioxidant status. Washing of RBC lysate was done with normal saline by diluting in a ratio 0f 1:1 and centrifuged for 10 min at 3,000 rpm. Washing was done thrice. 1% lysate (100 μL of lysate and 990 μL of distilled water) and 33% lysate (330 μL of lysate and 640 μL of PBS with pH 7.4) was prepared.

Lipid peroxidation in erythrocytes was determined by the evaluation of malondialdehyde (MDA) production by the method of [24]. In brief, 1 mL of 33% RBC haemolysate was taken and added to 1 mL of 10% w/v of trichloracetic acid. After thorough mixing, mixture was centrifuged at 3,000 rpm for 10 min and supernatant was extracted. To 1 mL of the extracted supernatant added 1 mL of 0.67% w/v of thiobarbituric acid and kept in water bath for 10 min, cooled and diluted with 1 mL of distilled water. The same reagents were used except the haemolysate for control samples. Absorbance was noted at 535 nm in the spectrophotometer (Biospectrometer; Eppendorf India Ltd., Country).

Calculations were done using the extinction co-efficient (EC, 13,100 M⁻¹cm⁻¹) and results expressed in mM MDA per mL of blood, using the following formula.

The plasma catalase activity was determined by the method described by Marklund and Marklund [25]. In brief, 2 mL of phosphate buffer along with 20 μL of 1% lysate were taken. This mixture then incubated with 1 mL of 30 mM of hydrogen peroxide at 37°C and the decrease in the absorbance was observed at every 10 s interval for 1 min at 240 nm OD. The catalase activity was expressed as μmoles of H₂O₂ utilized/min/mg Hb using 36 as molar extinction coefficient of H₂O₂.

The activity of erythrocytic SOD was determined by method of Hafeman et al [26]. In brief, 1.5 mL of 100 mM tris HCl buffer, 20 μL of 1% haemolysate, 0.5 mL of 6 mM EDTA and 1 mL of pyrogallol was added. The rate of auto-oxidation of pyrogallol was taken from the increase in absorbance at 420 nm in a spectrophotometer, every minute after a lag of 30 s up to 4 min. For the test, an appropriate amount of enzyme was added to inhibit the auto-oxidation of pyrogallol to about 50%. A unit of enzyme activity is defined as the amount of enzyme causing 50% inhibition of the auto-oxidation of pyrogallol observed in blank. The activity of SOD was expressed as SOD units/mg protein.

The activity of GPx in erythrocytes was performed as per method of Loor [27]. In brief, 0.1 mL of erythrocytes was taken. To this was added 1 mL of glutathion peroxidase, 1 mL of 0.4 M sodium phosphate containing EDTA, 0.5 mL of 0.01 M of NaN₃ and distilled water to make a volume of 5 mL. The solution was incubated for 5 min. After the incubation, 1 mL of 125 mM hydrogen peroxidase added and r incubation continued for about 3 min. 4 mL of meta-phosphoric acid was added to 1 mL of liquid from the incubation mixture. Then, 2 mL of this solution was added with 2 mL of 0.4 M sodium hypo-phosphate (NaHPO₄) and 1 mL of DTNB (5,5’-dithio-bis-2-nitrobenzoic acid) reagent. For the blank solution, water was used instead of hydrogen peroxidase and rest of the procedure was same. OD was taken at 412 nm against blank solution.

All statistical analyses were performed using SPSS (Statistical Package for the Social Sciences) statistical software version 16 (IBM SPSS Statistics 16). Data were analyzed using analysis of variance and the Duncan’s multiple range test was performed to provide significance levels (p< or >0.05) for the difference between the means.

Urine sample analysis of the selected buffaloes showed negative result for Rothera’s test and urine strips (SD Urocolor) for ketone bodies when compared with their positive control (Figure 1).

Serum level of BHBA (nmol/mL) increased significantly (p<0.05) from pre-partum to post-partum period; Gp-I to Gp-III (313.96±2.81 nmol/mL, 397.36±1.69 nmol/mL, and 483.69±1.98 nmol/mL, respectively) and glucose concentrations showed significant decline (p<0.05) from far off dry period to early lactation (Gp-I, 65.31±3.01; Gp-II, 59.44± 1.78; and Gp-III, 53.08±1.74), although buffaloes in these groups were non-ketotic (Figure 2).

Compared with normal non-ketosis samples, 2,121 genes were found to be differentially expressed in the ketosis samples. Among these 1,053 were found to be upregulated and 1,068 were found to be downregulated. Principal component analysis of the two groups revealed that all ketosis samples were sharing one component, and all healthy samples were sharing other principal components along with the heat map showing log fold changes of gene and hierarchal clustering of various healthy and ketosis samples (Figures 4 and 5).

Gene ontology analysis revealed that some of the genes were directly related to the immune response, inflammatory response, and chemotaxis and leukocyte cell-cell adhesion. Upregulated genes included interleukin-6 (IL6), interleukin-3 (IL3), C-C chemokine receptor type 1 (CCR1), interferon beta 1 (IFNB1), acetylserotonin O-methyltransferase like (ASTML), interleukin-12 receptor, beta 1 (IL12RB1), and leptin (LEP). Pathway analysis of the up-regulated genes was related to the immune response, with most genes involved in the cytokine-cytokine receptor interaction (Figure 6), induction, regulation of the local inflammatory response and oxidative stress related mechanisms. The group of down-regulated genes in the ketosis guanylate cyclase activator 1B (GUCA1B), general receptor for phosphoinositides-1 (GRTP1) found to involve in Glycosphingolipid biosynthesis (Figure 7), fat digestion and absorption pathways and in fatty acid βoxidation.

A total of 60 ketosis- specific genes were determined by Pearson correlation coefficient. The top 10 correlated genes determined with at least a 0.3 Pearson correlation coefficient were selected as the gene signature for each cell cluster (Figure 8). The top 20 genes identified as markers in Ketosis include IL6, IL3, CCR1, IFNB1, IL12RB1, ASTML, and LEP. On the other end, the top 10 genes identified as marker genes in healthy samples were beta-1,3-N-acetylglucosaminyltransferase (B3GNT5), beta-1,4-galactosyltransferase (B4GALT3), and fucosyltransferase 9 (FUT9) (Table 3).

Out of 20 potential markers of ketosis, two were selected (ASTML; upregulated and GUCA1B; downregulated) and analysed by qRT-PCR technique (Table 4). Additional two sets of genes (CPT1A; upregulated and IGF-1; downregulated) that were not a part of microarray assay were also subjected to qRT-PCR cycling condition (Table 4). They have role in β-oxidation during post-partum stimulation of hepatic fatty acid oxidation to maintain energy demand via gluconeogenesis [28] and uncoupling of growth hormone-insulin growth factor-1 (GH-IGF-1) axis causing downregulation of liver growth hormone receptor-1A (GHR 1A) and IGF-1 level, ultimately elevation in the growth hormone concentration that antagonizes the function of insulin and causes lipolysis and gluconeogenesis in early lactation [29] respectively.

Relative mRNA expression of marker ASTML showed significant (p<0.05) upregulation as transition period progressed from prepartum period (Gp-I) to calving (Gp-II) and post-partum period (Gp-III); 0.836±0.02, 1.268±0.03 and 1.624±0.01 respectively (Figure 8). The same trend in the expression level of CPT1A was seen from Gp-I to Gp-III (0.96±0.06, 1.336±0.02, and 1.372±0.02) (Figure 9). Whereas downregulation in the expression level of GUCA1B was seen from pre-partum period (1.548±0.05) to calving (1.474± 0.08) and post-partum period (1.014±0.04) (Figure 10). The expression fold of IGF-1 marker showed a significant (p<0.05) trend of downregulation from pre-calving (1.063±0.05) to near calving (0.839±0.02) term thereafter, gradual increase in the concentration as post-calving period progressed (1.123 ±0.02).

The imbalance between oxidative marker and antioxidant markers level causes oxidative stress. On analysis of these antioxidant and oxidative markers, our results showed a significant difference (p<0.05) amongst these groups providing sufficient evidence about oxidative biomarkers and antioxidants production during transition period in buffaloes.

Our study showed significant (p<0.05) increase in the mean±standard error (SE) values of MDA (nmoles of MDA produced/g of Hb/h) in Group-II (near parturition) with comparison to Group-I (−30 days) and Group-III (+30 days). Whereas, mean±SE values of SOD (U/mg of Hb) and catalase (CAT; μmoles) showed a significant decrease (p<0.05) in their values from Group-I (−30 days) to Group-III (+30 days) (Table 5). Blood GPx (mg/Hb) levels showed trend of significant decrease (p<0.05) from pre-partum period (−30 days) to near parturition time. Although, a significant increase (p<0.05) in the levels of GPx was seen in buffaloes when transition period progressed to +30 days (Table 5).