Skip to content


  • Original article
  • Open Access

Generation of EST and cDNA Microarray Resources for the Study of Bovine Immunobiology*

  • 1,
  • 1,
  • 1,
  • 1 and
  • 1
Acta Veterinaria Scandinavica200344 (Suppl 1) :S89

  • Received: 16 August 2001
  • Accepted: 28 November 2001
  • Published:


Recent developments in expressed sequence tag (EST) and cDNA microarray technology have had a dramatic impact on the ability of scientists to study the responses of thousands of genes to external stimuli, such as infection, nutrient flux, and stress. To date however, these studies have largely been limited to human and rodent systems. Despite the tremendous potential benefit of EST and cDNA microarray technology to studies of complex problems in domestic animal species, a lack of integrated resources has precluded application of these technologies to domestic species. To address this problem, the Center for Animal Functional Genomics (CAFG) at Michigan State University has developed a normalized bovine total leukocyte (BOTL) cDNA library, generated EST clones from this library, and printed cDNA microarrays suitable for studying bovine immunobiology. Our data revealed that the normalization procedure successfully reduced highly abundant cDNA species while enhancing the relative percentage of clones representing rare transcripts. To date, a total of 932 EST sequences have been generated from this library (BOTL) and the sequence information plus BLAST results made available through a web-accessible database Cluster analysis of the data indicates that a total of 842 unique cDNAs are present in this collection, reflecting a low redundancy rate of 9.7%.

For creation of first generation cDNA microarrays, inserts from 720 unique clones in this library were amplified and microarrays were produced by spotting each insert or amplicon 3 times on glass slides in a 48-patch arrangement with 64 total spots (including blanks and positive controls) per patch. To test our BOTL microarray, we compared gene expression patterns of concanavalin A stimulated and unstimulated peripheral blood mononuclear cells (PBMCs). In total, hybridization signals on over 90 amplicons showed upregulation (>3×) in response to Con A stimulation, relative to unstimulated cells. A second experiment with PBMCs from a different group of animals was performed to test reproducibility of microarray results. There was a high correlation between the 2 experiments (r = 0.72, P < 0.001). Resources described in this publication offer a highly efficient and integrated system to study gene expression changes in bovine leukocytes.


  • expressed sequence tag
  • bovine leukocytes
  • normalization
  • cDNA microarray
  • EST
  • database


New DNA microarray technologies have been developed that allow large-scale screening of gene expression patterns in many biological models (functional genomics). These technologies offer the prospect that all genes within an organism can be scrutinized to obtain integrated information on biological systems that is not currently obtainable using traditional reductionist approaches [4]. Complementary DNA (cDNA) microarray technology, in particular, has become a central technical platform for functional genomics experiments [18, 26, 27].

From a human health perspective, the volumes of recent biomedical literature on the subject and number of large biomedical corporations adopting microarray technology provide evidence of utility in human medicine and biology [1, 5, 9, 1115, 17, 20, 21, 23, 28]. However, relatively little attention has been drawn to the power of microarray technology to better understand diseases of economically significant livestock species. This is curious because study of the most significant livestock diseases is tremendously daunting due to the variety of complex factors that form the basis for susceptibility to production diseases. Some of these factors include genetic selection pressure for production over fitness, routine husbandry stress, application of growth and milk production-enhancing biotechnologies, and extreme variations in climate and nutrition [2, 8, 10, 19, 25]. Indeed, the widespread occurrence of very complex production diseases that have been difficult or impossible to resolve are of key economic and societal concern today, and will continue to be in the future.

With application of functional genomics tools to study complex livestock disease researchers may, for the first time, be able to study the intricate patterns of gene expression events leading to complex metabolic and infectious disease states. Armed with this knowledge, new tissue, cell-, and protein-specific therapies, novel husbandry practices, and new ways of controlling susceptibility to diseases may be applied. Perhaps one of the most significant opportunities afforded by functional genomics technology on the livestock is the discovery of many candidate genes affecting traits of economic importance. Ultimately, however, the real value of cDNA microarray technology to profile gene expression pathways in livestock diseases and disease susceptibilities will only be realized when excellent quality cDNA libraries and microarrays are available to a wide number of researchers [16]. The Center for Animal Functional Genomics (CAFG) at Michigan State University was created to address this critical need. We have begun by establishing an integrated program generating sequence-tagged, normalized cDNA libraries from cells and tissues relevant to studies of bovine immunobiology and by applying these resources to cDNA microarray analysis of important bovine disease problems ([7], this issue). Described in this paper is the development of a non-redundant bovine total leukocyte cDNA library, a web-accessible database for housing clone sequences, BLAST results, and cluster information, and the initial use of these EST resources to generate cDNA microarrays. In addition, we have preformed an initial set of experiments using a well-established model of immune cell activation to demonstrate the reproducibility and value of our functional genomics tools to study bovine immunobiology.

Materials and methods

Construction of the cDNA library

The bovine total leukocyte (BOTL) library was constructed using a commercial cDNA synthesis system (Invitrogen Life Technologies, Inc. Gaithersburg, MD). Poly (A)+ RNA was purified from total peripheral blood leukocytes isolated from pooled blood samples of 4 normal mid-lactation Holstein dairy cows. First-strand cDNA synthesis was performed using the oligonucleotide 5'-GACTAGTTCTAGATCG-CGAGCGGCCGC AATTAAT15-3', which contains a NotI site (underlined) and an AATTAA sequence as a tissue-specific tag. The subsequent steps for construction of the library were performed essentially according to the manufacturer's instructions. Double-stranded cDNA fragments were cloned directionally into the SalI and NotI sites of the plasmid vector pSPORT1 (Invitrogen Life Technologies) and the library was amplified in a semi-solid medium to minimize biased growth of primary transformants.

Normalization of the cDNA library

Normalization of the library was performed using a modification of a method (Method 4) described by [3], which is based on the different re-association kinetics for cDNA molecules of different relative abundance. Briefly, double-stranded plasmid library DNA was converted to single-stranded circles through the combined actions of Gene II protein and Exonuclease III provided within a Gene Trapper II kit (Invitrogen Life Technologies). A fraction of these single-stranded plasmids was used to generate PCR-amplified cDNA inserts using primers recognizing sequences on pSPORT1 flanking the inserts. The amplified products were purified using Sephadex spin-columns (Amersham Pharmacia, Inc., Buckinghamshire, UK) and ethanol precipitated. Approximately 3 μ g of the purified PCR products (denatured) were used to hybridize with 150 ng of single-stranded library plasmid in a 30 μ l reaction volume containing 50% formamide, 10 mM Tris-HCl pH 7.5, 1 mM EDTA, 0.5 M NaCl, and blocking oligonucleotides. The hybridization was performed at 30°C for 24 hrs. Single-stranded circles remaining after hybridization were purified by hydroxyl appatite (HAP) chromatography and converted to double-stranded plasmid by primer extension using an oligonucleotide complementary to the single-stranded plasmid DNA (M13-reverse primer). The resulting double-stranded plasmids were then electroporated into E. coli DH10B, generating the normalized BOTL library.

Southern blot analysis

To determine the relative abundance of cDNA inserts representing various transcripts in the BOTL library, 0.5 μ g of plasmid DNA from the original and normalized libraries were separately digested with NotI and SalI, separated on 1% agarose gels and blotted onto Zeta-probe membrane (Bio-Rad Inc., Hercules, CA). The membranes were separately hybridized at 68°C in PerfectHyb solution (Sigma Chemical Co., St. Louis, MO) with 32P-labeled PCR-amplified cDNAs (amplicons) for GAPDH, elongation factor 1, ribosomal protein L12, and ribosomal protein S3A to detect high-abundance transcript clones. To detect medium and low-abundance transcript clones, 32P-labeled PCR-amplified cDNAs representing Rab7 and CD40L, respectively were used as probes. Membranes were washed twice in 2 × SSC/0.1% SDS at room temperature and twice at 65°C in 0.2 × SSC/0.1% SDS. Washed membranes were wrapped and exposed to BioMax MS film (Fisher Scientific, Pittsburgh, PA) to visualize bands.

DNA sequencing and sequence analysis

The transformed normalized library was plated onto 150 mm plates containing LB agar supplemented with 50 μ g/ml ampicillin. Colonies were randomly picked from the agar plate and transferred to 96-well plates using a GeneTAC G3 robot equipped with a 48-pin pneumatic colony picking head (Genomic Solutions Inc., Ann Arbor, MI). Cultures were grown in the 96-well plates overnight and plasmid DNA isolated using the QIAprep 96 Turbo miniprep kit (Qiagen Ltd., Valencia, CA) according to the manufacturer's instructions. Sequencing reactions were performed using a BigDye terminator sequencing kit (Perkin-Elmer/ABI, Palo Alto, CA) and analyzed on an ABI 3700 DNA analyzer. Sequence data were outputed to Geospiza software (Geospiza, Inc., Seattle WA) and picked up via password-protected web access. Final sequences were exported in FASTA format into our database program that directs basic local alignment search tool (BLAST) searches against the GenBank nonredundant (nr) database using the BLASTN and BLASTX programs. For information on similarity between clones within our database and the Genbank dbEST, the entire dbEST was downloaded and sequences with bovine in the annotation parsed to a separate file, resulting in 877 Mb of information representing 170546 bovine ESTs. A BLAST searchable database was then created from this file, to allow local blast analysis. Finally, sequences from our EST database were compared to the 170546 sequences in the bovine EST-specific file.

EST database presentation and management

A Bioinformatics server running on a SUN Ultra2 workstation is provided for storage of the EST data, BLAST searches for sequence homology, storage of results from the BLAST search, and dissemination of stored results and data via a WWW interface Sequence data and annotation information are stored in an Oracle database (Oracle Corp., Redwood Shores, CA). Access to these data is through a Turnkey system (INCOGEN, Inc. Clemson, SC). The oracle data can be accessed via Oracle Client (Net8) software, The Statistical Analysis System, PHP OCI8, and Perl DBI:Oracle. On a routine basis, Perl and PHP scripts perform the tasks of the Turnkey System. These include: 1) Adding sequence data in FASTA format to the database system; 2) Executing BLASTN and BLASTX searches on the new data; 3) Updating the Turnkey Web BLAST data files in the Turnkey system; 4) Clustering the sequence data using a hashing algorithm, as implemented in the tl-cluster software; and 5) Loading clustered data into the database.

Production of cDNA microarrays

Clone inserts from the BOTL library were amplified by inoculating bacteria directly into PCR master mix (20 μ M each dNTP, 0.2 μ M M13-forward primer, 0.2 μ M M13-Reverse primer, 200 mM Tris-HCl (pH 8.4), 500 mM KCl, and 2.0 units of Taq DNA polymerase) in 96-well plates. Following a pre-heat step of 95°C for 10 min to disrupt the bacteria, PCR was performed in a PE 9700 Thermocycler (Perkin Elmer Corp., Palo Alto, CA) using the following conditions: 95°C for 30 sec; 55°C for 30 sec and 72°C for 1 min. Based on agarose gel electrophoresis of purified amplicons, our success rate of clone insert amplification with this protocol has typically been >90%. Insert amplicons were purified in 96-well format using Millipore filter plates (Millipore Corp., Bedford, MA) filled with Sephadex G-50 (Amersham Pharmacia Biotech.). Final insert amplicons were resuspended in 17 μ l of 50% DMSO for low-density (3000 to 4000 amplicons per slide) spotting on microarrays. Approximately 1 μ l of each purified insert amplicon was separated on a high-volume 1.2% agarose gel to ensure that each clone was represented on the final microarrays. This check also allowed final adjustment for differences in amplicon concentration just prior to microarray source plate set up. Finally, a total of 5 μ l of each purified amplicon was transferred to 384-well microarray source plates.

Microarrays were spotted using a GeneTAC G3 arraying robot (Genomic Solutions, Inc., Ann Arbor, MI) equipped with a 48-pin head, with each pin having a nominal end diameter of 200 microns. In 50% DMSO (low-density arrays), this arrangement yields spots of approximately 250–350 microns. The pin configuration of the G3 yields microarrays consisting of 48 "patchs" of spots. A microarray design was derived that allowed triple spotting of each clone within a patch and an overall 8 × 8 spot pattern in each patch (64 spots per patch). Microarray source plates were designed such that each patch would contain 3 GAPDH amplicons (controls for RNA loading differences and potential dye effects). In addition, each patch contained 2 synthetic lambda Q gene positive control spots, 17 blanks to control for background effects on a patch specific basis, and 1 DMSO only spot. Three blanks surround one lambda Q gene spot in the upper right corner of each patch, establishing a landmark for subsequent microarray image orientation and analysis (for microarray image, see our companion paper, [7], this issue).

Preparation of PBMCs, RNA extraction, labeling and hybridization

For assessment of microarray utility in studying bovine immunobiology, blood samples from 3 healthy mid-lactation Holstein cows were obtained from the coccygeal (tail) vein using 2.5 cm 21 guage multiple sample needles and a series of four 6 ml vacutainer tubes containing acid citrate dextrose (ACD) as an anticoagulant. Peripheral blood monocytes were prepared as previously described [6]. Briefly, blood samples were centrifuged at 4°C for 10 min at 1000 × g and the resulting buffy coat (approximately 2 ml) transferred to a new 50 ml conical tube containing 34 ml of ice-cold sterile PBS overlaid on a 10 ml cushion of Percoll (1.084 g/ml, Sigma Chemical Co., St. Louis, MO). Cells were centrifuged at 1000 × g for 40 min at room temperature to separate erythrocytes and polymorphonuclear leukocytes from mononuclear cells. Following careful aspiration of the PBS, PBMCs lying at the PBS/Percoll interface were transferred to a new 50 ml conical tube, rinsed once with 20 ml of sterile PBS and finally resuspended in maintenance medium RPMI-1640 containing 10% heat-inactivated fetal bovine serum (FBS), 2 mM L-glutamine, 100 U/ml penicillin, and 100 μ g/ml streptomycin. PBMCs from all animals were combined at this point and gently mixed. The total volume was adjusted to 20 ml with maintenance medium and the mixed PBMCs were split into 2 equal fractions of 10 ml. One fraction received 100 μ l of sterile PBS (unstimulated) while the second fraction received 100 μ l of PBS containing 2.5 mg/ml of concanavalin A (Con A stimulated), resulting in a final concentration of 25 μ g Con A per ml. Cells were incubated overnight (18–20 hours) in a humidified atmosphere of 5% CO2 and 95% air at 39°C.

RNA was extracted from the unstimulated PBS control and Con A-stimulated PBMCs using Trizol reagent (Invitrogen Life Technologies, Inc., Gaithersburg MD). Quantity and quality of extracted total RNA was estimated by UV spectrophotometry and electrophoresis on 1.2% native agarose gels. Integrity of separated 28S and 18S RNA bands on these gels serves as a good indicator of RNA quality and separation on agarose gels is a sensitive method for detecting degraded RNA.

For Con A Experiment 1, total RNA (20 ug) from Con A stimulated cells was used as template in a reverse transcription reaction incorporating an amino-modified dUTP into the cDNA (Clonetech Atlas Glass labeling system, Clonetech Inc., Alameda CA). Oligo (dT)15 was used as primer and 1–2 ng of synthetic lambda Q gene RNA containing an engineered poly A tail was spiked into the reaction as a control for cDNA synthesis and as a positive hybridization control. Following first-strand cDNA synthesis, the sample was split into 2 equal portions for differential labeling with Cy3 and Cy5 as described below. This process produced 2 differentially labeled samples that should have identical gene expression levels at each spot on the BOTL microarray.

For Con A Experiments 2 and 3, control unstimulated and Con A-stimulated PBMC cDNAs were differentially labeled using NHS-derivatized Cy3 and Cy5 dyes, respectively (Amersham Pharmacia, Ltd.). Labeled cDNAs were extensively purified following the Clonetech Atlas Glass labeling system instructions to remove unincorporated dyes. The purified labeled cDNAs were then combined and concentrated to 4–10 ul using Microcon 30 spin concentrators (Millipore Corp.). Microarray hybridizations were performed by addition of concentrated Cy3 and Cy5 labeled probe cDNAs to 45 μ l of GlassHyb (Clonetech, Inc.), supplemented with 10 μ g/ml BSA and 10 ug/ml denatured salmon sperm DNA. Hybridizations were conducted for 18–20 hours at 50°C in sealed and humidified chambers (Arrayit, TeleChem International, Inc., Sunnyvale, CA). Following hybridizations, the microarrays were washed twice at room temperature in 1 × SSC/0.05% SDS and once at 50°C in the same solution. Washed microarrays were rinsed in ddH2O and dried by centrifugation in a cushioned 50 ml conical centrifuge tube. Final microarrays were scanned immediately using a GeneTAC LS IV microarray scanner and GeneTAC LS software (Genomic Solutions, Inc., Ann Arbor, MI). GeneTAC analyzer software was then used to process microarray images, find spots, integrate robot-spotting files with the microarray image, and finally create reports of either spot intensity ratios or total spot reports. For Con A experiments 2 and 3, different groups of 3 healthy mid-lactation Holstein cows were used as a source of PBMCs.


Characterization of the normalized cDNA library

Using Poly (A)+ RNA isolated from bovine total leukocytes from 4 healthy mid-lactation Holstein cows, we constructed a directionally cloned cDNA library (BOTL) with a tissue-specific tag sequence (AATTAA) at the 3'end of each cDNA. The quality of this original library was first evaluated by titration of the original electroporated bacterial stock and restriction analysis of plasmid DNA prepared from 18 clones randomly picked from the library. The library was estimated to contain a total of 1.1 × 106 independent clones. Restriction analysis showed that 15 out the 18 randomly selected cDNA clones contained inserts with an average size of 1.2 kb (Fig. 1, Panel A). The library was then amplified in a semi-solid media and plasmid DNA prepared. Approximately 5 ug of the library plasmid DNA was used for the normalization process.

The original BOTL library was normalized as described in Materials and Methods to reduce the relative concentration of clones representing high-abundance transcripts and therefore limit redundancies in subsequent EST programs. To assess whether the normalization procedure caused any changes in size distribution of cDNA inserts, plasmid DNA from the normalized library were digested with NotI and SalI to release inserts and these digests were separated by agarose gel electrophoresis. The patterns of the inserts from the BOTL library before and following normalization were indistinguishable, indicating that there were no significant changes in size distribution of inserts after normalization (Fig. 1 panels A and B). In this analysis, 16 out of the 18 analyzed clones contained inserts with sizes, ranging from 0.2 to 3 kb (Fig. 1, Panel B). This range of insert sizes appears similar to the range of insert sizes isolated from clones of the original library (Fig. 1, Compare Panels A and B).
Figure 1
Figure 1

Comparison of random BOTL Library clone insert sizes before (A) and after (B) normalization.

The effectiveness of library normalization was evaluated by DNA sequence analysis of >100 clones randomly picked from the original and normalized libraries. Of 101 cDNA sequences from the original library, 27 were found to represent abundantly expressed genes; 21 of 27 were ribosomal protein genes, while the remaining 6 included genes encoding GAPDH, elongation factor 1, histone H3.3, and beta-globin. In contrast, none of 127 cDNA clones picked and sequenced from the normalized library represented abundantly expressed genes. This indicates that the normalization procedure significantly reduced the frequencies of clones representing abundant cDNAs in the normalized library. Furthermore, 74.3% (75/101) of the cDNA clones from the original library had matches with the known genes in the GenBank database (Cutoff E value = 10-15), while only 61.4% (78/127) of cDNA clones from the normalized library matched known genes. This suggests that there may be a higher probability of identifying unknown cDNA species in the normalized library.

The efficiency of normalization was further assessed by Southern blot analysis of the frequency changes of 4 abundant, 1 intermediate, and 1 rare cDNA clones in the original and normalized libraries. To accomplish this, total library DNA was digested with Sal I and Not I, releasing the inserts. Inserts and vector were separated by electrophoresis, transferred to nylon membranes, and hybridized as described in "Materials and methods". As shown in Fig. 2, the relative frequencies of inserts representing the 4 highly-abundant cDNAs, GAPDH, elongation factor 1, ribosomal protein L12, and ribosomal protein S3A, were dramatically decreased in the normalized library as compared to the original library. The relative frequency of inserts representing Rab7, a moderately-abundant mRNA, was also decreased, but to a lower extent than the high-abundance transcript clones. In contrast, the relative frequency of clone inserts representing CD40L, a rare transcript, was increased in the normalized library as compared to the original library. These data suggest that the normalization process successfully reduced the relative number of clones representing highly-abundant cDNA species, but enhanced the relative number of clones representing rare transcripts. Once the effectiveness of normalization was established, the final cDNA library was plated and individual colonies isolated by robotic picking into 96-well plates. Approximately 1200 colonies were analyzed by single-pass sequencing to generate a collection of expressed sequence tagged (EST) clones.
Figure 2
Figure 2

Southern blot analysis of the relative frequencies of 6 different cDNA clones (4 abundant, 1 intermediate and 1 rare) in the original and normalized BOTL libraries.

Analysis of the ESTs

A total of 932 EST sequences have been successfully generated from the 3' end of isolated cDNA clones from the normalized BOTL library. These sequences are available in our database under the BOTL library. Cluster analysis of the data using the tlcluster program indicated that there were 842 unique cDNAs in this collection representing a 90.3% degree of novelty in the ESTs. The average similarity in the clusters was 98.8%. A total of 9.7% of the clones were represented more than once.

Of the 932 ESTs in our database, 642 (68.9%) have significant similarity (based on BLASTN matches with E-value < 10-15) to known genes in the public Genbank database. Of the 642 clones matched to known genes, 103 (16%) of these matched known bovine genes, while the rest had similarity to known genes from other species. These results are important because they allow a tentative identification of genes represented in our database based on similarity to previously identified genes in cattle or other species. It was also important to assess the similarity of our EST clones to other bovine ESTs housed in the public Genbank dbEST. This information can be used to identify other clones representing the same transcript and to assemble longer sequences by aligning overlapping EST sequence files. Clones that fail to show significant similarity to previously identified ESTs may represent unique transcripts that have not been previously isolated or characterized and may add significant new information to the growing body of identified bovine gene transcripts. Comparison of our EST clone sequences using the BLASTN algorithm against a subset of the Genbank dbEST database containing over 170000 bovine EST sequences revealed that 761 (81.65%) of our EST clone sequences matched previously identified ESTs (E-Value > 10-15). Importantly, 171 (18.35%) of our EST clone sequences did not match any previously identified bovine ESTs (E-Value < 10-15), indicating the possibility of their specific expression in bovine total leukocytes. One caveat to this is that many of the bovine EST sequences in the public databases were derived through 5' end sequencing and ours were derived by 3' end sequencing. Thus, 2 opposing ESTs corresponding to the same gene that do not overlap would fail to show any homology. For many organisms, The Institute for Genome Research (TIGR) maintains a database of EST sequences that have been aligned based on similarity and overlap. Each group of EST sequences is referred to as a cluster and it is usually assumed that each cluster represents a unique transcript expressed in that particular organism. The recent addition of information on clones in our database matching clusters within the TIGR bovine gene index (bTGI) increases access to information on clones, links to clones within public bovine EST resources representing similar sequences, and allows access to greater sequence lengths through BLAST analysis of an entire cluster sequence. Of the 932 clones currently represented within our database, 634 show significant similarity to TIGR bTGI cluster sequences. This suggests that of the 932 clones in our current database, 298 represent novel bovine transcripts that have not previously been isolated or characterized. To date, we have not compared our sequences to singletons within the bTGI to determine if the remaining clones match or overlap any of these sequences, thus creating new clusters.

Use of BOTL microarrays to study bovine immunobiology

Our overall goal in developing the BOTL library, database, and cDNA microarrays was to create resources for functional genomics studies related to bovine immunobiology. To test our resources, inserts from 720 unique BOTL clones were amplified and purified as described in "Materials and methods". The final purified insert amplicons were aliquoted into the wells of two 384-well microarray source plates. A GAPDH amplicon that was similarly amplified and purified was added to each well of the first block of 48 wells in plate 1 (wells A1 through E12) to provide controls for RNA loading and dye effects. To improve reproducibility and allow statistical analysis of microarray results, BOTL insert amplicons were spotted in triplicate within a patch. In total, each patch contained 14 BOTL insert amplicons spotted in triplicate, 3 GAPDH spots, 16 blanks, 1 DMSO control spot, and 2 positive control spots in each of 48 patchs on the microarray. Because bovine cytokines were not adequately represented in the original 720 clone inserts originally selected from our BOTL library for microarray production, we produced amplicons representing bovine IL-1, IL-2, IL-4, IL-5, IL-6, IL-8, IL-10, IL-12, TNFα, CD40L, IFNγ, and TGFβ and used these to replace amplicons representing various ribosomal protein genes within the original microarray source plates. Likewise, amplicons representing various matrix metalloproteinases and receptors were also produced and added to the microarray source plates.

We chose Con A stimulation of mixed PBMCs as a well-characterized model of immune cell activation to assess the ability of our BOTL microarrays to detect an immune response. Con A stimulated PBMC are well known for their strong proliferative capacity and production of various cytokines [22, 24, 29, 30]. Since it was important to determine the background or "noise" associated with our microarrays, we first performed a control experiment (Con A Experiment 1) where the source RNAs for differential dye labeling were derived from the same sample of Con A stimulated PBMCs. Since there should be no differences in the cDNA produced from this experiment, we would anticipate microarray analysis to yield equal fluorescence intensities for both the Cy3 and Cy5 labels at every spot on the microarray. To accomplish this test experiment, PBMCs from 3 healthy mid-lactation Holstein cows were stimulated with Con A and RNA extracted as described in "Materials and methods". Following analysis of the raw data, relative fluorescence intensities for the Cy3 and Cy5 channels were plotted for every spot on the microarray (Fig. 3). As expected, there were few differences between the Cy3 and Cy5 signals for most of the microarray spots in this experiment. Subsequent analysis of the data revealed that only one amplicon showed significant differential expression of 2-fold or more with an associated P-value < 0.01. An additional 2 amplicons showed differential expression of 2-fold or more with a P-value < 0.05. Thus, using our BOTL microarrays under the conditions described herein resulted in only three amplicons out of 720 yielding false positive results (amplicons showing differential expression when their actual expression levels should be similar). This result demonstrated that the design of our microarrays and subsequent analysis procedures were rigorous.

We next wished to ascertain the ability of our BOTL microarrays to detect immune cell activation in response to Con A stimulation. To accomplish this, we performed an experiment (Con A Experiment 2) in which mixed PBMCs were isolated, separated into 2 equal aliquots, and each aliquot separately stimulated with PBS or Con A. Corrected relative fluorescence intensity data was plotted (Fig. 4) for every spot on the microarray in a manner similar to Con A Experiment 1. The resulting plot indicates that, as expected a number of PBMC genes were strongly activated (Cy5>Cy3) by Con A stimulation. Subsequent analysis of the microarray data indicated that expression of more than 90 genes was activated >3-fold in response to Con A, with expression of many genes (~25) activated >5-fold, relative to unstimulated cells. These data demonstrate that our BOTL microarrays are readily able to detect enhanced gene expression in response to a general mitogen.
Figure 3
Figure 3

Total sample plot for BOTL microarray ConA Experiment 1, comparison of Cy3 and Cy5 intensities for the same sample.

Figure 4
Figure 4

Con A Experiment 2, comparison of relative gene expression in PBMCs stimulated overnight with ConA in PBS (Cy5) or PBS alone (Cy3).

Because there have been many questions raised with regard to reproducibility of cDNA microarray data and slide-to-slide variation, it was important to determine if a duplicate experiment would yield results similar to Con A Experiment 2. Thus, a third experiment was performed to assess the reproducibility of our BOTL microarrays (Con A Experiment 3). In this experiment, differential labeling of cDNA, microarray hybridization, washing and scanning were all performed exactly as in Con A Experiment 2. When the results of this experiment were compared to those of Experiment 2, the overall correlation between the 2 experiments was 0.72 and was highly significant (P < 0.001). Table 1 lists 25 clones from Con A Experiment 3 whose relative expression ratio (Cy5/Cy3) demonstrated 5-fold or greater activation in response to Con A. Also listed in Table 1 are the expression ratios for these same clones obtained in Con A Experiment 2 and P-values associated with the Cy5 to Cy3 comparison for each clone in each experiment. Of the 25 clones selected on the basis of expression ratio from Con A Experiment 3, 23 (92%) showed a similar activation trend in Con A Experiment 2, although the magnitude of the change is different in some cases. The 2 clones not showing significant activation in Con A Experiment 2 (dark arrows, Table 1) showed either no (BOTL0100010_D08) or only a slight (BOTL0100001XC05R) activation. In no case did we observe an opposite trend between the 2 experiments, that is a gene activated in one experiment but depressed in the other. Thus, the greatest variation in results between Con A Experiments 2 and 3 was not in which genes were activated by Con A stimulation, but in the absolute degree of activation.
Table 1

Comparison of Expression Ratios for Selected Clones Between Con A Experiments 2 and 3.


Con A-2

Con A-3

Clone Name

Cy5/Cy3 Ratio


Cy5/Cy3 Ratio
















































































































scavenger R















Data for Con A Experiment 3 was filtered to show only those clones whose average expression ratio was ≥ 5. The resulting list of clones was used to parse matching clone data from Con A Experiment 2 using an outjoin comparison in Microsoft Access. The resulting data was organized to show Clone name, average expression ratio for Con A Experiments 2 and 3, along with the associated P-values from the Student's 2-tailed t-test. Clones showing different expression ratio trends between the 2 experiments are indicated by a solid arrow (←) to the right of Table 1. Clones exhibiting at least one P-value > 0.1 are indicated by an asterisk (*) to the right of Table 1.


Application of cDNA microarray technology to complex problems in animal and veterinary science will require development of high-quality integrated resources dedicated to particular species and perhaps to specific systems within species. The required resources include: 1) high-quality normalized cDNA libraries from the tissues of interest; 2) collections of ESTs or random clones representing transcripts expressed in the system of interest; 3) databases to track EST clone information and assist in interpretation of microarray data; 4) microarray designs that allow production of robust and reproducible data. In this manuscript, we have described development of such resources for study of bovine immunobiology.

Development of high quality cDNA libraries is an essential step towards the generation of ESTs for cDNA microarray experiments. Library normalization is of particular importance as the use of normalized libraries significantly increases the efficiency of EST generation (Soares 1994, Hiller et al. 1996). The BOTL library reported here is of high quality and will allow us to generate many unique ESTs for gene expression studies of bovine immunobiology using cDNA microarray technology. Initial analysis of ESTs generated from the BOTL library indicate that many of the clones already processed (171) are not yet represented in the public databases (dbEST) and may therefore represent genes transcribed preferentially in bovine blood leukocytes.

EST data from our libraries is stored in a custom designed web-accessible database that retrieves and displays the top BLAST results for each clone compared against sequences in the Genbank non-redundant database. Additional information, including clone sequence data and the TIGR cluster (TC) number are also displayed. Links to these public databases allows rapid retrieval of additional information on sequences matching particular clones. The database is searchable with regards to clone name, keywords derived from BLAST hits, and by direct BLAST searching within the database. The entire library contents as well as the results of any searches may be exported in a comma separated values (CSV) file that is compatible with most spreadsheet and database systems. These features are of primary importance when analyzing results of cDNA microarray experiments, since one can easily move from a list of clone names to a tentative identification of the transcript represented by each clone. Public access to the database ensures timely dissemination of data as well as providing a ready analysis tool for collaborators at distant sites.

Initial construction of cDNA microarrays with amplicons from a subset (720 out of 932) of BOTL EST clones was performed in a manner that accommodated triple spotting of each amplicon and incorporation of numerous controls. This design and subsequent analysis of the microarray data are rigorous, as only 3 amplicons showed false positive results in a control experiment (0.4%, Con A Experiment 1). These results suggest we can be confident that most differentially expressed genes highlighted in an experiment using the BOTL microarrays are, in fact, differentially expressed. The triple spotting pattern also allows statistical analysis of the data for each amplicon within a microarray (Table 1 and [7], this issue). This is a particularly important consideration since the standard errors associated with any one amplicon can vary widely.

We assessed the ability of our BOTL microarrays to detect the response of mixed PBMCs to a general mitogenic agent, Con A. Comparison of gene expression patterns for mixed PBMCs stimulated with Con A versus PBS alone revealed a dramatic upregulation of gene expression. In total, over 90 genes displayed enhanced expression in the Con A treated PBMCs relative to unstimulated PBMCs from the same animals. A separate experiment (Con A Experiment 3) comparing Con A versus PBS stimulation with PBMCs from a different group of animals was highly correlated (r = 0.72, P < 0.001) with the first Con A versus PBS experiment. Of 25 clones selected from Con A Experiment 3 on the basis of expression ratios (Cy5/Cy3) equal to or greater than 5-fold, 23 showed similar activation patterns in Con A Experiment 2. We attribute this high level of reproducibility to the microarray design, which includes triple spotting of all amplicons as well as numerous positive and negative controls.

In summary, we have described development of integrated resources that allow application of cDNA microarray technology to functional genomics studies of bovine immunobiology. Microarrays resulting from this work continue to be improved by addition of amplicons representing known genes involved in immunological responses as well as more ESTs from the BOTL library. We believe these resources will assist us and collaborators in advancing the status of bovine immunobiology, allowing application to a wide variety of production and infectious diseases.


*This contribution was published in Acta vet. scand. 2001, 42, 391–405.



The authors wish to acknowledge the excellent technical assistance of Dr. Patty Weber and Mr. Chris Colvin of the Immunogenetics and Molecular Pathogenesis Laboratories, respectively, at Michigan State University. We further acknowledge the laboratory contributions of Mr. William Nobis and Mr. Matthew J. Coussens. Dr. George Smith is acknowledged for kindly providing amplicons representing bovine matrix metalloproteinases and several cell-surface receptors for inclusion on the BOTL microarrays. This work was supported through funds provided by the Michigan Agriculture Experiment Station, the Michigan State University Office of the Vice President for Research and Graduate Studies, the Research Excellence Fund State of Michigan, and the Department of Animal Science at Michigan State University.

Authors’ Affiliations

Department of Animal Science and Center for Animal Functional Genomics, Michigan State University, East Lansing, USA


  1. Barrans JD, Stamatiou D, Liew C: Construction of a human cardiovascular cDNA microarray: portrait of the failing heart. Biochem Biophys Res Commun. 2001, 280: 964-9. 10.1006/bbrc.2000.4137.View ArticlePubMedGoogle Scholar
  2. Blecha F: Immunomodulation: a means of disease prevention in stressed livestock. J Anim Sci. 1988, 66: 2084-90.PubMedGoogle Scholar
  3. Bonaldo MF, Lennon G, Soares MB: Normalization and subtraction: two approaches to facilitate gene discovery. Genome Res. 1996, 6: 791-806. 10.1101/gr.6.9.791.View ArticlePubMedGoogle Scholar
  4. Bowtell DD: Options available – from start to finish – for obtaining expression data by microarray. Nat Genet. 1999, 21: 25-32. 10.1038/4455.View ArticlePubMedGoogle Scholar
  5. Bull JH, Ellison G, Patel A, Muir G, Walker M, Underwood M, Khan F, Paskins L: Identification of potential diagnostic markers of prostate cancer and prostatic intraepithelial neoplasia using cDNA microarray. Br J Cancer. 2001, 84: 1512-9. 10.1054/bjoc.2001.1816.PubMed CentralView ArticlePubMedGoogle Scholar
  6. Burton JL, Nonnecke BJ, Dubeski PL, Elsasser TH, Mallard BA: Effects of supplemental chromium on production of cytokines by mitogen-stimulated bovine peripheral blood mononuclear cells. J Dairy Sci. 79: 2237-46.Google Scholar
  7. Burton JL, Madsen SA, Yao J, Sipkovsky SS, Coussens PM: An immunogenomics approach to understanding periparturient immunosuppression and mastitis susceptibility in dairy cows. Acta vet scand. 2001, 42: 407-424.PubMedGoogle Scholar
  8. Collier RJ, Beede DK, Thatcher WW, Israel LA, Wilcox CJ: Influences of environment and its modification on dairy animal health and production. J Dairy Sci. 65: 2213-27.Google Scholar
  9. Ellisen LW, Palmer RE, Maki RG, Truong VB, Tamayo P, Oliner JD, Haber DA: Cascades of transcriptional induction during human lymphocyte activation. Eur J Cell Biol. 2001, 80: 321-8. 10.1078/0171-9335-00162.View ArticlePubMedGoogle Scholar
  10. Galyean ML, Perino LJ, Duff GC: Interaction of cattle health/immunity and nutrition. J Anim Sci. 1999, 77: 1120-34.PubMedGoogle Scholar
  11. Geraci MW, Moore M, Gesell T, Yeager ME, Alger L, Golpon H, Gao B, Loyd JE, Tuder RM, Voelkel NF: Gene expression patterns in the lungs of patients with primary pulmonary hypertension: a gene microarray analysis. Circ Res. 2001, 88: 555-62. 10.1161/01.RES.88.6.555.View ArticlePubMedGoogle Scholar
  12. Heller RA, Schena M, Chai A, Shalon D, Bedilion T, Gilmore J, Woolley DE, Davis RW: Discovery and analysis of inflammatory disease-related genes using cDNA microarrays. Proc Natl Acad Sci USA. 1997, 94: 2150-5. 10.1073/pnas.94.6.2150.PubMed CentralView ArticlePubMedGoogle Scholar
  13. Jin K, Mao XO, Eshoo MW, Nagayama T, Minami M, Simon RP, Greenberg DA: Microarray analysis of hippocampal gene expression in global cerebral ischemia. Ann Neurol. 2001, 50: 93-103. 10.1002/ana.1073.View ArticlePubMedGoogle Scholar
  14. Kallioniemi OP: Biochip technologies in cancer research. Ann Med. 2001, 33: 142-7. 10.3109/07853890109002069.View ArticlePubMedGoogle Scholar
  15. Kennedy GC: The impact of genomics on therapeutic drug development. Exs. 2000, 89: 1-10.PubMedGoogle Scholar
  16. Khan J, Bittner ML, Chen Y, Meltzer PS, Trent JM: DNA microarray technology: the anticipated impact on the study of human disease. Biochim Biophys Acta. 1999, 1423: M17-28.PubMedGoogle Scholar
  17. Khanna C, Khan J, Nguyen P, Prehn J, Caylor J, Yeung C, Trepel J, Meltzer P, Helman L: Metastasis-associated differences in gene expression in a murine model of osteosarcoma. Cancer Res. 2001, 61: 3750-9.PubMedGoogle Scholar
  18. Kurian KM, Watson CJ, Wyllie AH: DNA chip technology. J Pathol. 1999, 187: 267-71. 10.1002/(SICI)1096-9896(199902)187:3<267::AID-PATH275>3.0.CO;2-#.View ArticlePubMedGoogle Scholar
  19. Lewin HA: Disease resistance and immune response genes in cattle: strategies for their detection and evidence of their existence. J Dairy Sci. 1989, 72: 1334-48. 10.3168/jds.S0022-0302(89)79241-9.View ArticlePubMedGoogle Scholar
  20. Lewohl JM, Dodd PR, Mayfield RD, Harris RA: Application of DNA microarrays to study human alcoholism. J Biomed Sci. 2001, 8: 28-36. 10.1007/BF02255968.View ArticlePubMedGoogle Scholar
  21. Manger ID, Relman DA: How the host 'sees' pathogens: global gene expression responses to infection. Curr Opin Immunol. 2000, 12: 215-8. 10.1016/S0952-7915(99)00077-1.View ArticlePubMedGoogle Scholar
  22. Mastro AM, Sniezek MJ: The effect of removal of adherent cells in lectin and allogeneic cell stimulation of bovine lymphocytes. Vet Immunol Immunopathol. 1983, 5: 161-76. 10.1016/0165-2427(83)90017-X.View ArticlePubMedGoogle Scholar
  23. Meltzer PS: Spotting the target: microarrays for disease gene discovery. Curr Opin Genet Dev. 2001, 11: 258-63. 10.1016/S0959-437X(00)00187-8.View ArticlePubMedGoogle Scholar
  24. Okragly AJ, Hanby-Flarida M, Baldwin CL: Monocytes control gamma/delta T-cell responses by a secreted product. Immunology. 1995, 86: 599-605.PubMed CentralPubMedGoogle Scholar
  25. Preisler MT, Weber PS, Tempelman RJ, Erskine RJ, Hunt H, Burton JL: Glucocorticoid receptor expression profiles in mononuclear leukocytes of periparturient Holstein cows. J Dairy Sci. 2000, 83: 38-47. 10.3168/jds.S0022-0302(00)74852-1.View ArticlePubMedGoogle Scholar
  26. Schena M, Heller RA, Theriault TP, Konrad K, Lachenmeier E, Davis RW: Microarrays: biotechnology's discovery platform for functional genomics. Trends Biotechnol. 1998, 16: 301-6. 10.1016/S0167-7799(98)01219-0.View ArticlePubMedGoogle Scholar
  27. Schena M, Shalon D, Davis RW, Brown PO: Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science. 1995, 270: 467-70. 10.1126/science.270.5235.467.View ArticlePubMedGoogle Scholar
  28. Schena M, Shalon D, Heller R, Chai A, Brown PO, Davis RW: Parallel human genome analysis: microarray-based expression monitoring of 1000 genes. Proc Natl Acad Sci USA. 1996, 93: 10614-9. 10.1073/pnas.93.20.10614.PubMed CentralView ArticlePubMedGoogle Scholar
  29. Sentsui H, Satou K, Hatakeyama H: Establishment of an IL-2-dependent bovine T cell line and an assay of lymphocyte response inhibition in leukemic bovine serum. J Vet Med Sci. 1992, 54: 293-6. 10.1292/jvms.54.293.View ArticlePubMedGoogle Scholar
  30. Whist SK, Storset AK, Larsen HJ: The use of interleukin-2 receptor expression as a marker of cell-mediated immunity in goats experimentally infected with Mycobacterium avium ssp. paratuberculosis. Vet Immunol Immunopathol. 2000, 73: 207-18. 10.1016/S0165-2427(00)00155-0.View ArticlePubMedGoogle Scholar


© The Author(s); licensee BioMed Central Ltd. 2003

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.