The Journal of Immunology, 2002, 169: 3736-3743.
Copyright © 2002 by The American Association of Immunologists
T Cell Receptor-
Allele-Specific Selection of V1/V
4 Cells in the Intestinal Epithelium1
Kalliopi Grigoriadou,
Laurent Boucontet and
Pablo Pereira2
Unité du Développement des Lymphocytes, Center National de la Recherche Scientifique, Unité de Recherche Associée, Institut Pasteur, Paris, France
 |
Abstract
|
|---|
Previous genetic analyses have shown that the relative
representation of subsets of 
intestinal intraepithelial
lymphocytes (i-IELs) is influenced by genes linked to the TCR,
TCR, and MHC loci. Here, we have analyzed V-gene use in
i-IELs from C57BL/6 (B6) and C57BL/10 (B10) mice and from their
F1 and F2 progenies with a larger panel of
V- and V-specific mAbs and have shown that the influence of
TCR-linked genes operates at two levels: one influencing the
representation of V1 (or V7) i-IELs and other acting specifically
on the V1/V4 i-IEL subset, which represents 3% and 15% of the
i-IELs in B6 and B10 mice, respectively. Analysis of mice
transgenic for a rearranged V1J4C4 chain of B6 origin
demonstrated that the TCR-linked genes influencing the
representation of the V1/V4 i-IEL subset are the structural genes
of TCR chains. This influence is allele specific and cell
autonomous, as evidenced by the different behavior of V1/V4 cells
bearing either parental allele in F1 mice. The
representation of V1/V4 cells among thymocytes is similar
in B6 and B10 mice, demonstrating that the V4 chain can pair well
with both alleles of the V1J4C4 chain and strongly suggesting
that a cellular selection mechanism is responsible for the observed
differences. The V1-J4 junctional amino acid sequences of B6
V1/V4 i-IELs are diverse but display less variation in length
than those found in similar cells from B10 mice, indicating that B6
V1/V4 cells are the target of this cellular selection
event.
|
Introduction
|
|---|
All
throughout its life span, an 
T cell is submitted to selective
pressures based on the specificity of its TCR. During its intrathymic
differentiation, a combination of positive and negative selection
ensures that a mature T cell can recognize peptide Ags presented by
self-MHC molecules without overtly reacting with self-Ags
(1, 2, 3, 4). Once in the periphery, naive T cells need
continuous interaction with self-MHC molecules to survive and compete
with peripheral resident cells (5, 6, 7). Encounter with Ag
in the appropriate context will expand and differentiate Ag-specific
cells, whereas clearance of the Ag is accompanied by a drastic
elimination of the specifically activated cells by apoptosis (8, 9).
Several lines of evidence suggest that T lymphocytes also
undergo selection, although the mechanisms remain quite elusive.
Studies with mice transgenic
(Tg)3 for a TCR-
specific for the MHC class Ib molecule T10/T22b
have provided evidence for both positive and negative selection of
cells (10, 11, 12, 13). However, cells recognizing
MHC Ags appear to be the exception rather than the rule. Thus, in
studies of mice lacking the 2-microglobulin
molecule or MHC class II Ags, the presence at normal levels of
cells in various lymphoid organs has rather ruled out a strict and
direct role of MHC molecules in the development of cells
(14, 15). Furthermore, the frequency of cells
recognizing the molecule T10/T22b is 
0.5%
(16).
To analyze the impact of TCR selection in cells and as an
important step in the search for endogenous ligands for cells,
we (17) and others (18, 19) have used
available V- and V-specific mAbs to study the representation of
different T cell subsets in different organs and in different
strains of mice. In the intestine, the presence of a high frequency of
cells expressing the V4 chain correlated with the expression of a
gene closely linked to the I-E region of the MHC class II locus
(18), although additional experiments with mice Tg for the
I-E molecule showed that I-E expression was not sufficient to dictate
the V4-high phenotype (20, 21). Similarly, an influence
of MHC-linked genes in the repertoire of TCRs expressed by
intestinal intraepithelial lymphocytes (i-IELs) was also evident in
analyses of recombinant inbred strains generated from
C57BL/6 (B6) and DBA/2 founders, although cells not
expressing the V4 chain appeared to be the targets of the selective
event (17). In these experiments, and in similar
experiments performed in the same recombinant inbred strains but
analyzing the representation of different T cell subsets among
splenic cells (19), the influences of genes linked
to the TCR and TCR loci were also evident.
Although these experiments were interpreted as indicative of cellular
selection mechanisms operating on cells, they could not
precisely define the cell target of the selection. This was mostly due
to the limited number of V and V genes commonly used by
cells to form their TCRs and to the fact that only one V-specific Ab
was used in those experiments. Thus, an increased or decreased
representation of a given subset is always compensated by decreased or
increased representation of other subsets, respectively. Therefore,
differences in a given subset may result from a selective mechanism
operating directly on this subset or may reflect a compensatory
mechanism to a selective event acting on a different subset.
Furthermore, structural mechanisms rather than cellular selection
mechanisms may be at the basis of the influence of the structural genes
for TCR and in the representation of T cell subsets.
To circumvent these problems and to analyze further the role of TCR
polymorphism in the representation of i-IEL subsets, we have
used a larger panel of V-specific mAbs that we have recently
produced (22) to analyze the V/V usage by
i-IELs from B6 and C57BL/10 (B10) mice and their
F1 and F2 progenies. B6 and
B10 are genetically related strains of mice (differing by <10% of
their genome) (23), which share the same MHC and TCR
loci but differ in their TCR haplotype (24, 25).
|
Materials and Methods
|
|---|
Mice
C57BL/6JIco (B6), BALB/cByJIco, 129/SvPasIco, CBA/JIco,
DBA/2Jico, and FVB/Nico were obtained from Iffa Credo (L’Abresle,
France). C57BL/10SnJ (B10), C57BL/10J, and B10.D2/nSnJ mice were
obtained from The Jackson Laboratory (Bar Harbor, ME). C3H/HePas, A/J,
DBA/1J, CBA/N, and NOD/Lt were obtained from the Pasteur Institute
(Paris, France). AKR/OlaHsd, NZB/OlaHsd, NZW/OlaHsd, SJL/JHanHsd,
SWR/OlaHsd, and 129/OlaHsd were obtained from Harlan (Gannat, France).
B6 mice Tg for a rearranged V1J4C4 chain have been previously
described (26, 27). (B10 x
B6)F1 hybrid mice (B10B6F1)
and B10B6F2 mice were produced in our animal
facilities. All animals were used between 6 and 12 wk of age.
Cell preparation and cultures
Single-cell suspensions were prepared from thymus and inguinal
and axillary lymph nodes (LNs) according to standard procedures.
CD4-CD8- double negative
(DN) thymocytes were prepared by complement-mediated killing as
described (28), and i-IELs were prepared as described
(29). To expand cells, total LN cells (5 x
106 cells/ml) from individual mice were cultured
in 24-well plates previously coated with 10 µg/ml anti-C mAb
(3A10) in 2 ml of either DMEM or RPMI 1640 with Glutamax-1 (Life
Technologies, Gaithersburg, MD) supplemented with sodium pyruvate,
5 x 10-5 M 2-ME, nonessential amino acids,
and antibiotics (all from Life Technologies), 10% FCS (Boeringer
Mannheim, Meylan, Germany) and 100 U/ml of mouse rIL-2. After 3 days
cells were further expanded for 2–4 days in flasks in the presence of
rIL-2.
Abs, staining, and cell sorting
Anti-CD4 (RL.174), anti-CD8 (HO 2.2), anti-C (3A10),
anti-V1 (2.11), anti-V6.3/4 (clone 9D3)
(27), anti-V7 (F2.67), anti-V6B (F4.22;
V6B refers to the Vd6 genes that have >90% identity at the
nucleotide level with the previously described p
12)
(30), and anti-V5 (F45.145) were prepared and used
as described (22). These V Abs were described by
Pereira et al. (22). The 7C10 mAb was obtained in the same
fusion as 9D3 (27) and its specificity was partially
described by Azuara and Pereira (31). PE- and
biotin-labeled anti-V4 (GL2) and PE-labeled anti-V6.2/6.3
(8F4H7B7) were purchased from BD PharMingen (San Diego, CA).
Cells (105–106) were
incubated in staining buffer (PBS, 3% FCS, 0.1%
NaN3) with the indicated labeled mAbs for 30 min
on ice and were washed twice. When biotin-conjugated Abs were used, the
cells were further incubated with PE-labeled streptavidin (Southern
Biotechnology, Birmingham, AL) or allophycocyanin-labeled streptavidin
(BD PharMingen) for another 15 min on ice. After two washes, cells were
analyzed using a FACScan or a FACSCalibur flow cytometer (BD
Biosciences, Mountain View, CA). Dead cells were gated out by their
forward and side scatter profiles and by their staining with propidium
iodide. Data were analyzed using the CellQuest program.
For FACS sorting, i-IELs and DN thymocytes were prepared as above,
incubated with the appropriate mAb as described above, and sorted in a
FACStarPlus (BD Biosciences). Purity of the
sorted populations was >98%.
PCR, cloning, and sequencing
Total cellular RNA from FACS-sorted cells was extracted with
RNA-B (Bioprobe Systems, Montreuil, France). cDNA was synthesized as
described by Azuara et al. (28). The sequences of the
V1-, C-, V4-, C-, and FAM-labeled J1 primers are
described by Azuara et al. (28). The sequence of the
FAM-labeled J4 primer used was CAAATATCTTGACCCTGA and that of
the C4 was CTTTCCAATACACCCTTA. PCR and run-off reactions were
performed as in Refs. 28 and 32 . The TOPO TA
cloning kit (Invitrogen, San Diego, CA) was used for cloning and the
ABI PRISM dye terminator cycle sequencing ready reaction kit
(PerkinElmer, Wellesley, MA) was used for sequencing according to the
manufacturer’s instructions. The C primer was used to sequence the
V1-J4 junctions.
Genetic and statistical analyses
Male and female B10B6F2 mice (n =
83) were separated into three groups according to their TCR
haplotype as defined by the anti-V1 and 7C10 mAbs. Mice having
all of their V1+ cells stained by the 7C10 mAb
were defined as having both TCR loci of B10 haplotype. Mice in which
a fraction of the V1+ cells were stained by
the 7C10 mAb were considered to have inherited one TCR allele from
the parental B10 strain and the other from the B6 strain. Mice whose
V1+ cells did not stain with the 7C10 mAb were
defined as having inherited both TCR alleles from the
parental B6 strain. The TCR haplotype thus defined was confirmed
by the analyses of the same F2 mice with the D13
Mit3 marker. This marker is polymorphic between the B6 and the B10
strains (23) and maps at the same genetic distance from
the centromere as the TCR locus. Consistent with this predicted
genetic localization, only one recombinant between the D13 Mit3 marker
and the TCR haplotype defined by the 7C10 mAb was found among the 83
F2 animals tested. These data provided genetic
evidence indicating that the gene encoding the protein recognized by
the 7C10 mAb is located in the vicinity of the TCR locus.
Polymorphism at the D13 Mit3 marker was analyzed by PCR as described
(31). For each of these three groups the mean values
± SD of the percentages of each i-IEL subset were calculated
and compared by ANOVA followed by unpaired Student’s t
tests using the StatView program.
|
Results
|
|---|
Representation of different i-IEL subsets in B6, B10,
B10B6F1, and B10B6F2 mice
To estimate the relative proportion of different V/V i-IEL
subsets in B6 and B10 mice, we performed three-color immunofluorescence
analysis with mAbs recognizing the C chain together with available
mAbs specific for different V (V1 and V7) and V (V4,
V5, V6.3, and V6B) chains (22). The V1 and
V7 mAbs recognize >85%, whereas the four V-specific Abs
recognize 70% of the i-IELs in B6 and B10 mice. A
representative staining experiment of B6 and B10 i-IELs is shown
in Fig. 1
, whereas the relative
proportions of each V/V subset defined by these mAbs in B6
(n = 32), B10 (n = 24), and
B10B6F1 (n = 18) mice are
compared in Fig. 2, A–C. Similarly to DBA/2 i-IELs
(17), B10 i-IELs contain a relatively high
proportion (>15%) of V1/V4 cells, whereas the same cell subset
represents only 2% of the i-IELs in B6 mice (Fig. 2B, top left panel). This different
representation of the V1/V4 subset between the B6 and B10 mouse
strains is likely to result from two mechanistically different events.
One is a V-independent event influencing the relative representation
of cells bearing different V chains in a strain-specific manner. The
existence of such an event is supported by the fact that the
representations of the V1/V5 and of the V1/V6 subsets are
also higher in B10 than in B6 mice (Fig. 2B, top
panels). Accordingly, B10 mice contain a higher proportion of
V1+ i-IELs than do B6 mice (Fig. 2A). The second event operates specifically on V1/V4
cells and is readily evident when the representation of the different
V1/V subsets in the parental strains are calculated as a fraction
of the total V1+ i-IELs present in each strain
(thus minimizing the differences in the representation of the
V1/V4 subset due to the V-independent mechanism previously
defined). Thus, whereas the representation of the V1/V5 and the
V1/V6 subsets among V1+ i-IELs in B6 and
B10 mice are comparable, that of the V1/V4 subset is threefold
higher in B10 than in B6 mice (Fig. 2C, top
panels).
View larger version (59K):
[in this window]
[in a new window]
|
FIGURE 2. Representation of different T cell subsets in i-IELs from B6,
B10, B10B6F1, and B10B6F2 mice. i-IELs from B6
(n = 32) and B10 (n = 24) mouse
strains, B10B6F1 hybrids (n = 18), or
B10B6F2 progeny (n = 83) either were
incubated with FITC-labeled anti-V and biotin-labeled
anti-C or anti-V mAbs followed by streptavidin-PE and
analyzed in a FACScan or were labeled as described in Fig. 1 and
analyzed in a FACSCalibur. A and D,
Fraction of + i-IELs expressing the V1 or the
V7 chains in the indicated strains (A) or in
F2 animals separated on the basis of their TCR haplotype
(D). B and E, Fraction of
+ i-IELs expressing the V1 (top) or
the V7 chain (bottom) together with the indicated
V chains. C and F, Fraction of
V1+ (top) or V7+
(bottom) i-IELs expressing the indicated V chains.
Results are shown as the mean ± SEM
(A–C) or as the mean ± SD of the
F2 animals grouped according to whether they have inherited
their TCR locus from the B6 parental strain (B6), the B10 parental
strain (B10), or both (F1)
(D–F). Statistical analyses were
performed by ANOVA followed by Student’s t tests.
***, p < 0.0001; **, 0.0001 <
p < 0.001; *, 0.001 <
p < 0.01.
|
|
For both phenotypic traits (levels of V1 i-IELs and of V1/V4
cells), the frequencies observed in B10B6F1
animals were intermediate between the two parental strains (Fig. 2),
indicating that nonrecessive B10 genes are responsible for high V1
and V1/V4 phenotypes. Furthermore, both phenotypic traits are
genetically controlled by an element(s) physically linked to the TCR
locus, as indicated by the correlation between the levels of V1 and
V1/V4 cells and the inheritance of the TCR haplotype observed
in B10B6F2 (n = 83) mice (Fig. 2, D–F). Thus, F2 animals
that inherited two B10 alleles, two B6 alleles, or one allele from each
parental strain at the TCR locus show levels of V1 and
V1/V4 i-IELs similar to those found in B10, B6, and
B10B6F1 mice, respectively (Fig. 2).
The representation of the V7/V i-IEL subsets in the two parental
strains is not exactly the reciprocal of that of the V1/V
subsets, suggesting that additional "selective" events operate
specifically on V7/V cells (likely V7/V6B and/or
V7/V4 subsets) (Fig. 2, B and C,
lower panels). The representation of both subsets in
B10B6F1 animals is intermediate between B6 and
B10 mice, similar to that of the V1/V subsets. In contrast, their
representation is not influenced by an element(s) linked to the TCR
locus, further indicating that their differences result from different
selective events (Fig. 2, E and F, lower
panels). Such selective events appear to be specific to the B10
substrain used in these experiments (C57BL/10SnJ) because the
representation of the V7/V subsets in C57BL/10J mice was
comparable to those found in B6 mice (data not shown). The mechanisms
responsible for the different representation of V7/V subsets
between B10 and B6 were not studied further.
Altogether, these experiments extend previous observations on the
genetic influence of an element(s) closely linked to the TCR locus
in the representation of i-IEL subsets (17). They
demonstrate that this influence operates at two different levels: one
influencing the representation of V1 (or V7) i-IELs and the other
acting specifically on the V1/V4 i-IEL subset. The following
experiments were aimed at characterizing the selective mechanisms
responsible for the different representation of the V1/V4 i-IEL
subset in B6 and B10 mice.
Evidence for V1 allele-dependent selection of the
V1/V4 i-IEL subset
Obvious candidate genes that could influence the representation of
the V1/V4 i-IEL subset are the structural genes of TCR chains.
If that were the case, the different levels of V1/V4 i-IELs
observed in B6 and B10 mice must relate to polymorphism at the regions
coding for their V1J4C4 chains, implying that the B6 and B10
V1J4C4 alleles must behave differently with regard to whatever
mechanism is responsible for high or low levels of V1/V4 i-IELs.
Alternatively, the representation of the V1V4 i-IEL subset may be
influenced by a gene(s) unrelated to the structural genes of TCR
chains but closely linked to them. The analysis of the V repertoire
expressed by V1+ i-IELs isolated from B6 and
B10B6F1 mice Tg for a rearranged V1J4C4
chain of B6 origin (Tg-) may distinguish between these two
alternative possibilities, because the vast majority of the
cells in these mice express the transgenic chain (Ref. 31
and data not shown). Thus, if the former hypothesis is correct, the
levels of V4+ i-IELs among
Tg+ i-IELs would be expected to be similar in B6
and in B10B6F1 Tg- mice, and it should compare
well with the fraction of V4+ cells among
V1+ i-IELs found in normal B6 mice. In
contrast, if the latter hypothesis is correct, the levels of
V4+ i-IELs among Tg+
cells would differ in B6 and B10B6F1 Tg- mice
and should compare well with the fraction of
V4+ cells among V1+
i-IELs found in normal B6 and B10B6F1 mice,
respectively. As shown in Fig. 3, the
V1/V4 subset represents 9% of the
V1+ i-IELs in B6 and
B10B6F1 Tg- mice as well as in wild-type B6
mice, whereas the same population represents >20% of the
V1+ i-IELs in non-Tg
B10B6F1 mice. These results strongly suggest that
the allelic form of the V1J4C4 chain dictates the
representation of the V1V4 subset among i-IELs in these
strains.
The previous statement would predict that the intermediate phenotype
observed in F1 mice results from a different and
independent behavior of the V1J4C4 chains encoded in the
parental TCR haplotypes. We have recently produced a mAb (7C10) that
appears to recognize the V1J4C4 chain present on the
TCRa haplotype (i.e., B10) but not the
V1J4C4 chain present in any other TCR haplotype tested,
including the B6 TCRb haplotype (Ref.
31 and Fig. 4). To further
characterize the specificity of this mAb, we isolated seven
V1-bearing hybridomas (four 7C10+ and three
7C10-) originated from (B6 x
DBA/2)F1 thymocytes and sequenced their
functional V1C4 chains. All four 7C10+
hybridomas expressed a functionally rearranged V1C4 chain of the
parental DBA/2 allele, whereas the three 7C10-
hybridomas expressed a V1C4 chain of the parental B6 allele
(p < 0.02, according to a Spearman rank
correlation coefficient test with the null hypothesis that the two
variables, reactivity to the 7C10 mAb and expression of the
V1a allele, are independent of each
other).
View larger version (19K):
[in this window]
[in a new window]
|
FIGURE 4. The 7C10 mAb recognizes the V1J4C4 chain present in the
TCRa haplotype. DN thymocytes from B10
(TCRa; see Refs. 24 , 25 , and
33 for TCR haplotypes), B6 (TCRb), AKR
(TCRc), and 129/Sv (TCRe) were stained
with PE-labeled anti-C, FITC-labeled 7C10, and biotin-labeled
anti-V1 (2.11) followed by allophycocyanin-labeled streptavidin
and were analyzed in a FACSCalibur. Dot plots represent 7C10 and V1
staining among electronically gated + thymocytes.
Numbers represent percent of cells in each quadrant. Other
TCRa strains tested and positive for 7C10 were
C57BL/10J, B10.D2, C3H/He, CBA/J, CBA/N, DBA/1, and DBA/2. Also, NOD/Lt
mice were positive for 7C10. Tested 7C10-negative strains were BALB/c,
SJL, SWR, NZW, NZB (TCRb), A/J, and 129/Ola
(TCRe) and FVB (unknown TCR haplotype).
|
|
With this mAb we separated the V1a-
(V1+7C10+) and the
V1b-expressing
(V1+7C10-) i-IELs from
B10B6F1 animals, analyzed independently their
TCR repertoires, and compared them with those found in V1 i-IELs
present in B10 and B6 mice. The results of these analyses are shown in
Fig. 5. In B10B6F1
mice, different levels of V4+ cells among
V1+7C10+ and
V1+7C10- i-IELs were
readily evident, whereas the levels of V5+ or
V6+ cells among
V1+7C10+ and
V1+7C10- i-IELs were
comparable. Interestingly, similar levels of
V4+ cells were found among
V1+7C10+ i-IELs in
B10B6F1 and B10 mice on one hand and among the
V1+7C10- cells in
B10B6F1 and B6 mice on the other.
Altogether, these results demonstrate that polymorphism at the regions
coding for the V1J4C4 chain influence the representation of
the V1/V4 i-IEL subset and, consequently, strongly suggest that
the TCR-linked gene influencing this phenotypic trait is the
structural gene encoding the V1J4C4 chain.
Sequence polymorphism at the V1J4C4 chains of B6 and B10
mice
Nucleotide sequences of the B10 and B6 V1J4C4 chains
obtained in our laboratory were identical with those previously
published (34, 35). They differ by two nucleotides, which
result in two amino acid substitutions: an Ala to Glu in the V-domain
(position 12) and a Lys to Glu in the C-domain (position 157), with the
B6 V1J4C4 chain containing Glu at these positions. In a
homology-based modeling of the B6 V1J4C4 chain, these two
polymorphic residues between the B6 and the B10 V1J4C4 alleles
are predicted to be solvent exposed but far away from any
complementarity-determining region (CDR). Thus, Glu12 is
predicted to lie near the peptide loop that separates the V and the C
domains, whereas Glu157 is located within the
contact area between the V and C domains (Fig. 6).
The V4 chain can pair equally well with V1a and
V1b chains
The different representation of the V1/V4 i-IEL subset in B6
and B10 mice and its dependence on the V1J4C4 allele could be
easily explained without the need for invoking a cellular selection
mechanism if the V4 chain could not pair correctly with the B6
V1J4C4 chain. Although the fact that 2% of i-IELs
in B6 mice express the V1/V4 TCR seems incompatible with
this hypothesis (see Fig. 2), it could be argued that this low fraction
of V1/V4 i-IELs may result from the expansion of even rarer
V1+V4+ cells.
Therefore, we sought to analyze the levels of V1/V4 cells in a
situation in which mature cell expansion is limited, which is
usually the case in the organs where the cells develop. Because the
origin of the i-IEL is still a controversial matter, we decided
to analyze this issue in the thymus. As shown in Fig. 7, B6, B10, or
B10B6F1 thymocytes contain similar levels
(8%) of V1/V4 cells, strongly suggesting that the V4 chain
can pair well with both alleles of the V1J4C4 chain.
B6 V1/V4 i-IELs are the target of the selective event
Lack of pairing constraints for the B6 V1 and V4 chains
makes it likely that cellular selection events are responsible for the
differences in the representation of V1/V4 i-IELs in B6 and B10
mice and for the strong overrepresentation of the B10 allele of the
V1J4C4 chain among V1/V4 i-IELs in
B10B6F1 animals (>90% of the V1/V4 i-IELs
in B10B6F1 animals expressed the
V1a allele, whereas the same allele was
expressed in 65% of the
V1+V4- i-IELs; data
not shown). However, the experiments presented here are compatible with
an increased expansion and/or survival of V1/V4 i-IELs in B10
mice as well as with a specific deletion or decreased survival of
V1/V4 i-IELs in B6 mice. Although intermediate levels of
or cell populations expressing defined TCR V-gene segments in
F1 mice have been currently used as
suggestive of positive selection of the cells studied
(19, 36, 37), the fact that the V1/V4 cells
expressing distinct V1J4C4 alleles are submitted to different
selective pressures precludes such a conclusion.
The target cell at which selection operates and, consequently, the
nature of the selection may be inferred from the analyses of the
junctional diversity of the V1 and V4 chains expressed by the
V1/V4 i-IEL subsets present in B6 and B10 mice. These analyses
can be performed at the population level by studying the distributions
of the length of the CDR3 in TCR and TCR junctions
(38). In polyclonal populations, CDR3 lengths distribute
in a Gaussian-like curve with a well-defined number of peaks that is
characteristic of each TCR chain, whereas such Gaussian distribution is
lost in less diverse populations.
As shown in Fig. 8 (left
panels), the profiles of the distribution of CDR3 lengths observed
on V4-(D)-J1 junctions expressed by sorted
V1+V4+ i-IELs
isolated from B6 and B10 mice are quite similar. Both contain 12
defined peaks differing in length by three nucleotides, thus
corresponding to in-frame rearrangements. These profiles are typical of
polyclonal TCR junctions. In contrast, CDR3 length distributions of
V1-J4 junctions expressed by the same sorted populations in both
mouse strains were clearly different (Fig. 8, right panels).
In B10 mice, the typical profile of polyclonal V-J junctions was
evident, with three to five defined peaks forming a Gaussian-type curve
and centered on a CDR3 of 9 aa in length (39). In
contrast, the profile obtained from sorted
V1+V4+ i-IELs from B6
mice was not Gaussian and showed a prominent peak at a CDR3 length of
10 aa. This profile was unique to the
V1+V4+ i-IEL subset,
as indicated by the Gaussian-type profile centered on a CDR3 of 9 aa
obtained from sorted V1+ i-IELs from B6 mice
(which contains >97%
V1+V4- cells).
From these experiments we conclude that B6 V1/V4 i-IELs, rather
than the B10 V1/V4 cells, are the target of the cellular
selection event described here. Because B6 mice contain very low levels
of V1/V4 i-IELs, these experiments suggest that the selective
event results in the reduction of the V1/V4 subset among B6
i-IELs. Such a selective event appears to spare a fraction of
V1/V4 cells that are enriched in V-J junctions with
relatively long CDR3.
To investigate whether, besides their CDR3 length, other structural
features were apparent in the V-J junctions present in the B6
V1/V4 i-IELs, their expressed V1-J4 rearrangements were
cloned and sequenced. Consistent with the population analyses shown
above, 11 of 26 clones (42.3%) contained a CDR3 of 10 aa in length.
Their junctional nucleotide sequences are shown in Fig. 9A, whereas their predicted
amino acid sequences are shown in Fig. 9B. Most nucleotide
and amino acid sequences were unique. The frequent presence of Ile at
position 4 of the CDR3 is likely due to the usual presence of the last
codon of the V1 (ATA) in these junctions. Similarly, that of Ser at
position 6 results from the presence of the first codon of the J4
(TCA) in the junctions. Finally, the overrepresentation of Gly and Arg
at position 5 is mainly due to the presence in the junctions of the P
nucleotides of the J4 segment (GA). Comparable frequencies of the
same amino acids at these positions were also found in V1-J4
junctions of the same CDR3 length isolated from B10 V1/V4 i-IELs
or from B6 V1 i-IELs (data not shown).
Tissue-specific selection of V1/V4 cells
To investigate whether similar selective events also operate in
other peripheral sites, we evaluated and compared the representation of
the V1+ and the V1/V4 populations among
LN cells in B6, B10, and B10B6F1 mice
(Fig. 10A). The
representation of both cell populations among LN cells was
significantly lower in B6 than in B10 mice, although the differences
were less pronounced than those previously observed among
i-IELs. Thus, V1+ cells represented 62% and
42% of the LN cells in B10 and B6 mice, respectively (Fig. 10A, left panel), whereas less than a
twofold difference was observed in the representation of the
V1/V4 subset in these mouse strains (Fig. 10A,
middle panel). B10B6F1 animals
displayed intermediate levels of these two populations, and analyses of
B10B6F2 mice showed that the different
representation of these populations is influenced by
TCR-linked genes (data not shown). The representation of
V4+ cells among V1+
LN cells in both strains of mice was also significantly different,
although these difference appeared too small to be of major biological
significance (V4+ cells represented 32% and
26% of the V1+ cells in B10 and B6 mice,
respectively; Fig. 10A, right panel). Taken
together, these data indicate that most of the differences observed in
the representation of the V1/V4 LN subset between these two
strains of mice can be attributed to the V-independent selective
event. Consistent with this interpretation, the CDR3 length
distribution of the V1-J4 junctions present in V1/V4 LN
cells in B6 mice contains a predominant peak at a CDR3 of 9 aa (Fig. 10B), although a slight increase in the frequency of
junctions with a CDR3 of 10 aa was noticed in some mice (data not
shown). Altogether, these results suggest that the V-dependent
selection mechanism operating on V1/V4 cells is more prevalent
in, if not confined to, the intestinal epithelium.
|
Discussion
|
|---|
Previous experiments analyzing V-gene use by i-IELs or
splenocytes with a limited number of V- and V-specific mAbs have
shown that genes closely linked to the TCR, TCR, and MHC loci
influence the representation of cell subsets
(17, 18, 19). These experiments provided indirect evidence
suggesting that peripheral cells may be submitted to cellular
selection mechanisms. However, neither the selective mechanisms nor the
target cell of the selective events could be unambiguously defined in
those experiments. By analyzing V-gene use in i-IELs from B6 and
B10 mouse strains and in their F1 and
F2 progenies with a large panel of V- and
V-specific mAbs, we have shown that the linkage to the TCR locus
of the representation of i-IEL subsets results from two
mechanistically different events possibly acting on two distinct target
cells. The first event influences the representation of cells
expressing different V chains independently of the chain and is
evident not only in the different representation of cells expressing
the V1 or the V7 chains among i-IELs (Fig. 2), but also in
the representation of cells expressing the V1 or the V4 chains
among thymocytes (data not shown) and LN cells (Fig. 10). Such
an event may be responsible, at least in part, for the observed
influence of the TCR locus in the representation of
V4-V4+ splenocytes
(likely representing V1/V4 cells) in crosses between B6 and DBA/2
parental strains (19). A mechanistically simple hypothesis
to explain these results postulates that rearrangements involving the
J1 or the J4 gene segments occur at different frequencies in
precursors from different mouse strains. This possibility is
currently under investigation. Alternatively, a V-dependent,
V-independent selective mechanism can also be invoked.
The second event is likely a cellular selection event acting
specifically on V1/V4 i-IELs, and it has two major consequences.
First, animals carrying the TCRb haplotype
contain a reduced proportion of V1/V4 i-IELs. Second, the
remaining V1/V4-bearing i-IELs in TCRb
mice contain V1J4C4 chains that are selectively enriched for
CDR3 lengths longer by 1 aa than the prevalent size of V1J4C4
chains observed in other V1-bearing cells in the same mice. Our
experiment excluded structural pairing constraints as a possible
mechanism to explain the low abundance of V1/V4 in the intestinal
epithelium of B6 mice, because V1/V4 cells expressing a very
diverse repertoire of TCRs are present in similar numbers in the thymus
and in the periphery of B6 and B10 mice. There are at least three
possible explanations for these observations. First,
V1b/V4 i-IELs may be actively deleted.
Alternatively, V1a/V4 i-IELs may be truly
positively selected by an endogenous ligand, whereas
V1b/V4 i-IELs fail to react with the same
ligand and are passively deleted. Finally,
V1a/V4 i-IELs but not
V1b/V4 i-IELs may be expanded in situ via
specific Ag stimulation. Analysis of germ-free mice and mice deficient
in molecules involved in death and survival of lymphocytes (currently
underway) may help to clarify this issue. Whatever the mechanism, our
results indicate a differential reactivity of
V1a/V4 and
V1b/V4 i-IELs that has a physiological
consequence.
The different reactivities of the V1a/V4
and the V1b/V4 i-IELs, as well as the
overrepresentation of V1 chains containing CDR3s of 10 aa in length
in the V1/V4 i-IEL subset in B6 mice must relate to polymorphisms
at the coding region of the V1J4C4 chain. Based on the
sequence identity and on the expected structural homology of the V and
C domains of the mouse V1J4C4 chain with the V9C chain
of the recently crystallized human receptor TCR V9V2
(40), a model of the V1J4C4 chain was simulated
(Fig. 6). In this model, the two polymorphic residues between the B6
and the B10 V1J4C4 alleles are disposed far away from any CDR.
Given that the human V9C chain and the mouse V1C4 share
47% identity at the amino acid level, it is very likely that the model
predicts accurately the position of these residues in the mouse
V1C4 chain. How polymorphism at any of these two residues affects
CDR3 length in B6 V1/V4 i-IELs is not readily evident from these
analyses. Given their predicted localization, the possibility that they
exert a direct effect in the structure of the CDR3 appears unlikely.
However, because their side chains are predicted to be exposed to the
solvent, they could participate in the overall recognition through a
putative interaction with a distinct site of the ligand.
The fact that the V1-J4 junctions present in the V1/V4
i-IELs of B6 mice are enriched in relatively long but highly diverse
in-sequence CDR3 suggests that selection operates more on the overall
CDR3 structure than on their primary amino acid sequence. This suggests
that germline-encoded elements present in the Ag-binding site of the
V1V4 TCR are determinant in ligand recognition and specificity,
as has been proposed for the recognition of prenyl pyrophosphates by
human V9V2 cells (40, 41). This may be a general
feature of ligand recognition by T cells, which often correlates
with V and/or J segment usage but allows a high degree of diversity at
the V to J junction (42, 43, 44, 45). The experiments presented
here should provide a good experimental model to analyze T cell
reactivity in vivo. Unraveling the specificity of cells, their
mechanisms of Ag recognition, and the consequences of such recognition
may be the keys to the understanding of their unique functions.
|
Acknowledgments
|
|---|
We thank P. Alzari and G. Bentley for their help in the
construction of the structural model of the V1J4C4 chain.
|
Footnotes
|
|---|
1 This work was supported by institutional grants and by grants from the Association pour la Recherche sur le Cancer. K.G. was supported by a fellowship from the Fondation pour la Recherche Médicale. 
2 Address correspondence and reprint requests to Dr. Pablo Pereira, Unité du Développement des Lymphocytes, Centre National de la Recherche Scientifique, Unité de Recherche Associée 1961, Institut Pasteur, 25 Rue du Dr. Roux, 75724 Paris cedex 15, France. E-mail address: ppereira{at}pasteur.fr
3 Abbreviations used in this paper: Tg, transgenic; i-IEL, intestinal intraepithelial lymphocyte; LN, lymph node; DN, double negative; CDR, complementarity-determining region.
Received for publication April 22, 2002.
Accepted for publication August 1, 2002.
|
References
|
|---|
Top
Abstract
Introduction
