Direct measurement of lymphocyte receptor diversity

Abstract

The ability to mount an immune defense against infectious microorganisms and their products, and against tumors is believed to be a direct function of lymphocyte diversity. Because the diversity of lymphocyte receptor genes is >1000-fold more diverse than the entire genome and varies between genetically identical individuals, measuring lymphocyte diversity has been a daunting challenge. We developed a novel technique for measuring lymphocyte diversity directly using gene chips. We reasoned and here demonstrate that the frequency of hybridization of nucleic acids coding for lymphocyte receptors to the oligonucleotides on a gene chip varies in direct proportion to diversity. We applied the technique to detect changes in lymphocyte diversity in mice with known B cell alterations and in persons with known T cell repertoire defects. This approach is the first to provide direct analysis of lymphocyte receptor diversity and should facilitate fundamental study of the adaptive immune system and clinical efforts to assess immunological diseases. In addition, this approach could be more broadly applied, for example to measure diversity of viral quasi-species.

INTRODUCTION

While the total number of lymphocytes in the blood can be directly measured, the diversity of the lymphocyte compartment, on which immunocompetence is based, cannot. In the absence of direct measures of lymphocyte diversity, various indirect means for estimating diversity have been used. For example, antibodies against variable (V) region families have been used to characterize lymphocyte populations by flow cytometric analysis (1,2). Since this approach detects ‘constant’ antigenic determinants shared by many lymphocyte receptor clones, diversity is at best inferred from the result. As another example, nucleic acids encoding lymphocyte receptors can be amplified by PCR using constant region (C) and V family-specific primers (3). Like FACS analysis, this approach does not differentiate between individual clones of the same family and may fail to detect balanced narrowing (or expansion) of the repertoire.

Diversity can also be estimated by spectra typing (also called immunoscope) or by the heteroduplex method (4–6). These methods employ electrophoretic separation of amplified lymphocyte receptor V family PCR products according to the junctional (J) sequences; diversity is inferred from the number and electrophoretic separation pattern of amplified and re-annealed V region PCR products. Spectra typing and heteroduplex methodologies effectively detect clonal expansion within a V family; however, because several thousand V-J family combinations for lymphocyte receptors exist, all V-J combinations cannot be analyzed routinely. Since only a small fraction of V-J combinations are analyzed, the choice of which is random, the actual diversity of the T cell receptor (TCR) repertoire may not be quantified.

Still another means to assess lymphocyte diversity is based on the tenants of limiting dilution analysis and detects the frequency of a given TCR clone (7). This method is quite laborious and is based on the assumption that the frequency of the selected clonotype is representative of the frequency of all clones.

While techniques in current use offer value, they also have limitations, the most vexing of which is the inability to directly measure diversity. The approach we describe addresses this limitation by directly probing the entire population of lymphocyte receptors. This is accomplished by hybridization of all lymphocyte receptor-specific RNAs in a given sample to oligonucleotides on a gene chip; the number of sites undergoing hybridization out of the >400 000 available sites on a gene chip corresponds to the level of diversity. This approach sidesteps analysis of specific receptor families or clones and the limitations associated therein.

MATERIALS AND METHODS

Isolation of RNA

All mouse strains were raised and maintained with protocols approved by the Institutional Animal Care and Use Committee of the Mayo Clinic, Rochester, MN. All human samples were obtained in accordance with the institutional review board of the Mayo Clinic. Spleens harvested from mice were placed in RPMI and pushed through a 70 µm cell strainer. Lymphocytes were isolated from the resulting suspension of splenocytes or from peripheral blood using Ficoll-paque™ (Amersham Biosciences, Piscataway, NJ) gradient. Total RNA was isolated from the lymphocytes using the Qiagen RNeasy Kit™ (Qiagen Inc., Valencia, CA) as per the manufacturer’s instructions. Isolated RNA was resuspended at a concentration of 2 µg/µl.

Generation of lymphocyte receptor-specific cRNA

First-strand cDNA was constructed first as follows. In an RNase-free microcentrifuge tube, 10 µl of total RNA (20 µg) was mixed with 1 µl (100 pmol/µl) of either T7+Cβ (5′-GGCCAGTGAATTGTAATACGACTCACTATAGGGAGGCGGCGGGCTGCTCCTTGAGGGGCTGCG-3′) for T cell receptor analysis or T7+CJ_H4 (5′-GGCCAGTGAATTGTAATACGACTCACTATAGGGAGGCGGGAGGAGACGGTGACTGAGGTTCCTTG-3′) for B cell receptor analysis.

This mixture was incubated at 70°C for 10 min followed by a quick spin and chill on ice. To this reaction, 4 µl of 5× first-strand cDNA buffer (Invitrogen Inc., Carlsbad, CA), 2 µl of 0.1 M dithiothreitol and 1 µl of 10 mM dNTP mix were added and incubated at 37°C for 2 min. Next, 2 µl of SuperScript II Reverse Transcriptase™ (Invitrogen Inc.) was added and the total mixture was further incubated at 37°C for 1 h. Following incubation, the first-strand product was placed on ice. For second-strand synthesis, the following reagents were added to the first-strand product: 91 µl of DEPC-treated water, 30 µl of 5× Second Strand Reaction Buffer (Invitrogen Inc.), 3 µl of 10 mM dNTPs, 1 µl of 10 U/µl DNA ligase, 4 µl of 10 U/µl DNA polymerase I and 1 µl of 2 U/µl RNase H. The reaction was incubated at 16°C for 2 h in a cooling water bath. Following incubation, 2 µl of 10 U T4 DNA polymerase was added and the entire mixture was incubated for an additional 5 min at 16°C. Finally, 10 µl of 0.5 M EDTA was added to the mixture. The completed double-stranded cDNA was purified using a phase lock gel followed by phenol/chloroform extraction. The double-stranded cDNA product was then biotinylated with an Enzo, BioArray High Yield RNA Transcript Labeling Kit™ as per the manufacturer’s instructions. The in vitro transcription (IVT) product (cRNA) was purified using RNeasy spin columns (Qiagen Inc.) as per the manufacturer’s instructions. The purified product was quantified using spectrophotometric analysis applying the convention that 1 OD at 260 nm equals 40 µg/ml of RNA. cRNA was resuspended at a concentration of 1 µg/µl. cRNA was then fragmented to 50–200 bp sizes by combining with 5 µl of 5× fragmentation buffer (Invitrogen Inc.) in 15 µl of water. The mixture was incubated at 94°C for 35 min and put on ice following incubation.

Application of cRNA to the gene chip

Equal amounts of cRNA from different samples were hybridized to U95B gene chips (Affymetrix Inc., Santa Clara, CA). While the ideal gene chip might be constructed using random oligonucleotides, we reasoned that chips containing known but unselected expressed sequence tags from human genes would share less homology with mouse lymphocyte receptor RNA and could be used instead. And, in fact, duplicate experiments performed on U95C chips yielded comparable results, suggesting that a random oligonucleotide chip may not add benefit.

Data analysis

For each gene chip experiment, we obtained raw data corresponding to oligonucleotide location and hybridization intensity. Data were arranged in order of ascending hybridization intensity. The number of oligonucleotide locations with intensity above background (i.e. number of hits) was summed. First, the standard curve was generated (from hybridization of samples with known numbers of different oligos). Next, test samples were assessed and, based on the number of hits, the diversity was extrapolated from the standard curve.

ELISA for detection of anti-keyhole limpet hemocyanin (KLH) antibodies

Mice were immunized by i.p. injection with 25 µg of KLH (Sigma, St Louis, MO) in incomplete Freund’s adjuvant. A boost of 10 µg of KLH was administered 20 days later. After an additional 2 weeks the mice were killed and serum and splenocytes were isolated. Purified KLH [3 µg/ml in phosphate buffered saline (PBS), 50 µl/well] was added to wells of 96-well flat bottom microtiter plates (Nunc-Immuno 96 Micro Well–Maxisorp™; Nalge Nunc International, Rochester, NY). ELISA was developed as described (8). Plates were read using a microplate reader (Power Wave X™; BioTek Instruments, Winooski, VT) and analyzed using KC4 Kineticalc software. Samples were analyzed in triplicate in three independent experiments.

Statistical analysis

Slope and y intercept of the standard curves were compared between experiments using a single factor analysis of variance for a random effects model. Differences were considered significant at a value of P < 0.05. Immune responsiveness was compared between control (C57B1/6) and various mutant mice (QM, J_H^–/–) using unpaired Student’s t-test data. Differences were considered significant at a value of P < 0.05. All gene chip experiments were performed twice; representative data are shown.

RESULTS

As a first test of the concept, we asked whether the diversity of random oligonucleotides is predicted by the number of sites hybridized on a gene chip. Since the variable sequences of the complementarity-determining region of lymphocyte receptors are relatively random, the sequences should be represented by random oligonucleotides. We designed a random oligonucleotide and then inserted random point assignments at specific locations along the oligonucleotide. For example, to generate a sample with ∼10⁶ different oligonucleotides, we synthesized the oligonucleotide with 10 sites of random base assignment (4¹⁰ = 1 048 576). We synthesized samples with 1, 10³, 10⁶ and 10⁹ different oligonucleotides per sample. The oligonucleotides were biotinylated and then 10 µg of each was hybridized to separate gene chips under similar stringency conditions to those for conventional applications. As anticipated, the number of hybridized sites increased with increasing diversity of oligonucleotides (Fig. 1A). Due to the exponential nature of the relationship, the natural logarithm of both variables was taken and plotted to yield a linear relationship (Fig. 1B and C). This relationship represents a standard curve that can be used for assessment of diversity in physiological samples. To our knowledge, this approach is the only means to generate such a curve in the absence of physiological samples of known diversity. To ascertain the reproducibility of this relationship (i.e. the standard curve), we repeated the above experiment six times (Fig. 2). The trend (i.e. slope of the standard curve) was highly reproducible (P > 0.1), however, overall hybridization intensity (i.e. y intercept of the standard curve) varied between experiments (P < 0.05; Fig. 2). This variability requires use of a standard curve for each experiment conducted.

Figure 1.

Establishing the relationship between number of gene chip hybridization sites and sample diversity. Samples consisting of 1, 10³, 10⁶ and 10⁹ different oligos were biotinylated and hybridized to gene chips. Hybridization intensity data were arranged in ascending order. The number of probe locations with intensity above background (i.e. number of hits) was summed and compared to the number of different oligos initially applied to the gene chip (i.e. number of variants). (A) Relationship between number of hits and number of variants. The number of hits increases with the number of variants, indicating that the human gene chip can be used to detect random oligos. (B) Linear relationship between number of hits and number of variants. The natural log of both axes yielded a linear relationship between hits and variants. (C) Visual hybridization of random oligos to gene chips. Scans of the gene chips afford rapid inspection of the ‘hit’ profile.

Figure 2.

Reproducibility of the method for analysis of receptor diversity. Samples obtained as described in Figure 1 were studied in six separate experiments (filled circle, open circle, filled square, open square, filled triangle, open triangle) to test reproducibility. The slopes of the standard curves were the same statistically; however, the y intercept varied from experiment to experiment.

To test whether variations in lymphocyte diversity could be measured directly, we applied the method to the study of B cells in mice. We used murine B cells for this purpose, because diversity of these cells can be measured, at least in principle, through analysis of Ig gene expression and because we have available mutant mice with defined variations in B cell antigen receptor repertoire. We compared diversity of B cell antigen receptors in wild-type (C57Bl/6) mice with the diversity of B cell receptors (BCRs) in quasi-monoclonal (QM) and J_H^–/– mice. The QM mice were generated by gene-targeted replacement of the endogenous J_H elements with a VDJ rearranged region from a 4-hydroxy-3-nitrophenylacetate (NP)-specific hybridoma (9). The κ light chains in these mice are non-functional and, therefore, the knock-in heavy chain can only pair with endogenously rearranged λ chains. All B cells in QM mice start out with the same heavy chain (QM B cells); however, secondary rearrangements and hypermutation change the specificity of B cell receptors (non-QM B cells) (9). Thus, 80% of the peripheral B cells in the QM mice express NP-specific antibodies that contain the same single heavy chain, and the remaining 20% express antibodies expressing diverse heavy chains and diverse antigen specificities. J_H^–/– mice have a targeted deletion of the J_H and J gene segments and, therefore, cannot assemble Ig heavy or κ light chains (10). These animals are B cell deficient, although they do have surface Ig-negative precursor B cells (B220⁺/CD43⁺ pro-B cells) that assemble λ light chain genes at a low level.

To test changes in lymphocyte diversity, we compared gene chip hybridization of B cell heavy chain-specific cRNA from splenocytes of wild-type, QM and J_H^–/– mice. The heavy chain specificity of the cRNA was gained via generation of first strand cDNA from isolated RNA using a J_H4-specific custom primer. A primer to the J_H4 region of the heavy chain was used to detect diversity because in the QM mice all heavy chains encode J_H4 segments. Our results thus compare the diversity of this heavy chain segment rather than all heavy chain regions. Following second-strand DNA synthesis, the double-stranded product was biotinylated via IVT. The IVT product (cRNA) was purified and 10 µg of each sample and 10 µg of each standard were hybridized to individual gene chips. The hybridization intensities obtained from the J_H^–/– mice (which lack B cell receptors) were used to set the background threshold, above which hybridization sites (hits) were counted. Sample diversity was extrapolated from the standard curve. As the results shown indicate, wild-type mice expressed more than 10⁵ (2.8 × 10⁵) different B cell J_H4-positive heavy chain clones and QM mice expressed 3.9 × 10² different J_H4-positive heavy chain clones (Fig. 3A). Thus, as expected, QM J_H4-positive heavy chain diversity was much less than wild-type diversity, though well above background levels.

Figure 3.

Analysis of B cell diversity using the gene chip method. Splenocytes were harvested from 3- to 4-week-old J_H^–/–, QM and WT mice and mononuclear cells were isolated on Ficoll-paque gradients. Total RNA was isolated from the lymphocytes and first-strand cDNA was generated using a primer designed to bind the constant region of the mouse heavy chain J_H4 region plus the T7 polymerase promoter. The custom primer promoted amplification of J_H4 heavy chain-specific RNA only. Equal amounts of the in vitro transcription product (cRNA) from each mouse and standards were hybridized to gene chips and then the chips were stained and analyzed as described in Materials and Methods. (A) B cell heavy chain diversity in mutant mice before and after immunization with KLH. Pre-immunization, WT diversity (filled circle) was more than 2-fold higher than QM (open circle) diversity. Post-immunization, WT diversity (filled circle) decreased (4.0 × 10⁴ different B cell heavy chain clones) while QM (open circle) diversity increased (7.5 × 10³). Background hybridization was established using J_H^–/– RNA (filled square). (B) Immune responsiveness to KLH. An ELISA was used to detect levels of anti-KLH antibodies in the serum following immunization. The QM anti-KLH antibody titer was ∼40% of the wild-type following immunization. *P < 0.05.

We then tested the ability of this system to detect changes in diversity following antigenic challenge with KLH. Following immunization with proteins such as KLH, heavy chain diversity is thought to decrease due to oligoclonal expansion of high affinity clones. In contrast, immunization of QM mice with KLH (an antigen that does not bind with the QM antibody) causes diverse non-QM B cells to expand at the expense of the predominant QM B cells, thereby increasing heavy chain diversity (11). Consistent with these theoretical concepts, J_H4-positive heavy chain diversity in wild-type mice decreased by greater than one order of magnitude, from 2.8 × 10⁵ to 4.0 × 10⁴, and diversity in QM mice increased (7.5 × 10³) following immunization with KLH (Fig. 3A). Immune responsiveness was confirmed by quantitation of anti-KLH Ig in the serum (Fig. 3B).

We next asked whether this approach could distinguish medically relevant conditions in humans via assessment of T cell diversity. Peripheral blood lymphocytes were isolated from two normal individuals (35 and 55 years of age) and from two individuals who had cardiac transplants in infancy (3 and 5 years of age; transplantation conducted at <1 year of age). Those who receive cardiac transplants in infancy undergo removal of the thymus as part of the surgical procedure and depletion of T cells with anti-T cell antibodies (12). These patients would be expected to have a notably contracted TCR repertoire because they lack a source of new naïve T cells. RNA was isolated from these lymphocytes and from Jurkat cells and first-strand cDNA was generated using a primer designed to bind the constant region of the TCR β chain. We could have designed a primer to bind the constant region of the TCR α chain or combined the α and β chain primers to isolate all TCR RNA. However, the majority of lymphocyte diversity is conferred by the β chain (13) and so, in an effort to maintain the simplicity of the system, here we show β chain diversity only. Jurkat cells express only one TCR (Vα1.2Vβ8.1), and so the hybridization intensities of this sample were used to establish the background threshold. Normal human T cell receptor β chain diversity was 4.4 × 10⁶ and 5.1 × 10⁶ (Normal, Fig. 4). These values are consistent with estimates of β chain diversity deduced by other means (7,14), and taken with estimates of α chain diversity and pairing (13) would place overall T cell diversity at minimally 1.1 × 10⁸ (4.4 × 10⁶ × 25 α families). And, as expected, the T cell receptor β chain diversity of the thymectomized/T cell-depleted subjects (2.2 × 10³ and 1.8 × 10²; Thymectomized/T cell depleted, Fig. 4) was more than two orders of magnitude lower than that of normal individuals. To determine whether the assay could detect smaller changes in T cell diversity, we tested an individual with inflammatory bowel disease (25 years of age), which in experimental systems has been associated with decreased T cell diversity (15). We found a decrease in TCR β diversity of more than one order of magnitude in this subject (2.6 × 10⁵; IBD, Fig. 4). Our results thus suggest that human inflammatory bowel disease may be linked with decreased lymphocyte diversity.

Figure 4.

Analysis of human T cell diversity using gene chips. Total RNA was isolated from human peripheral blood lymphocytes or Jurkat cells and first-strand cDNA was generated using a primer designed to bind the constant region of the TCR β chain. Equal amounts of the in vitro transcription product (cRNA) from each sample and standards were hybridized to gene chips and then the chips were stained and analyzed as described in Materials and Methods. Two normal individuals (filled circle, open circle), two thymectomized/T cell-depleted individuals (filled square, open square) and one individual with IBD (filled triangle) were analyzed. The TCR β chain diversities of the normal individuals were 4.4 × 10⁶ and 5.1 × 10⁶, respectively (similar results were found even after diluting the normal sample with Jurkat cells at various ratios up to 20:80), of the thymectomized individuals 2.2 × 10³ and 1.8 × 10², respectively, and of the individual with IBD 2.6 × 10⁵. Background hybridization was established using Jurkat cell hybridization.

DISCUSSION

Here we report the direct measurement of lymphocyte diversity and illustrate potential applications. The system we devised can estimate diversity of the entire lymphocyte repertoire (i.e. all gene segment combinations) at once and avoids the complications of indirect estimates of lymphocyte diversity. With only minor adjustments, this approach is equally capable of measuring B cell and T cell diversity and can be adjusted via primer design to include or exclude lymphocyte receptor subsets. The system is sufficiently simple and effective to allow widespread application.

There is much interest in the relationship between lymphocyte diversity and immunocompetence. Loss of diversity has been implicated in various disease states (7,16,17), and so changes in diversity might be used to track the progression or remission of disease. For example, this approach might be used to monitor immune reconstitution following bone marrow transplantation or intensive anti-retroviral therapy. In these settings, a small number of clones might expand by homeostatic proliferation to yield normal lymphocyte numbers, but diversity might be altered (18). The technique reported here might also be used to track expanded T cell clones or clusters of clones over time based on gene chip hybridization pattern. Certainly, the numbers of individual T cell clones in normal blood are too few to allow clone tracking by any means; however, the expanded T cell clones seen in response to infection, transplantation or homeostatic proliferation might in principle be tracked over time with this method. Preliminary experiments done in our laboratory support this concept (data not shown).

While we have focused on the measurement of lymphocyte receptor diversity, the technique discussed here might also be applied more broadly to the daunting task of quantifying biological diversity of, for example, viral quasi-species. Because persistence of some viral infections such as hepatitis C is positively related to the diversity of the virus (19), quantifying quasi-species diversity may be critical to guide therapeutic choices and prognostic assessments. Diversity of hepatitis C virus quasi-species may be accomplished by generating viral envelope gene-specific cRNA followed by hybridization to gene chips and analysis similar to the one described above for the lymphocyte receptor genes.

The primary challenge of using gene chips to measure lymphocyte diversity is determining the background correction for analysis of extremely narrow repertoires. We are exploring the usefulness of intensity data from non-lymphoid cells for this purpose as well as enhancing our panel of standards to include all single-order variations in diversity (i.e. 10¹, 10², 10⁴, 10⁵, etc.). An additional caveat warrants mention. It is difficult to predict whether the number of hits generated by a diverse oligonucleotide mixture is equivalent to the number of hits generated by the same diversity of lymphocyte receptor cRNA. We controlled for the potential difference in hybridization by including a sample of known diversity (i.e. Jurkat with one TCR and MBT with one BCR) in the generation of the standard curve. Thus the number deduced by this approach, while likely not the actual number of different receptors in a population, is an approximation of the actual number of receptors in a population. The number therefore serves both as an estimate of the receptor diversity within a population and a basis for comparison with other populations.

Overall, the advent of high throughput hybridization technology has made it possible to directly assess lymphocyte receptor diversity and thus perhaps immune fitness. This approach should facilitate fundamental study of the physiology of the adaptive immune system and clinical efforts to assess and follow immunological diseases.