Positive Selection Is Not Required for Thymic Maturation
ofTransgenic ~8 T Cells
By Edina Schweighoffer and B.J. Fowlkes
From the Laboratory of Cellular and Molecular Immunology, National Institute of Allergy and
Infectious Diseases, National Institutes of Health, Bethesda, Maryland 20892-0420
Previously published reports describing thymic differentiation in two TCtL~/8 transgenic
mouse models have suggested that ~/8 T cells require MHC-mediated positive selection to
reach full maturity. Recent studies indicate that recognition of antigen by mature ~/8 T cells is
not MHC restricted, raising the issue of why developing ~/8 T cells would even require MHC-
driven positive selection. Therefore, we have reinvestigated the requirements for development
and selection in G8 "y8 T cell receptor (TCtL) transgenic mice. Analyses of absolute cell num-
bers, phenotypic subsets, and functional competence of thymic and peripheral G8 ~/~ T cells
indicate that these cells can fully mature in class I MHC-deficient mice. Moreover, mixed bone
marrow chimeras demonstrate that y5 T cells of mutant B2-microglobulin ([32m ~ origin are
partially deleted in the presence of H-2d-bearing thymocytes (previously beheved to be the
haplotype mediating positive selection). We conclude that there is no requirement for class I-like
molecules for the maturation/development of these transgenic y8 T cells and that the differ-
ences in thymocyte phenotype and number observed are, instead, attributable to effects of
defined, while the maturation process for ~/8 T cells is less
well understood. Since the proportion of y8 T cells in the
thymus is <1% (1), studies on the development of these
cells has been greatly aided by ~/8 TCIL transgenic mice.
For our investigation, we have used the G8 mouse (2) that
carries a transgene containing productively rearranged
and 8 TCtL genes derived from a T cell clone (3, 4) pro-
duced by immunization of BALB/c nu/nu mice with
B10.BP,. spleen cells. The reactivity of the original T cell
clone and of the "y5 T cells from the transgenic mice has
been mapped to T22/T10 in the nonclassical (class Ib) re-
gion of the MHC (TL-region). The strongest reactivity
was observed with H-2 b, intermediate with H-2 k, and no
reactivity with H-2 a antigen-presenting cells (APCs) (4).
Like classical MHC class I (class Ia) molecules, the putative
class Ib ligand of the G8 TCIL associates with [32-micro-
globulin (132m) 1 (5); therefore, the absence of[32m may in-
fluence the developmental fate of the transgenic ",/8 T cells.
Consistent with this idea were the reported phenotypic dif-
ferences between the G8 [32m + and G8 [32m ~ thymuses that
demonstrated a higher proportion of HSA-CD44hiCD451LB hi
~/8 T cells in the class I + mice (6). Moreover, the G8 ~32m ~
he major events in the intrathymic differentiation of
ot[3 T cells have been intensely investigated and well
~ Abbreviation used in this paper: 132m, [~2-microglobulin.
thymocytes did not proliferate to allogeneic stimulation,
and these mice had very few, if any, peripheral lymphoid
~/~ T cells (7). These data lead to the conclusion that ~/B T
cells, like ot[3 T cells, require MHC-dependent positive se-
lection to complete their maturation. Similar conclusions
were reached using another TCtL~/B transgenic strain,
KN6, which has a similar pattern of recognition to G8 and
was shown to react to T22, as well (8, 9). In contrast to
these results, an investigation of ~/~ development in non-
transgenic ~2m ~ mice indicated that many or most yB cells
do not require class I or class I-like MHC for their devel-
opment (10). These conflicting results could be explained if
only a subset of ~/8 T cells use MHC for positive sel .~ction.
In our efforts to investigate the cellular and molecular in-
teractions involved in thymic ~/~ T cell development, sur-
prisingly, we found no requirement for [32m-associated
molecules for full ~/8 T cell maturation in G8 mice. Since
the only other example of MHC requirement for ~/8 devel-
opment arises from a transgenic strain with very similar
ligand reactivity to G8 (11), our findings raise the issue of
whether any ~/8 T cells require MHC-driven positive se-
Materials and Methods
Mice. G8 transgenic mice (2) were backcrossed five times to
B10.D2 mice and crossed two times to 132m ~ (H-2 b) mice (previ-
ously backcrossed five times to C57BL/10 (B10) mice) to gener-
J. Exp. Med. y The Rockefeller University Press i 0022-1007/96/05/2033/09 $2.00
Volume 183 May 1996 2033-2041
ate TClq, y8 transgenic, [32m ~ o~pring (H-2 b or H-2a). C57BL/6
mice were obtained from the National Cancer Institute (Freder-
ick, MD) and B10.D2 mice from The Jackson Laboratory (Bar
Harbor, ME). Mice were bred and housed in an NIAID P,e-
search Animal Facility according to AAALAC specifications.
R_AG-2 ~ (H-2 a) (12) and 132m ~ (13) mice (both backcrossed five
times to C57BL/10) were bred and maintained on NIAID con-
tract at Bioqual Inc. (Rockville, MD).
Monoclonal Antibodies And Flow Cytometry.
ric analyses, cells were stained according to standard protocols
For flow cytomet-
(14) using the following labeled antibodies: anti-TCRyS-biotm
(GL3), anti-TC1LVy2-biotin or -FITC
TCR.otlB-FITC or-PE (H57-597), anti-CD5(Ly-1)-biotin or-lYE
(53-7.3), anti-CD44(Pgp-I)-biotin or -PE (IM7), anti-CD45R.B-
biotin or-PE (16A), anti-CD24/HSA-biotin or-PE (M1/69),
anti-Ka-FITC (SF1-1.1), all obtained from PharMingen (San Di-
ego, CA); anti-TCRyS-FITC (GL3), streptavidin-TC or-PE,
from Caltag Labs. (San Francisco, CA); and avidin-APC (Molec-
ular Probes, Eugene, OR). For cell sorting, cells were stained
with anti-TClLyS-biotin/APC- avidin, anti-HSA-PE, and anti-
TCRoL~-FITC and sorted into yS+ot[3-HSA + and yS+o~[3-HSA
cell populations, respectively. To avoid FcR-mediated binding of
antibodies, cells were pretreated with anti-Fc-receptor mono-
Figure 1. Two-color flow cytometric analyses reveal phenotypic dif-
ferences between thymocytes from G8 j32m + (H-2 d) and G8 j32m ~ mice.
Freshly prepared thymocytes were stained with anti-TCR~8 mAb and
the indicated antibodies. FACS ~ density plots are shown; data were col-
lected using live gating for the TCR~/8 + cells. (Data are plotted in two-
color format to facilitate comparison with data from chimeras shown in
Figure 2. Thymocyte numbers in G8 J32m + (H-2 a) and G8 J]2m ~
mice. (Top) Total number of3~8 thymocytes isolated from G8 (32m + (H-2 a)
and G8 j32m ~ mice, respectively. (Bottom) Number of HSA- y~ thy-
mocytes per mouse derived from the product of the percentage of HSA +
and HSA 3'8 thymocytes and the total number ofthymocytes. Each cir-
cle represents one mouse. Average values for each group are shown as col-
umns: (top) 2.9 X 106 vs. 15.3 • 106; (bottom) 31.5 • 104 vs. 25.3 X 104.
2034 Thymic Selection ofy~ T Cells
clonal antibody (mAb), 24G2. Data were acquired and analyzed
on a FACScan | using Cellquest, Lysys II, and PC-Lysys software,
all from Becton-Dickinson (Mountain View, CA). Electronic cell
sorting was performed on a FACStar | Plus (Becton-Dickinson).
Radiation Bone Marrow Chimeras.
made according to standard protocols (14). Briefly, NK cells of
the recipients were depleted by injecting PK136 antibody (15) in-
traperitoneally the night before bone marrow transfer. Recipients
were lethally irradiated (1,000 rads) 2-4 h before bone marrow
transfer. Bone marrow suspensions were prepared from femurs
and tibias of donors and were T cell depleted using anti-Thy-l.2
(JIJ) (16) and anti-Ly-l.2 (C3PO) (17) antibodies with low-tox
rabbit complement (Cedarlane Labs, Westbury, NY). Recipients
were injected with 2 • 107 donor bone marrow cells, intrave-
nously. 5 wk after bone marrow transfer, the chimeric mice were
killed and their thymuses were removed and analyzed.
Responder cells were prepared from G8
TCt( transgenic thymocytes or lymph node cells (depleted of B
cells using anti-mouse Ig-coated plates ) that were enriched
for double negative (CD4-CD8-) cells using anti-CD8 (3-t55)
and anti-CD4 (RL-172) antibodies with low-tox rabbit comple-
ment (Cedarlane Labs.). Irradiated (3,000 rads) stimulator cells
were prepared from C57BL/6 (H-2 b) or B10.D2 (H-2 a) spleen
cells by T cell depletion (using anti-Thy-l.2 [J1J]  antibody
and rabbit complement), and 2 • 104 sorted responder cells were
added to 3 • l0 s stimulator cells per well in U-bottom 96-well
plates. Proliferation was measured after 3 d of culture by pH]thy-
midine incorporation (1 IxCi/ml pulse for 18 h). Samples were
harvested with a 96-well harvester (Brandel, Gaithersburg, MD),
and [3H]thymidine incorporation was measured in a Betaplate
counter (Pharmacia, Uppsala, Sweden). All values represent
means of triplicate wells.
Bone marrow chimeras were
To define the MHC requirements for y~ T cell develop-
ment, we compared the phenotype and function of G8
TCR~/8 + T cells of normal and MHC class I-deficient
([32m ~ mice. We analyzed thymocytes of these strains for
surface expression of HSA, CD44, CD45, and CD5 (Fig.
1). In agreement with the previously published results (6),
MHC class I + thymuses have a higher percentage of
of class I ~ mice. It is noteworthy that only class I + mice
show a relative increase in ~/~ thymocytes bearing HSA-
and CD5 hi, both considered to be markers of thymocyte
maturation. Also of note is the enrichment in the class I +
thymuses for cells bearing high levels of the activation
markers CD44 (18) and CD45RB (19).
Surprising were our analyses of actual thymocyte num-
bers. Although there is a large variation in the number of
y8 thymocytes, we find on average five to six times more
of these cells in the [32m-deficient G8 mice (Fig. 2, top). This
difference in cell numbers can be due to negative selection
in the [32m + mice or to a failure of positive selection in the
class I-deficient mice, leading to an accumulation of imma-
ture (HSA +) cells at the "pre-positive selection" stage. If
the latter interpretation were correct, we would not expect
to see any (or very few) HSA- ~/8 T cells in the class I ~ thy-
muses. The percentages of HSA- thymocytes in Fig. 1 are
deceiving, however, since, as determined by sampling a
large pool of mice, the [32m + and [32m ~ thymuses harbor
comparable numbers of HSA- y8 T cells (Fig. 2, bottom).
~/8 thymocytes than those
Figure 3. Three-color flow cytometric analyses show
HSA § and HSA- ~/~ T cells in the peripheral lymph nodes
of G8 132m + (H-2 a) and G8 [32m ~ mice. Freshly isolated
peripheral lymph node cells were stained with anti-
TCR~/~-FITC and anti-HSA-PE antibodies after pretreat-
ment with anti-Fc-receptor antibody (24G2) to prevent
FcP, binding. FACS | density plots are shown; data in c-d
were collected live gated for TCR.~/~ + cells. Equivalent re-
sults are obtained with ~2m ~ mice of either the H-2 d 'or
H-2 b haplotype.
2035 Schweighoffer and Fowlkes
Lymph Node Cells
Proliferation of Sorted TCR T 3+ G8 Thymocytes and
H-2 b H-2 d H-2 d
1Kesponders APC + IL-2 APC APC + IL-2
G8]~2m + HSA § 318 35,958
G8]32m ~ HSA + 33 21
HSA- 37,065 1,198
G8 Lymph node cells
G8132m + HSA + 3,114 101,923
G8]32m ~ HSA + 4,646
Thymocytes and peripheral lymph node cells from G8 ]32m + (H-2 a)
and G8 [32m ~ mice were isolated, stained with anti-TC1K'/& anti-
TCR~]3, and anti-HSA antibodies, and then sorted into 3,8+R[B-HSA +
and y8+cx]3-HSA- cell populations. Cells were assayed as described in
Materials and Methods. IL-2 (100 U/ml) was added where indicated.
Cpm values show means of triplicate wells. Data are representative of
six experiments. No response is elicited by ]32m ~ APCs, equivalent to
results with H-2 a APCs (not shown).
A distinct population of TCR3'8+TCRot[3 - T cells is
able to mature and reach the periphery of both the [32m +
and 132m ~ G8 mice (Fig. 3, a and b). These 3'8 T cells are
85-90% CD4-8- and 10-15% CD4-8 + (not shown). In
addition, there is a small population of CD4 + T cells ex-
pressing both ot~ and 2*8 TC1K (Fig. 3, a and b). Although
there has been no report of HSA + y8 T cells in the periph-
eral lymphoid organs, we observe both a HSA + and HSA-
y8 T cell subpopulation in the lymph nodes of G8 mice
([32m + or [32m ~ (Fig. 3, c and d). The presence of liSA on
a subset of peripheral 3'8 T cells raises the issue of whether
HSA on y8 T cells is a marker of maturational stage, state
of activation, or a lineage marker--a question that has not
To demonstrate that the 3'8 T cells of the [32m ~ mice are
functional and bear the transgenic ~8 TCR, TCR3'8+ot[3 -
cells from G8 [32m ~ thymuses and lymph nodes were sorted
and assayed for specific alloreactivity (7). As shown in Table
1, regardless of whether ",/8 T cells developed in the presence
or absence of [32m and whether they are from the thymus
or from lymph nodes, they all make specific responses to
H-2 b, but no response to H-2 a or 132m ~ APCs. Although
the HSA- y8 cells proliferate without exogenous cytokines,
the HSA + "y8 T cell response is dependent on added IL-2.
Thus, in marked contrast to previous reports, where no
y8 T cells appeared in the [32m ~ peripheral lymphoid or-
gans, these results demonstrate that 3'8 T cells of G8 mice
can functionally mature and reach the periphery in the ab-
sence of ]32m. This argues against a developmental block
due to the absence of class I and suggests, instead, that neg-
ative selection may be responsible for the reduced number
of 3'8 thymocytes and the altered proportions of subsets in
the H-2 d mice.
To address this latter possibility, we constructed radia-
tion bone marrow chimeras using stern cells from G8 j32m ~
or G8 [32m + donors, or a mixture of these two bone mar-
rows. Irradiated B10.D2 or [32m ~ mice served as recipients.
These experiments allowed us to follow interactions be-
tween the ]32m + and [32m ~ G8 hematopoietic cells in the
thymus, since the class I expression of the chimeric thy-
mocytes could be used to identify the cells derived from
each bone marrow donor. Table 2 shows that repopulation
with either unmixed donor marrow (G8 132m~
G8 [32m+--+J32m ~ results in similar 3'8 thymocyte numbers
to those of intact animals (Fig. 2 a); that is, the G8 [32m ~
donors give rise to about five to six times more 3'8 thy-
mocytes than do their class I + counterparts. More impor-
tantly, the lower number of G8 ",/8 thymocytes in the
mixed chimeras, equivalent to those that received unmixed
132m + marrow only (Table 2), reveals a dominant effect
imposed by the [32m + hematopoietic cells, a property nor-
really associated with negative selection.
Phenotypic analysis provided evidence that a subset of
3'8 T cells was the target of this deletional effect (Fig. 4).
]32m+-derived thymocytes are distinguished from ]32m ~
derived ones by staining with labeled anti-K d class I anti-
body. The phenotype of the thymocytes in unmixed bone
marrow chimeras (Fig. 4, a--c and g-0 corresponds to that
of intact donors (Fig. 1) and is independent of the recipi-
ent's genotype, such that there is a higher percentage of the
y8 thymocytes that are HSAI~
G8 [32m+-+[32m ~ chimeras (Fig. 4, g-0 than in the G8
132m~ ~ chimeras (Fig. 4, a-c). In contrast, the G8
132m~ thymocytes (K d-) in the mixed bone mar-
row chimeras, which can interact with class I + hematopoi-
hl in the
TCR T 8+ G8 Thymocyte Numbers in Chimeric Mice
Recipients G8 ]32m ~ G8 [~2m ~ + G8 ~2m + G8 [32m +
B 10.D2 8.8 1.8
5 wk after bone marrow transfer, thymuses were removed and ana-
lyzed. Numbers of~/8 cells are displayed for each thymus in millions.
2036 Thymic Selection of'y8 T Cells
etic elements, acquire the phenotype of the G8 [32m +
(Kd+)-derived thymocytes; that is, they contain a higher
percent of the HSAl~
From these altered distributions, it appears that it is the im-
primary target of this negative selection. This deletion
seems to be much less dramatic than the negative selection
of ~/8 T cells originally described in H-2 b mice (2), in that
the remaining cells are TCR~/8 hi (as in the intact mice ana-
lyzed in Figs. 1 and 3), and they respond to stimulation
with H-2 b APC. Similarly, there are well-documented ex-
amples of incomplete deletion in ot[3 T cell development,
where coreceptor quantity or MHC levels, as well as quan-
tity ofpeptide, can affect the degree of selection (20).
The question arises as to what cell types are mediating
cells (Fig. 4, d-f).
1~ population that is the
the deletion in the class I + mice. Epithelial elements do not
seem to play a role, since the class I haplotype expressed on
the radioresistant cells of the recipients does not seem to af-
fect the developmental pattern in the bone marrow chime-
ras. Among the hematopoietic cells, the most obvious can-
didates are the macrophages, the dendritic cells, or the
thymocytes themselves. To investigate this issue, we made
radiation chimeras using bone marrow of G8 [32m ~ mice
mixed with RAG-2 ~ bone marrow. This latter strain was
chosen because mice carrying mutations in both of their
RAG-2 alleles lose the ability to generate T and B cells (12)
but retain other hematopoietic cells. Fig. 5 shows that, un-
like mixes made with nonmutant bone marrow, RoAG-2 ~
hematopoietic cells do not mediate phenotypic changes in the
",/~ T cells derived from [32rn ~ marrow (Fig. 5, a-c vs. d-f).
Figure 4. Thymic phenotype of radiation chimeras, a-c and g-i show three-color flow cytometric analysis ofthymocytes from chimeras made with un-
mixed bone marrow stained and analyzed similarly to the intact G8 [32m ~ and G8 [32m + (H-2 d) mice, respectively (shown in Fig. 1). d-fare analyses of
mixed bone marrow chimeras. Cells derived from the G8 [32m ~ vs. G8 [32m + (H-2 d) bone marrow can be distinguished based on class I (K d) expression.
All plots were gated for 3'~ thymocytes. The nttmbers in parentheses represent percentages of each quadrant normalized for the percentages of [32m ~ or
[32m+-derived ~/~ thymocytes. FACS | plots are representative of six experiments.
2037 Schweighoffer and Fowlkes
These latter results suggested that either thymocytes or
thymic B cells could be inducing negative selection. To in-
vestigate this possibility, we bred G8 with R_AG-2-deficient
mice to produce G8 RAG-2 ~ offspring selected to be ho-
mozygous for H-2 d. The G8 RAG-2 ~ mice contain about
the same number of~ thymocytes (3.1 - 1.8 X 106) as
the G8 RAG-2 + animals, indicating that whatever the de-
leting element is in the H-2 a mice, it must be present on
the "y~ cells themselves. Accordingly, chimeras made from
G8 [32m ~ bone marrow mixed with G8 RAG-2 ~ marrow
induced deletion in the [32m~
(Fig. 5, g-0. Although these findings demonstrate that ~/~
thymocytes are able to promote deletion, they do not ex-
clude the Ix[3 thymocytes or the thymic B cells as additional
mediators of negative selection. We plan to assess the rote
(K a-) ~/8 thyrnocytes
of a[3 thymocytes using H-2 d TCR-[3 or TCR-ot mutant
mice since, in the absence ofTCtk-[3, T cell maturation is
blocked at the double negative (CD4-CD8-) stage, while
in the TC1k-et knockout mice, thymocytes proceed to the
double positive (CD4§ +) stage (21).
TCR transgenic mice have been used extensively to study
development and selection in cx[3 T cells. In contrast, re-
quirements for ~/~ selection using such mice has been lim-
ited to two strains (G8 and KN6). The ligand for the ~/8
TCR in both cases is class Ib, T22. It was somewhat per-
plexing that these transgenic ~/8 T cells failed to develop in
~2m mutant mice whereas y~ development, in general, ap-
Figure 5. Thymic phenotype of chimeras constructed with mixtures of G8 [32m ~ bone marrow and either RAG-2 ~ (H-2 d) (d-f) or G8 KAG-2 ~ (H-2 a)
(g-i) bone marrow, a-c show the unmixed (G8 ~2m~ ~ chimera for comparison. Cells derived from the G8 ~2m ~ vs. R.AG-2 ~ (H-2 d) or G8
RAG-2 ~ (H-2 d) bone marrow can be distinguished based on class I (K d) expression. All plots shown are gated for "yB thymocytes. The numbers in paren-
theses represent percentages of each quadrant normalized for the percentages of ~2m ~ or [~2m+-derived ~B thymocytes. FACS | plots are representative
of three experiments.
2038 Thymic Selection of~/~ T Cells
peared normal in nontransgenic class II ~ and [32m mutant
mice (10, 22). It was possible, nevertheless, that the trans-
genic ~/~ T cells represented a subclass of~/8 T cells that re-
quired MHC for their development, implying that the re-
quirements for ",/8 development were different and perhaps
even more complex than those for the ot[3 lineage.
In an effort to compare and contrast the developmental
requirements for the ot[3 and ~/~ T cell lineages, our inten-
tion was to identify selecting hgands and to determine
whether thymic stromal elements also mediate ~/8 selec-
tion, as they do for Ix[3 cells. In our studies of positive se-
lection, we were surprised to find transgenic "yB T cells that
were phenotypically and functionally mature in the thymus
and periphery of class 1-deficient mice (Figs. 1-3 and Table
1). Moreover, our prehminary data show that they are
present in the lymph nodes and spleen of G8 scid 132m ~
mice, as well, ruling out any role for endogenous receptors.
Although we cannot pinpoint the cause of the discrepan-
cies from the pubhshed data (6, 7), we suggest that what-
ever is responsible for the maturational block previously
observed to be linked to [32m, this association is disrupted
in the genetic combinations we used. The fact that both
the class I + G8 mice and the [32m ~ mice used to generate
G8 132m ~ offspring were five times backcrossed to the B10
strain raises the issue of whether background genes in the
129 (the strain used to produce the 132m mutation) are re-
sponsible for the earlier resuks and is an issue under investi-
Not only did we find mature ~ T cells in our G8 [32m ~
mice, but the higher (five to six times) number of~/8 T cells
in the ~2m ~ thymus suggested that there may even be some
deletion in the presence of class I or class I-hke molecules
(Fig. 2). Evidence for this negative selection was provided
by mixed radiation bone marrow chimera experiments,
showing that, in the presence of class I + hematopoietic
cells, a portion of the immature class I-deficient ~/8 T cells
disappeared (Fig. 4). Radiation chimeras made with mix-
tures of KAG-2 ~ and G8 RAG-2 ~ marrow indicated that
the ligand mediating the deletion is present on the ~/8 cells
themselves (Fig. 5). Whether a[3 thymocytes also play a
role in this process remains to be established.
The described ligand of the G8 clone appears to be en-
coded by the T22 b (23) and T10 b (24) genes of the MHC.
The corresponding allele of T22 b in the H-2 d haplotype is
structurally defective (5, 8); T10 a, however, appears to be
intact (8). In addition, T22 seems to be widely expressed
(spleen, thymus, hver, kidney, etc.), while T10 is more re-
stricted to spleen, thymus, and peritoneal exudate cells (8).
Based on these data, T10 could be a potential candidate
ligand causing the negative selection of transgenic ~/~ T
cells that we observe in the 132m + mice. Whatever the
hgand is, our data suggest that it is not expressed on the epi-
thelium or on the thymic APCs but rather on the thymocytes,
as has been shown for other TL family members (25).
In an earlier pubhcation (6), the enrichment for a subset
of~/8 thymocytes with a CD45KBraMel-14~~ phenotype in
G8 H-2 d mice that was not observed in [32m ~ mice seemed
to be consistent with the notion that activation markers
2039 Schweighoffer and Fowlkes
could be expressed by cells undergoing positive selection.
Since we show here that G8 ~/b T cells in the H-2 a mouse
are undergoing some negative selection, these activation
markers could be restricted to cells engaged in clonal dele-
tion. The fact that we do not detect proliferation of G8 ~/~
T cells from ~2m ~ mice in response to H-2 a APCs could
be explained as it is for the otl3 lineage; that is, that negative
selection of thymocytes can be triggered by lower avidity
interactions than those required for the activation of ma-
ture T cells (26, 27). The relative depletion of the HSA hi
~/8 cells in the H-2 a thymus, compared with those in [32m ~
mice, indicates that the HSA hi cells are the target of this de-
letional mechanism. Thus, depletion of HSA hi ~/8 thy-
mocytes would increase the relative proportion of HSA-
2r cells, making it appear that more mature ~/~ thymocytes
were present in the H-2 a than in the class I-deficient mice.
The failure to see diminution of HSA- "/~ thymocytes in
the thymus and in the periphery of the class I + G8 mice ei-
ther could be explained by the accumulation of mature
cells that survive deletion or implies that the HSA + cells are
not precursors to the HSA- cells.
Although MHC-restricted antigen responsiveness of Ix[3
T cells is well documented, the rules for antigen recogni-
tion by ~/~ T cells remain elusive. ~/~ T cells have been re-
ported to recognize a wide variety of antigens, including
MHC class I (5, 28) and class II (29) molecules, mycobacte-
rial peptides, heat shock proteins (30, 31), and small non-
peptide ligands (32). At the same time, recent studies inves-
tigating class II- and class Ib-specific ~/~ T cell clones have
demonstrated that antigen processing and binding of pep-
tides to the MHC molecules are not necessary for ~/~ T cell
activation (23, 24). Taken together, these conflicting data
suggest that the nature of ot[3 and ~/~ TCIK recognition is
quite distinct. Moreover, the requirements for T cell matu-
ration in these two lineages seems to be different, in that
nontransgenic [32m ~ (10) and class II-deficient (22) mice
seem to have normal ~/g T cell populations. Ifyg T cell an-
tigen recognition is not MHC restricted, and if the purpose
of thymic positive selection is to produce a large cohort of
mature T cells that recognizes foreign antigens in the con-
text of self-MHC, there should be no purpose for MHC-
mediated positive selection in ~/g development. This view
is consistent with the data presented here. We have shown
that the more mature, HSA l~ ~/~ thymocytes exist in the
132m ~ mice in numbers comparable to those in 132m + thy-
muses and that functional transgenic cells are present in the
periphery. We do not find any evidence for a developmen-
tal block in class I-deficient G8 "y~ TCP.. transgenic mice.
Indeed, we have demonstrated that class I-mediated posi-
tive selection is not necessary for complete ~ maturation.
Given the similarities in MHC recognition of G8 to KN6
(the only other TC1K'y~ transgenic strain reported to require
positive selection), we question whether MHC-driven pos-
itive selection is required in any ~/8 development. While
we cannot exclude the possibility, of course, that [32m-
independent hgands interact with "y~ thymocytes at some
point in their development and influence their fate, the ex-
istence and nature of such ligands have yet to be elucidated.
We thank Drs. J. Bluestone and L. Matis for the G8 mice; B. Koller and O. Smithies for the ~32m; F. Alt for
the R.AG-2 gene-targeted mutant mice; D. Raulet, L. Marls, E. Robey, F. Macchiarini, and C. Fteischacker
for critically reading the manuscript; E.O. Matechak for her expert assistance in many aspects of the work;
and C. Eigsti for cell sorting.
Address correspondence to B.J. Fowlkes, LCMI, NIAID, Bldg. 4, Room 111, 4 Center Dr., MSC-0420,
NIH, Bethesda, MD 20892-0420.
Received for publication 27 December 1995 and in revised form 8 March 1996.
1. Penninger, J., V.A. Wallace, K. Kishihara, T. Molina, H.
Krause, and T.W. Mak. 1991. Molecular organization, on-
togeny and expression of murine 0el3 and 3'8 T cell receptors.
Exp. Clin. Immunogenet. 8:57-74.
2. Dent, A.L., L.A. Matis, F. Hooshmand, S.M. Widacki, J.A.
Bluestone, and S.M. Hedrick. 1990. Self-reactive ",/8 T cells
are eliminated in the thymus. Nature (Lond.). 343:714-719.
3. Matis, L.A., R. Cron, andJ.A. Bluestone. 1987. Major histo-
compatibility complex-linked specificity of 3'8 receptor-bear-
ing T lymphocytes. Nature (Lond.). 330:262-264.
4. Bluestone, J.A., P,.Q. Cron, M. Cotterman, B.A. Houlden,
and L.A. Matis. 1988. Structure and specificity of T cell re-
ceptor y/8 on major histocompatibility complex antigen-spe-
cific CD3+,CD4-,CD8 - T lymphocytes. J. Exp. Med. 168:
5. van Kaer, L., M. Wu, Y. Ichikawa, K. Ito, M. Bonneville, S.
Ostrand-R.osenberg, D.B. Murphy, and S. Tonegawa. 1991.
Recognition of MHC TL gene products by 3'8 T cells. Im-
munol. Rev. 120:89-115.
6. Wells, F.B., Y. Tatsumi, J.A. Bluestone, S.M. Hedrick, J.P.
Allison, and L.A. Matis. 1993. Phenotypic and functional
analysis of positive selection in the 3'/8 T cell lineage._/. Exp.
7. Wells, F.B., S.J. Gahm, S.M. Hedrick, J.A. Bluestone, A.
Dent, and L.A. Matis. 1991. Requirement for positive selec-
tion of y8 receptor-beating T cells. &ience (Wash. DC). 253:
8. Ito, K., L. van Kaer, M. Bonneville, S. Hsu, D.B. Murphy,
and S. Tonegawa. 1990. Recognition of the product of a
novel MHC TL region gene (27b) by a mouse y8 T cell re-
ceptor. Cell. 62:549-561.
9. Haas, W., P. Pereira, and S. Tonegawa. 1993. Gamma/delta
cells. Annu. Rev. Immunol. 1 l:637--685.
10. Correa, I., M. Bix, N.S. Liao, M. Zijlstra, R. Jaenisch, and D.
P, aulet. 1992. Most 3'8 T cells develop normally in ]32-micro-
globulin-deficient mice. Proc. Natl. Acad. Sci. USA. 89:653-
11. Pereira, P., M. Zijlstra, J. McMaster, J.M. Loring, R. Jae-
nisch, and S. Tonegawa. 1992. Blockade of transgenic 3,8 T
cell development in [32-microglobulin deficient mice.
EMBO (Eur. 21/loi. Biol. Organ.) J. 11:25-31.
12. Shinkai, Y., G. Rathbun, K.P. Lain, E.M. Oltz, V. Stewart,
M. Mendelsohn, J. Charron, M. Datta, F. Young, A.M. Stall,
and F.W. Alt. 1992. R.AG-2-deficient mice lack mature lym-
phocytes owing to inability to initiate V(D)J rearrangements.
13. Koller, B.H., P. Marrack, J.W. Kappler, and O. Smithies.
1990. Normal development of mice deficient in ~-2M,
MHC class I proteins and CD8 + T cells. Science (Wash. DC).
14. Coligan, J.E., A.M. Kruisbeek, D.H. MarguLies, E.M. She-
vach, and W. Strober, eds. 1994. Current Protocols in Im-
munology. John Wiley & Sons, Inc., New York.
15. Koo, G.C., F.J. Dumont, M. Tutt, J. Hackett, and V. Kumar.
1986. The NK-I.I(-) mouse: a model to study differentia-
tion ofmurine NK cells.J, lmmunol. 137:3742-3747.
16. Bruce, J., F.W. Symington, T.J. McKeam, and J. Sprent.
1981. A monoclonal antibody discriminating between subsets
ofT and B cells.J. Immunol. 127:2496-2501.
17. Mark, C., F. Figueroa, Z.A. Nagy, andJ. Klein. 1982. Cyto-
toxic monoclonal antibody specific for the Lyt-l.2 antigen.
18. Budd, R.C., J.C. Cerottini, and H.P,. MacDonald. 1987.
Phenotypic identification of memory cytolytic T lympho-
cytes in a subset ofLyt-2 + cells.J. ImmunoL 138:1009-1013.
19. Wallace, V.A., W.P. Fung-Leung, E. Timms, D. Gray, K.
Kishihara, D.Y. Loh, J. penninger, and T.W. Mak. 1992.
CD45RA and CD45RB high expression induced by thymic
selection events._/. Exp. Med. 176:1657-1663.
20. Robey, E., and B.J. Fowlkes. 1994. Selective events in T cell
development. Annu. Rev. Immunol. 12:675-705.
21. Mombaerts, P., A.R. Clarke, M.A. Rudnicki,J. Iacomini, S.
Itohara, J.J. Lafaille, L. Wang, Y. Ichikawa, R. Jaenisch, M.L.
Hooper, and S. Tonegawa. 1992. Mutations in T-cell antigen
receptor genes fx and [3 block thymocyte development at dif-
ferent stages. Nature (Lond.). 360:225-231.
22. Bigby, M., J.S. Markovitz, P.A. Bleicher, M.J. Grusby, S.
Simha, M. Siebrecht, M. Wagner, C. Nagler-Anderson, and
L.H. Glimcher. 1993. Most y8 T cells develop normally in
the absence of MHC class II molecules. J. Immunol. 151:
23. Weintraub, B.C., M.IK. Jackson, and S.M. Hedrick. 1994.
Gamma delta T cells can recognize nonclassical MHC in the
absence of conventional antigenic peptides. J. Immunol. 153:
24. Schild, H., N. Mavaddat, C. Litzenberger, E.W. Ehrich,
M.M. Davis, J.A. Bluestone, L. Matis, R.K. Draper, and
Y.H. Chien. 1994. The nature of major histocompatibility
complex recognition by 3'8 T cells. Cell. 76:29-37.
25. Flaherty, L., E. Elliott, J.A. Tine, A.C. Walsh, and J.B. Wa-
ters. 1990. Immunogenetics of the Q and TL regions of the
mouse. Crit. Rev. Immunol. 10:131-175.
26. Pitcher, H., U.H. Rohrer, D. Moskophidis, R.M. Zinkerna-
gel, and H. Hengartner. 1991. Lower receptor avidity re-
quired for thymic clonal deletion than for effector T-cell
function. Nature (Lond.). 351:482-485.
27. Vasquez, N.J., J. Kaye, and S.M. Hedrick. 1992. In vivo and
in vitro clonal deletion of double-positive thymocytes. J. Exp.
2040 Thymic Selection of y8 T Cells
28. Vidovic, D., M. Roglic, K. McKune, S. Guerder, C.
Mackay, and Z. Dembic. 1989. Qa-1 restricted recognition
of foreign antigen by a ~/8 T-cell hybridoma. Nature (Lond.).
29. Matis, L.A., A.M. Fry, R.Q. Cron, M.M. Cotterman, R.F.
Dick, and J.A. Bluestone. 1989. Structure and specificity of a
class II MHC alloreactive ~8 T cell receptor heterodimer.
Science (Wash. DC). 245:746--749.
30. Born, W., M.P. Happ, A. Dallas, C. Reardon, R. Kubo, T.
Shinnick, P. Brennan, and R. O'Brien. 1990. Recognition of
heat shock proteins and ~/B cell function. Immunol. Today 11:
31. Rajasekar, R., G.K. Sim, and A. Augustin. 1990. Self heat
shock and ~/B T-cell reactivity. Pro& Natl. Acad. Sci. USA. 87:
32. Tanaka, Y., S. Shigetoshi, E. Nieves, G. DeLibero, D. Rosa,
R.L. Modlin, M.B. Brenner, B.R. Bloom, and C.T. Morita.
1994. Nonpeptide ligands for human ~/8 T cells. Proc. Natl.
Acad. Sci. USA. 91:8175-8179.
2041 Schweighoffer and Fowlkes