Abstract
Chronic rejection represents a major cause of long-term kidney graft loss. T cells that are predominant in long-term rejected kidney allografts (35 ± 10% of area infiltrate) may thus be instrumental in this phenomenom, which is likely to be dependant on the indirect pathway of allorecognition only. We have analyzed the variations in T cell repertoire usage of the Vβ chain at the complementary determining region 3 (CDR3) level in 18 human kidney grafts lost due to chronic rejection. We observed a strongly biased intragraft TCR Vβ usage for the majority of Vβ families and also a very high percentage (55%) of Vβ families exhibiting common and oligoclonal Vβ-Cβ rearrangements in the grafts of patients with chronic rejection associated with superimposed histologically acute lesions. Furthermore, Vβ8 and Vβ23 families exhibited common and oligoclonal Vβ-Jβ rearrangements in 4 of 18 patients (22%). Several CDR3 amino acid sequences were found for the common and oligoclonal Vβ8-Jβ1.4 rearrangement. Quantitative PCR showed that biased Vβ transcripts were also overexpressed in chronically rejected kidneys with superimposed acute lesions. In contrast, T lymphocytes infiltrating rejected allografts with chronic rejection only showed an unaltered Gaussian-type CDR3 length distribution. This pattern suggests that late graft failure associated with histological lesions restricted to Banff-defined chronic rejection does not involve T cell-mediated injury. Thus, our observation suggests that a limited number of determinants stimulates the recipient immune system in long-term allograft failure. The possibility of a local response against viral or parenchymatous cell-derived determinants is discussed.
Despite nonimmunological risk factors such as initial nephron load (1), uncontrolled blood pressure, hyperlipemia, and drug-related nephrotoxicity, all of which contribute to late deterioration of kidney graft function, chronic rejection (CR)3 of kidney allografts represents the main cause of long-term graft failure (see Ref. 2 for review). Clinically, CR corresponds to a slow deterioration of graft function (usually over months to years), which correlates with typical histological changes (2). Histologically, CR is defined as a “chronic transplant nephropathy,” and the grading of the lesions (Banff grades I, II, III) is based on severity of interstitial fibrosis, tubular atrophy, and typical vascular lesions (3, 4). Several immunological risk factors (5) for CR have been identified such as histoincompatibility (6), early acute rejection episodes (7), and posttransplant CMV infection (8).
Acute rejection is histologically characterized by an intense and destructive leukocytic inflammation of graft tissues with a predominant Th1 cell type in the graft infiltrate (9). Alloreactive T cells involved in acute allograft rejection predominantly recognize donor MHC Ags on donor APC (“direct” pathway). In contrast, recognition of donor determinants presented by recipient APC (“indirect” pathway) is more likely to be involved in long-term rejected grafts (see Ref. 10 for review). Alloreactive T lymphocytes derived from acutely rejected human kidney allografts were shown (by amplification of the Vβ transcripts (11, 12) and Southern blot-based studies (13)) to exhibit some restricted repertoire. In addition, a bias in TCR fragment usage, in all cases involving Vβ8, was detected in 14 of 19 cases of kidney transplant recipients when using mAbs against different Vβ families (14). However, these studies were performed following in vitro culture and after stimulation (11, 13) and therefore may not reflect the in vivo situation. In addition, biopsy material of limited size, which has provided a better understanding of cellular transplant immunity (15), may favor clonal expansion (16). Because the occurrence of acute rejection episodes is highly predictive of chronic rejection, alloreactive T cells could represent the common immunological link between these two pathological entities. This is also suggested by the role of T cells in graft neointima formation in CR (17).
Diversity of the TCR (18) αβ is generated by random combination of V, (D), J, and C region segments, nibbling of junctional V, (D), and J regions, addition of nontemplate nucleotides, germlime polymorphism between individuals, and pairing of one α-chain with one β-chain (19). The complementary determining region 3 (CDR3), the hypervariable region that carries most of the fine specificity of Ag recognition by T cells, varies in length due to the addition or removal of nucleotides at the VD and DJ junctions by nucleotide transferases during the recombination process. Therefore, analysis of CDR3 lengths and sequences would indirectly reflect the selection of the T cell repertoire usage by alloreactive T cells involved in CR and may bring new and important information concerning the mechanism of this poorly understood process to light (20). In this paper, we report on an analysis of T cell infiltration and, for the first time, on the variations of Vβ T cell repertoire usage at the CDR3 level in transplantectomy pieces from 18 human kidney-graft recipients with a clinical history and histological lesions of long-term CR with or without superimposed histologically acute lesions. Our findings show that a highly altered T cell response is involved in CR with superimposed acute lesions and suggest the presence of a restricted set of Ags.
Materials and Methods
Patients
We analyzed a population of 18 kidney recipients who underwent transplantation between 1983 and 1996 and who were transplantectomized due to a chronic graft dysfunction, all with characteristic histological lesions of CR (Table I⇓). The mean graft survival was 69.8 ± 38.8 mo (range, 6–126). The time interval between graft loss (initiation of chronic dialysis) and transplantectomy was 243 ± 248 days (range, 0–750). Fourteen patients received a first kidney graft, twice combined with a pancreas. The mean recipient age at grafting was 39.1 ± 17.5 years (range, 8–66). Eleven recipients (66%), vs 14 donors (78%), were male. The initial disease was a glomerulonephritis in nine patients, three had urological malformations, one a vascular disease, and five were not histologically characterized before grafting. Twelve patients had no “historical” anti-T panel-reactive Abs (PRA) before grafting, five patients had anti-T PRA between 4 and 58%, and only one was hyperimmunized, with a PRA level above 75%. Sixteen patients received an induction therapy consisting of anti-thymocyte globulins or anti-lymphocyte globulins (Pasteur-Mérieux et Vaccins, Lyon, France), one received an anti-IL-2 receptor mAb (21), and one received HLA-derived peptide (Allotrap; Sangstat Medical Corporation, Menlo Park, CA) (22). Ten patients (55%) presented at least one early acute rejection episode, three presented two episodes, and one presented four episodes. All were CMV seronegative before grafting. During the follow up, six patients (33%) were treated for an overt CMV disease. CMV infection was detected in peripheral blood leukocytes 1–3 mo after grafting, either by CMV antigenemia assay (n = 4) or by amplification of CMV-DNA (n = 2) depending on the length of the survey. At transplantectomy, patients had been free of immunosuppressive therapy for 12 ± 9 mo, except for four who received cyclosporin A (n = 1), cyclosporin A and azathioprine (n = 1), prograf (n = 1), or steroids (n = 1). Histologically, according to the Banff classification (3), CR lesions were observed in all patients (grades II, III), associated with significant superimposed acute lesions in 14 of them (types I, II, III) (Table I⇓). None of the 18 patients had a recurrent disease on the transplant. PBMC, collected at the day of but before grafting from five recipients, were also available and used for the analysis of the β-chain repertoire. Finally, TCR β-chain repertoires were also studied using resting PBMC from healthy volunteers. PBMC were isolated by density gradient (Eurobio, Les Ullis, France), according to standard procedures.
Clinical characteristics and graft histology
Immunohistology and quantitative analysis of cellular infiltrate
Transplantectomy pieces from long-term rejected kidney allografts were immediately snap-frozen in liquid nitrogen, embedded in OCT compound (Tissue Tek; Miles Laboratories, Elkhardt, IN), and stored at −80°C until used for immunohistology. Immunoperoxydase staining of 5-μM sections of frozen tissue was performed as follows: sections were cut, fixed in acetone, and labeled using a three-step indirect immunoperoxydase technique (23). Primary Abs were the following mouse IgG anti-human mAbs: anti-CD3 (Dako, Trappes, France), anti-CD11b, anti-CD16, anti-CD19, anti-HLA-DR, anti-RIL-2, and anti-TCR PAN γ/δ (Immunotech, Marseille, France). All of these Abs were used at the dilution recommended by the manufacturer. The mouse mAb directed against human TCR δ- and γ-chain determinants was pretested on human lymph nodes as a positive control to predetermine the optimal dilution. The nonspecific staining was taken into account by omission of the first Ab. The secondary Ab used was a biotin-conjugated anti-mouse IgG (Vector Laboratories, Burlingame, CA). Finally, tissue sections were incubated with HRP streptavidin (Vector Laboratories) and developed with “intense purple” (VIP kit; Vector Laboratories).
The area of each immunoperoxydase-labeled tissue section that was infiltrated by cells was determined by quantitative morphometric analysis (24, 25). Briefly, positively stained cells in the interstitium of each section were counted by morphometric analysis using a point counting method with a 121-intersection squared grid in the eyepiece of the microscope. Results were expressed as the percentage of area infiltrate of each graft section occupied by cells of a particular antigenic specificity (±SD). The percentage of area infiltrate was calculated as follows: [(number of positive cells under grid intersections) ÷ (total number of grid intersections = 121)] × 100. The graft sections were examined at a magnification of ×400. The accuracy of the technique is proportional to the number of points counted. Thus, to maintain a SD of <10%, 15 fields were counted for each labeled section of high density and 40 fields for sections with low density (<10%). Point counting was scored by two observers.
RNA extraction and cDNA synthesis
Total RNA from kidney transplantectomies was isolated by the guanidium isothiocyanate procedure and purified on a cesium chloride gradient (26). Total RNA from PBMC was extracted by the Chomczynski and Sacchi method (27). Before retrotranscription, RNA samples were systematically treated with DNase (Promega, Charbonniéres, France). Ten micrograms of RNA were reverse-transcribed into cDNA using 14 μg/ml of oligo(dT)12- 18, 10 mM DTT, 0.5 mM of each dNTP, 40 U RNAsin (Promega), and 200 U M-MLV reverse transcriptase in 5× first-strand buffer (Life Technologies, Gaithersburg, MD). The cDNA synthesis reaction was brought to a final reaction volume of 100 μl.
TCR β-chain CDR3 fragment size determination and sequencing
PCR amplification and elongation reactions for analysis of the CDR3 lengths.
Aliquots of the cDNA synthesis reaction (2 μl) were amplified in 50-μl reactions with 1 of the 20 Vβ human family-specific primers and a Cβ primer recognizing both Cβ1 and Cβ2 regions (28, 29). The Vβ10, 20, 21, and 24 families were not represented because no transcripts were available. The PCR amplification conditions were as previously described (30). Aliquots (2 μl) of the 20 unlabeled Vβ-Cβ PCR products were then subjected to elongation with either Cβ or 13 Jβ fluorophore-labeled specific primers using the same PCR conditions as previously described (30). The elongation reaction products were then heat-denatured and loaded onto a 6% acrylamide, 8 M urea gel and run on an ABI 373A DNA sequencer (Applied Biosystems, Foster City, CA) for size and fluorescence intensity determination.
TCR Vβ chain transcript sequencing.
When common and dominant Vβ-Jβ expansions were found, the Vβ transcripts were then reamplified for 35 cycles with one of the 13 unlabeled Jβ primers (0.5 μM) using the same PCR conditions as described above. Amplification products (2 μl) were then ligated into the PCR topo vector of the Topo Ta Cloning Kit (Invitrogen, Carlsbad, CA) in accordance with the manufacturer’s instructions. The ligation product was transfected into Top10F′ competent cells (Invitrogen). Twenty white growing colonies (containing individual clones) were randomly selected. Plasmid DNA was then recovered by the alkaline lysis method and digested with EcoRI to confirm insertion of clones. Forty microliters of plasmid DNA were purified, and 5 μg of the preparation were lyophilized and sequenced using standard methods (Appligene, Illkirch, France).
Analysis of TCR repertoire alterations
Several approaches were used. The Immunoscope software (28) was used to obtain a semiquantitative analysis of the β-chain of the TCR repertoire at the CDR3 level. The distribution profile of CDR3 lengths in base pairs for each Vβ family was represented, typically with 7–11 peaks each separated by 3 nt. Each peak corresponds to a TCR transcript with a given CDR3 length that may contain multiple sequences. The number of TCR transcripts with a specific CDR3 length is proportional to the area under each peak. An increase in the height and area of a size peak usually signals an oligoclonal or monoclonal expansion in the polyclonal T cell background. An oligoclonal expansion with the same dominant CDR3 length shared by several individuals is defined as common. When only one sequence is found for the same CDR3 length, the peak signals a monoclonal expansion, which is defined as “public” if reproducibly found in several samples.
The Reperturb software (25, 31) was used to further identify and quantify the alterations in CDR3 length distributions of each Vβ family and in each sample. Briefly, each CDR3 profile obtained using the Immunoscope software was translated into a probability distribution using the fraction of the area under the profile for each CDR3 length, as described in detail elsewhere (25, 31). A control profile, representing the unaltered Gaussian repertoire, was established for each Vβ family by calculating the average probability distribution for the corresponding Vβ in PBMC of three healthy volunteers. The average of these distributions, for each Vβ family separately, was then used as a control distribution for analysis of the other samples. The alterations in each Vβ family were defined as the sum of the absolute values of difference for all CDR3 lengths in that profile. The sum of the alterations in the 20 Vβ families studied in a sample gives the alterations in the entire Vβ TCR repertoire in this sample. Differences between controls and patients were statistically assessed using the Student’s t test, significance being established at p < 0.05.
Relative quantification of Vβ transcripts
Principle.
The ABI PRISM 7700 Sequence Detection Application program (Applied Biosystems) is used to detect and measure fluorescence emitted during PCR amplification of a given target sequence in a 96-well reaction plate. This detection is conducted in real-time because data is collected during each PCR cycle. Fluorescence is emitted following binding of the dye labeler, SYBR Green, to double-stranded DNA (32). Thus, the level of fluorescence is directly proportional to the level of PCR product. Samples are measured as well as a standard prepared using a dilution range of the same target sequence, from which a standard curve is derived. Using this standard curve, the quantity of the target sequence in samples is calculated. Normalization of the levels of the target sequence is then conducted using a reference house-keeping gene. In this case, the levels of each Vβ family were measured separately and were normalized against the level of Cβ, which is a reference gene for T cells (33).
Oligonucleotides and standard construction.
To prepare each Vβ standard curve, the respective Vβ primer (Vβ1, 5, 8, 13, 14, 16, 18, 19, 23) was used in the forward position together with the Cβr primer in the reverse position (Table II⇓ and Fig. 1⇓). To prepare the Cβ standard curve, the Cβf and Cβr primers were used in the forward and reverse positions, respectively (Table II⇓ and Fig. 1⇓). The target sequence was amplified in cDNA derived from a healthy volunteer using forward (one Vβ or Cβf) and reverse (Cβr) oligonucleotides, which were electrophoresed and purified using a gel extraction kit (QIAquick Gel Extraction Kit; Qiagen, Germany). For each Vβ family and Cβ, the standard concentration was derived from its absorbance value at 260 nm, and the number of copies per ml was then calculated using the m.w. of the cDNA. Subsequent dilutions of each Vβ and Cβ standard DNA were performed to obtain 107, 106, 105, 104, 103, and 102 copies per well to establish a range of concentration similar to that of the target in the biological samples.
Amplifications were performed through the CDR3 hypervariable region, which corresponds to the recombined VDJ segements and varies in length due to the addition or removal of nucleotides (n) during the recombination process.
Sequences of primers used for relative quantitation of Vβ and Cβ transcriptsa
PCR amplification and analysis.
A constant amount of cDNA (200 ng of total RNA) for each dilution of each Vβ standard, or Cβ standard, was amplified in 25 μl of SYBR Green PCR Core Reagent (Applied Biosystems) with 0.6 U of AmpliTaq Gold polymerase, 0.25 U of AmpErase uracil-H-glycosylase, 200 μM of each dNTP, 300 nM of each primer, and 3 mM of MgCl2, in 10× SYBR Green PCR buffer (1× final concentration). Amplifications were performed by an ABI Prism 7700 Sequence Detection System Perkin-Elmer machine (Applied Biosystems). The PCR started with an initial step of 2 min at 50°C, followed by 10 min at 95°C and then 40 cycles each consisting of 15 s at 95°C and 1 min at 60°C. Each sample was analyzed in duplicate. The exact number of copies of the cDNA target sequence was deduced from the comparison of the measured fluorescence with the standard curve. Results were expressed as a ratio of Vβ/Cβ transcript number (×100) to compare the different samples.
Detection of virus genome
PCR assays were used to detect the presence of DNA of the HSV (34), human herpes virus 6 (HHV6) (35), EBV (36), CMV (37), and human papillomavirus (38) in long-term rejected kidney allografts. EBV lytic-phase Zebra mRNA in the grafts was also analyzed by reverse transcription and PCR amplification (39).
Results
αβ+ T lymphocytes are, together with monocytes and macrophages, the predominant cell type in long-term rejected kidney allografts
Subtypes of inflammatory cells in the area infiltrate, including T lymphocytes, were quantified in 18 long-term rejected kidney human allografts with CR, either associated (n = 14) or not associated (n = 4) with superimposed acute lesions at Banff scoring. Chronically rejected kidneys with superimposed histologically acute lesions displayed an area infiltrate represented by 35 ± 10% of T lymphocytes (CD3+), 17 ± 10% of monocytes-macrophages (CD11b+), 20 ± 10% of B lymphocytes (CD19+), and 1.8 ± 0.6% of NK cells (CD16+), as quantitatively assessed by the point counting method (Fig. 2⇓). Chronically rejected kidneys with restricted CR lesions showed an area infiltrate composed of 30 ± 7% T lymphocytes, 35 ± 9% monocytes-macrophages, 20 ± 9% B lymphocytes, and 1 ± 0.4% NK cells (Fig. 2⇓). However, despite a roughly similar number of mononucleated cells (i.e., T/B lymphocytes plus monocytes/macrophages) infiltrating kidneys with CR or CR with acute lesions on frozen sections (immunochemistry), only Banff scoring discriminates the “active infiltrate,” particularly tubulitis, restricted to kidneys with superimposed acute lesions.
Leukocyte infiltration in long-term rejected kidney allografts. Anti-CD3 (T cells), anti-TCR PAN γδ (γδ T lymphocytes), anti-CD11b (monocytes, macrophages), anti-CD16 (NK cells), anti-CD19 (B lymphocytes), anti-HLA-DR, and anti-IL-2 receptor α-chain mouse anti-human mAbs were tested on the 18 transplants. Results are expressed as the percentage (±SD) of positive cells in the area infiltrate in chronically rejected kidneys (CR) with (n = 14) or without (n = 4) superimposed acute lesions (SAL).
In addition, because γδ T lymphocyte expansion has been reported in some acutely rejected human kidney allografts (40) and in PBMC of CMV-infected patients (41), we also examined the presence of this T lymphocyte subtype in the grafts. Less than 0.1% of T lymphocytes were stained by anti-PAN γ/δ TCR (Fig. 2⇑). Finally, a high proportion (45 ± 9% for CR plus superimposed acute lesions, 50 ± 7% for CR) of cells in the area infiltrate exhibited HLA-DR activation marker, and the IL-2 receptor α-chain activation marker was expressed in only a few cells (8 ± 6% and 9 ± 5%, respectively).
Highly altered Vβ-Cβ repertoire in chronically rejected human kidney allografts with superimposed histologically acute lesions
The length of the CDR3 region in each Vβ family, in patients and healthy volunteers, was amplified using specific primers for the Cβ and 20 Vβ human families. The degree of alteration in the T cell repertoire was further studied using the Immunoscope and Reperturb softwares. Fig. 3⇓ shows the Immunoscope analysis of the CDR3 size distribution of 20 Vβ families in the transplant of a representative patient (patient CI), as compared with PBMC from patient CI before grafting, as well as from a healthy individual. The comparative analysis of CDR3 size for each Vβ family in pregraft PBMC showed a Gaussian distribution pattern (7–11 peaks), similar to the healthy volunteer profile and characteristic of a resting population. In contrast, the Vβ patterns in the graft of the same patient displayed numerous oligoclonal or monoclonal T cell expansions in terms of CDR3 length (Fig. 3⇓).
CDR3 length profiles for TCR transcripts in PBMC from one healthy volunteer in PBMC from patient CI before transplantation and in his graft (29 mo after transplantation). TCR β transcripts were reverse-transcribed and amplified using Vβ- and Cβ-specific primers. A dye-labeled Cβ-specific primer was then used to visualize the amplified products. After electrophoresis and subsequent computer analysis, the different size peaks were separated and the CDR3 size in base pairs calculated. The graph represents the intensity of fluorescence in arbitrary units (y-axis) as a function of the CDR3 length in base pairs of Vβ-Cβ elongation reaction products (x-axis). The Vβ1-Vβ23 profiles in PBMC of one healthy volunteer, in PBMC of one representative patient (patient CI) at the day of grafting, and in his graft are respectively represented at the top, middle, and bottom of each box.
Total alterations in the TCR repertoire in the grafts of patients with CR with (n = 14) or without (n = 4) superimposed histological acute lesions were then further analyzed using the Reperturb software. Total alterations in kidneys with restricted CR lesions (n = 4) ranging from 7.8 to 10.6% (as compared with the control profiles) were smaller than those observed in the 14 kidneys with CR associated with superimposed histologically acute lesions (from 7.8 to 19.5%) (Fig. 4⇓). No correlation was observed when the percentage of total TCR alterations was plotted against the Banff grade I/II/III or its “i” component. Details of the alterations for each Vβ family in the PBMC of three healthy volunteers, compared with those observed in three representative kidneys with CR lesions only, and with superimposed acute lesions are illustrated in Fig. 5⇓. These alterations are represented by tridimensional views in which smooth landscapes indicate an unaltered repertoire, while “mountains” and “valleys” denote over- and underamplified CDR3 lengths. The profiles in kidneys with lesions restricted to CR showed only a slightly altered TCR Vβ usage as compared with the patterns observed in the PBMC of the three healthy volunteers (Fig. 5⇓B). In contrast, the patterns observed in the kidneys of the patients with CR associated with superimposed acute lesions displayed a very strongly altered TCR Vβ gene usage (Fig. 5⇓C). In this group, every patient profile was characterized by the presence of altered Vβ families.
Analysis of total TCR repertoire graft alterations in PBMC of healthy volunteers and in CR with or without superimposed acute lesions. Percentages of total alterations in all Vβ family patterns were calculated using the Reperturb software. The magnitude of the alterations in the repertoire of TCR in PBMC from each healthy volunteer (n = 3), in the graft from each patient with chronic rejection (CR) lesions (n = 4), and for each patient with CR associated with superimposed acute lesions (CR + SAL) (n = 14) are represented respectively at the left, middle, and right panel on the x-axis. Each dot represents the average of the alterations for all the 20 Vβ families for each healthy volunteer and for each patient, compared with the theoretical Gaussian profile.
Comparison of the alterations in each TCR Vβ family in PBMC of healthy volunteers and in long-term rejected kidney allografts. The alterations (y-axis) for a specific Vβ family (x-axis) with a given CDR3 length (z-axis) are calculated by comparing the area between each patient’s and control’s distribution. These alterations are represented for each Vβ family (x-axis) by tridimensional views in the PBMC of three healthy individuals (A) compared with the patterns observed in the kidneys of representative patients with CR lesions (n = 3) (B) and with CR associated with superimposed acute lesions (n = 3) (C). Patient names are indicated inside each graph.
Several patients used common and oligoclonal Vβ-Cβ and Vβ-Jβ rearrangements
Oligoclonal expansions were present in several patients and involved 11 Vβ families. First, two common and dominant rearrangements were found for the Vβ1 (180, 183 bp), Vβ5 (159, 162 bp), and Vβ23 (272, 278 bp) families and were found in 5 (28%), 6 (33%) and 6 (33%) of 18 patients, respectively. Immunoscope profiles for these three Vβ families and for one representative patient are shown in Fig. 6⇓A. Other oligoclonal expansions with one dominant size rearrangement were also shared by several patients and involved Vβ8 (271 bp, patients VJ, BC, LJM, MP, MC, CC, GM), Vβ11 (140 bp, patients TS, BA, DF, PE, CI), Vβ13 (327 bp, patients VJ, BC, LJM, LB, MP), Vβ14 (258 bp, patients MC, CK, GM), Vβ15 (174 bp, patients TS, VJ, BC, CI, BR, MR), Vβ16 (136 bp, patients TS, BR), Vβ18 (200 bp, patients CK, MP, BR), and Vβ19 (153 bp, patients PE, BR). Representative profiles are shown in Fig. 6⇓B. These common rearrangements were predominantly found in patients with CR associated with superimposed acute lesions. Nevertheless, no correlation was found between these oligoclonal responses and degree of HLA compatibilities, donor/recipient HLA specificities, history of CMV disease, presence of viral genome in the graft (see later), sex matching, time between graft loss and nephrectomy, or number of acute rejection episodes.
Oligoclonal Vβ-Cβ rearrangements reproducibly found in long-term rejected kidneys allografts. Each graph represents the intensity of fluorescence, which is proportional to the number of transcripts for each Vβ family, in arbitrary units (y-axis), as a function of the CDR3 length, in base pairs (x-axis), of three Vβ families, respectively (Vβ1, Vβ5, Vβ23), with two common and dominant size expansions (A) and for eight Vβ families (Vβ8, Vβ11, Vβ13, Vβ14, Vβ15, Vβ16, Vβ18, Vβ19) with one dominant size expansion (B). The length of each dominant expansion is indicated in base pairs under each box (x-axis). The name of each patient is indicated inside each Immunoscope profile. At the top of the figure, for comparison, are the Immunoscope profiles of one healthy volunteer for the corresponding Vβ families. The length of the dominant peak, for each family, in healthy volunteers is indicated at the bottom of each control Immunoscope profile.
All of the common Vβ-Cβ rearrangements were then elongated using the 13 Jβ primers. Two Vβ families showed common Vβ-Jβ rearrangements between several patients (Fig. 7⇓). Dominant Vβ8-Jβ1.4 expansions were observed in 4 of 18 (22%) grafts and involved patients VJ, BC, LB, and MP. A dominant Vβ8-Jβ1.4 peak corresponding to a CDR3 length of 9 aa (229 bp for Vβ-Jβ PCR product) was present in rejecting allografts from patients VJ and BC, whereas the length of this expansion consisted of 8 aa (226 bp) in rejecting allografts from patients LB and MP (Fig. 7⇓A). This common Vβ8-Jβ1.4 rearrangement was only present in patients with CR associated with superimposed acute lesions but without correlation with HLA compatibilities or specificities, history of CMV disease or viral genome in the graft, sex matching, time between graft loss and nephrectomy, or number of acute rejection episodes. An additionnal common Vβ23-Jβ2.5 rearrangement corresponding to a CDR3 length of 11 aa (240 bp for Vβ-Jβ PCR product) was also expanded in the grafts of patients BA, TS, CI, and MP (Fig. 7⇓B). Because alterations in common Vβ-Jβ rearrangements were likely to be those supporting the highest constraint, sequence analysis was restricted to these products. The CDR3 motive of Vβ8-Jβ1.4 oligoclonal and common expansions in patients VJ, BC, LB, and MP are listed in Table III⇓. Although sharing the same oligoclonal and dominant Vβ-Jβ usage, the sequences from the Vβ8-Jβ1.4 peaks were different in the transplants of all patients tested (VJ, BC, LB, and MP).
Oligoclonal Vβ-Jβ rearrangements reproducibly found in the rejected kidneys. Each graph represents the intensity of fluorescence, in arbitrary units (y-axis), as a function of the CDR3 length, in base pairs (x-axis), of Vβ-Cβ common and dominant rearrangements and the corresponding Vβ-Jβ oligoclonal expansions. The lengths of Vβ8-Cβ and Vβ8-Jβ1.4 (A) or Vβ23-Cβ and Vβ23-Jβ2.5 (B) common and oligoclonal peaks are indicated in base pairs under each Immunoscope profile. Patient names are indicated inside each graph.
CDR3 amino acid sequences of Vβ8-Jβ1.4 oligoclonal expansionsa
Analysis of viral genome in long-term rejected kidney allografts
Because oligoclonal Vβ-Cβ or Vβ-Jβ expansions could be correlated to viral infections (31, 42, 43), we also examined, by PCR amplification, the presence of CMV, HSV, HHV6, human papillomavirus, and EBV genome in long-term rejected kidneys. Some viral DNA was detected in the grafts, involving CMV (patient VJ), HSV (patient PE), HHV6 (patients MR, LB, TS), and EBV (patients MR, CC, VJ, LB, CI, PE, BC, MC, DD). Only one kidney (patient BC) presented EBV lytic-phase mRNA. However, no significant correlation (χ2 test) could be found between the presence of a Vβ-Cβ or a Vβ-Jβ common and oligoclonal expansion and the presence of a viral genome in the grafts.
Vβ transcripts of altered patterns accumulated in CR with superimposed histologically acute lesions
Because the level of expression of a given mRNA species could be even more important than its qualitative Immunoscope profile and gives further information on the magnitude of the selective pressure on the Vβ repertoire, we quantified by PCR (SYBR Green methodology, see Materials and Methods) the accumulation of the altered Vβ transcripts in the grafts, with primers devoted to nine Vβ families. Vβ/Cβ transcript ratios were found to be variously increased in the grafts as compared with the ratios observed in the PBMC of healthy volunteers, except for the Vβ13 family (Table IV⇓). The average quantity of Vβ8/Cβ transcripts in the grafts from patients VJ, BC, LB, and MP who presented a Vβ8-Jβ1.4 oligoclonal expansion was five times higher than other patients without this rearrangement (Table IV⇓). However, transcripts of some common expansions such as Vβ8 (see patient MP in Table IV⇓ and his corresponding Immunoscope profile in Fig. 7⇑A) were 120 times higher than in pregraft PBMC or in healthy volunteers.
Relative quantification of Vβ transcripts in CR with superimposed acute lesions
Continued
Discussion
In this study, we showed that αβ+ T cells infiltrating long-term rejected kidney allografts exhibit a strongly altered TCR Vβ usage particularly when CR is associated with superimposed histologically acute lesions. Moreover, besides vigorously skewed “private”-type TCR biases, common and oligoclonal Vβ-Cβ rearrangements were also frequently observed. In contrast, transplants presenting a restricted CR at histological Banff grading displayed a resting-type Gaussian TCR profile that, together with the absence of an “active” infiltrate (i.e., tubulitis) in this CR-restricted pattern, does not suggest a major involvement of T cells in this specific process. Our data suggest that a few antigenic determinants only may stimulate the immune system of recipients of long-term kidney grafts.
The analysis and understanding of the T cell repertoire alterations involved in local or systemic responses has been crucial in several clinical situations including alloimmunity, tumor immunity, autoimmune, and infectious diseases. For instance, oligoclonal T cell expansions at the CDR3 level have been reported in melanoma tumors (30), in skin lesions of acute graft-vs-host disease (44) and in synovial lesions from patients with rheumatoid arthritis (28). However, despite most of these studies not using the Reperturb approach to quantify and compare the average global TCR alterations, their magnitude, as assessed by the number of affected Vβ families and of common oligoclonal expansions (i.e., shared by several patients) for a given Vβ-Cβ (or Vβ-Jβ) rearrangement, was much less remarkable. Only one report on TCR alterations during AIDS progression, using the same method of assessment of total alterations of nine Vβ families, showed a comparably strongly altered pattern, but without a common profile being shared between patients (31), a frequent characteristic in our study of long-term rejected grafts.
Restricted or altered Vβ repertoires of T lymphocytes infiltrating kidney allografts during acute and chronic rejection have been already described in humans but with semiquantitative PCR or by Southern blot methods that do not discriminate CDR3 biases within each of the Vβ families (11, 12, 13, 45). Barth et al. have shown that the TCR repertoire in biopsies of acutely or chronically rejected kidney allografts was restricted to one to three dominant Vβ families, without public expansions (12). However, in these reports of overrepresentation of some Vβ families, the fact that the tissue samples used were biopsies casts a doubt on their relevance in representing the actual global alterations in a graft. Indeed, focal T cell clonal expansion may be selected by the use of very small tissue samples, likely not representative of the total immune response occurring in a graft (11, 13, 16, 46). In addition, most of these studies were performed on alloreactive T cell lines derived from grown graft-biopsy infiltrating cells (11, 13, 46). Despite the fact that one cannot exclude that the in vitro step allows the most relevant alloreactive cells to emerge from the heterogenous infiltrate, it is highly likely that a selection operates during the culture, generating biases per se, and that again, the overall observed pattern may not reflect the in vivo situation. Other artifactual biases can be further amplified by the use of polyclonal stimuli in the culture, promoting uncommitted cells. Moreover, Hall and collaborators have shown that alloreactive clones derived from rejected kidney grafts presented an altered TCR repertoire that was positively correlated with the length of time cells were maintained in culture (11). In contrast, in our study, we directly amplified transcripts of Vβ families from large pieces of surgically removed grafts (without a culture step) to reflect as much as possible the in vivo situation.
Interestingly, the TCR Vβ repertoire of kidneys that underwent slow functional degradation with histological lesions fitting with the definition of CR only (i.e., without superimposed acute lesions) presented a Gaussian pattern of Vβ family CDR3 lengths, which usually reflects uncommitted T lymphocytes at the resting stage, can also be observed in the context of a strong polyclonal activation (25). However, despite a roughly similar number of mononucleated cells found in kidneys with restricted CR and in CR with superimposed acute lesions, the absence of “active” localization of T cell infiltrate (i.e., tubulitis) in restricted CR patterns does not favor such an alternative. Rather, our data strongly suggest that T cells may not be involved in this restricted CR process and that other nonimmunological factors (initial nephron load, blood pressure, hyperlipemia, drug related chronic toxicity) may operate in this situation (2). The superimposed histologically acute lesions associated with CR bring some semantic complexities to the interpretation of our findings because the pattern could be more related to late acute rejection rather than to restricted CR. However, it is noteworthy that most of these kidneys were removed several years after grafting (11 of 14 after 4 years) and that such a profile is likely to be restricted to the indirect pathway of allorecognition (47) and not to direct recognition of donor MHC motives. Therefore, the superimposed histologically acute lesions are likely to reflect an emphasized indirect recognition process that would not have the possibility to emerge when a normal immunosuppressive regimen was still being administered. In this context, such transplants might be characteristic of the processes operating in long-term kidneys and would allow an interesting basis for characterization of the relevant Ags involved, which for now remain elusive.
Allorecognition occurring through the indirect (or cross-priming) pathway can proceed from a variety of peptides from donor MHC, graft tissues, or microorganisms. However, no correlations between oligoclonal alterations and donor sex, donor or recipient MHC alleles, or MHC incompatibilities were noted. Could donor T cells have contributed to this pattern (i.e., altered Vβ repertoire)? Thus far, there is no direct evidence available in the literature. However, donor cell chimerism has been detected years after kidney transplantation (48) at levels as low as 1:100,000 donor to recipient cells by PCR sequence-specific primers (SSP) using the Y chromosome as a target. In the present study, no circulating donor cells could be detected by PCR-SSP performed on HLA-DR targets in four patients for whom PBL harvested at the time of the experiment were available (patients CS, LJM, VJ, and DD, data not shown). Based on this finding, it is likely that the altered Vβ repertoire observed in this study was due to recipient T cells alone.
Nonrestricted public CDR3 sequences found in the common Vβ8-Jβ1.4 alteration in several recipients does not preclude the possibility that the corresponding Vβ chains do not interact with identical determinants. Indeed, the cross-reactive nature of TCR recognition is now a well-accepted concept (49, 50), and situations where TCR with similar CDR3 sequences interact with different peptides or, alternatively, where Vβ-chains of different CDR3 sequences interact with a single determinant are increasingly reported (51, 52). Furthermore, identical Vβ TCR sequences shared by T lymphocytes infiltrating the graft of several animals, even in a congeneic combination, were the exception (53, 54) and were almost restricted to a single donor-specific transfusion-induced tolerance (55). However, the above-mentioned high cross-reactive nature of the TCR makes any interpretation at this stage difficult, and one cannot exclude the fact that positively selected clones with the observed Vβ alterations could also recognize viruses known to frequently infect long-term graft recipients. Highly altered Vβ repertoires with oligoclonal Vβ expansions have also been described in the peripheral blood of patients infected by HIV (31), of healthy individuals infected by EBV (42), and in patients with chronic infection with hepatitis B virus (43). Because CMV infections are frequent after transplantation (8), oligoclonal and common TCR expansions could also have been directed against CMV-derived peptides. However, γδ TCR clones have only been found to be expanded during CMV infection (41), and γδ-positive T cells represented a tiny minority (<1%) of infiltrating cells in our patients. Furthermore, we were unable to identify TCR-αβ alterations mimicking the common and oligoclonal patterns found in the grafts studied when peripheral TCR repertoires in four patients were analyzed, with the same method, following an unambiguous overt CMV disease. Finally, neither the clinical CMV infections nor the finding of viral genome (CMV, HSV, HHV6, human papillomavirus, and EBV) correlated with a Vβ-Cβ or a Vβ-Jβ oligoclonal expansion found in the kidneys could sustain, at this stage, a rationale for the hypothesis of viral-driven TCR alterations. Furthermore, none of the published characteristics or sequences of virus-driven TCR alteration reported in humans (42) were noted in the altered Vβ transcripts derived from the rejected grafts studied.
Nevertheless, it is noteworthy that the most altered family (with common Vβ-Jβ rearrangement) was found to be the Vβ8 family. An increased Vβ8 usage has been reported not only in an autoimmune disease with a potentially viral etiology (56), or following superantigen stimulation (57), but also in peripheral T lymphocytes of kidney graft recipients (14). However, no analysis of Vβ CDR3 lengths or of association with specific Jβ was performed in the referred studies. Furthermore, our study not only showed an altered Vβ8 family profile, but also provided a quantitative indication of the strong accumulation of Vβ8 mRNA (increased up to 120-fold), further suggesting a positive selection of the corresponding clones.
Finally, an elegant demonstration that in an experimental model of heart allotransplantation the presence of the chronically rejected grafts could elicit an autoimmune process by peptide molecular mimicry between donor MHC and parenchymatous motives (cardiomyosin) has been recently proposed in mice (58). In addition, the autoreactive responses, triggered by the graft, were, in turn, instrumental in worsening the CR process (58). Furthermore, chronic inflammatory lesions induce a local tissue stress response that may trigger additional T cell activation. Such T cell might react through stress protein-related mechanisms (59). Whether the strong skewed common Vβ-Jβ TCR repertoire we observed in long-term rejected kidneys is related to such a process also remains to be tested.
Acknowledgments
We thank Dr. Rene Duquesnoy (Pittsburgh) for his critical analysis of this work and Dr. Marc Bonneville (Nantes) for comments on CDR3 sequences. In addition, we thank Bernard Besse for his excellent technical assistance with virus DNA amplifications and Joanna Ashton for her help in the editing of the manuscript.
Footnotes
-
↵1 This study was supported in part by the “Fondation Transvie.”
-
↵2 Address correspondence and reprint requests to Dr. Jean-Paul Soulillou. Institut National de la Santé et de la Recherche Médicale, Unité 437, 30 Boulevard Jean Monnet, 44093 Nantes Cedex 01, France. E-mail address: jps{at}nantes.inserm.fr
-
↵3 Abbreviations used in this paper: CR, chronic rejection; CDR3, complementary determining region 3; HHV6, human herpes virus 6; PRA, panel-reactive Ab.
- Received August 27, 1999.
- Accepted November 11, 1999.
- Copyright © 2000 by The American Association of Immunologists
References
In this issue
