Analysis of the Paired TCR α- and β-chains of Single Human T Cells

Analysis of the paired i.e. matching TCR α- and β-chain rearrangements of single human T cells is required for a precise investigation of clonal diversity, tissue distribution and specificity of protective and pathologic T-cell mediated immune responses. Here we describe a multiplex RT-PCR based technology, which for the first time allows for an unbiased analysis of the complete sequences of both α- and β-chains of TCR from single T cells. We validated our technology by the analysis of the pathologic T-cell infiltrates from tissue lesions of two T-cell mediated autoimmune diseases, psoriasis vulgaris (PV) and multiple sclerosis (MS). In both disorders we could detect various T cell clones as defined by multiple T cells with identical α- and β-chain rearrangements distributed across the tissue lesions. In PV, single cell TCR analysis of lesional T cells identified clonal CD8+ T cell expansions that predominated in the epidermis of psoriatic plaques. An MS brain lesion contained two dominant CD8+ T-cell clones that extended over the white and grey matter and meninges. In both diseases several clonally expanded T cells carried dual TCRs composed of one Vβ and two different Vα-chain rearrangements. These results show that our technology is an efficient instrument to analyse αβ-T cell responses with single cell resolution in man. It should facilitate essential new insights into the mechanisms of protective and pathologic immunity in many human T-cell mediated conditions and allow for resurrecting functional TCRs from any αβ-T cell of choice that can be used for investigating their specificity.


Introduction
T cells are essential for protective immunity against infections and malignant tumors. Furthermore, they may confer pathogenic immune responses in autoimmune disorders such as multiple sclerosis, psoriasis vulgaris, rheumatoid arthritis, type I diabetes, inflammatory bowel disease and others. Advances in understanding the mechanisms of protective and pathogenic Tcell responses in man have been hampered by the fact that often neither the relevant T cell populations nor their antigen specificity could be defined. T-cell receptor (TCR) analysis has been used to address these issues but provided only incomplete insights. Most techniques to analyze the human TCR repertoire were limited to TCR b-chains, i.e., they revealed only ''half the truth'', while an unbiased characterization of complete TCR molecules could not yet be achieved [1]. Monoclonal TCR antibodies are only available for a limited number of TRBV regions. RT-PCR-amplification, spectratyping, and random cloning of TCR b-chain transcripts provided only estimates of the heterogeneity of T-cell populations. A technique to analyse a-chains from single T cells obtained by laser-microdissection was limited to clonally expanded T cell populations that could be identified by anti-TCR Vb-chain antibodies [2]. An unbiased characterization of the paired, i.e. complementary TCR aand TCR b-chain from single human T cells, which would be required to define the specificity of T-cell responses [3], however, failed hitherto due to the huge diversity of genetic elements from which functional TCR ab transcripts are generated [4]. As a consequence, the pathomechanism of many human T-cell mediated conditions have remained elusive.
To overcome this obstacle in the characterization of T-cell mediated immunity we developed a multiplex RT-PCR-based protocol to determine paired ab-TCR rearrangements from any single T cell of choice without prior knowledge of their particular TRAV or TRBV usage. We assessed its capacity to characterize pathologic T-cell infiltrates in two T cell-mediated autoimmune disorders, psoriasis vulgaris (PV) and multiple sclerosis (MS).
PV is a chronic, recurrent inflammatory skin disease resulting from a T-cell mediated immune response in the skin [5]. Former analyses of TCR b-chain repertoires of lesional psoriatic T-cell infiltrates by random amplification, cloning and sequencing of TCR b-chain cDNA from lesional biopsies had suggested select oligoclonal T-cell expansions [6,7,8,9], but the precise clonality, subtype and tissue distribution of the pathogenic T cells has remained unknown. MS is believed to be an autoimmune disease where clonally expanded lymphocytes can be detected in different regions of the brain [10,11] and CD8 + T cells are thought to mediate an inflammatory attack on the central nervous system [12,13]. Expanded CD8 + T cell clones were shown to dominate the lesion-infiltrating immune cell population [10]. However, in these studies only TCR b-chain rearrangements were analysed, while the matching TCR achains could not be determined.
In both disorders we could identify T-cell clones, characterize their matching TCR aand b-chain rearrangements and show their tissue distribution on the single cell level. Accordingly, our technology should have wide applications and promote research into the nature and antigen-specificity of many medically important protective or pathologic T-cell responses, allowing for the development of novel diagnostic, therapeutic or preventive strategies.

Strategy and Primer Design
We established an RT-PCR-based method for the molecular analysis of the matching i.e. paired ab-TCR rearrangements of single human T cells. For this purpose we developed nine degenerate PCR primers (termed Vp1 to Vp9, Table 1), which cover the entire repertoire of functional TCR b-gene V-elements and, together with 24 TCR Va-out primers for the Va repertoire developed recently [2], can simultaneously amplify the cDNA of all functional human TCR Va-and Vb-region gene transcripts by multiplex RT-PCR.
The Vb-primers were designed as degenerate primers utilizing sequence homologies of the various Vb-gene families. To maintain a high specificity we allowed only one mismatch with the primary nucleotide Vb-gene sequence. Except Vp1, which is located in the leader segment, all primers are positioned within the Vb genes. Primer interactions with the Va-primer set were excluded first by in silico analysis using several software programs and then experimentally by testing numerous primer combinations by RT-PCRs on mRNA from peripheral blood T cells.
The strategy for the molecular characterization of ab-TCR rearrangements of single T cells is outlined in Figure 1. It starts with a one step multiplex pre-amplification RT-PCR, where reverse transcription is done by primers specific for the aand bconstant region genes (step 1A). Then the TCR aand b-chains are amplified with our novel Vb-gene specific primer set (Table 1) and the ''outer primer'' set for the Va-region genes [2] (step 1b). The pre-amplification PCR products then served as templates to amplify the b-chains by anchor PCR (step 2) and the a-chains (step 3) by nested PCR.
For anchor PCR we introduced a unique 22-mer oligonucleotide as a universal anchor sequence (termed ''UP'') to the 59 end of the nine different Vb-gene forward primers ( Table 1). The preamplification RT-PCR product was subjected to a run-off PCR using these nine anchored primers (termed Vp1-UP to Vp9-UP, Figure 1, step 2A). Next we amplified the respective single cell TCR b-chain rearrangement independent from the TCR Vb-gene family using the ''UP'' anchor primer together with a nested Cb-in primer in a third PCR (step 2B). When a PCR product for the TCR b-chain was obtained, we amplified the corresponding TCR a-chain rearrangement from the pre-amplification multiplex RT-PCR product in five different nested PCRs (step 3A) using five Vain primer pools for the Va repertoire described recently [2]. The amplified TCR aand b-chain rearrangements were characterized by direct sequencing (steps 2C and 3B).

Functional Verification of the Primer Sets
We validated the specificity and compatibility of the 9 Vb-gene primers and the Cb-out primer in various PCRs together with the Va primers. Each Vp-primer yielded a PCR product of the expected size. In Figure 2 we show an experiment where we reverse transcribed mRNA of peripheral blood T cells with the Ca-out and Cb-out primers and amplified the transcripts with the 24 Va-and 9 Vb-primers in the pre-amplification multiplex RT-PCR. We then selectively amplified the rearrangements of the different Va-and Vb-gene families in a second (Vb) and third (Va) round of nested PCRs using the nine anchor-forward Vb-primers with the Cb-in reverse primer, or the 24 Va-in forward primers with the reverse Ca-in primer [2] in individual reactions. Each Vb-( Figure 2A, lanes 1-9) and Va-primer ( Figure 2B, lane 1-24) yielded a PCR product of the expected size.
In a second, parallel approach we amplified the pre-amplification multiplex RT-PCR product with 23 Vb-primers specific for the different functional Vb-region gene families together with the FAM-labelled Cb-in reverse primer [9]. The fragment-lengths distribution of the hypervariable N(D)N-regions of the PCR products was then determined by PAGE [14]. All amplified TCR Vb-gene rearrangements revealed spectratypes with Gaussian-like distributions typical of polyclonal T-cell populations in peripheral blood ( Figure 2C). Together, these results document that the combined Va-and Vb-primer sets were capable of simultaneously amplifying the full functional repertoire of both TCR aand bchains.

Analysis of TCR aand b-chains of Single T Cells from Peripheral Blood
We tested the efficiency of this approach for single cell TCR analysis using CD4 + and CD8 + blood T cells. To maintain an optimal viability the T cells had been labelled with magnetic beads and aspirated manually with a pipette in an inverted microscope. In 82 of 96 CD4 + (85.4%) and in 76 of 96 (79.2%) CD8 + T cells we could amplify and sequence a functional TCR b-chain rearrangement. Among these 158 b-chain rearrangements all functional TCR Vb-gene families were represented except Vb25, which is rarely rearranged in general. We did not find a single chain twice, which provides evidence that PCR contaminations can be excluded. We then characterized the TCR a-chain rearrangements exemplarily in 20 CD4 + and 20 CD8 + T cells where we had identified functional b-chains. We obtained the corresponding TCR a-chains from each individual T cell investigated. Thus, our technique allows the molecular characterization of the paired ab-TCR rearrangements from single viable T cells with high yields.
When we tested single anti-CD3 labelled T cells from peripheral blood isolated by flow cytometry, the yield for b-chains in 139 analyzed T cells was 27.3% (38/139), and for matching a-chains was 39.5% (15/38). The decreased yields are presumably due to the expected loss and damage of cells during sorting. However, these data show that our method is also suited for high throughput screening experiments.

TCR Analysis of Single T-cells from the Pathologic Infiltrate in Psoriasis
Next we examined viable, putatively pathogenic T cells from explant cell culture from lesional biopsies of five patients with chronic plaque PV. We divided each biopsy specimen in two parts. From part ''A'' we isolated single CD4 + or CD8 + T cells for TCR analysis. To ensure that T cells with identical b-chains belonged to a clonally expanded population, we applied two criteria: Firstly, they had to carry the same Va rearrangement; Secondly, we tested for select TCR clones whether we could identify the respective bchain transcript by PCR and direct DNA sequencing in the second part ''B'' of the biopsy.
In patient #1 we analysed 50 CD4 + and in patient #2 50 CD8 + T cells. They had been isolated from ''full thickness'' biopsies consisting of dermis and epidermis. In 40/50 CD4 + and 37/50 CD8 + T cells functional ab-TCRs were identified. Three of the 40 CD4 + and two of the 37 CD8 + T cells had identical TCR Vb-and Va-rearrangements (Table 2). Further, we could amplify their TCR b-chain transcripts using clone-specific primers from part ''B'' of the biopsy. According to the above criteria, these T cells represented clonally expanded populations in the tissue lesion. The other T cells had unique b-chains.
In patient #3 we analysed both lesional CD4 + and CD8 + T cells. This patient had been tonsillectomized due to constant PV flares in association with a recurrent streptococcal angina, which is the main infectious psoriasis trigger [9,15]. cDNA from blood lymphocytes and fractions of the tonsils were available for analysis as well. 100 CD4 + and 50 CD8 + T cells were analyzed. In 82 CD4 + and 41 CD8 + T cells functional ab TCRs were identified. Two out of the 82 CD4 + T cells had identical ab TCRs. Among the 41 CD8 + T cells, we found twice two identical ab TCRs and once three identical ab TCRs ( Table 2). The respective b-chain rearrangements were also identified within the corresponding part ''B'' of the biopsy, in peripheral blood and/or the tonsillar tissue of the patient (Table 2). Thus, our technology may identify clonally expanded ab-T cells in complex T cell populations and track their distribution across different organs.

Differential Distribution of Psoriatic CD8 + T Clones in Epidermis and Dermis
While the majority of the psoriatic T-cell infiltrate is located within the dermis, the development of PV is crucially dependent on the accumulation of CD8 + T cells within the epidermal keratinocyte layer [16]. In patients #4 and #5 we therefore separated dermis and epidermis of the lesional biopsies and analysed the differential distribution of CD8 + T cells in these skin compartments. In patient #4 (Table 3) 52 of 60 analyzed epidermal and 36 of 45 analyzed dermal T cells yielded functional ab TCRs. We found nine different ab-TCR rearrangements in duplicates or triplicates among the 52 epidermal CD8 + T cells. Four of these ab-TCR rearrangements (clones #1, 2, 6, and 9) were also seen in the dermal T cells. The 36 dermal T cells contained only one TCR clone (#10), which was not seen in the epidermis. In patient #5 (Table 4), 33 of 40 analyzed epidermal and 25 of 30 analyzed dermal CD8 + T cells yielded functional ab TCR rearrangements. Three different ab-TCR rearrangements (clones #1-3) were selectively present in three, four or seven of the 33 epidermal CD8 + T cells. Two TCRs (clones #4, 5) were seen in each one epidermal and dermal T cell. Two other T-cell clones (#6 and #7) were only seen in the dermis. For select T-cell clones the Vb signature was also confirmed in part ''B'' of the biopsy.   Accordingly, our technology of single cell TCR-analysis may define the distribution of particular T-cell clones within heterogeneous T-cell infiltrates. For PV our results indicate that CD8 + T-cell clones may predominate within the epidermis. A total of 425 T cells isolated by manual aspiration from the PV biopsy specimens was analyzed by single cell TCR analysis. The overall yield was 346 functional ab TCR rearrangements (81.4%).
Direct microscopic analysis of the beads-labelled T cells allowed an interesting observation. In patient #4 each two CD8 + T cells from two clones (#2 and #3, Table 3) were seen in tight contact with target cells of either dendritic (DC) or epithelial-like morphology ( Figure 3A, B). The contacts between the T cells and antigen presenting cell were so tight that they even survived the isolation procedure.

TCR Analysis in Cryosections from Brain Lesions of Multiple Sclerosis
In many conditions only frozen biopsy specimens are available. Furthermore, it may be of interest to precisely define the localisation of particular pathogenic T cells in tissue lesions by immunohistochemistry. Therefore, we tested whether our technology would be suited to analyze T cells from frozen biopsy samples using brain tissue sections from a patient with MS. We stained sections from three different tissue blocks with antibodies specific for the CD8 beta-chain and the T cell activation marker CD134 to ensure that we selectively isolated activated effector T cells, which are more likely to be the mediators of inflammatory demyelination ( Figure 4).
We analysed the TCRs from 643 activated, i.e. double-stained CD8 beta + /CD134 + single effector T cells isolated by laser microdissection (Figure 4) from three different tissue blocks of one big lesion. In 68 of these formerly frozen and then immunostained T cells functional TCR b-chain rearrangements could be identified (yield: 10.6%) ( Table 5, left panel). Of these, four TCR b-chain rearrangements were detected more than once. We found a Vb8.1-chain 46 times (clone #8), a Vb6.5-chain 8 times (clone #6) and a Vb13.1-chain 3 times (clone #10) ( Table 6). In 13 of the T cells we could also identify the paired a-chain (Table 5, right panel, Table 6). Of note, cells of the same T-cell clones were found not only in different tissue blocks, but were equally distributed in several morphologically distinct regions of Step 1: one step multiplex pre-amplification RT-PCR composed of step 1A, reverse transcription using a Ca-out and Cb-out primer and step 1B, amplification of the a and b TCR rearrangements by a pool of 9 oligonucleotide Vb primers (Vp1-Vp9) for the Vb repertoire and a pool of 24 oligonucleotide Va primers for the Va repertoire. Subsequently, the PCR products of the Vb (step 2) and Va (step 3) gene rearrangements are handled separately.
Step 2: Introduction of a universal primer (UP) site at the 59site of each Vb-rearrangement using primers Vp1-UP to Vp9-UP by a run off reaction (step 2A), followed by amplification of the Vb gene rearrangement by seminested PCR with the UP primer and a Cb-in primer (step 2B) and sequencing (step 2C).
Step 3: Amplification (step 3A) and sequencing (step 3B) of the Va gene rearrangement by nested PCR from the pre-amplification PCR product using five different nested Va-in primer pools and a Ca-in primer. doi:10.1371/journal.pone.0037338.g001 the lesion (Table 6). This widespread clonal expansion points to the processing and presentation of identical or similar antigenic epitopes by antigen presenting cells in different parts of the brain. Accordingly, our technology may not only define clonal ab TCRs in situ but also follow their precise localisation across pathologic T-cell infiltrates.
Two clones, #10 and #11, had been detected as expanded clones in earlier studies by other techniques [17,18]. According to homology searches using the protein query tool of Blast (http:// blast.ncbi.nlm.nih.gov/Blast.cgi?PAGE = Proteins) all other PV and MS b chains represented unique rearrangements not identified before.

Identification of T Cells with Dual TCRs
Several expanded clones expressed one b-chain in combination with two different a-chains. In PV, the T cells of two CD8 + T-cell clones of patient #3 (Vb6S5 clone, Vb6S3 clone), two CD8 + clones of patient #4 (clones #5, #10) and of the CD8 + clone #1 of patient #5 showed two different functional Va rearrangements each (Tables 2, 3, 4).
The two strongly expanded MS TCR chains (Vb8.1-and Vb6.5-chains, MS clones #6 and #8) were each found in combination with two different a-chain rearrangements as well (Table 5), however, in different T cells. The Vb6.5-chain of MS clone 6 was seen either in combination with a Va20S1/Ja30 rearrangement (subclone 6a) or a Va20S1/Ja37 rearrangement (subclone 6b). The Vb8.1-chain of clone 8 was identified either with a Va1S5/Ja34 (8a) or Va30S1/Ja18 (8b) rearrangement. Dual TCRs may have two different specificities. This is of particular interest, because T cells with dual ab-TCRs have been implicated in T-cell mediated autoimmunity [19,20].

Discussion
Here we describe a PCR-based technology which to our knowledge for the first time allows for an unbiased molecular analysis of matching i.e. paired TCR aand b-chains from single human T cells. It employs a set of degenerate Vb-gene primers and the introduction of an anchor primer site at the 59 end of the PCR products of Vb-gene rearrangements for further transcript amplification. Together with previously designed primer sets for the complete Va repertoire [2] this allows us to characterize heterodimeric TCRs from single ab T cells. The sensitivity of the method correlated with the method of isolation and accordingly mRNA preservation of the single T cells. It reached 80% for freshly isolated viable T cells aspirated by manual pipetting, but even pre-stained and laser-dissected T cells from frozen brain sections yielded an output sufficient to characterize the T cell infiltrates.
The technology has several obvious advantages. It is independent from the availability of monoclonal anti-Vb region antibodies. Neither a prior knowledge of the TCR aand b-chain regions nor preceding TCR spectratyping or other methods for the detection of expanded clones are required. This is documented by the analysis of the MS lesion: Two of the TCR rearrangements described here, #10 and #11 (Table 5), had already been identified formerly by single cell PCR using rearranged genomic DNA as PCR-template [17] or CDR3 spectratyping of the TCR b-chain repertoire [18]. In contrast to these earlier studies, however, we now could also characterize the corresponding achains, and we identified other clonal T cells from the same tissue blocks, which formerly had escaped detection.
T cell clones prevailed in skin lesions from psoriasis, as well. Similar to MS [10,18,21,22] this had formerly been proposed from To validate that all TCR Vb-gene families were covered, the preamplification multiplex RT-PCR product was amplified using 23 Vb-primers specific for the functional Vb-gene repertoire together with the FAMlabelled Cb-in reverse primer in individual reactions as described [14]. The PCR-products were analyzed by spectratyping their fragment-lengths on a genetic sequencer. doi:10.1371/journal.pone.0037338.g002 Table 2. ab-TCR rearrangements and frequency of clonal CD4 + or CD8 + T cells isolated from lesional full thickness skin biopsies of PV patients #1 to #3.  Table 3. ab-TCR rearrangements, frequency and tissue localization of clonal CD8 + T cells isolated from lesional dermis and epidermis of PV patient #4.

CDR3a
Clone # TCR b-chain repertoire studies [6,7,8,9,14]. We could now document that indeed numerous T cells of the psoriatic T-cell infiltrate carry identical ab TCRs, and in accordance with the HLA-Cw6 association of PV [23] this clonality was particular evident for CD8 + T cells. Identification of ab-heterodimers by our technology provided definite evidence that completely identical clones may be present in morphologically distinct regions of human autoimmune tissue lesions. In MS considerable numbers of identical T cells were observed (Table 6). Such pervasive T cell clones [11] are obviously a quite general phenomenon in MS [10]. In psoriasis, the CD8 + Tcell clones predominated in the epidermis, indicating that the epidermis may be the actual site of psoriatic T-cell activation. And indeed, blocking T cell transmigration from the dermis into the epidermis by a1b1 integrin antibodies in a human-mouse skin transplant model may suppress psoriatic inflammation [16]. Identification of the clonal Vb signature in blood and tonsils of a patient with streptococcal driven psoriasis is in line with our former observation that the pathogenic T cell clones are not primary skin-resident T cells but may be recruited from lymphoid organs such as the tonsils via the blood stream [9].

Epidermal T cells identical/analyzed Dermal T cells identical/analyzed
Thus, our technology facilitates the characterization of complex T-cell populations and helps to distinguish pathologically relevant T-cell clones from abundant bystander T cells of inflammatory infiltrates. It may therefore be used to characterize the distribution and expansion of particular T-cell clones across tissues or body fluids. Such T-cell clones may even be identified from low numbers of cells, where statistical techniques like CDR3 spectratyping [24] are not applicable. In PV less than 40 cells were sufficient to detect clonal T-cell expansions, as verified by analysis of an adjacent part of the biopsy.
ab-TCR analysis may help to identify the respective antigenpresenting events. In PV different T cells with identical TCRs were seen in direct tight contact with cells of dendritic or epithelial morphology. So far direct evidence of an antigen-presenting event in situ, i.e. in the autoimmune tissue lesion, has been rare. The ''in flagranti'' observation of a likely antigen-presenting event in a psoriasis explant (Figure 4) now provides evidence that distinct antigen-presenting cells in psoriatic lesions may present a dominant antigen to different T cells of the same clonal origin.
The high sensitivity of our technology is reflected by its ability to detect dual TCRs on the single cell level. We identified several CD8 + T-cell clones from skin lesions of the PV patients #3, 4 and 5, which had two different a-chains associated with the same Vbchain rearrangement, respectively (Tables 2, 3, 4). Likewise, clones #6 and #8 from the MS lesion had two different a-chains associated with a Vb6S5-or Vb8S1-chain (Table 5). In the MS patient two possibilities are conceivable: Either the antigen-driven T-cell expansion selected two different T-cell clones each with the same TCR b-chain but different a-chains, or the T-cell clones expressed two different Va-chains in a single dual TCR T cell, but only one a rearrangement was identified due to a low mRNA preservation. Dual-TCR T cells have been described, and it is assumed that up to 30% of all T cells may carry dual TCRs [19]. In case of the MS clones, we may not distinguish these possibilities, because the two a-chains were cloned from different cells with the same b-chain. However, the psoriasis clones were classical dual TCR T cells, as the two different a-chains were definitely cloned from the same T cells. This is the first time that clonal T cells with dual ab-TCRs were identified directly from human autoimmune tissue lesions. These T cells likely have two different antigenspecificities, which may be relevant for the induction of autoimmunity. In a mouse model of autoimmune encephalitis employing a dual TCR specific for an exogenous viral antigen and Table 4. ab-TCR rearrangements, frequency and tissue localization of clonal CD8 + T cells isolated from lesional dermis and epidermis of PV patient #5. myelin basic protein viral infection could brake tolerance to MBP and induce autoimmune pathology [20] hinting to such a mechanism in humans as well.
One of the greatest challenges in clinical immunology is the identification of antigens of T-cell mediated autoimmune disorders or tumor-specific immune responses. The technical hurdles,   Table 5. TCR rearrangements of microdissected T cells from MS biopsy tissue.  however, are high [1]. The molecular characterization of the paired ab heterodimers as described here and the identification of T-cell clones on the single cell level clearly supports that the immune response in both disorders, PV and MS, is antigen-driven. Thus, our technology will now allow us to recombinantly reconstruct functional TCRs from practically any ab-T cell of choice. Such ''resurrected'' T cells may then be used to identify the original antigen-specificity by a most recently developed generally applicable technology [25] or to test the functional outcome of Tcell mediated immune reactions in TCR transgenic animals. This application may significantly promote investigations on T-cell mediated autoimmune disorders, tumor-specific protective immunity, and infections.

Patients
The ethics committee of the medical faculty of the Ludwig-Maximilian-University of Munich had approved the study. Patients with type 1 chronic plaque psoriasis (early disease onset, positive family history for PV and/or presence of HLA-Cw6) [23] participated voluntarily and gave written informed consent. Clinical data of the MS patient are described [17,18]. The biopsy was frozen in liquid nitrogen immediately upon excision.

Isolation of Single T Cells from Peripheral Blood or Lesional Psoriatic Biopsies
Small spindle shaped biopsies from chronic PV plaques were cut in half. One part (''A'') was minced completely mechanically and cultured for 12 to 24 h in RPMI medium (PAA Laboratories) with standard supplements to allow T cells to segregate from the tissue threads. In some experiments, dermis and epidermis were dissociated by dispase treatment (2.4 U/ml, Roche) for 1 h at 37uC [26] and the T cells isolated from these two different compartments separately. Part ''B'' of the biopsy was homogenized in Trizol reagent for the preparation of mRNA and TCR bchain cDNA. Peripheral blood lymphocytes (PBL) were prepared from heparinised venous blood by Ficoll density gradient centrifugation.
To isolate single CD4 + or CD8 + T cells, explant culture cells or PBL were suspended in PBS pH 7.2, 2 mM EDTA (auto-MACS TM Rinsing Solution, Miltenyi Biotec), 0.5% bovine serum albumin (BSA) and incubated with anti-CD4 or anti-CD8 magnetic beads (Dynabeads, Invitrogen). Beads-coated cell suspensions were transferred into 6 well microtiter plates (Costar, Corning) and adjusted to a concentration of 1 to 2 beads-coated T cells per visual field at 200-fold magnification in an inverted microscope. Single beads-coated T cells were aspirated in a volume of ,0.5 ml using a variable 2 ml pipette, transferred into PCR tubes containing 5.5 ml 1x OneStep RT-PCR buffer (Qiagen) and stored at 280uC. For FACSorting of single T cells, PBMC were stained with the anti-CD3 antibody UCHT1 (Dako), and single cell were sorted into PCR tubes using a FACSVantage SE cell sorter.

Isolation of Single Cells from Frozen Brain Tissue by Laser Microdissection
PET membrane slides (P.A.L.M Microlaser) were baked at 180uC for 4 h, UV irradiated and coated with poly-L-lysine hydrobromide (Sigma). 10 mm thick cryostat sections from frozen tissue specimens were mounted onto these slides and stored at 280uC. Prior to staining, the slides were thawed, fixed briefly in 100% acetone rehydrated in PBS for 10 seconds, and blocked for 3 minutes in PBS containing 2% BSA (Sigma). Next, the sections were co-incubated for 5 minutes with an FITC labelled anti CD134 Antibody (clone ACT35, BD-Pharmingen) and a Cy3 labelled anti-CD8 beta chain antibody (clone 2ST8.547, Beckman Coulter). The sections were then rinsed with 1 ml of PBS and incubated for 3 minutes with a 1:100 diluted Alexa-488 labelled anti-FITC antibody (A11096, Invitrogen) to enhance the fluorescence signal.
After a second PBS rinse the sections were covered with 1propanol to retard RNase activity and prevent specimen drying. They were immediately analyzed under a P.A.L.M Microbeam-Z microscope (Zeiss). Cells that were either double positive for both T cell markers or single positive for CD8b were marked electronically using the PalmRobo Software (V3.0, Zeiss). After evaporation of 1-propanol, the cells were cut and catapulted by laser pressure into the mineral oil coated lids of single reaction 200 ml PCR tubes.
We use the TCR nomenclature of Arden et al. [27] throughout.
Reverse Transcription and Pre-amplification of TCR aand b-chains by Multiplex PCR cDNA synthesis was performed directly from single T cells using the OneStep RT-PCR Kit. 6.5 ml of RT reaction mix with 10 mM of the Ca-out and Cb-out oligonucleotide primers (Table 1) were added to the 6.0 ml sample containing the single T cell. Reverse transcription of TCR mRNA was performed for 35 min at 50uC. For cells isolated by laser microdissection, 12.5 ml of the RT mix were added directly into the cap of the reaction tube followed by centrifugation at 14,000 rpm, 4uC for 3 minutes.
After reverse transcription 12.5 ml PCR mix from the OneStep RT-PCR Kit were added to the sample. The PCR mix contained 3 mM Va-and Vb-oligonucleotide primers each. The primer set for the Va-repertoire consisted of the 24 oligonucleotide primers described recently [2]. The primer set for the Vb repertoire consisted of 9 oligonucleotide primers (Vp1 to Vp9) given in Table 1. After an initial incubation for 15 min at 95uC to activate hot start polymerase, PCR was run for 10 cycles at 94uC for 30 sec, 60uC for 90 sec and 68uC for 60 sec. Then another 30 cycles were run at 94uC for 30 sec, 53uC for 90 sec and 68uC for 60 sec, followed by a final extension at 68uC for 15 min. This preamplification product served as template for the subsequent characterization of TCR aand b-chains.

Characterization of the TCR b-chains
To facilitate a further amplification of the TCR b-chains by PCR, a universal primer (UP)-sequence was added as an anchor sequence to the 59-end of each Vp1 to Vp9 primer (Vp1-UP to Vp9-UP). The nucleotide sequences of Vp2-UP and Vp9-UP were slightly modified to avoid primer interactions (Table 1). 1 ml of the multiplex PCR product was subjected to a run-off reaction in a PCR mix composed of 1 ml 10x PCR buffer (Roche), 0.2 ml dNTP (10 mM each), 7.65 ml H 2 O, 0.1 ml Vp1-UP to Vp9-UP primers (11.1 mM each), 0.05 ml Taq DNA Polymerase (5 U/ml, Roche). The run off conditions were 94uC for 5 min, 53uC for 150 sec, and 68uC for 15 min.
The TCR b-chain transcripts were then amplified by seminested PCR. 1 ml of the run-off reaction product was used in 20 ml PCR reaction mix that contained 2.5 mM of the Cb-specific nested primer (Cb-in) and 2.5 mM of the universal primer (UP) (Roche). After pre-incubation at 94uC for 2 min, PCR was run for 50 cycles at 94uC for 30 sec, 58uC for 1 min, 68uC for 1 min. After a final elongation step at 68uC for 15 min, PCR products were analyzed by agarose gel electrophoresis and sequenced.

Characterization of the Matching TCR a-chains
Only such cells were tested, which had yielded a PCR product for the b-chain. The TCR a-chains were amplified by nested PCR from the pre-amplification product as described [2] with minor modifications: The concentrations of the Va-in primers were raised to 1.0 mM, and the Ca-in primers were used at 5 mM. Further, the touch-down PCR was run with 4 cycles and 1 min each for the annealing temperatures 61uC, 58uC, and 56uC. Then 40 cycles were run at 53uC. Denaturation and extension steps were for 30 sec at 95uC and for 1 min at 68uC. PCR products were analyzed by agarose gel electrophoresis and sequenced.

Analysis of the TCR aand b-chain Repertoires from Blood T Cells
The Va-and Vb-TCR repertoires from PBMC were reverse transcribed and pre-amplified using the multiplex PCR conditions described above. Individual TCR Va-and Vb-gene families were then amplified from the multiplex PCR product in separate PCR reactions and analyzed by agarose gel electrophoresis or by CDR3 fragment lengths spectratyping as described [2,14]. Only the primers for TRBV8 (59 CCAGCCCTCAGAACCCAG 39), TRBV21 (59 CTCTCAGGATCCAGCCTGCA 39) and TRBV25 (59 GATTTTCAGCTAAGTGCCTCC 39) differed from these protocols, because the original primers for TRBV8 and TRBV21 were located outside the multiplex b-chain PCR product, and the primer set did not include a primer for the TRBV25 family.

Detection of Particular TCR Chains by Clone-specific PCR
For select clonal TCR rearrangements detected by single cell PCR in sample ''A'' of PV biopsies we investigated by clonespecific semi-nested RT-PCR whether their particular TCR bchain transcripts were also present in cDNA of part ''B''. The first PCR employed the respective Vb forward primer and the Cb-out primer. Then, in a semi-nested PCR, the Vb forward primer and a CDR3 specific primer, which extended 4 to 6 nucleotides from the J-region gene into the respective N(D)N-sequence were used. PCR products were sequenced.