T-Cell Receptor Variable/3 Genes Show Differential Expression in CD4 and CD8 T Cells Michael P. Davey, Mary M. Meyer, Dimitri D. Munkirs, Darcie Babcock, Marcus P. Braun, James B. Hayden, and Antony C. Bakke

ABSTRACT: Studies in transgenic and inbred strains of mice have shown that the critical molecular interactions controlling positive selection involve major histocompatibility complex (MHC), T-cell receptor (TCR), and CD4 or CD8 coreceptor molecules. Correlations have been established between MHC gear products and the percentage of CD4 or CD8 T cells that express specific variable (V) B-gene products as part of the a/3 heterodimer. These studies have important implications regarding potential mechanisms of HLA-linked autoimmune diseases in humans. If similar interactions are required for positive selection in humans, one would predict that the TCR repertoire expressed by mature, peripheral blood CD4 and CD8 T cells would vary. To test this hypothesis the expression of specific TCR

Va-region genes by CD4 and CD8 T cells from healthy individuals was compared using both triple-color flow cytometry and polymerase chain reaction based experimental approaches. The results show that the TCR repertoire does vary as a function of CD4 and CD8 T-cell subsets. Among unrelated individuals certain V# genes were consistently overrepresented in the CD4 population (V#-5.1, -6.7a, and -18); some were skewed to the CD8 population (V#-14) while others showed variable patterns (V~-12 and -17). Deletion of entire V# gene families was not observed suggesting that this is a rare event in humans. Attempts to correlate the expressed TCR rep. ertoire in humans with HLA alleles will require consideration of these differences in expression as a function of subset. Human Immunology 32, 194-202 (1991)

ABBREVIATIONS HPLC high performance liquid chromatography mAb monoclonal antibody MHC major histocompatibility

PCR TCR V

polymerase chain reaction T-cell receptor variable

INTRODUCTION The antigen-specific T-cell receptor (TCR) expressed on the majority of peripheral T cells is a heterodimer consisting of o and/3 chains. The amino-terminus of each chain is encoded by variable (V) region genes that have been sequenced and labeled by numbers corresponding to specific families [1]. An important recent advance in immunology has been the development of monoclonal antibodies (mAbs) that make it possible to From the Department of VeteransAffairs Medical Centerand Oregon Health Sciences University, Portland, Oregon. Address reprint requeststo Michael P. Davey, VA MedicalCenter,P.O. Box 1034, Portland, OR 97207. ReceivedSeptember 14, 1990; acceptedJuly 19, 1991.

194 0198-8859/91/$3.50

identify populations of T cells based on the V region families that they express. Experiments using these reagents to study inbred and transgenic strains of mice demonstrated that both positive and negative selection are dependent on the direct interaction between TCK with major histocompatibility complex (MHC) determinants expressed by cells residing within the thymus [2]. In a series of experiments using T C R a/3 transgenic mice it was shown that for positive selection to occur the M H C type of the animal receiving the transgenes must match the M H C type of the T-cell clones from which the transgenes were derived [3, 4]. In the course of these studies it became apparent that the eventual HumanImmunology32,194-202 (1991) © AmericanSocietyfor Histocompatibilityand Immunogenetics,1991

Va Expressionby T-Cell Subsets

CD4 or CD8 phenotype of a T cell was likewise determined as a result of the interaction of TCR with MHC [3, 4]. In transgenic mice with an appropriate MHC match, it was observed that the resulting peripheral T-cell population that expressed the o~3 transgene was heavily skewed in favor of the CD4 or CD8 type of the original T-cell clone from which the TCR was derived. The same molecular interactions that allow peripheral T cells to behave in a MHC restricted manner also occur within the thymus; i.e., the positive selection of CD4 and CD8 T cells is directed by MHC class II and class I molecules, respectively [2-7]. Positive selection can be blocked with antibodies to TCR, MHC, and CD4 or CD8 molecules [2]. Thus, the interaction of TCR, MHC, CD4, and CD8 coreceptor molecules regulates positive selection, controls differentiation into CD4 or CD8 subsets, and ultimately shapes the peripheral T-cell repertoire. The mechanism by which the inheritance of specific HLA genes increases the risk of developing certain human autoimmune diseases is unknown. Since studies in mice have shown that the peripheral T-cell repertoire is influenced by MHC haplotypes, attempts to correlate the expressed TCR repertoire with specific HLA alleles may provide important insights into the mechanism of HLA-associated diseases. Given the enormous differences in complexity between an analysis involving inbred mouse strains and a random human population, it was necessary to test the feasibility of TCR repertoire analysis in humans on a relatively straightforward question before attempting to correlate V// expression with specific HLA alleles. The purpose of the studies described here was to test the hypothesis that human TCR V48 gene usage would differ between CD4 and CD8 T-cell subsets. Since the eventual CD4 or CD8 phenotype of a T cell is determined as part of the process of positive selection, one would predict that the TCR repertoire expressed by these subsets would not be identical. Indeed, it has been shown that certain V# gene families are skewed to one population compared to the other in several inbred strains of mice [2, 6-9]. To test this postulate in humans we have compared the level of expression of V~8gene families in CD4 and CD8 T cells from unrelated individuals of known HLA type using two different approaches. Using mAbs that recognize specific V~8gene families, a triple-color flow cytometric analysis was performed. In order to increase the number of V~8gene families that could be analyzed, V3 expression was also compared using the polymerase chain reaction (PCR) with oligonucleotide primers specific for 20 different V~8 gene families [10]. The results show that certain V~ families are preferentially expressed by CD4 or CD8 T cells. While the overall pattern of V/3 expression is unique for each individual, among unre-

195

lated individuals certain V~8 genes are consistently expressed to a greater degree in a specific subset. These data are consistent with the concept that the eventual CD4 or CD8 pheuotype of a cell is regulated by MHCTCR interactions. In addition, the data show that when comparing TCR V/3 expression between different individuals, careful attention must be given to the CD4 and CD8 content of the T-cell populations being compared. MATERIALS AND METHODS Cell preparations. Healthy laboratory personnel served as donors for this analysis. HLA typing was performed by standard serologic techniques. CEM and CD8 cell populations were prepared by negative selection from peripheral blood samples. Briefly, mononuclear fractions were collected by isopaque/Ficoll centrifugation, divided into two equal aliquot portions, and incubated with saturating concentrations of anti-CD4 or anti-CD8 mAb (Coulter, Hialeah, FL). The cells were then incubated with goat anti-mouse antibodies coupled to magnetic beads (Dynal, Great Neck, NY) according to the manufacturer's instructions.The preparations were then placed in a magnedc field, superaatant fluids were collected and analyzed by flow cytometry. Cell preparations treated with anti-CD4 contained less than 1q~ residual CD4 cells. Similar results were obtained for anti-CD8 treated preparations.

Oligonudeotide primers. Twenty-two different V/3 primers, designed as previously describ~.d [!91 were synthesized. The Vfl primers exactly match published sequences for the 20 Vfl families previously reported [1] and correspond to sequences located 200-250 base pairs 5' to sequences in the C region (codons 129-134) of the transcript to whkh a Cfl primer was designed. The primers are referred to by numbers corresponding to the Vfl gene family that they were designed to amplify. Two primers were required for the V~ 5 and 13 families. For many of the V# families, subfamilies have been identified based on sequence homology criteria. The names of the primers and the subfamily members that they are predicted to amplify are as follows. V~I: 1.1 and 1.2; V/32: 2.1, 2.2, and 2.3; V~3:3.1 and 3.2; va4: 4.1, 4.2, and 4.3; Va5.1: 5.1; V~5.2/3:5.2 and 5.3; V/36: 6.1, 6.2, and 6.3; V/~7:7.1 and 7.2; VB8: 8.1, 8.2, 8.3, and 8.4; V/39: 9.1; V/310:10.1 and 10.2; V~I 1: 11.1 and 11.2; V/312:12.1 and 12.2; VB13.1: 13.1; V~13.2: 13.2. V#14-20 each amplify the single member reported for these families. This panel of 22 different primers is predicted to amplify at least 38 of the 46 known unique V~ gene families and subfamilies previously reported [10]. Subfamilies not pcedicted to be amplified include V# 5.4, 6.4, 6.5, 6.6, 6.7, 6.8, 6.9, and

196

18.2. V//13.1, 13.2, and 14 have also been called 12.3, 12.4, and 3.3, respectively [1]. The 3' C//primer recognizes both C//I and C//2. Two additional primers were designed that correspond to sequences located 237 base pairs apart within the 3' carboxy-terminus region of TCR//-chain transcripts. This primer pair, which recognizes sequences in C//1 and C//2, was used to amplify a fragment that served as an internal standard for total TCR//-chain content of each cDNA preparation. The sequences were: 5'-CGCTGTACCGTCCAGTTCTA-3' (codons 211-217) and 5'-TCTC'ITGACCATGGCCATCA-3' (codons 284-290). All V//values were normalized with the value obtained with this C//primer pair.

V// analysis by PCR. RNA was prepared by ultracentrifugation in guanidine isothiocyanate/cesium chloride. First strand cDNA was generated from 2 /zg of total RNA using random hexanucleotides and reverse transcriptase in a reaction volume of 50/~l. PCR was performed using a master mix containing reaction buffer (Perkin-Elmer-Cetus), nucleotides (final concentration 0.2 raM), Taq polymerase (Perkin-Elmer-Cetus, 2.5 units per reaction), and the C// primer (0.3 v.M) corresponding to codons 129-134. The master mix was split into two equal portions and added to the CD4 or CD8 cDNA preparation (1/zl of cDNA per reaction). An aliquot portion of each cDNA/master mix preparation was then placed into tubes containing a unique Vb primer (0.3/zM). The final reaction volume was 100/zl. For each cDNA preparation a separate set of PCR reactions were performed using the primer pair homologous to the 3' carboxy terminus of//-chain transcripts. PCR was performed using a Coy TempCycler (Ann Arbor, MI). The cycle profile was an initial incubation at 94°C for 2 minutes, followed by 30 cycles of 94°C for 1 minute, 55°(2 for I minute, and 72°C for 1 minute. The profile concluded with a 5-minute incubation at 72°C. Quantitation of amplified fragments. Amplified DNA fragments were quantitated using high performance liquid chromatography (HPLC). Separation was achieved with a TSK-DEAE-NPR column (Perkin-Elmer, 35 × 4.6 mm internal diameter) using a Hewlett Packard liquid chromatographic system (HP1090M) with a diode array detector and analytical workstation (HP79994A). The gradient was prepared and run as previously described [11]. Briefly, mobile phase A was 1M NaCl, 25 mM Tris-HCl, pH 9.0, and mobile phase B was 25 mM Tris-HCl, pH 9.0. The gradient run was: 30%-40% A in 0.1 minute, 40%-52% A in 2.9 minutes, 52%-62% A in 7 minutes, 60%-100% A in 0.5 minute, 100% A for I minute, 100%-30% A in 0.2 minute, followed by a 5-minute postgradient reequilibration at 30% A. The

M.P. Davey et al.

detector was set at 260 nm, band width 4nm, and the flow rate was 1.0 ml/min. Measurement of area under each peak was determined using Hewlett Packard integrating software.

Flow cytometric analyses. Analyses were performed on a FACScan or a dual laser FACStar+ (Becton Dickinson). The following unconjugated mouse anti-V//mAbs were used (T-Cell Sciences): cz//V(a) (specific for V//5.1), BV6(a) (mainly specific for VB6.7a [12]), 3V8(a) (recognizes V//8), and//V12(a) (recognizes V//12, subfamily specificity unknown). Fluorescein isothiocyanate (FITC) conjugated goat anti-mouse IgG (Tago) was used as a secondary reagent. Additional antibodies included: phycoerythrin (PE)-conjugated anti-CD4 and anti-CD8, allophycocyanin (APC)-conjugated streptavidin and perCP-conjugated anti-CD3 (all from Becton Dickinson), and biotin-conjugated anti-CD2 (Gentrak). PerCP is a new fluorochrome detected in the FL3 channel of the FACScan (deep red, > 630 nm) and was the kind gift of Andrew Blidy (Becton Dickinson). RESULTS

Quantitation of DNA fragments by HPLC. The most common method used to quantitate amplified DNA fragments involves gel electrophoresis and radioisotopes. In our hands several different variations of this approach resulted in significant intraassay and interassay variation. Therefore, we chose to quantitate DNA fragments generated by PCR using a recently described HPLC approach [11]. Initially, different parameters of the detection system were analyzed. Using the ion exchange column and elution gradient described above, DNA fragments that vary in size up to approximately 1000 base pairs can be identified as separate peaks (Fig. 1). The gradient proved to be ideal for analyzing the 200- to 250-base pair fragments generated by PCR using the different TCR//-chain primers (Fig. 2). Note in the examples shown, which are representative of all of the V// primers, that a single peak was obtained. Thus, under the PCR conditions described, DNA fragments resulting from nonspecific amplification were undetectable. To assess the intraassay variation of the detection system, five separate PCR reactions containing a 236base pair DNA fragment were pooled and five successive HPLC injections were performed. The average area under the curve was 221 -+ 7.85, resulting in a 3.5% error. Interassay variation of the detection system was assessed by repeat injections of 15 ng of a 194-base pair fragment at six different times over a 1-month period. The average area under the curve was 20.23 -+ 0.82, resulting in a 4% error. The area under the curve is exactly proportional to the amount of fragment loaded

VB Expression by T-Cell Subsets

197

Ij 20 18

H

18 14 12

B

TI

C

=

(

~3

i

FIGURE 1 Elution profile of PhiX DNA digested with Haclll. A commercially availabb preparation of DNA (BRL) was applied to a TSK-DEAE-NPR column and eluted as described in Materials and Methods. Absorbance units are shown on the y-axis, elution time is shown on the x-axis. The size of the doublestranded DNA fra~nents (in base pairs) are follows: peak A, 72; B, 118; C, 194; D, 234; E, 271 and 281; F, 310; G, 603; H, 872; i, 1078; J, 1353.

G

F

1

)

more complete spectrum of V~ usage [9, 14, 15]. It has been demonstrated that V # expression can accurately be studied by analyzing T-celt m R N A transcripts [7t. A recently described PCR assay [ 10] capable of detecting transcripts encoding most of the V 3 genes described to date was used here in conjunction with the HPLC detection system. To assess the degree of variability

on the column (Fig. 3). The sensitivity of the HPLC system permits approximately 10 ng of a 236obase pair fragment to be detected (Fig. 3).

Comparing V~ usage by PCR. Since there are few mAbs available that recognize human V ~ gene products; indirect approaches have been developed in order to study a

2511 1

FIGURE 2 Elution profile of DNA fragments amplified with oligonucleotide primers specific for TCR V~8gene families. PCR reactions were performed on cDNA prepared from peripheral blood lymphocytes of a healthy donor and analyzed by HPLC. The xand y-~xes are as in Fig. 1. The V~ primer and elution time is shown above each peak. The V~6 fragment (top) is 195 base pairs, V~13.2 (bottom) is 249 base pairs. The peaks prior to 3 minutes represent unincorporated primers and nucleotides.

®

:i

1l 2

4 Time

(mlm,]

4 Tlrne

(mln,)

e

8

G

8

25

20'

15

10

5

oZ 2

198

M.P. Davey et al.

lo 4 -

30 / o!

. . . . . . .

0

25

50

75

, . . . . . . . i00 12.5 150 175 20(i

cone

Olo2_

Cng)

1o -11 !

FIGURE 3 Quantitation of a known amount of DNA by HPLC. Serial dilutions of a known amount of PhiX DNA digested with Haelll were applied to the ion exchange column and eluted. The area under the 236-base pair peak (yaxis) was calculated using the Hewlett Packard integrating program and is graphed as a function of concentration (x-axis) of the 236-base pair fragment in the digest (R 2 = 1). A similar analysis of the 196-base pair fragment in the digest likewise produced a straight line (R 2 = 1, data not shown). present in the reverse transcriptase reaction, three identical reactions were prepared using R N A purified from peripheral blood T cells of a healthy individual. V 3 6 and the C/3 internal standard were then amplified in triplicate PCR reactions using c D N A fi'om each preparation. The means of the normalized values were 0.44, 0.47, and 0.46, indicating that the reverse transcriptase assay produces a relatively uniform number of c D N A templates under the synthesis conditions utilized. In the original description of this PCR approach to study the expressed V/3 repertoire, it was shown that the relative values obtained with PCR correlated with the frequency of T cells expressing V/3-specific gene products as determined by analysis with V/3-specific mAb [10]. A similar flow cytometric-PCR comparison was performed here. Since anti-V/3 mAbs react with a small percentage of peripheral blood T cells, we performed a three-color flow cytometric analysis [13] whereby CD2- or CD3-positive T cells were analyzed for the simultaneous expression of a specific V/3 family and CD4 or CD8. O u r approach was to gate on CD2- or CD3-positive cells and collect 75,000 events within this window for the expression of the other determinants. The feasibility of this approach was first determined by mixing serial dilutions (10%, 5%, 2.5%, 1.25%, and 0.625%) o f a V/38.1-positive T-cell line (HPB-ALL) in a non-V/3-expressing T-cell line (HUT-102). In three separate experiments performed in triplicate, the expected percentage of V/38.1-positive cells was reproducibly detected (data not shown). Importantly, there was a significant difference in the measured values for mixtures containing 1.25% and 0.625% V/38.1-positive T cells (p < 0.0001, Student's t test). An example of this three-color approach using sorted peripheral blood T cells is shown in Fig. 4. By setting an electronic gate on

!

101

I

102 103 anti-V88

104

lo4-

103 -

o, °~

0

102 -

101-

0

I

1@

I

1@ 104

anti-V88 FIGURE 4 Triple-color immunofluoresence analysis of V~08expression. Peripheral blood mononuclear cells from donor 1 were sorted by negative selection into CD8 (top) and CD4 (bottom) enriched populations. Cells were stained in sequence with unconjugated anti-Vfl8, FITC-conjugated goat anti-mouse, either PE conjugated anti-CD8 (top) or anti-CD4 (bottom), and perCP-conjugated anti-CD3. The flow cytometric analysis was performed on a FACScan by first gating on small lymphocytes that were positive for CD3, followed by counting 75,000 CD3 positive events per tube for the expression of V~08and CD4 or CD8. The x- and y-axes represent the log of green and red fluorescence, respectively. Based on controls performed using these reagents individually, the contour plots have been divided into quadrants that identify cells negative for CD4 or CD8 and V~08(lower left), positive for CD4 or CD8 but negative for V/38 (upper left), positive for both CD4 or CD8 and V//8 (upper right), and positive for VB8 but negative for CD8 or CD4 (lower right). The lower right quadrant in each panel is clear since this analysis was performed on sorted cells. The lower left quadrants represent CD3+ cells that lack both CD4 and CD8 and are presumed to be y8 T cells as previously shown l 13].

V~ Expression by T-Cell Subsets

199

ences in the number o f V 3 1 2 subfamilies recognked by the two techn/ques. It is not known (T-cell Sciences, personal communication) if the anti-Vf112 mAb recognizes both V~-12.1 and 12.2. Both subfamilies are predicted to be recognized by the Vf112 primer. MAb to Vfl5.2/.3 (which may recognize Vb5.4 as well) and Vfl6 (which predominantly recognizes V~6.7) were not included in this analys/s.

A o.4

~ o.2 ol

oo

~.i

n

12

Vll Farail~

R=*LS

5.1

a=O.~,S

x

a=I.Z~'

12

VI~ Family

FIGURE 5 Comparison of V~8expression analyzed by both PCR and flow cymmetry. Peripheral blood mononuclear cells from donor i were sorted into CD4- (hatched bars) and CD8enriched (open bars) populations. A portion of each sorted population was analyzed in triplicate by PCR (A) using the primers shown on the x-axis. The normalized PeR value obtained for each primer is shown on the y-axis. Error bars represent standard deviation. A statistically significant difference in expression between subsets, calculated by the Student's t test, was found for V~-5.1 and -8 (p < .001). V~12 expression did not vary between subsets (p -~ .566). At the top of each pair of columns the ratio (R) of the PCR value obtained for the CD4 compared to the CD8 subset is shown. The remainder of each sorted population was analyzed by flow cytometry (B) as described in Fig. 4. The anti-V~ antibodies are shown on the x-axis. The percentage of CD4 or CD8 T cells that coexpress the specific V~ antibody is shown on the y-axis. The ratio (R) of CD4 to CD8 T cells simultaneously staining with each Vfl antibody is shown at the top of each pair of columr.s. These data were obtained by counting 75,000 CD3+ events per sample. A repeat three-color analysis counting 25,000 events per sample for V#8 and Vf112 resulted in a ratio of 0.74 and 1.17, respectively.

gfl usage by CD4 and CD8 subsets. CD4- and CDSenriched cell populations were prepared from a heahhy individual (donor no. 1), R N A extracted, c D N A synthesized, and PCR performed using the complete panel of Vfl primers (Fig. 6). Based on this survey, specific Vfl families were selected for further analysis (Fig. 7). A statistically significant difference favoring e x p r e ~ o n by CD4 cells was found for V#-2, -5.1 (Fig. 5), and -18. Expression favoring the CD8 subset was found for Vfl-3, -8 (Fig. 5), -14, -16, and -17. Other Vfl gene families were expressed to the same degree between subsets (e.g., V#-4, -6, -20, Fig. 7). Donors 2 and 3 were analyzed using the complete panel of Vfl primers and certain Vfl primers were selected for further analysis (Fig. 8). Vfll4 expression ~gain favored the CD8 population from donor 2. This trend was suggested but not statistically significant for donor 3. Both donors 2 and 3 showed preferential V~18 expression in the CD4 subset. In contrast, differences in expression were noted between these two donors for the subset that favored Vfl17 expression. These data indicate that for unrelated individuals, certain V3

FIGURE 6 Comparison of TCR V~ arnplification from CD4 (hatched bars) and CD8 (open bars) T cells from donor 1. PCR using Vfl geae family specific primers (x-axis) was performed on cDNA obtained from CD4- and CDS-enriched T-cell preparations. The y-ax~sshows the area under the curve for each peak normalized to the Cfl standard as described in text.

o.6o,

o,4o C D 3 + cells, it was possible to identify a discrete population of cells that were positive for both Vfl8 and CD8 (Fig. 4, top) or CD4 (Fig. 4, bottom). A direct flow cytometry-PCR comparison is shown in Fig. 5. An excellent correlation was found between the two techniques for VflS. 1 and Vfl8 expression. The ratio obtained for Vfll2 was different between the two techniques. This discrepancy may be the result of differ-

~. 0:0

~co VB ~

Fa~ly

M . P . Davey et al.

200

ore)

TABLE 1 Comparison of HLA type and Vfl-5.1, -6.7, and -12 expression by CD4 and CD8 T cells

.,~= i L

HLA Donor no. O.31)

~

0.20

i m

:

i)-~tl)1

1 2 3

p=