Microarray analysis of gene expression in blood neutrophils of parturient cows

Sally A. Madsen, Ling-Chu Chang, Mary-Clare Hickey, Guilherme J. M. Rosa, Paul M. Coussens, Jeanne L. Burton


It is well documented that blood neutrophils from parturient dairy cows do not perform as well as neutrophils from nonparturient cows in laboratory assays of adhesion, migration, or phagocytosis-induced respiratory burst. However, little is known about the possible molecular basis for parturition-induced changes in neutrophils. cDNA microarray analysis was used in the current study to explore parturition-induced changes in gene expression profiles in bovine blood neutrophils. Total RNA from isolated blood neutrophils of four parturient Holstein cows was obtained before, during, and after parturition, reverse transcribed into cDNA, and sequentially labeled with Cy3 or Cy5 dyes prior to paired hybridizations to 1,056 member bovine total leukocyte (BOTL-3) microarrays in a loop design. Resulting gene expression data were LOWESS normalized by array and analyzed using a mixed model approach. Results showed that expression profiles for 302 BOTL-3 genes were influenced by parturition. BLASTn analysis and preliminary clustering of affected genes by biological function indicated that the largest proportion (14%) of changed genes encode proteins critical to regulation of apoptosis. Independent confirmation of altered expression for 16 of these genes was achieved using quantitative real-time RT-PCR (Q-RT-PCR). A predominantly survival phenotype inferred from the microarray and Q-RT-PCR results was substantiated by monitoring apoptosis status of blood neutrophils from castrated male cattle cultured in the presence of sera from parturient cows. Thus our combined gene expression and apoptosis phenotyping results suggest that bovine parturition may induce prolonged survival in normally short-lived blood neutrophils.

  • bovine total leukocyte microarray
  • neutrophil biology
  • apoptosis
  • inflammation

mammals undergo varying degrees of immunosuppression and disease susceptibility during late pregnancy and parturition (8, 15, 17, 27). Leukocytes from dairy cows provide an excellent model for studying parturient immune suppression, as these cells exhibit impaired inflammatory responses that associate with leukocytosis and increased susceptibility of the animals to opportunistic bacteria, such as gram-negative coliforms that cause mastitis in infected mammary glands (12, 15). Negative effects of parturition are readily detected in bovine blood neutrophils (3, 23), which normally provide the main immunologic defense against mastitis-causing bacteria (reviewed in Ref. 1). Specifically, neutrophil adhesion, migration, and phagocytosis-induced respiratory burst activities become depressed in some parturient cows to such an extent that intramammary bacteria get the upper hand in the host-pathogen battle (reviewed in Refs. 1 and 15). Recent studies have begun to elucidate potential molecular bases for certain parturition-induced neutrophil dysfunctions, showing that transcripts encoding key adhesion molecules, mitochondrial proteins, and ribosomal proteins are significantly decreased in neutrophils following the surge in blood steroids at parturition (22, 33). In the future, a broader examination of gene expression and corresponding phenotypic changes in neutrophils of periparturient cows may help researchers better understand and circumvent innate immune dysfunction and disease in female mammals.

Recently, we have developed bovine-specific cDNA microarrays for studies on bovine immunobiology (2, 5, 38). Our preliminary microarray experiment used populations of bovine total leukocytes collected before and just after parturition to extend our previous findings (22, 33) that genes involved in leukocyte trafficking, phagocytic killing, maintenance of memory lymphocytes, energy metabolism, transcription, and translation are influenced by parturition (2). However, alterations in the proportions of circulating leukocyte subpopulations in blood samples used for that study precluded our ability to determine whether the observed gene expression changes were real or simply due to changing leukocyte populations that contributed mRNA for the microarray analysis. Therefore, in the current study, we elected to use isolated populations of blood neutrophils for gene expression profiling during the periparturient period using our third generation bovine total leukocyte (BOTL-3) cDNA microarray. We were particularly interested in understanding gene expression changes in neutrophils that might explain the pronounced increase in circulating neutrophil counts that occur for ∼2 days around parturition (28, 33). Our novel results suggest that patterns of expression for Bcl-2 family genes and Fas-related genes were changed by parturition in such a way as to suggest delayed apoptosis in the cells. Enhanced neutrophil survival was confirmed by follow-up phenotyping experiments and could partly explain parturition-induced neutrophilia.


Animals and sample collection.

Blood neutrophils utilized for all gene expression experiments of this study were obtained from periparturient Holstein cows (5 primiparous and 2 multiparous). Additional neutrophils were collected for phenotyping experiments from three young castrated male Holsteins (weighing between 225 and 325 kg). All animals were fed and housed according to standard operating procedures at the Dairy Teaching and Research Facility, and their use for the described experiments was approved by the All University Committee for Animal Use and Care, both of Michigan State University. Blood samples (30 ml) for neutrophil purifications intended for RNA isolations were collected by tail venipuncture into commercial acid citrate dextrose (ACD)-containing evacuated tubes (Vacutainer; BD Biosciences, San Jose, CA) from periparturient cows before (day −7), at (day 0), and after (days 0.25 and 1) parturition. An additional tube of blood (no anticoagulant) was taken from cows at each sampling for serum harvesting. Blood samples (60 ml) for neutrophil purifications intended for phenotyping assays were collected from male cattle through indwelling jugular cannulas into ACD-containing tubes. All blood samples were placed on ice as soon as they were collected and transported to the laboratory (∼7 min drive) for immediate processing.

Sample preparation.

Upon arrival at the laboratory, 200-μl aliquots of whole blood from each ACD-anti-coagulated sample were reserved for monitoring total leukocyte counts (no. cells/ml) by electronic counting (Beckman Coulter Z1 Coulter Particle Counter and Zap-Oglobin lytic reagent; Beckman Coulter, Miami, FL) and neutrophil differential counts [percent G1+ cells determined by immunostaining (clone MM20A; Veterinary Medical Research Diagnostics, Pullman, WA) and fluorescence-activated flow cytometry (FACSCalibur with CellQuest software; Becton-Dickinson, San Jose, CA)], as in Ref. 33. Numbers of circulating neutrophils (no. cells/ml whole blood) were then calculated as total leukocyte counts multiplied by percent G1+ cells. Remaining blood was used for neutrophil isolations (≥93% purity) using Percoll density gradients (1.084 g/ml; Amersham Biosciences, Piscataway, NJ) followed by hypotonic lysis of red blood cells, as in Ref. 33. For mRNA abundance profiling, neutrophils were immediately lysed in TRIzol Reagent (Invitrogen, Carlsbad, CA) for 10 min at room temperature and stored in the same reagent at −80°C until use. For apoptosis phenotyping, the cells were suspended at a concentration of 1 × 107 cells/ml in culture media and treated as described below. Total time from blood collection to lysis in TRIzol or suspension for culture was <3 h.

Total RNA was isolated from stored neutrophils according to the TRIzol manufacturer’s instructions, and concentration and purity were determined using a spectrophotometer (model DU-650; Beckman, Schaumburg, IL) and the 260- and 280-nm readings. RNA quality was checked in several randomly selected samples by 28S and 18S rRNA band visualization following gel electrophoresis and ethidium bromide staining (33).

Additional tubes of blood collected for serum harvesting were allowed to clot overnight (∼18 h) at 4°C. Samples were then centrifuged at 1,000 g for 30 min at 4°C. Following centrifugation, serum was transferred to microcentrifuge tubes in 1-ml aliquots and stored at −20°C until use in cell culture (see below).

cDNA microarray experiment.

Neutrophil RNA from four of the primiparous Holstein cows was used for the microarray experiment. The general cDNA spotting design (including all control genes) used on the BOTL-3 microarrays is described in detail elsewhere [6; National Center for Biotechnology Information Gene Expression Omnibus (NCBI GEO; http://www.ncbi.nlm.nih.gov/geo/) platform accession number GPL363] and included 3 spots for each of 1,056 leukocyte genes present on the arrays. Expressed sequence tags (ESTs) of these spotted genes were obtained from three sources: 710 from our group’s BOTL cDNA library (38), 14 from previous differential display RT-PCR experiments (22), and 332 from PCR-generated amplicons of targeted genes with known functions in immunobiology (5). These BOTL ESTs can be viewed on our website at http://nbfgc.msu.edu. Control genes spotted within and across the 48-patch microarrays (9 × 9 series of spots per patch) included 96 spots of synthetic lambda Q cDNA (external control), 144 spots of GAPDH, 75 spots of β-actin, and 75 spots of RPL-19 (internal controls).

To preliminarily screen for parturition-induced neutrophil gene expression changes, RNA was pooled by day relative to parturition across the four test cows such that 20 μg per sample was available for cDNA synthesis and two dye coupling reactions. Accordingly, there was no biological replication (n = 1) in this screening assay. Assay replication per test gene was n = 6 (3 spots per gene by 2 dyes). The RNA pools were separated into two aliquots of 10 μg each for reverse transcription and dye labeling using a loop design (16). The experiment included four BOTL arrays with each sample labeled with Cy3 (green) and Cy5 (red) fluorescent dyes (Amersham Pharmacia Biotech; Piscataway, NJ) and run in paired hybridizations on the arrays. Array no. 1 included a comparison between the day −7 and day 0.25 samples, array no. 2 between the day 0.25 and day 0 samples, array no. 3 between the day 0 and day 1 samples, and array no. 4 between the day 1 and day −7 samples. We employed this pairing strategy to directly compare the day −7 sample with the day 0.25 (array no. 1) and day 1 (array no. 4) samples because results from our previous studies suggested that the most pronounced differences in neutrophil gene expression may occur between day −7 and days 0.25 or 1 (22, 28, 33). Data from these microarray experiments can be found in GEO, with series accession number GSE544.

For cDNA synthesis, 10 μg of sample RNA was converted to cDNA (42°C for 60 min) using the Atlas glass fluorescent labeling kit (Clontech, Palo Alto, CA). Included in this reaction was 1.25 ng of lambda Q synthetic mRNA, which served as a cDNA synthesis and dye labeling control. Dye couplings and labeled cDNA purifications were performed according to the manufacturer’s instructions. The cDNAs were incubated at 70°C for 5 min just prior to array hybridization using a GeneTAC Hybridization Station (Genomic Solutions, Ann Arbor, MI). Arrays were incubated 2 min at 75°C prior to probe addition. Following probe addition at 75°C, hybridization conditions included 3 h at 65°C, followed by 3 h at 55°C, and finally 12 h at 50°C. This was immediately followed by two medium-stringency washes at 50°C, two high-stringency washes at 42°C, and two washes with postwash buffer at 42°C (wash buffers and postwash buffer were from Genomic Solutions). Finally, arrays were removed, rinsed in 2× SSC, and centrifuged in open 50-ml conical tubes (500 g for 3 min at room temperature) to dry them. Array scanning was done using a GeneTAC LS IV (Genomic Solutions) and accompanying software (version 3.01). Spot analyses were performed using GeneTAC Integrator microarray analysis software, version 3.3.0 (Genomic Solutions). Total intensity values for each dye channel were stored as comma-separated value data files and exported into Excel spreadsheets for subsequent loading into SAS for data normalization and analysis (see below).

Statistical analysis of microarray data.

Potential dye intensity biases in the microarray data sets were visualized using M vs. A scatter plots constructed for each array, where log intensity ratios M = log(Cy3/Cy5) = logCy3 − logCy5 were plotted against mean log intensities A = (logCy3 + logCy5)/2 for each array spot, as described by Yang et al. (37). Array-specific data normalization was then performed considering a robust local regression technique (4) using the LOESS (also known as “LOWESS,” for “locally-weighted regression and smoothing scatter plots”) procedure of SAS (30). The efficiency of LOWESS normalization was assessed by monitoring M-A plots for data from each array before and after LOWESS normalization. The normalized data were then back-transformed prior to further statistical analyses using the following formulas: logCy3* = A + /2 and logCy5* = A/2, where logCy3* and logCy5* are the normalized log intensities. Here, represents each of the normalized M values with being the LOWESS predicted value for each spot.

LOWESS-adjusted log intensities were then analyzed statistically using a mixed model approach consisting of two steps (36). The first step involved array-specific spatial variability normalization, and the second step involved gene-specific analyses, to test the effect of day relative to parturition on expression profiles for individual genes. The normalization model in the first step was as follows Math where yijklmn represents each observed fluorescent intensity signal; μ is an overall mean value; Ti is the main effect of time i (day relative to parturition); Dj is the main effect of dye j; Al is the main effect of array l; P(A)lm is the effect of patch m within array l; and εijklmn is a stochastic error (assumed to be normally distributed with mean 0 and variance σ2). The second step of the statistical analysis consisted of gene-specific models for the estimated residuals (ε^ijklmn) obtained from the normalization approach discussed above. These models were as follows Math where all the effects have the same definitions as for the normalization model, except that now they are specific for each gene so they carry an additional index k. Moreover, the error terms eijklmn were assumed to have independent normal distributions with gene-specific variances σk2. These analyses were computed using the MIXED procedure of SAS (30). In the gene-specific analyses, the day relative to parturition effect was declared significant if P < 0.01. Overall effects of parturition on gene expression profiles were first visualized using GeneSpring software (Silicon Genetics, Redwood City, CA). To do this, day relative to parturition least squares means of genes for which a parturition effect was suggested in the primary statistical analysis were loaded into the software, and expression levels on days 0, 0.25, and 1 were plotted as ratios to expression levels on day −7 (our “normal” expression control). Resulting profiles were then clustered using K-means clustering by general pattern of expression change across the four test days. In this way, clusters of genes that showed parturition-induced decreases were readily differentiated from genes that were induced by parturition. Next, the spotted cDNA sequences representing genes whose expression profiles were significantly influenced by day relative to parturition in this analysis were subjected to BLASTn analysis to reveal identities, and the biological functions of these genes were determined through an extensive PubMed literature search (http://www.ncbi.nlm.nih.gov:80/entrez/query.fcgi?CMD=search&DB=PubMed). This information was used to form an ontological clustering of affected genes into broad functional categories, which was then used to determine which genes we would confirm as changed in expression due to parturition using an independent assay on individual RNA from neutrophils of three additional cows.

Confirmation of altered mRNA abundance profiles using quantitative real-time RT-PCR.

Confirmation of altered expression for several genes in the predominant ontology cluster that were identified as differentially expressed in our preliminary microarray screening experiment was pursued through the use of quantitative real-time RT-PCR (Q-RT-PCR) in an Applied Biosystems 7000 DNA sequence detection system (PerkinElmer Applied Biosystems, Foster City, CA). The 17 genes selected for real-time PCR validation were chosen based upon three criteria: 1) documented importance to apoptotic cell death in neutrophils or other immune cells; 2) P value of effect of day relative to parturition in the microarray analysis; and 3) a mixture of upregulated and downregulated genes. Individual RNAs from Percoll-purified blood neutrophils of two multiparous and one primiparous periparturient Holstein cows (different from the 4 cows used for the microarray experiment) sampled on days −7, 0, 0.25, and 1 relative to parturition (on day 0) were converted into first-strand cDNA by combining 2 μg of the RNA with 10 mM oligo(dT)12–18 primer and sterile water in a 10-μl volume that was incubated 5 min at 70°C followed by 5 min at 20°C. Master mix containing 4 μl of buffer (supplied by the RT manufacturer; final reagent concentrations of 50 mM Tris·HCl, pH 8.3, 75 mM KCl, and 3 mM MgCl2), 200 U of SuperScript II RNase H reverse transcriptase (Invitrogen Life Technologies), and a final concentration of 10 mM DTT and 0.5 mM dNTP were added to achieve a final reaction volume of 20 μl. Reverse transcription was allowed to proceed at 42°C for 60 min, heated to 70°C for 15 min, and cooled to 37°C prior to the addition of 2 U of DNase-free RNase H (Invitrogen Life Technologies). Incubation at 37°C was continued for 20 min in the presence of RNase H to remove the original RNA template followed by enzyme inactivation via addition of 0.5 μl of 0.5 M EDTA (pH 8.0). First-strand cDNAs were purified with QuickClean resin (Clontech) followed by precipitation with sodium acetate and ethanol. Purified cDNAs were suspended in DNase/RNase-free sterile water, quantified spectrophotometrically, diluted to a final concentration of 10 ng/μl, and stored at −20°C until use. Q-RT-PCR was performed using the SYBR Green PCR Master Mix (PerkinElmer Applied Biosystems) and 17 gene-specific primer pairs (see results) designed using Primer Express Software (PerkinElmer Applied Biosystems) and synthesized at a commercial facility (Qiagen-Operon, Alameda, CA). Primers for β-actin were also made, and this gene was included in all Q-RT-PCR analyses for the purpose of data normalization (22). Results were recorded as relative gene expression changes after normalizing for β-actin gene expression, computed using the 2−ΔΔCt method described in detail by Livak and Schmittgen (21). This method monitors relative gene expression changes across treatments (days relative to parturition in this case) based on differences in the PCR amplified target reaching a fixed threshold cycle (CT) number at a set treatment (day −7) vs. other treatments (days 0, 0.25, and 1). Thus, for our 2−ΔΔCt analysis, the CT for day −7 was the calibrator used to determine relative gene expression changes on all other days for each β-actin-normalized test gene. Statistical analysis of these data was performed by comparing days 0, 0.25, and 1 individually to day −7 using t-tests with pooled standard errors on a log ratio scale. Gene-specific standard errors were estimated using independent analyses of variance (ANOVA), which included the effects of cow and time (0, 0.25, and 1 day). All the analyses were performed using the SAS System (30).

Neutrophil apoptosis phenotyping.

Based on results from our microarray screening and Q-RT-PCR experiments, we performed ex vivo apoptosis phenotyping on neutrophils from three Holstein steers. The periparturient blood environment was simulated in neutrophil cultures by adding heat-inactivated (56°C for 30 min) blood sera collected at days −7, 0, 0.25, and 1 (relative to parturition) from the three cows used for neutrophil collections for the Q-RT-PCR experiment. Apoptosis phenotyping was assessed by two-color fluorescence-activated flow cytometric analysis of cultured neutrophils stained with annexin V-FITC and propidium iodide (PI) (35). Briefly, Percoll-isolated blood neutrophils from each of the three steers were reconstituted at 1 × 107 cells/ml in RPMI-1640 medium (Invitrogen) containing 0.25% penicillin-streptomycin (Invitrogen), of which 0.1 ml was seeded into wells of 96-well cell culture plates (Fisher Scientific, Pittsburgh, PA). Neutrophil cultures were supplemented with 20% or 40% of individual sera from three periparturient cows, such that the final culture volumes were 0.2 ml/well. Neutrophils were then incubated in moist 5% CO2 air at 39°C (normal body temperature for cattle) for 24 h. After incubation, cells were centrifuged at 500 g for 5 min at 4°C, washed twice with cold phosphate-buffered saline, pH 7.2, and stained with FITC-conjugated annexin V and PI following the protocol contained in a commercial kit (annexin V-FITC apoptosis detection kit; BD Biosciences Pharmingen, San Diego, CA). Cells were then transferred to 5-ml polystyrene round-bottom tubes (Becton-Dickinson), and apoptosis data acquisition of 5,000 cells per sample was performed using a FACSCalibur flow cytometer and CellQuest software (Becton-Dickinson). Quadrants were set on resulting two-color flow cytometric density plots effectively separating annexin V/PI non-apoptotic cells (bottom left quadrant) from annexin V+/PI early apoptotic cells (bottom right quadrant), annexin V+/PI+ late apoptotic cells (top right quadrant), and annexin V/PI+ necrotic cells (top left quadrant). For simplicity of data reporting, we selected cells in the lower left and right quadrants (i.e., percent non-apoptotic and early apoptotic) for further statistical analyses and data presentation.

Statistical analysis was performed to test the fixed effect of day relative to parturition on the percent annexin V/PI and the percent annexin V+/PI neutrophils using the MIXED procedure of SAS (30). The statistical model also included random effects of steer (neutrophil donors) and cow (serum donors) as well as all two-way interactions (steer × day, cow × day, and cow × steer). Significance of the effect of day relative to parturition was declared when P < 0.05.


Characterization of neutrophils and RNA used.

Blood samples collected from periparturient cows used in this study were subjected to electronic counting of total leukocytes and to G1 immunostaining and flow cytometric analysis of neutrophils. As expected (22, 28, 33), parturition caused leukocytosis (Fig. 1A; P < 0.012) and neutrophilia (Fig. 1B; P < 0.001) in the test cows that were particularly striking 6 h postpartum (day 0.25). Small aliquots of Percoll-isolated neutrophils from each blood sample were also subjected to G1 immunostaining and flow cytometric analysis to determine purity, which was always >93% (not shown). The quantity of RNA averaged ∼5 μg per sample and was considered of high quality based upon spectrophotometric and agarose gel electrophoretic analysis. However, the quantity of RNA we obtained from individual samples of purified neutrophils was relatively low, requiring that we pool them across cows within sample time to have sufficient RNA for the preliminary screening of gene expression changes using a loop design in the microarray experiment.

Fig. 1.

Whole blood counts (number of cells/ml) of total leukocytes and neutrophils from cows sampled for the microarray and Q-RT-PCR experiments. A: leukocytosis (P = 0.012) at and 6 h after parturition (days 0 and 0.25) compared with prepartum (day −7) and 24 h postpartum (day 1). B: leukocytosis was driven by neutrophilia (P = 0.001). *Daily means significantly different (P < 0.05) from the day −7 mean.

LOWESS normalization of the microarray data.

Representative M-A plots for fluorescence intensities of all spot data from array no. 3 (day 0 vs. day 1), are shown before (Fig. 2A) and after (Fig. 2B) LOWESS normalization, which effectively adjusted spot intensities for the small Cy5 dye bias that was present at lower average intensities (Fig. 2A). The LOWESS-normalized data from each array were back-transformed and subjected to GeneSpring and statistical analyses to visualize and test the effect of day relative to parturition on gene-specific expression profile changes.

Fig. 2.

LOWESS normalization effectively removed a slight Cy5 dye bias that was apparent for low-intensity spots (negative control and some test spots) in the BOTL microarray data sets. Yellow dots are internal GAPDH control spots, pink dots are external lambda Q control spots, green dots are internal β-actin control spots, blue dots are internal RPL-19 control spots, and red dots are assay blanks (negative controls). Black dots represent the 1,056 thrice-spotted leukocyte genes. A: prenormalization M-A plot for array no. 3, which was hybridized with RNA from day 0 (Cy3 labeled) and day 1 (Cy5 labeled) neutrophils. B: post-LOWESS normalization M-A plot for the same array shown in A.

GeneSpring analysis of neutrophil gene expression profiles.

Statistical analysis of the microarray data suggested that 302 neutrophil genes of 1,056 genes spotted on the BOTL microarray (i.e., ∼30%) had expression profile changes (P < 0.01) induced by parturition. GeneSpring analysis of these genes revealed four main profiles based on the general type of expression changes over days −7, 0, 0.25, and 1. Clusters included genes that were highly induced on days 0, 0.25, and (or) 1 relative to expression on day −7; genes with moderately induced expression on days 0, 0.25, and (or) 1 relative to expression on day −7; genes whose expressions were inhibited on days 0, 0.25, and (or) 1 relative to expression on day −7; and genes with fluctuating expression changes indicating no particular pattern over time (data not shown). This analysis also demonstrated that expression of approximately one-third of the genes affected by parturition were upregulated on at least one test day, whereas the remaining two-thirds were downregulated or fluctuated with no definitive pattern across test days.

In silico determination for function of genes affected by parturition.

BLASTn searches of the GenBank and EST databases revealed that 36% of neutrophil genes affected by parturition were unknown (i.e., BLASTn hits were for cloned genomic DNA and specific chromosomal regions with no genes identified). A combination of BLASTn analysis and exhaustive literature searching revealed clear identities for the remaining affected genes with expectation values (E values) always <10−4. These could be grouped into 12 broad ontological clusters based on the best known function of their protein products. The clusters included apoptosis (14% of affected genes), leukocyte activation (7%), adhesion/trafficking (6%), signal transduction (5%), transcription/translation (5%), energy metabolism (5%), growth factors (4%), genes affecting cellular organelles (endoplasmic reticulum, cytoskeleton, ribosomes, and mitochondria; 4%), genes affecting steroid hormone receptors (2%), matrix metalloproteinases and their tissue inhibitors (2%), angiogenesis (1%), and one cluster we called “other” because the genes in it were seemingly unrelated and thus without an ontological theme (9%). We found it interesting that expression of 42 apoptosis-regulatory genes were putatively affected by parturition (Table 1), and we selected 17 key genes from this ontological cluster for further independent validation by Q-RT-PCR.

View this table:
Table 1.

Forty-two apoptosis regulatory genes detected in blood neutrophils by cDNA microarray analysis as putatively altered in expression by parturition

Q-RT-PCR confirms expression changes in 16 of 17 tested genes.

As shown in Table 2, parturition-induced expression changes were confirmed by Q-RT-PCR for 16 of 17 tested apoptosis regulatory genes (0.0001 < P < 0.07). β-Actin, selected as the normalizing gene based on previous findings (22, 33), was not differentially expressed across sample days. Many of the genes confirmed to be affected by parturition influence spontaneous apoptosis in neutrophils (e.g., A1, Bag-1, Bak, Bax, Mcl-1) through actions on mitochondrial membrane stability and release of mitochondrial cytochrome c. Several confirmed genes are also involved in the regulation of death signaling through plasma membrane death receptors (e.g., FADD, Daxx, FLASH, RIP). Other confirmed genes encode predominantly cytosolic proteins that influence neutrophil survival through effects on other apoptosis proteins [e.g., putative inhibitor of apoptosis (IAP)], oxidative stress responses (thioredoxin-like 2), and expression of proinflammatory genes that also induce survival (e.g., IL-8). Expression profiles for 10 confirmed genes are shown in Fig. 3.

Fig. 3.

Mean expression changes (±SE) of key neutrophil apoptosis regulatory genes during the peripartum period. Data were derived from Q-RT-PCR assay of RNA from three cows sampled on days −7, 0, 0.25, and 1 relative to parturition (on day 0) and expressed as ratios of expression on day −7. Parturition increased A1 expression and decreased Bak expression, effectively increasing both A1:Bak and A1:Bax ratios (AC). Parturition also decreased expression of genes encoding key Fas-signaling death proteins (DG) and induced pronounced expression of the pro-survival chemokine, IL-8 (H). Parturition decreased expression of IAP and NF-κB p65 genes (I and J, respectively), which have anti-apoptotic roles in many cell types but are not well described in neutrophils. **P ≤ 0.01. *P ≤ 0.05. ‡0.05 < P < 0.07.

View this table:
Table 2.

Quantitative real-time RT-PCR confirmation of 16 of 17 apoptosis regulatory genes altered in expression by parturition

Apoptosis phenotyping of neutrophils supports microarray and Q-RT-PCR results.

Because we confirmed that 16 key apoptosis regulatory genes of blood neutrophils were influenced in expression by parturition, we were curious to know how parturient blood affects the apoptosis phenotype of cultured neutrophils. We treated isolated neutrophils from donor steers with sera collected from the periparturient cows used for the gene expression experiments and assessed their apoptosis status flow cytometrically (Fig. 4). Cultures treated with parturient sera (day 0) had higher percent annexin V/PI (non-apoptotic) neutrophils and lower percent annexin V+/PI (early apoptotic) neutrophils at 12 and 24 h than cultures treated with sera from days −7, 0.25, or 1 (P < 0.01). The main differences in both non-apoptotic and early apoptotic cells were between neutrophils cultured in day −7 sera vs. day 0 sera. Identical results were observed when sera were used at 20% (Fig. 4) and 40% (not shown) of the culture volume. Thus the parturient blood environment induced temporary survival in otherwise normal blood neutrophils.

Fig. 4.

Blood serum from parturient cows prolongs neutrophil survival ex vivo. Data in A are representative two-color flow cytometric density dot plots for neutrophils from one steer incubated for 24 h in periparturient sera from one cow collected on day −7 (top left), day 0 (top right), day 0.25 (bottom left), and day 1 (bottom right) relative to parturition (on day 0). Annexin V-FITC fluorescence of the cells is on the x-axes, and propidium iodide (PI) fluorescence of the cells is on the y-axes. Color changes within each quadrant of these plots represent a 50% change in density on the log10 scale of the two-color-stained cells. Percentage non-apoptotic neutrophils (i.e., %annexin V/PI in bottom left quadrants) was higher, and % early apoptotic neutrophils (i.e., %annexin V+/PI in bottom right quadrants) was lower when cells were cultured in day 0 sera compared with sera from days −7, 0.25, or 1. B: means (±SE) for percent non-apoptotic and percent early apoptotic neutrophils cultured for 12 and 24 h in the various periparturient sera are summarized across all steers and cows tested (P values indicate significant effects of day relative to parturition).


Neutrophils are mature, terminally differentiated leukocytes that normally survive for a short time in the circulation (6–12 h) before undergoing apoptosis and clearance from blood by the body’s phagocytic cell network (11). Thus balance between production of new neutrophils in bone marrow and apoptosis in circulating cells largely determines blood neutrophil counts in healthy animals. In the current study, 42 neutrophil genes that encode apoptosis-regulatory proteins were found to be changed in expression during parturition (Tables 1 and 2). Noteworthy were expression profiles for key Bcl-2 family member genes (Fig. 3, AC), Fas-signaling genes (Fig. 3, DG), and the IL-8 gene (Fig. 3H), all of which suggested a pro-survival gene expression pattern at parturition relative to 7 days prepartum. Subsequent apoptosis phenotyping of normal neutrophils treated with sera from periparturient cows substantiated this by demonstrating enhanced survival in the presence of serum collected at parturition (day 0) vs. serum collected before (day −7) or after (days 0.25 and 1) parturition (Fig. 4). Extended survival could partly explain the pronounced neutrophilia we observed in parturient cows (Fig. 1B), because non-apoptotic neutrophils would not be cleared from circulation (11).

The normally short life span of circulating neutrophils is thought to be due to their lack of Bcl-2 gene expression and relatively high level of Fas gene expression (11). Bcl-2 is the prototypic anti-apoptosis protein in most cells that protects mitochondrial membranes from attack by pro-apoptotic Bcl-2 family members such as Bax and Bak (29). While lacking Bcl-2 expression, mature neutrophils do express other anti-apoptotic Bcl-family member genes, such as A1, Bag-1, and Mcl-1 in addition to pro-apoptotic Bax and Bak (26, 34). A1 is the Bcl-2 homolog in mature neutrophils, and its increased expression via NF-κB activation in cells treated with survival-inducing pro-inflammatory factors (e.g., LPS and G-CSF) rescues the cells from Bax/Bak-induced spontaneous apoptosis (10, 18). A1 protein works by protecting mitochondrial membranes from Bax/Bak-induced pore formation, preventing release of cytochrome c and subsequent activation of caspases 9 and 3 that effect DNA fragmentation and cell death (26, 32). Thus expression ratios of A1 to Bax and Bak are the main determinants of whether neutrophils live or die by the mitochondrial cytochrome c release pathway (18, 34). Our Q-RT-PCR experiment confirmed that parturition temporarily increased the A1:Bax and A1:Bak ratios in bovine blood neutrophils (Fig. 3, AC), supporting the survival induction observed in normal neutrophils cultured in medium containing parturient serum (Fig. 4).

Although Bcl-2 family members are intracellular proteins, Fas (CD95/APO-1) is expressed on the plasma membranes of neutrophils (19) and is a prototypic death receptor from the TNF receptor superfamily of molecules. Ligand-activated Fas induces aggressive apoptosis in neutrophils and other cells by recruiting adaptor proteins (e.g., FADD, Daxx, FLASH, and RIP) to its cytoplasmic death domain to form potent signaling complexes known as DISC (“death-inducing signaling complex”) (31). DISC recruits and activates caspase 8, which then cleaves and activates downstream caspases (e.g., 7 and 3) that inactivate normal survival signals (e.g., NF-κB) (13, 20) and cleave cytoskeletal and DNA repair proteins, leading to DNA fragmentation and irreversible cell death (reviewed in Refs. 14 and 25). Circulating neutrophils normally express higher levels of Fas than other leukocytes, easily explaining their sensitivity to apoptosis induction and short half-life in blood compared with lymphocytes, monocytes, and eosinophils (11). Given that expression of genes encoding the four main DISC proteins were dramatically downregulated in blood neutrophils of our study (Fig. 3, DG), it would appear that the potent Fas-induction pathway of cell death may be disabled in the cells and support their temporary survival. If future studies prove that parturition interrupts formation of DISC at the death domains of Fas, then expression changes in these proteins could become potential targets for drug development to manipulate neutrophil survival and inflammatory responses.

Given our results, it appears that factor(s) in parturient blood influence expression of genes that change the apoptosis status of neutrophils. Although these factors have yet to be identified, biomedical literature has documented that glucocorticoids effectively induce survival in cultured human, fish, and rodent neutrophils (7, 24, 35). Glucocorticoids are dramatically increased in blood of cows at parturition (28), including the cows used in the current study (data not shown), so this is one possibility that may explain the survival induction we observed. Given this, we are not certain how to reconcile the inhibited IAP (Fig. 3I) and NF-κB p65 (Fig. 3J) gene expression in neutrophils around parturition. The products of these genes are best known for their ability to confer survival by inhibiting caspases (IAP; reviewed in Ref. 9) and inducing pro-survival gene expression (NF-κB; see above and Ref. 18) in most other leukocytes. Although our BOTL microarrays are valuable tools for studying gene expression in neutrophils, their limitation is that they do not contain all possible genes expressed by neutrophils, including apoptosis regulatory genes. Thus other apoptosis genes not present on our microarrays that may override putative IAP and NF-κB survival systems may have been induced in the neutrophils in favor of extended survival. It is also possible that, in addition to acute survival induction via altered A1:Bax/Bak and DISC protein gene expression ratios, parturition induces a more chronic pro-death expression signature in genes such as IAP and NF-κB to ensure that the cells do eventually die by apoptosis. Supporting this possibility, most of the anti-apoptotic gene expression changes induced by parturition began to return to normal pro-apoptotic expression patterns observed 7 days prepartum by day 1 postpartum (Fig. 3), when neutrophil numbers in blood were relatively normal (Fig. 1B) and blood serum added to neutrophils ex vivo no longer supported prolonged survival (Fig. 4). In fact, A1 expression was significantly repressed and Bax expression induced on day 1 postpartum relative to day 7 prepartum, suggesting that this pathway was even more active after parturition than before parturition.

Finally, our results have posed an apparent anomaly. It is not intuitive why parturition would induce a temporary pro-survival gene expression pattern in blood neutrophils or how this would relate to heightened mastitis susceptibility in newly calved cows. One could postulate that neutrophils in survival mode would actually augment the supply of available neutrophil defense around parturition due to resulting neutrophilia. Why is it then that parturient dairy cows tend to lack efficient inflammatory responses in peripheral tissues making them susceptible to clinical mastitis (1)? Alternatively, could altered expression profiles of additional genes related to trafficking, migration, and energy metabolism suggested by our microarray experiment and other studies (22, 33) better explain mastitis at parturition? If survival induction is a byproduct of altered inflammatory phenotypes resulting from changed expression in these other genes, then it may occur simply to protect animals from the growing numbers of neutrophils that would cause harmful systemic inflammation if they died by necrosis or outnumbered the capacity of the phagocytic network to clear them from the circulation. These questions are important to answer because the balance between neutrophil survival and programmed cell death ultimately determines the outcome of all inflammation, including in the bovine mammary gland. Guided by results of this study, our future work will be aimed at addressing these questions.


We thank Dr. Patty Weber, Dr. Anantachai Chaiyotwittayakun, Jennifer Jacob, and Rebecca Darch for assistance during blood sampling, cell preparations, and RNA isolations; we thank Sue Sipkovsky and Erin McCandless for assistance during the microarray experiment; and we thank Dr. Robert Tempelman for assistance with statistical analysis of the apoptosis phenotyping experiments.


This work was supported by the Michigan Agricultural Experiment Station (MICL01691-USDA Multistate Research Project NC-209) and by USDA-IFAFS Grant 2001-52100-11211.


  • Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).

    Address for reprint requests and other correspondence: J. L. Burton, Immunogenetics Laboratory, Dept. of Animal Science, 1205E Anthony Hall, Michigan State Univ., East Lansing, MI 48824 (E-mail: burtonj{at}msu.edu).



View Abstract