International Immunology Advance Access originally published online on February 28, 2006
International Immunology 2006 18(4):613-626; doi:10.1093/intimm/dxl001
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
A comprehensive SAGE database for the analysis of 
T cells
Veterinary Molecular Biology, Montana State University, Bozeman, MT 59717, USA
1 Present address: Molecular Biosciences Building, 960 Technology Boulevard, Bozeman, MT 59718, USA
Correspondence to: M. Jutila; E-mail: uvsmj{at}montana.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionSupplementary dataReferences |
|---|
T cells have been conserved since the adaptive immune system arose, yet their importance is still unclear. In an attempt to compensate for the lack of a broad knowledge base of T cells across species, global analyses of T cell transcriptomes have been performed using serial analysis of gene expression (SAGE). Twelve new SAGE libraries were generated from the following bovine lymphocyte populations: magnetic bead-sorted blood T cells, spleen T cells and enriched spleen 
ß T cells from a single calf, both rested and Con A/IL2 stimulated, and flow cytometry-sorted blood and ß T cells each either rested, Con A/IL2, or phorbol 12 myristate 13-acetate/ionomycin stimulated. These new libraries complement two earlier SAGE libraries of circulating T cell subsets. These databases were analyzed using new web-based bioinformatic tools, which allow the user to rapidly compare gene expression patterns within these and other SAGE and standard expressed sequence tag libraries generated from different cell types and different species. These analyses revealed striking differences between blood and spleen T cells and how these cells respond to mitogenic stimulation. These analyses also confirm previous studies that suggested that global gene expression in and ß T cells is quite similar; however, a 5-fold increase in T cell-specific transcripts could be induced by Con A/IL2 stimulation. These new public databases provide additional resources for the annotation/analysis of global gene expression in T cells, which will facilitate studies of the biology of this enigmatic lymphoid cell.
Keywords: serial analysis of gene expression (SAGE), ß T cells
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionSupplementary dataReferences |
|---|
T cells represent only a minor population of circulating T lymphocytes in most animals, but are evolutionarily conserved. T cells have been the focus of many studies over the last two decades, yet their specific functions within the tissues to which they localize remain unclear (1). Unique biological and immunological functions of T cells have possibly been overlooked because many of the earlier studies of these cells were designed based on known ß T cell biology. The unique functions of T cells also remain vague due to differences identified between the various species studied (2, 3, 4). For the most part, sequential analysis of single functional attributes has been used to study these cells in various animals, including humans. This approach is effective in studying T cells in a single species but does not efficiently reveal clusters or patterns of activities of T cells, which may be conserved across species.
Recently, global analyses of T cell transcriptomes have been performed using both microarrays and serial analysis of gene expression (SAGE; 5), which have provided unique insights into the function of these cells (6). In one study (7), oligonucleotide microarrays were used to examine murine intraepithelial lymphocytes ( IELs) in a Yersinia infection model. Though a role for T cells in the Yersinia infection was not identified, some unique gene expression patterns were revealed. For example, T cells were implicated in intestinal lipid metabolism and cholesterol homeostasis. Another group compared and ß T cells isolated from the same tissue (murine IELs) using SAGE (8). The analysis showed that these two T cell populations had highly similar gene expression profiles, though some T cell-specific transcripts were detected. A similar study compared murine T cells to unconventional T cells also using SAGE (9). It was concluded that T cell gene expression profiles were similar to those of unconventional CD8+ ß T cells.
We have previously compared specific subsets of T cells isolated from bovine peripheral blood (10, 11). SAGE was used to compare CD8– and CD8+ T cell subsets stimulated with phorbol 12 myristate 13-acetate (PMA)/ionomycin, and cDNA microarrays were used to study the same subsets following Con A/IL2 stimulation. Results revealed that CD8– T cells expressed genes that correlated with inflammatory and proliferative profiles, while CD8+ T cells expressed genes of more quiescent and apoptotic profiles. A complementary study was just completed that compared human V1+ and V2+ T cell subsets under various activation conditions (12). These subsets were found to have similar profiles as the bovine CD8+ and CD8– T cell subsets, respectively. These findings show the utility of global gene expression analyses in identifying trends between species.
Though the global analysis of gene expression in T cells has provided new insights into the biology of these cells, transcriptomes of only a few select populations of T cells in limited activation states and that identify few unique gene expression patterns are available to date. To increase the potential of identifying unique T cell transcripts and to provide a more comprehensive data set, we employed SAGE to construct 12 additional SAGE libraries. SAGE was chosen for these analyses because data from any system, including data from different species and laboratories, are directly comparable due to the unique advantage of the SAGE technique in that the data is internally normalized (5). These new libraries represent the following bovine lymphocyte populations: magnetic bead-sorted blood T cells (>90% pure), spleen T cells (>90% pure) and enriched spleen ß T cells (>80% pure) from a single calf, both rested and Con A/IL2 stimulated for 6 h, and flow cytometry-sorted blood and ß T cells (pooled from three calves; >95% pure) each either rested, Con A/IL2, or PMA/ionomycin stimulated as above. This study complements our original CD8+ and CD8– T cell SAGE data sets.
We hypothesized that many T cell-specific/selective transcripts and gene expression patterns would be identified by comparing these libraries against each other and with other data sets. Results indicate that these libraries do exhibit several gene expression differences between rested and mitogen-stimulated blood T cells, between blood and spleen T cells, as well as differences between spleen and blood and ß T cells. A cross-comparison of these libraries revealed unique T cell-specific transcripts, as well as tissue-specific gene expression patterns. In addition, a web-based resource for the use of these libraries is described.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionSupplementary dataReferences |
|---|
Cell extraction, magnetic bead cell sorting and culturing of calf spleen and peripheral blood lymphocytes
A bovine calf spleen was surgically removed and kept on ice in HBSS (Mediatech, Inc., Herndon, VA, USA). Tissue was homogenized using tissue grinders. Single-cell suspensions of mononuclear cells were isolated using Histopaque 1077 (Sigma Aldrich, St Louis, MO, USA), as per the manufacturer's instructions and filtered through a 70-µm filter. PBMCs were also isolated using Histopaque 1077. PBMCs were monocyte depleted by incubation in plastic flasks and all non-adherent peripheral blood lymphocytes (PBLs) were collected.
T cells were isolated from both PBLs and single-cell suspensions of spleen lymphocytes using magnetic bead cell sorting (MACS) LS Separation Columns (Miltenyi Biotec, Auburn, CA, USA) as per the manufacturer's instructions and GD3.8 (anti-bovine TCR), as previously described (13). Briefly, PBLs were first depleted of B cells using the magnetic bead separation and CC21 antibody (specific for bovine B cells). Purified T cells and ß T cell-enriched populations were obtained [purity of >90% spleen and >95% blood T cells and >80% spleen ß T cells (GD3.8– population); data not shown]. Cells were rested overnight in complete RPMI (10% fetal bovine serum in RPMI supplemented with 1% each of essential amino acids, penicillin/streptomycin, L-glutamine and 10 mM HEPES; Mediatech, Inc.) at 37° C, 10% CO2 at a concentration of 1 x 107 cells ml–1. After approximately 16 h of rest, lymphocytes were either stimulated for 6 h with 5 µg ml–1 Con A (Sigma Aldrich) and 1 ng ml–1 IL2 (Pepro Tech, Inc., Rocky Hill, NJ, USA) or rested for an additional 6 h.
Flow cytometric analysis
A small fraction of cells were stimulated or remained resting for 24 h for use in flow cytometric analysis. Previously described techniques were used to perform single- and two-color flow cytometric analyses using a FACSCalibur (BD Biosciences, San Jose, CA, USA; 10, 11, 14). Staining with the following antibodies was used to determine the extent of stimulation: GD3.8 directly labeled with FITC and DREG56 (anti-bovine L-selectin), H58A (anti-bovine MHC class I; VMRD, Inc., Pullman, WA, USA), CACT1165A (anti-bovine IL2R; VMRD, Inc.) and CAT82A (MHC class II; VMRD, Inc.). Second-stage reagents included PE-conjugated anti-mouse IgG (Jackson ImmunoResearch Laboratories, West Grove, PA, USA). The indirect stain was performed first, followed by incubation in 10% mouse serum to block all reactive sites of the immobilized second-stage antibody and then the FITC-labeled GD3.8 mAb stain. A successful activation profile resulted in decreased expression of L-selectin and increased expression of MHC class I, MHC class II and/or IL2R.
Flow cytometric cell sorting and culturing of PBLs
High-speed cell sorting on a Vantage SE cell sorter (BD Biosciences) was used to obtain >95% pure peripheral blood and ß T cell populations from three calves, as described previously (10, 11, 14). Briefly, PBMCs were isolated as described above, and stained using CC42 mAb (anti-bovine CD2) and a FITC-labeled GD3.8 mAb. All GD3.8+ cells were included in the T cell-sorted population, and only GD3.8–/CD2+ cells were included in the ß T cell population. Cells were rested overnight and then stimulated as above with Con A/IL2 or 20 ng ml–1 PMA (Sigma Aldrich) and 0.5 µg ml–1 ionomycin (Sigma Aldrich) or remained resting for 6 h prior to RNA extraction.
SAGE library construction
RNA was extracted from cells using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) as per the manufacturer's protocol. SAGE libraries were constructed using the SAGE-lite method, utilizing template switching and PCR amplification, as previously described (15). Briefly, a biotinylated oligo(dT) primer (biotin-5'-AAGCAGTCCTAACAACGCAGAGTAC(T)30VN-3', where V = A, C or G, and N = T, C, G or A) and Superscript II reverse transcriptase (RT) (Invitrogen) were used to make first-strand cDNA from 100 ng of total RNA. Second-strand synthesis and cDNA amplification were completed by PCR using Advantage2 Polymerase Mix (Clontech, Palo Alto, CA, USA) and a switching primer (5'-AAGCAGTGGTAACAACGCAGAGTACGCGGG-3') in combination with the original biotinylated oligo(dT) primer. SAGE library construction was completed using 5 µg of the double-stranded cDNA following standard protocols (5, 16, 17). FACS-sorted PBL libraries were built with equal amounts of cDNA from cell preparations from three different calves. Large-scale sequencing of concatemerized ditags was completed by Agencourt Bioscience Corporation (Beverly, MA, USA).
SAGE data analysis
Ditags were pulled from sequences and duplicate ditags that arise during PCR amplification were eliminated (18). Ditag sequence was processed using software previously developed to extract individual SAGE tag information, record tag frequency and correct sequence error in the raw data set by nearest neighbor analysis (18; http://195.70.0.115/bioinf/sage/). Tag frequencies for each library were adjusted for slight differences in sequencing depth by multiplying the tag count by the adjusted library size, divided by the actual size—where the adjusted size was equal to 25 000 tags. This corrected data set provides the highest level of confidence and, therefore, was used in all of the analyses presented. Tags were compared with The Institute for Genomic Research (TIGR) bovine gene index release 11.0 sequence (http://www.tigr.org/). Annotation of SAGE/expressed sequence tag (EST) matches was performed by BLAST comparisons to the Human 15.0 and Mouse 14.0 TIGR databases using a 1 x10–6 cut-off. Uncorrected and adjusted tag frequencies and annotations can be viewed and analyzed on the public Bovine SAGE database (http://vmbmod10.msu.montana.edu/vmb/jutila-lab/sagebov.htm).
SAGE tags with adjusted frequencies of 10 or greater within all 12 libraries were imported into GeneSpring 7.2 (Agilent Technologies Inc., Palo Alto, CA, USA) to build visual representations of the 12 libraries. Data were analyzed in GeneSpring 7.2 using the log of ratios mode. These data sets were compared by cluster analysis using the Pearson correlation as a similarity measurement to generate gene trees. Condition trees were also generated based on the Pearson correlation. Data sets united by the shortest branches were the most similar. Correlation coefficients can be determined by subtracting the length of the branches (represented on the condition trees) from 1. Gene lists were generated by comparing fold changes of the adjusted tag frequencies.
Reference long SAGE library construction
A long SAGE library was constructed as previously described (19). Briefly, RNA was collected as described above from lymphocytes isolated from calf spleen, lymphatic fluid, lymph nodes and peripheral blood and pooled together. The long SAGE library was built as described for the SAGE libraries, with the following modifications. Linkers containing restriction sites for MmeI were used rather than linkers containing BsmFI restriction sites. Cleavage of cDNA with MmeI resulted in tags of 21 bp in length rather than only 14 bp. Elimination of linkers from the ditags was performed with four washes over streptavidin magnetic beads. Concatemerization and sequencing were performed as described above for the SAGE libraries. Data analysis was performed using in-house programs designed to isolate 17- to 22-bp tags. Long SAGE tags were annotated as described above. In-house programs were also used to compare short SAGE tags with the corresponding long SAGE tags.
Web database platform
The web site (http://vmbmod10.msu.montana.edu/vmb/jutila-lab/sagebov.htm) is served from a Dell 4400 running RedHat Linux (http://www.redhat.com/) and Apache web server (http://apache.org/). All scripts created to implement and serve the web site were written in Perl. The Perl DBI and CGI modules are used, respectively, to connect to the MySQL (http://dev.mysql.com/) database and serve the site to web browsers.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionSupplementary dataReferences |
|---|
SAGE libraries/annotation of SAGE tags/bioinformatic tools
Twelve new SAGE libraries were generated from the following bovine lymphocyte populations: MACS-separated blood
T cells, spleen T cells and enriched spleen ß T cells from a single calf, both rested and Con A/IL2 stimulated for 6 h and FACS-sorted blood and ß T cells (pooled from three calves) each either rested, Con A/IL2, or PMA/ionomycin stimulated as above. Activation of the blood and spleen populations was confirmed by increased staining by anti-IL2R-, MHC class I- and/or MHC class II-specific antibodies and/or decreased staining by an L-selectin-specific antibody in flow cytometric assays after 24 h of stimulation (Supplementary Figure 1, available at International Immunology Online). The 12 libraries were separated for comparative analysis based on the cell isolation procedure: MACS or FACS.
Over 10 000 quality SAGE tags were generated from each library (with the exception of the PMA/ionomycin-stimulated T cell library with only 7638 tags) for a total of over 205 000 tags to provide an initial ‘corrected’ data set for comparison (Table 1, Bovine SAGE database: http://vmbmod10.msu.montana.edu/vmb/jutila-lab/sagebov.htm). Confirmation of SAGE as a predictor of gene expression in bovine blood T cells was previously performed in our laboratory (10). Here, differences in MACS-sorted rested and Con A/IL2-stimulated blood T cells were confirmed using real-time RT–PCR in 9 out of 13 candidate genes, with only one gene showing opposite regulation (Supplementary Figure 2, available at International Immunology Online). This level of confirmation was similar to that shown in our previous study (10), further validating SAGE in this system and also illustrating the consistency in gene expression profiles in blood T cells from different animals. Spleen cell preparations were not as consistent in follow-up analyses of specific genes, which is discussed below.
|
Between all 12 SAGE libraries, there were over 88 000 unique tags sequenced; however, 71.2% were represented only once in a single library (Table 2). In theory, each tag represents a unique gene; however, some unique tags are created by PCR or sequencing error due to a single base pair change (18). Highly abundant tags were likely to be the source of several singletons (tags represented only once) that differ by only a single base pair from the abundant tag. Therefore, singletons were not used in the analyses reported here, but are made available in the bovine SAGE database. Classification of tags by increasing frequency cut-offs greatly decreased the number of tags in each category (Table 2). For example, tags with an adjusted frequency of 5 or greater between all 12 libraries represented only 5.2% of the unique tags, correlating with approximately 4600 genes. Approximately, 41% of the unique tags from each library matched a known expressed bovine sequence for annotation. As expected, a greater percentage of the medium- to high-abundance tags matched a sequence, increasing confidence that they represent ‘real’ genes (Table 2). For this reason, our global comparisons use only tags with an adjusted frequency of 10 or greater between all 12 libraries.
|
GeneSpring 7.2 software was used to build visual representations of each SAGE library, facilitating comparison to the others. Using the high-confidence cut-off of tags represented by an adjusted frequency of 10 or more, all 12 libraries were compared (Fig. 1). Many tags were expressed at similar levels in all libraries, most likely representing housekeeping genes. However, as predicted, similarities and differences became apparent between the various activation states and cell types.
|
Effect of mitogen stimulation on gene expression in MACS-separated blood and spleen
T cellsWe hypothesized that several genes would be identified as differentially expressed between rested and mitogen-stimulated blood
T cells. For an initial comparison, SAGE libraries were constructed from MACS-sorted rested and Con A/IL2-stimulated blood T cells. Magnetic bead separation was used because others have previously used the same technique to study gene expression in blood T cells (13) and it allows for isolation of a sufficient number of cells from the spleen (see below). Several tags were identified that were differentially expressed between the two libraries. Specifically, 43 tags were expressed at a 10-fold or greater level in rested cells, and 60 tags were expressed at a 10-fold higher level in Con A/IL2-stimulated cells (Tables 3 and 4). Notably, some tags were differentially expressed at levels over 100-fold.
|
|
Many tags could not be uniquely annotated as a specific gene. Some tags did not match any known bovine ESTs or matched a non-annotated bovine EST, and thus may represent novel genes. Other tags matched several known bovine ESTs and could not be uniquely identified as a specific gene. Curiously, tags that were represented at a greater level in the stimulated population were much more difficult to uniquely annotate because on average those tags matched 24 different ESTs (data not shown). However, tags expressed at higher levels in rested cells matched fewer than three sequences on average and could be annotated with a greater level of confidence.
An experiment was performed to examine how T cell subsets localized to different microenvironments responded to mitogenic stimulation. T cells from a bovine calf spleen were isolated for SAGE using the single-step MACS bead separation protocol and stimulated with Con A/IL2 or remained resting. Spleen T cells had a much less robust change in gene expression after Con A/IL2 stimulation when compared with the changes seen in the corresponding blood populations (Fig. 2). Specifically, only two tags were expressed at a 10-fold higher level in activated spleen cells than in the rested population, and only one tag was expressed at a 10-fold higher level in rested spleen T cells (data not shown).
|
All four populations of blood and spleen
T cells (rested and Con A/IL2 stimulated) shared expression of many of the same genes (Fig. 2). However, cells from both locations expressed tissue-specific genes. This study identified 70 genes that had an adjusted frequency of at least 10 when both blood libraries were combined and were not present in the spleen libraries, including CD205, a dendritic and myeloid cell antigen-recognition molecule (20; Table 5). There were 14 genes identified as specific to spleen T cells (Table 5), including nuclear receptor co-repressor 1 (N-CoR1), which has been implicated in regulation of T cell development and macrophage activation (21, 22).
|
T cells localized to the spleen were seemingly in a more transcriptionally active state than were circulating T cells. Several SAGE tags representing transcription factors were identified in activated blood T cells, which had a gene expression pattern more similar to both the rested and activated spleen cells than to the rested blood cell population (Fig. 2). The 5-fold greater number of blood-specific genes than spleen-specific genes was partially due to the rested population of blood T cells, which displayed a very different gene expression profile than the other three populations. Tables 3 and 4 illustrate that genes expressed at high levels in stimulated blood T cells were also expressed at high levels in both spleen populations. Likewise, genes that were expressed at higher levels in rested blood T cells were expressed, on average, at low levels in the spleen. These results suggest that T cells isolated from various microenvironments can have dramatically different gene expression profiles, thus warranting future studies on T cells from different tissues.
As an initial experiment to compare T cells to the predominant ß T cell population, SAGE libraries were constructed from rested and Con A/IL2-activated ß T cell-enriched preparations from the same spleen used in the analyses described above. A shortcoming in the bovine model is that there are no reagents available to efficiently purify ß T cells using a single mAb and the MACS bead separation protocol. At best, we achieved in the non- T cell fraction, ß T cell preparations which were approximately 80% pure versus >90% pure T cells (data not shown). The contaminating cells in the ß T cell preparation included B cells (5%), residual T cells (10%) and a few monocytes and/or NK cells (<5%) (data not shown).
As expected, both spleen T cell populations expressed many of the same genes, though at slightly different levels (Fig. 3A). However, several genes unique to the T cell population were observed, such as N-CoR1 (also spleen specific; data not shown). As another means of illustrating this data, the percentage of expressed genes that were similar or different in two-way comparisons of the four libraries are shown in pie graph form (Fig. 3B). When the rested T cell populations were compared, only 1% of the unique tags were specific to T cells, suggesting a minimal impact of the MACS bead isolation protocol on generating artificial differences between the cell populations. However, following Con A/IL2 stimulation, which provided a robust signal for both populations, the percentage of tags specific to T cells increased to 5%.
|
Comparisons of blood and spleen
T cells and spleen ß T cells, regardless of activation state, indicated that the gene expression profiles of spleen and ß T cells were more similar to each other than are T cells isolated from different microenvironments (Supplementary Figure 3, available at International Immunology Online). These data suggested that microenvironments, or the T cell subsets that localize to them, may play a greater role in determining gene expression profiles in T cells than does specific TCR chain expression. It should be noted that the cell isolation procedure needed for isolating tissue cells is much more extensive than that used to isolate blood cells. The difference in cell health (due to sorting procedures) of the isolated populations may affect the quality of RNA isolated for the library construction, which would affect the resulting SAGE data. However, these data also suggested that more insight into unique functions of T cells may be gained when these cells are analyzed following their response to specific agonists in vivo and in vitro.
Analysis of FACS-sorted PBLs
To test the conclusions drawn from the SAGE libraries built from MACS-separated lymphocytes, another series of SAGE experiments was performed using FACS separation to isolate purer preparations of and ß T cells (>95%). To minimize the effect of gene expression variability between animals, these SAGE libraries were generated from equivalent, pooled cDNA preparations from three animals. Also, to compare the impact on gene expression by different activating procedures, Con A/IL2 and PMA/ionomycin stimulation were tested.
Comparison of the data from Con A/IL2 and PMA/ionomycin-stimulated total T cells suggested that T cells responded to the different mitogenic stimuli by up- and down-regulating different sets of genes. The gene expression profiles of Con A/IL2-stimulated T cells were more closely related to those of rested blood T cells, whereas PMA/ionomycin-stimulated cells expressed a greater number of unique genes (Supplementary Figure 4 and Supplementary Tables 1 and 2, available at International Immunology Online). Specifically, there were 57 genes that were 15-fold down-regulated in PMA/ionomycin-stimulated T cells and only 14 genes down-regulated in Con A/IL2-stimulated T cells. Ten of these genes were down-regulated over 15-fold under both conditions. Likewise, PMA/ionomycin induced up-regulation of 58 genes, whereas Con A/IL2 induced only 18 genes 15-fold over rested T cells. This difference is readily seen in Supplementary Figure 4b, available at International Immunology Online, in that rested and Con A/IL2-stimulated T cells share 70% of their expressed genes, whereas rested and PMA/ionomycin-stimulated T cells share only 43% of their expressed genes. Interestingly, of the genes expressed by Con A/IL2 or PMA/ionomycin-stimulated T cells, nearly one-third of them were unique to each activation state.
Both stimuli induced up-regulation of transcription factors, including nuclear erythroid 2-related factor 1. Integrin- expression was also up-regulated in both populations. The activation marker CD69 was dramatically up-regulated in cells stimulated with PMA/ionomycin. Con A/IL2 also induced increased expression of CD69 on T cells, but to a lesser extent, supporting the observation that PMA/ionomycin had a more robust effect on gene expression profiles (Supplementary Table 2, available at International Immunology Online).
Like T cells, ß T cells responded to the mitogens by regulating the expression of different sets of genes. PMA/ionomycin also induced a greater change in gene expression profiles than did Con A/IL2. ß T cells down-regulated the expression of approximately the same number of genes with respect to PMA/ionomycin and Con A/IL2, as did T cells, 38 and 12, respectively (Supplementary Figure 5 and Supplementary Table 3, available at International Immunology Online). However, ß T cells did not up-regulate nearly as many genes in response to the stimuli (PMA/ionomycin and Con A/IL2) as did T cells, 22 and 5, respectively (Supplementary Table 4, available at International Immunology Online). In fact, the three different ß T cell populations shared a much higher percentage of their expressed genes than did T cells, illustrating less of a transcriptional response to the stimulants.
ß T cells, similar to T cells, up-regulated the expression of transcription and translation factors in response to mitogenic stimulation (Supplementary Table 4, available at International Immunology Online). Several genes encoding cytokines were up-regulated over 10-fold in PMA/ionomycin-stimulated ß T cells, including IFN-, macrophage inflammatory protein (MIP-1) and MIP-3 (data not shown). Both stimuli also dramatically down-regulated some genes, including CD2 and connexin 43 (Supplementary Table 3, available at International Immunology Online).
In comparing and ß T cell gene expression patterns, predictions based on the spleen analysis held true. For example, the majority of expressed genes were the same between blood T cell populations (81% in the rested populations; Supplementary Figure 6, available at International Immunology Online). However, upon stimulation with either Con A/IL2 or PMA/ionomycin, the lymphocyte populations became dramatically more dissimilar and shared only 71 and 59% of their expressed genes, respectively. The greater level of dissimilarity between FACS-separated T cells than in MACS-separated populations suggested that the MACS-separation protocol may have masked differences in gene expression profiles. Fig. 4 demonstrates the more robust response to mitogenic stimulation by T cells than ß T cells in that in all three conditions T cells shared only 29% of their expressed genes, whereas ß T cells shared 41%. This study also uncovered nearly 20 genes that were potentially specific to T cells and another 20 that were potentially specific to ß T cells (Supplementary Figure 7, available at International Immunology Online, and Table 6). The transcript that was expressed at the highest level in ß T cells, but was not present in the T cell population, was the alpha chain of the TCR. Another confirmatory gene was the specificity of CD2 to the ß T cell library, as the large majority of bovine T cells do not express the CD2 molecule (23). Other T cell-specific genes worthy of note included vascular endothelial growth factor, prohibitin (involved in repression of cell proliferation; 24, 25) and a low-density lipoprotein receptor.
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionSupplementary dataReferences |
|---|
To gain a better understanding of unique
T cell responses to mitogenic stimuli in different microenvironments and in comparison to ß T cells, and to provide new genome resources to those who study these cells, 12 new SAGE libraries were constructed. These libraries complement two previously published T cell SAGE libraries from our laboratory (10) and other global gene expression studies of T cells (6, 7, 8, 9, 11, 12). An initial experiment was performed with MACS-sorted T cells to compare T cells from blood and spleen in both rested and Con A/IL2 activation states and to a spleen ß T cell-enriched population. To test the conclusions drawn from the analysis of the MACS-separated PBL and spleen and ß T cells and the impact of activation, another series of SAGE experiments was performed using FACS to isolate purer preparations of and ß T cells from three animals. These 12 libraries will continue to expand as we sequence additional tags and the bovine genome sequence and annotation becomes more complete. To facilitate the use of the available SAGE libraries in the bovine SAGE database, we present the results as uncorrected sequencing data and adjusted data in a web-based interface. The user can search for a specific gene of interest, identify highest abundance tags, identify uniquely expressed tags and identify tags based on fold-expression differences (Table 7). Importantly, although the presented data were obtained from only bovine cells, they are comparable across species because the data sets were annotated against human and mouse gene resources.
|
Occasionally, a SAGE tag with an interesting expression pattern cannot be annotated as a single gene. To help overcome this obstacle, a reference long SAGE library was constructed to provide additional sequence information for tags that could not be uniquely identified or for tags that did not match any known bovine sequence. In many cases, the additional 7 bp of sequence obtained with a long SAGE tag can uniquely identify which gene a short tag represents (Supplementary Table 5, available at International Immunology Online). In the case of short SAGE tags that do not match any known sequence, long SAGE can provide enough sequence information to perform reverse SAGE (26) or 3'-RACE to gain additional sequence for gene identification. This reference long SAGE library is only beginning to be sequenced, but is made available as a web resource as well.
T cells exhibit diverse responses to different infections, including immunoregulatory functions, direct anti-microbial responses or they can be seemingly irrelevant to infection clearance (1). To begin examining responses of T cells to mitogenic stimuli, we compared global gene expression profiles of MACS-sorted T cells in rested and Con A/IL2 activation states, which revealed dramatic differences. These findings were confirmed by data from the SAGE libraries built from very pure FACS-separated blood T cells (in the absence of potential cross-linking by the anti-TCR mAb, as may occur in the MACS bead protocol) from three animals. Additionally, as expected, these libraries demonstrated unique responses of and ß T cells to Con A/IL2 and PMA/ionomycin. Though both mitogens are powerful stimulants, they act on the cells through different mechanisms. Con A stimulates T cells by cross-linking a variety of cell-surface proteins displaying -D-mannosyl and -D-glucosyl residues, including CD3, which signals through the TCR and induces changes similar to those seen when lymphocytes contact specific antigens (27). PMA/ionomycin enter the cell and directly induce calcium-dependent and protein kinase C pathways, thus activating multiple signal-transduction mechanisms (28). As expected, the cells responded with different gene expression profiles to the two stimuli, with the response to PMA/ionomycin being greater. However, the extent of the differences was unexpected. This finding emphasizes the need to study T cells in response to various stimuli, both in vitro and in vivo.
Unique subsets of T cells localize to specific tissues (1). To begin the investigation of how the microenvironment to which a T cell localizes affects gene expression profiles, we compared T cells from blood and spleen in both rested and Con A/IL2-activated states. We chose to examine spleen T cells because our previous work demonstrated that these cells were different than circulating T cells (14, 23). Specifically, bovine CD8+ T cells accumulate in large numbers in the spleen, unlike in the blood where T cells are predominately CD8–. Also, the CD8+ subset remains in the red pulp, whereas CD8– T cells appear to quickly re-enter circulation (14). Upon comparing the blood and spleen T cells, it became immediately obvious that the spleen cells were in a more transcriptionally active state. Following Con A/IL2 stimulation, blood T cell gene expression was more similar to that of spleen cells than to that of rested blood cells. This observation is consistent with the fact that immune cells function within organs and tissues and use the blood simply to gain access to these sites. Once within the spleen, T cells are exposed to an array of environmental factors, including potential antigens, which drive significant changes in gene expression. Upon leaving the spleen, significant transcriptional repression likely takes place, leading to the profile seen in the rested blood T cells. Since many of the changes in gene expression are likely tissue specific, it will be important to extend these types of analyses to cells harvested from other tissues in order to gain an even more comprehensive picture of the T cell transcriptome. However, these initial libraries provide a resource for identification of potential spleen- and blood-specific transcripts.
In support of the finding that gene transcription is suppressed in rested blood T cells, we have identified message for a transcriptional repressor, B lymphocyte-induced maturation protein-1 (Blimp-1; 29), which is expressed in rested blood T cells and is down-regulated upon activation (data not shown). Blimp-1 expression in T cells was first identified by SAGE analysis (10). Further functional studies are needed to determine if Blimp-1 is a major transcriptional repressor in T cells, as it is in B cells and monocytes, and if it contributes to the differences seen in the SAGE libraries. A SAGE tag representing CD37 (involved in repression of T cell proliferation; 30) was also identified in rested blood T cells, but not in stimulated blood T cells or either population of spleen T cells (data not shown). Therefore, Blimp-1 and CD37 gene expression profiles correlate and their down-regulation is consistent with the transcriptionally activated profiles of spleen T cells.
Additional real-time RT–PCR follow-up experiments have been performed on cells isolated from five additional calf spleens and indicate that expression profiles of specific genes vary between spleen preparations, unlike blood cell preparations, which are more consistent (data not shown). We have not determined whether the variation seen is due to calf variability, the extensive cell isolation procedure needed to generate the spleen cell preparations, or the variability in the purities of the cell preparations. Although the regulation of specific genes seen in these SAGE libraries may not hold true for all calf spleen T cells, we believe that the global gene expression patterns are representative of lymphocytes in the spleen microenvironment.
Even though and ß T cells express distinct TCRs, localize in tissues differently and likely perform distinct, yet complementary functions, the limited gene expression analyses performed to date suggest that, overall, they express mostly the same genes (8, 9). Here, we expand on these earlier studies by first examining and ß T cells isolated from the same tissue (spleen) in both rested and activated states. We chose to examine spleen T cells because there are known differences between the two cell types within the spleen. For example, T cells localize to sites of T cell traffic within the spleen (marginal zone, red pulp and the marginal sinus), but are rarely found in the conventional T cell regions of the white pulp (31). Due to the lack of quality reagents for isolating bovine ß T cells, the spleen ß T cell populations in this study were merely enriched for ß T cells (80% ß T cells; the GD3.8– population) and contained B cells, residual T cells and likely monocytes and/or NK cells. As expected, we found that the large majority of expressed genes were shared between spleen and ß T cells. In fact, the rested spleen populations were nearly identical, with only 1% of the transcripts being unique to T cells. Interestingly, total spleen and ß T cells became less similar when stimulated with Con A/IL2 than they were in the rested state. The majority of the differences arose due to a selective 5-fold increase in T cell-specific transcripts. Though it might have been predicted that stimulation via the TCR would lead to and ß T cell-specific gene expression profiles, the magnitude of the effect was striking.
To test the findings of the spleen T cell comparison, an additional experiment comparing SAGE libraries built from very pure FACS-separated blood and ß T cells was performed. This experiment confirmed the findings from the spleen libraries that and ß T cells share the majority of their expressed genes in the rested state and that the two populations become more dissimilar upon stimulation. Interestingly, the two T cell types did not respond by regulating the same sets of genes. Cross-linking of the TCRs by Con A likely induces lineage-specific signaling, thus one might expect different transcriptional responses as seen in this study. However, PMA/ionomycin likely activates the same signal-transduction pathways in both and ß T cells, yet even a greater array of lineage-specific responses occurred. This suggests that ‘pre-wired’ differences in baseline genes involved in downstream signaling initially expressed in each cell type dictate these differences or possibly that the chromatin packaging of specific target genes vary between the cell types, thus preventing specific transcription of some genes. Both of these possibilities are currently being examined. These findings emphasize the need to compare and ß T cells in their rested states, and more importantly, following in vivo and in vitro activation. A more thorough analysis of various T cell agonists will likely reveal more examples of lineage-specific gene expression patterns. Importantly, our current studies confirmed that even though most expressed genes are the same in the two T cell populations, T cells have unique gene expression profiles compared with ß T cells, thus emphasizing unique roles for each cell type.
An important consideration in any gene expression analysis of isolated cell populations is the impact of the isolation procedure on subset-specific gene expression. In comparing ß to T cells, differences could be due, in part, simply to the different approaches used to isolate the cells (e.g. negative versus positive sorts, impact of different antibodies used in positive sorts, etc). Our initial experiment used cells sorted using the magnetic bead isolation procedure, which others have previously used in the analysis of bovine T cell gene expression (13), and we have found it does not drive a positive signal in the cell as measured by proliferation or IL2R up-regulation (data not shown). However, the potential cross-linking event caused by the magnetic bead could still be expected to alter basal level gene expression, generating artificial T cell-specific transcripts. Since this would only occur on T cells, conclusions from these experiments would be suspect. Our testing of the conclusions from our magnetic bead sorts used very pure flow cytometry-sorted cell populations in which the anti-TCR mAb was directly conjugated to FITC to minimize cross-linking. Interestingly, the trends and overall conclusions were similar regardless of what separation technique was used; however, there were differences in specific gene profiles associated with each technique (Supplementary Figure 8 and Supplementary Table 6, available at International Immunology Online). Thus, in comparing gene expression data from different cell populations, as one would expect, it is important that procedures used to isolate the cells are the same and, perhaps, multiple methods should be compared. Because the two T cell populations are treated in a more similar fashion in the FACS versus the MACS bead protocol, we predict that subset-specific genes identified in these studies would be more reliable and predictive, and it is these data sets in which we are focusing our gene discovery efforts.
T cells clearly represent a diverse population of lymphocytes based on their tissue localization and activation states. As predicted, the analysis of these SAGE libraries has identified unique gene expression patterns of T cells in the various conditions studied, which has resulted in insights into these cells and has provided a foundation for more comprehensive studies of T cell gene expression. As additional gene expression studies are performed, greater insight into T cells will be gained, allowing for a better understanding of the role of this enigmatic lymphoid cell.
|
Supplementary data |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionSupplementary dataReferences |
|---|
Supplementary data are available at International Immunology Online.
|
Acknowledgements |
|---|
This project was supported by the Initiative for Future Agricultural and Food Systems Grant no. 00-52100-9612 from the United States Department of Agriculture (USDA) Cooperative State Research, Education, and Extension Service, and with additional Federal funds from the National Institute of Allergy and Infectious Diseases, National Institutes of Health, Department of Health and Human Services, under Contract No. HHSN266200400009/N01-A140009. This project was also supported by USDA Animal Health and Hatch funds and the Montana State Agriculture Experiment Station. We would like to thank Nicole Meissner for her technical expertise and interesting discussions and Jodi Hedges for her critical review of the manuscript.
|
Abbreviations |
|---|
| Blimp-1 | B lymphocyte-induced maturation protein-1 |
| EST | expressed sequence tag |
| IEL | intraepithelial lymphocyte |
| MACS | magnetic bead cell sorting |
| MIP | macrophage inflammatory protein |
| N-CoR1 | nuclear receptor co-repressor 1 |
| PBL | peripheral blood lymphocyte |
| PMA | phorbol 12 myristate 13-acetate |
| RT | reverse transcriptase |
| SAGE | serial analysis of gene expression |
| TIGR | The Institute for Genomic Research |
|
Notes |
|---|
Transmitting editor: S. M. Hedrick
Received 21 March 2005, accepted 27 January 2006.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionSupplementary dataReferences |
|---|
- Hayday, W. 2000. cells: a right time and a right place for a conserved third way of protection. Annu. Rev. Immunol. 18:975.[CrossRef][Web of Science][Medline]
- Salerno, A. and Dieli, F. 1998. Role of T lymphocytes in immune response in humans and mice. Crit. Rev. Immunol. 18:327.[Web of Science][Medline]
- Elbe, A., Foster, C. A. and Sting, G. 1996. T-cell receptor ß and T cells in rat and human skin—are they equivalent? Semin. Immunol. 8:341.[CrossRef][Medline]
- Cahill, R. N., Kimpton, W. G., Washington, E. A. and Walker, I. D. 1996. Origin and development of the gamma delta T-cell system in sheep: a critical role for the thymus in the generation of TcR diversity and tissue tropism. Semin. Immunol. 8:351.[CrossRef][Medline]
- Velculescu, V. E., Zhang, L., Vogelstein, B. and Kinzler, K. W. 1995. Serial analysis of gene expression. Science 270:484.
[Abstract/Free Full Text] - Hedges, J. F., Graff, J. C. and Jutila, M. A. 2003. Transcriptional profiling of T cells. J. Immunol. 171:4959.
[Free Full Text] - Fahrer, A. M., Konigshofer, Y., Kerr, E. M. et al. 2001. Attributes of intraepithelial lymphocytes as suggested by their transcriptional profile. Proc. Natl Acad. Sci. USA 98:10261.
[Abstract/Free Full Text] - Shires, J., Theodoridis, E. and Hayday, A. C. 2001. Biological insights into TCRgammadelta + and TCRalphabeta + intraepithelial lymphocytes provided by serial analysis of gene expression (SAGE). Immunity 15:419.[CrossRef][Web of Science][Medline]
- Pennington, D. J., Silva-Santos, B., Shires, J. et al. 2003. The inter-relatedness and interdependence of mouse T cell receptor + and ß+ cells. Nat. Immunol. 4:991.[CrossRef][Web of Science][Medline]
- Meissner, N., Radke, J., Hedges, J. F. et al. 2003. Serial analysis of gene expression in circulating gamma delta T cell subsets defines distinct immunoregulatory phenotypes and unexpected gene expression profiles. J. Immunol. 170:356.
[Abstract/Free Full Text] - Hedges, J. F., Cockrell, D., Jackiw, L., Meissner, N. and Jutila, M. A. 2003. Differential mRNA expression in circulating gammadelta T lymphocyte subsets defines unique tissue-specific functions. J. Leukoc. Biol. 73:306.
[Abstract/Free Full Text] - Kress, E., Hedges, J. F. and Jutila, M. A. 2006. Distinct gene expression in human V1 and V2 T cells following non-TCR agonist stimulation. Mol. Immunol. Epublished ahead of print.
- Deng, M., Liu, J., Pelak, C. N., Lancto, C. A. and Abrahamsen, M. S. 2005. Regulation of apoptotic pathways in bovine / T cells. Vet. Immunol. Immunopathol. 105:15.[CrossRef][Web of Science][Medline]
- Wilson, E., Aydintug, M. K. and Jutila, M. A. 1999. A circulating bovine gamma delta T cell subset, which is found in large numbers in the spleen, accumulates inefficiently in an artificial site of inflammation: correlation with lack of expression of E-selectin ligands and L-selectin. J. Immunol. 162:4914.
[Abstract/Free Full Text] - Peters, D. G., Kassam, A. B., Yonas, H., O'Hare, E. H., Ferrell, R. E. and Brufsky, A. M. 1999. Comprehensive transcript analysis in small quantities of mRNA by SAGE-lite. Nucleic Acids Res. 27:e39.
[Abstract/Free Full Text] - Powell, J. 1998. Enhanced concatemer cloning: a modification to the SAGE (serial analysis of gene expression) technique. Nucleic Acids Res. 26:3445.
[Abstract/Free Full Text] - Powell, J. 2000. SAGE. The serial analysis of gene expression. Methods Mol. Biol. 99:297.[Medline]
- Colinge, J. and Feger, G. 2001. Detecting the impact of sequencing errors on SAGE data. Bioinformatics 17:840.
[Abstract/Free Full Text] - Saha, S., Sparks, A. B., Rago, C. et al. 2002. Using the transcriptome to annotate the genome. Nat. Biotechnol. 20:508.[CrossRef][Web of Science][Medline]
- Kato, M., Khan, S., Gonzalez, N. et al. 2003. Hodgkin's lymphoma cell lines express a fusion protein encoded by intergenically spliced mRNA for the multilectin receptor DEC-205 (CD205) and a novel C-type lectin receptor DCL-1. J. Biol. Chem. 278:34035.
[Abstract/Free Full Text] - Ishaq, M., DeGray, G. and Natarajan, V. 2003. Protein kinase C
modulates nuclear receptor-corepressor interaction during T cell activation. J. Biol. Chem. 278:39296.[Abstract/Free Full Text] - Ogawa, S., Lozach, J., Jepsen, K. et al. 2004. A nuclear receptor corepressor transcriptional checkpoint controlling activator protein 1-dependent gene networks required for macrophage activation. Proc. Natl Acad. Sci. USA 101:14461.
[Abstract/Free Full Text] - Wilson, E., Walcheck, B., Davis, W. C. and Jutila, M. A. 1998. Preferential tissue localization of bovine gamma delta T cell subsets defined by anti-T cell receptor for antigen antibodies. Immunol. Lett. 64:39.[CrossRef][Web of Science][Medline]
- Darmon, A. J. and Jat, P. S. 2000. BAP37 and prohibitin are specifically recognized by an SV40 T antigen antibody. Mol. Cell Biol. Res. Commun. 4:219.[CrossRef][Medline]
- Wang, S., Zhang, B. and Faller, D. V. 2002. Prohibitin requires Brg-1 and Brm for the repression of E2F and cell growth. EMBO J. 21:3019.[CrossRef][Web of Science][Medline]
- Polyak, K., Xia, Y., Zweier, J. L., Kinzler, K. W. and Vogelstein, B. 1997. A model for p53-induced apoptosis. Nature 389:300.[CrossRef][Medline]
- Reeke, G. N., Becker, J. W., Cunningham, B. A., Gunther, G. R., Wang, J. L. and Edelman, G. M. 1974. Relationships between the structure and activities of concanavalin A. Ann. N. Y. Acad. Sci. 234:369.[Web of Science][Medline]
- Metcalfe, J. C., Hesketh, T. R., Smith, G. A., Morris, J. D., Corps, A. N. and Moore, J. P. 1985. Early response pattern analysis of the mitogenic pathway in lymphocytes and fibroblasts. J. Cell. Sci. Suppl. 3:199.[Medline]
- Turner, C. A. J., Mack D. H. and Davis M. M. 1994. Blimp-1, a novel zinc finger-containing protein that can drive the maturation of B lymphocytes into immunoglobulin-secreting cells. Cell 77:297.[CrossRef][Web of Science][Medline]
- van Spriel, A. B., Puls, K. L., Sofi, M. et al. 2004. A regulatory role for CD37 in T cell proliferation. J. Immunol. 172:2953.
[Abstract/Free Full Text] - Hein, W. R. and Mackay, C. R. 1991. Prominence of T cells in the ruminant immune system. Immunol. Today 12:30.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  S. Nemali, D. W. Siemsen, L. K. Nelson, P. L. Bunger, C. L. Faulkner, P. Rainard, K. A. Gauss, M. A. Jutila, and M. T. Quinn Molecular analysis of the bovine anaphylatoxin C5a receptor J. Leukoc. Biol., August 1, 2008; 84(2): 537 - 549. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. K. Lahmers, J. F. Hedges, M. A. Jutila, M. Deng, M. S. Abrahamsen, and W. C. Brown Comparative gene expression by WC1+ {gamma}{delta} and CD4+ {alpha}{beta} T lymphocytes, which respond to Anaplasma marginale, demonstrates higher expression of chemokines and other myeloid cell-associated genes by WC1+ {gamma}{delta} T cells. J. Leukoc. Biol., October 1, 2006; 80(4): 939 - 952. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||