Antigen receptor V-segment usage in mucosal T cells

Abstract

In the accepted model of lymphocyte intestinal homing, naïve T cells recirculate via organized lymphoid tissues, whilst induced effector/memory cells home to the intestinal mucosa. In order to assess the T-cell-receptor repertoire in the intestine and gut-associated lymphoid tissue (GALT), spectratyping was performed on the proximal and the distal intestine, spleen and mesenteric lymph node tissue from six PVG rats. The products were analysed with an automated sequencer and statistical analyses were performed with hierarchical cluster analysis. This demonstrated the presence of a restricted T-cell repertoire in the small intestine compared with that in the mesenteric lymph nodes and the spleen. It also demonstrated marked differences in repertoire between individual, fully inbred rats maintained under apparently identical conditions in the same cage and fed identical diets. In addition, this work demonstrated marked differences between repertoires in the proximal and the distal intestine. Such marked differences are likely to reflect the end result of increasing divergence over time produced by relatively subtle effects of environment and antigenic load. Equally, marked differences in repertoire between small intestinal segments within individual rats indicate selective recruitment or retention of specific clones, presumably antigen-driven.

Introduction

Mature, naïve T cells continuously recirculate between the peripheral blood and the efferent lymph via secondary lymphoid tissue. Following activation with antigen, these cells develop an increased propensity to home to non-lymphoid tissues. More specifically, activated cells appear to migrate preferentially to tissues associated with the lymphoid microenvironment where the antigen was first encountered.¹ This concept of selective homing was first suggested 30 years ago as a result of cell-transfer experiments.² In support, studies involving cannulation of the afferent and efferent lymphatics of sheep popliteal lymph nodes demonstrated that naïve CD45 RC⁺ T cells migrate through lymph nodes and that the memory CD45 RC^– cells cross from blood into tissues.³

Thus, the current model is that naïve cells circulate through the spleen, mesenteric lymph nodes (MLN) and Peyer's patches (PP). Upon meeting luminal antigens, the induced effector/memory cells home to the intestine.⁴ This hypothesis is supported by the fact that the majority of intraepithelial and lamina propria lymphocytes are of the effector/memory phenotype (CD45 RO⁺, CD25^lo)⁵ and that they express specific adhesion molecules (α4β7, αEβ7, CCR9) whose ligands are selectively expressed on mucosal endothelium or epithelium.^6–8

This phenomenon of selective gut homing has the effect that effector/memory cells gain access to the extra lymphoid anatomical site where they are most likely to meet their cognate antigen. Traditionally, it was considered that the wide range of antigens in the intestine would require a wide repertoire of T-cell specificities to recognize them.⁹ However, T-cell-receptor spectratyping studies in human, pig and mice intestinal tissue have demonstrated oligoclonality amongst the various populations of intraepithelial lymphocytes (IELs) and lamina propria T cells.^10–14 Despite this, there has been no direct comparison of the mucosal and systemic repertoire in a range of individual animals.

The actual repertoire of T-cell specificities in the intestine of an individual at any time-point is likely to be determined by a combination of fixed and variable factors, including major histocompatibility complex (MHC) haplotype, past and current antigenic exposure, available ‘space’ in the mucosal immune system and rate of transit through the mucosal compartments.

Thus, an increasingly restricted repertoire with increasing age may be caused by progressive change in antigen diversity in the intestinal microenvironment.^15,16 Equally, different segments of the intestine will be exposed to different antigenic loads: thus, food antigens are more likely to predominate in the lumen of the proximal small intestine and bacterial antigens in the distal. Although the CCR9 ligand, CCL25 (TECK), is not expressed in the colon, suggesting an involvement of this chemokine/receptor in regional specialization of the mucosal immune system of the intestine,¹⁷ no segment-specific localization has been reported within the small intestine. We predict, first, that the T-cell repertoire in the small intestine of the rat will be restricted compared with that in the MLN and the spleen; and, second, that subtle differences in previous and current antigenic exposure will result in interanimal differences in repertoire, even in MHC-inbred animals. Finally, if repertoire primarily reflects recruitment, proximal and distal intestinal repertoires should be similar: if local antigen exposure affects retention, repertoires should be different.

Materials and methods

Tissue retrieval

All animal work was performed under valid Home Office personal and project licences. Tissue was retrieved from six male PVG rats, aged between 19 and 24 weeks of age (supplied in a single batch from Harlan UK, Bicester, UK), stabilized on a chow diet fed ad libitum for at least 2 weeks (Banton and Kingman, Hull, UK). Prior to surgery, the animals were starved of solid food for up to 12 hr. They underwent gaseous induction and maintenance of anaesthesia with halothane. Analgesia and fluids were provided subcutaneously as necessary.

From each animal, tissue was retrieved from standard sites in the jejunum and terminal ileum, free of PP, and from the spleen and MLN. Each sample was about 2 mm² in size. Intestinal samples were full thickness and therefore included the serosa, muscularis, mucosa and epithelium. The samples were immediately snap frozen in liquid nitrogen and then stored at −80°.

RNA isolation

RNA was isolated from all tissue samples using the SV Total RNA isolation system (Promega, Southampton, UK), according to the manufacturer's recommendations.

Both on-column and in-solution DNase steps were performed using RNase-free DNase (Promega). From each sample, 100 µl of RNA was isolated, which was stored at −80°.

Reverse transcription

Reverse transcription reactions were performed using the ImProm-II™ Reverse Transcription System (Promega), according to the manufacturer's instructions, using 9 µl of isolated RNA from each sample. The 20 µl of cDNA produced was diluted to 50 µl with nuclease-free water and then stored at −20°.

Polymerase chain reaction

All polymerase chain reaction (PCR) amplifications were performed using the HotStarTaq® Master Mix (Qiagen, Crawley, UK). Each reaction used 1 µl of template and 24 µl of master mix (composed of 12·5 µl of 2× HotStarTaq® Master Mix, 200 nm of each primer and water to 24 µl).

Whilst establishing the technique, some reactions were run with forward and reverse C-region primers (C and C con, respectively) to amplify a 310-base pair fragment of the constant region from genomic DNA only and to confirm that T-cell receptor (TCR) DNA had been extracted.

All subsequent PCR procedures involved the amplification of cDNA using a labelled C-region primer and one of 22 rat Vβ-region primers. Primer sequences were taken from Zhai et al.,¹⁸ except Vβ7 and Vβ6, which were taken directly and with modification, respectively, from Gold et al.¹⁹ C-region primers were designed using primer 3 software (http://fokker.wi.mit.edu/cgi-bin/primer3/primer3_www.cgifrom) from the rat C-region nucleotide sequence (accession number X14319). Primers were supplied by either Sigma Genosys (Haverhill, UK) or MWG Biotech (Milton Keynes, UK). Amplification was carried out in a PTC-200 Peltier Thermal Cycler (MJ Research, Waltham, MA). PCR conditions included an initial activation at 95° for 15 min followed by 40 cycles of denaturation at 95° for 30 seconds, annealing at 55° for 30 seconds and extension at 72° for 30 seconds. On completion of the PCR procedure, PCR products were stored at −20°.

Where necessary, PCR products were run on 2% agarose gels. Gels were imaged using an ultraviolet (UV) transilluminator and a UV CCD camera (BIODOC-IT Imaging System; UVP, Upland, CA) and the images were captured and viewed using UVI photo MW software (version 99·03s for Windows; Cambridge, UK).

Precipitation and analysis

PCR products were ethanol precipitated into 96-well plates, with two separate products per well. Each well contained one Vβ that had been amplified with a fam (6-carboxyfluorescein)-labelled C-region primer and another with a hex (hexachloro-fluorescein)-labelled C-region primer. After centrifugation, the supernatant was removed and the pellet air-dried at room temperature for about 30 min before being resuspended in 20 µl of water. Five microlitres of this resuspended sample was removed from each well and transferred to the corresponding well of another 96-well plate, containing MegaBACE™ET-900-R Size standard (Amersham Biosciences). Plates were stored at 4° until analysis using a MegaBACE™ automated sequencer (Amersham Biosciences). Data were compiled in MegaBACE™genetic profiler, version 1·5 (Amersham Biosciences). Tissue samples in which the PCR product was not detected were processed again. If this still did not produce a PCR product, it was assumed that that particular Vβ family was not represented in that particular sample of tissue.

For each Vβ studied, the range of base pair lengths of products was identified. Data regarding peak height were extracted from the genetic profiler software and manually transferred into Microsoft®Excel. The data were normalized for each spectratype and statistical analysis was performed with hierarchical cluster analysis, by Wards method of Euclidean squared distances, using spss, version 12, for windows (SPSS, Chicago, IL). This method has previously been used to analyse spectratyping data.²⁰

Results

Validation of the technique

PCR reactions using the forward and reverse C-region primers, C and C con, respectively, amplified a 310-base pair-long section of the TCR β-chain constant region, demonstrated in lane 2 of Fig. 1. Products from template that had undergone reverse transcription clearly showed a band that was not present in the corresponding lane in which the isolated RNA had not been reverse transcribed, indicating amplification of mRNA-derived cDNA and not genomic DNA.

Figure 1.

Agarose-gel electophoresis of polymerase chain reaction (PCR) products from proximal intestine amplified using primers in the constant region (lanes 2 and 3) and using an example Vβ primer (Vβ5, lanes 4 and 5). Templates used in lanes 3 and 5 were not subjected to reverse transcription, as a negative control. C, C-region primer; C con, reverse C-region primer.

Repertoire analysis in the normal rat

In order to assess the repertoire in the normal rat intestine, tissue samples were taken from six donor animals, from the previously specified four sites (proximal small intestine, distal small intestine, MLN and spleen). A total of 24 samples were analysed for each Vβ family. Spectratyping of these samples demonstrated that all Vβ families were represented in at least one tissue sample for every animal. Vβ9 only produced a product in one tissue sample, the MLN of rat 2. To confirm that this was a biological phenomenon and not the result of a PCR error or primer problem, the process of reverse transcription, PCR with the Vβ9 primer, precipitation and analysis were repeated for all 24 samples. This confirmed that the Vβ9 family was only present in the same single tissue sample of the 24 analysed, as an identical spectratype was produced when repeated.

In order to assess the extent of restriction of repertoire in the intestine compared with that in the spleen and MLN, the complete spectratypes of the 24 tissue samples were analysed and compared. This was initially carried out by visual inspection (Fig. 2b). These figures were created by extracting the peak height data from each of the Vβ families for each animal. These data were then normalized, so that for each Vβ family and tissue sample, the proportion that each peak contributed to the total spectratype was calculated. When this data is plotted, as each peak is a proportion of the total, the height of peaks also relates to spread. Thus, a peak height of 1 represents a highly restricted repertoire with a single CDR3 length (Fig. 2a).

Figure 2.

(a) Normalizing of spectratype peak data. Peak height data were extracted from ‘raw’ spectratypes, as shown on the left. This data was then normalized, so that for each Vβ family, for each tissue sample, the proportion that each peak contributed to the total spectratype was calculated. When this data was plotted, as on the graphs on the right, as each peak is a proportion of the total, the height of peaks also relates to spread. Thus, when a peak height is 1, this represents a highly restricted repertoire, with a single CDR3 length. (b) Plots of normalized data, from all six animals, with the proportion of the spectratype contributed to on the y-axis and each of the 22 Vβ families on the x-axis. Red, proximal intestine; green, distal intestine; blue, mesenteric lymph node; grey, spleen.

Analysis of these plots strongly suggested that the repertoire of CDR3 length was highly restricted in the intestine, compared with the spleen and MLN, often being oligoclonal or monoclonal in nature. In fact, closer inspection suggested that the repertoire was most restricted in the proximal intestine, compared with the distal. In contrast, the repertoire in MLN samples appeared largely unselected: the spectratypes were mainly polyclonal, with a distribution that appeared to be almost normal in many samples. In general, the repertoire in splenic samples was also unselected, although there was more variability in this, with two of six rats having a few oligoclonal or monoclonal spectratypes in their splenic repertoire.

Hierarchical cluster analysis was performed on the entire data set (Fig. 3). Cluster analysis demonstrated that, based upon the Vβ repertoire, most of the organized lymphoid tissue (MLN and spleen) clustered together. Apart from rats 1 and 2, clustering was not obviously by rat. In contrast, TCR Vβ repertoire analysis of proximal intestinal tissues demonstrated not only differences from the repertoires in the organized tissues but, more importantly, marked differences between the animals, indicating markedly different repertoires between individuals. Analysis of distal intestinal samples further demonstrated a marked degree of heterogeneity, some samples clustering strongly with the unselected repertoires of the organized lymphoid tissues, whereas others showed highly selected repertoires as in the proximal intestine. However, proximal and distal intestinal repertoires did not cluster by animal, indicating segment-specific differences in repertoire generation or maintenance.

Figure 3.

Hierarchical cluster analysis of normalized peak height data for animals numbered 1–6. Based upon the Vβ repertoire, most of the organized lymphoid tissue (mesenteric lymph nodes and spleen) cluster together. In contrast, proximal intestinal tissues do not cluster with the organized tissues, with marked differences observed between the animals, indicating markedly different repertoires between individuals. Some distal intestinal samples clustered strongly with the unselected repertoires of the organized lymphoid tissues, whereas others showed highly selected repertoires, as in the proximal intestine. s.i., small intestine.

Discussion

This work described here clearly supports the hypothesis that the T-cell repertoire in the small intestine is restricted compared with that in the MLN and the spleen, based on both visual inspection of the spectratypes and hierarchical clustering of the data. Although previous work has shown a restricted repertoire in the intestines of humans^21,22 and mice,²³ and a polyclonal repertoire in rat splenocytes,²⁴ paired samples from the same individual animals have not been reported. This work therefore also provides strong evidence to support the accepted model of lymphocyte homing to the intestine, in which naïve, unstimulated T cells, with a more polyclonal repertoire, recirculate via the spleen and MLN, whilst induced effector/memory cells, with a restricted oligoclonal or monoclonal repertoire, home to the intestine.

However, the data clearly demonstrated two further important points. First, there were marked differences in repertoire between individual, fully inbred rats maintained under apparently identical conditions (i.e. in the same cage and fed identical diets). Second, there were marked differences between repertoires in the proximal and the distal intestine. These findings strongly support the hypothesis that luminal antigens progressively shape the Vβ and CDR3 repertoire of gut-associated lymphoid tissue (GALT) T cells. This may be a result of selective activation followed either by selective retention or recruitment of specific T cells. Thus, the interindividual differences in intestinal repertoire could reflect differences in current antigenic exposure between rats or reflect the end result of sequential accumulation of small differences in the level of antigen exposure over time. Such differences are likely to be small at any one time-point: studies of intestinal flora suggest some effect of genetics, but a strong effect of environment, such that animals from the same source had very similar flora.^25,26

Intra-individual differences in TCR repertoire, between the proximal and the distal intestine, may be caused by the presence of regional homing mechanisms. Such mechanisms may exist: for example, the CCR9 ligand, CCL25 (TECK), is not expressed in the colon.²⁷ However, this hypothesis would require differential expression of selective adhesion molecules or chemokines between the proximal and the distal intestine, and such molecules have not been identified.

In contrast, evidence does exist to support a role for antigen in driving segment-specific localization. Work with Thiry-Vella loops in sheep demonstrated that antigen-specific immunoglobulin A (IgA) only appeared in antigen-challenged loops, never in control loops.²⁸ Subsequent work in rats, using single Thiry-Vella intestinal loops, showed that luminal antigenic challenge with cholera toxin led to migration of anti-toxin antibody-containing cells into the lamina propria.²⁹ Although loops that had not received such a challenge also had some of these cells migrating through the lamina propria, the numbers of cells accumulating, and their persistence in the intestine, was much greater in the loops where antigen was present. Despite the underlying inadequacies of using the Thiry-Vella loop model, this work suggests that luminal antigen may drive the homing of antigen-specific cells in the intestine.

More recently, work has demonstrated the importance of commensal bacteria in the development of the gut immune system. Studies in rats have shown that the composition of IEL in the small intestine may be altered by microbial colonization.³⁰ Conventionalization of germ-free rats led to an increase in CD4⁺ and CD8⁺ αβ T cells, and a decrease in the double-negative population, as demonstrated by using flow cytometry. In addition, three-colour flow-cytometric analysis demonstrated a change in the cell-surface expression of some Vβ chains within the IEL populations, suggesting that luminal microbial antigens can shape the Vβ repertoire. However, this result could have been caused by the binding of pattern recognition molecules, such as Toll-like receptors, rather than a response driven by the presentation of luminal antigen.

In addition to the differences in luminal antigenic load, there are differences in enterocyte MHC II expression between the the proximal and the distal intestine.³¹ Immunohistological studies in rats have demonstrated that in the duodenum, MHC class II-positive enterocytes are only found in the upper parts of the villi, the villi bases and the crypts expressing no MHC class II. Along the length of the small intestine there is a progressive increase in this expression, such that all terminal ileum enterocytes express MHC class II. If enterocyte presentation occurs in vivo, this regional difference in MHC class II expression may also contribute to the differences in repertoire between the proximal and the distal intestine. As enterocytes do not express conventional costimulatory molecules and are capable of secreting a range of immunoregulatory cytokines,^32,33 it is possible that the difference in the ratio of enterocyte to formal antigen-presenting cell presentation in the proximal and the distal intestine may lead to differential T-cell retention.

In summary, we have demonstrated marked differences in the intestinal T-cell repertoire between inbred rats maintained in the same environments and on the same diets. Such marked differences are likely to reflect the end result of increasing divergence over time produced by relatively subtle effects of environment and antigenic load. Equally, marked differences in repertoire between small intestinal segments within individual rats indicate selective recruitment or retention of specific clones, presumably antigen-driven. The biological significance of these observations is that the responses of individual animals or humans to mucosal antigen may be difficult to predict: small differences between individuals in antigen dose, in the site of antigen targeting (proximal or distal intestine), or in previous antigenic exposure, are likely to result in increasing divergence between individuals in the epitopes recognized and potentially therefore in subsequent responses to other antigens.