| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
*
Ghost Lab, Section on T cell Tolerance and Memory, Laboratory of Cellular and Molecular Immunology, National Institute of Allergy and Infectious Diseases/National Institutes of Health, Bethesda, MD 20892;
Genetics Institute, Andover, MA 01810; and
Biomedical Research Institute, Rockville, MD 20852
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Although evidence that newly developing immune systems are tolerance-prone dates back to the experiments in newborn mice from Medawar and colleagues (2), recent evidence revealed that these experiments may have been misinterpreted. Newborn mice are perfectly able to respond if given Ags in appropriate doses, with appropriate adjuvants, or on appropriate APCs (3, 4, 5, 6). Furthermore, adult mice can be rendered tolerant if given large doses of Ag-bearing cells (4) or smaller doses of cells from which professional APCs have been removed (7, 57, 58). One interpretation from these results is that newborn and adult immune systems are both tolerizable and immunizable and that the decision to respond or become tolerant must therefore be governed by features other than age or timing of Ag expression (8). However, another interpretation, recently elaborated by Bretscher (9), is that the newborn is too old and has already passed through the tolerance-prone period. Therefore, tests for a tolerogenic period must be done earlier, at the inception of immune competence.
A number of experiments fitting these test criteria have shown that MHC or multiple minor histocompatability-mismatched grafts given before the development of immunocompetence can be rejected by a newly developing immune system (10, 11, 12, 13, 14, 15). However, it could be reasonably argued that these experiments are not a valid test of time-based models. A number of models, including Bretscher’s, are based on the hypothesis that antigenic exposure early in ontogeny is tolerogenic because of the low frequency of effector Th cells in the periphery at this time (9, 16, 17, 18). The number of Ags in MHC or multiple minor mismatched grafts could be greater than the number of tissue-specific self Ags to which the immune system must normally establish peripheral tolerance (19, 20) and consequently, the frequency of Th cells against these Ags would be high enough that any tolerance-prone period would be very short. Therefore, Bretscher suggested an experiment in which female recipients would be given APCs presenting the single minor Ag H-Y before the recipient has developed immunocompetence (9); Brestcher’s prediction being that such early presentation of a single minor-histocompatability Ag would induce tolerance rather than the immunity that was seen when neonatal females were challenged with H-Y expressing APCs (4). Therefore, we grafted male and female skin to immunodeficient female recipients, allowed the grafts to heal for several months, then reconstituted the mice with fetal liver cells or a fetal thymus and followed the fate of the grafts.
Different models predict different outcomes of this experiment. Some time-based models, in which tolerance is based on the low frequency of helper T cells early in ontogeny, predict that the male skin, because it preexists, should be tolerated by the newly arising T cells (9, 16, 17, 18). Another recent time-based model, in which tolerance to skin is based on a particular property of neonatal skin that allows traffic of naive T cells, would predict that both the male graft and the normal adult female skin graft should be rejected (1). Ignorance models would predict that the male graft should be ignored until the animals are immunized (21). We found that the results fit with none of these predictions. The newly arising T cells rejected the male grafts while simultaneously becoming tolerant of peripheral self tissues. These data are strong evidence against time-based models and indicate that tolerance to skin Ags can occur readily without a period of T cell trafficking into neonatal tissues.
The Danger model would predict that recent male skin grafts would be rejected because of alarm signals sent by cells damaged during the surgical procedures, but that the long-healed skin should be accepted (22, 23). Thus, we were left with the question of what allowed the newly developing T cells to discriminate between the long-healed skin of the male donor and the normal peripheral tissues of the recipient, rejecting the former and becoming tolerant of the latter. In other studies of "well-healed" MHC or multiple minor mismatched grafts (10, 11, 15), the stimulus for rejection could have been cross-reactive environmental Ags (which may themselves be associated with danger signals) rather than the grafts themselves, making these studies difficult to interpret. To resolve these difficulties, Bingaman et al. (10) suggested that studies with H-Y may be enlightening. Though we used grafts differing only H-Y, an Ag that seems not to have any environmental mimics (7, 24), the healed grafts nevertheless were rejected. Thus, cross-reactive Ags were not the explanation for the rejection of healed male skin grafts. We also tested the distribution of the male Ag in the draining lymph nodes. Similar to studies with Ag expressed in normal kidney and islet tissue (25) we found that the Ags in the graft were continuously cross-presented in the draining lymph nodes, but unlike normal tissues, this cross-presentation was immunogenic. Therefore, we analyzed the long-healed grafts by immunohistology and quantitative real time PCR for molecules associated with inflammation. When compared with GAPDH, but not cyclophilin, the long-healed grafts appeared to have raised levels of several messages associated with stress, inflammation, and wound healing. These data indicate that some yet to be discovered aspect of the context of Ag presentation rather than the timing of Ag exposure is the critical factor in determining immunity vs tolerance.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Adult C57BL/6 (B6),2 B6
nude, B6 nu/+, B6-MHC-II-knockout (KO), B10, B10.D2, normal and timed
pregnant B10-RAG-2-KO (B10-RAG), and B6-TCR
-KO mice were obtained
from Taconic Farms (Germantown, NY). C.B-17 SCID, as well as timed
pregnant B6 and BALB/c mice, were obtained from the Frederick Animal
Facility (National Cancer Institute, National Institutes of Health,
Frederick, MD). All mice were housed in specific pathogen-free
conditions, and B10-RAG mice received antibiotics in their drinking
water (trimethoprim-sulfa). The National Institutes of Health is an
American Association of Laboratory Animal Care-accredited facility.
Surgery, grafting, and reconstitution
Full-thickness skin grafts of 
2 cm in diameter were placed on
the lateral thoracic wall of each mouse. They remained at 1–2 cm in
diameter after healing. In experiments in which mice were given two
simultaneous skin grafts, the grafts were placed on either side of the
body and were 0.5–1.0 cm in diameter after healing. Grafts were
considered rejected when at least 90% of graft tissue had disappeared
or in the case of grafts undergoing "chronic" rejection, when the
graft tissue became "shiny" subsequent to hair loss
indicative of scar tissue. Adult syngeneic ovary grafts were placed in
the ear pinnae by the same technique commonly used for cardiac grafts
(26).
Thymectomy was performed by aspiration of both thymus lobes through a
small incision in the skin just above the sternum and was confirmed by
autopsy. Thymus grafting was performed by placing two to three lobes of
fetal (day 15 or 18) or neonatal (
24 h old) thymus under the
left kidney capsule.
Mice were immunologically reconstituted in the following four ways: 1) by grafting of neonatal thymuses to nude recipients; 2) by i.v. injection of 20 x 106 day-15 B6 or day-12 BALB/c fetal liver cells, respectively, to B10-RAG or C.B-17 SCID recipients; 3) by both thymus grafting and fetal liver cell injection of thymectomized B10-RAG recipients or 4) by i.v. injection of 50 x 106 spleen and lymph node (axillary, inguinal, cervical, popliteal, and mesenteric nodes) cells from unimmunized mice, and flow cytometry confirmed the presence of T cells.
Assay for Ag on APC in lymph nodes
Lymph nodes draining the skin grafts were made into a single
cell suspension by squeezing the node apart between two layers of nylon
mesh. The cells were irradiated (1500 rad) and cultured in titrated
numbers with 20 x 103 Rachel cells in
0.2-ml round-bottom microwells. Rachel is a CD4+
Th1 clone specific for H-Y/Ab generated from a
(B6 x CBA/N) female mouse. Cultures were pulsed with
[3H]thymidine at 48 h and harvested 15
h later to measure thymidine incorporation by liquid scintillation
counting.
Detection of anti-graft CTL priming
To assess CTL priming in skin graft recipients and controls we gave the recipients i.v. 107 B cells from B6 male donors. The B cells were rigorously purified as described previously (7). The injection of purified B cells boosts the response of primed T cells but instead tolerizes naive T cells (7). CTL were assayed by the JAM Test (27). Briefy, 6 x 106 spleen cells from recipients (responders) were stimulated in vitro per 2-ml well (2 wells) for 6 days with 2 x 106 irradiated male spleen cells per well and then tested for ability to kill [3H]thymidine-labeled male and female activated splenocyte Con A blasts at various responder to target ratios. The responder to target ratios are calculated from the number of responders originally cultured.
Histology
Heart, lung, kidney, liver, pancreas, small intestine, ovary, thyroid/parathyroid, and skin were fixed in 10% buffered formalin (Baxter) and embedded in paraffin. Serial sections were stained with hematoxylin and eosin by standard methods at Histo-Path of America (Millersville, MD).
Quantitative RT-PCR
Total RNA was extracted from the skin samples by using RNA STAT-60 (Tel-Test, Friendswood, TX ). RNA was treated with 10 U of RQ1 DNase I (Promega, Madison, WI) for 30 min at 37°C. Samples were extracted with phenol/chlorofrom and RNA was precipitated with 0.3 M NaOAc and 2 vol of 100% ethanol. RNA was resuspended in diethyl pyrocarbonate-treated sterile water, and the RNA concentration was determined by measuring the OD260 nm. rTth DNA polymerase was then used to reverse transcribe and amplify 25–50 ng of total RNA in a single-tube assay with the Perkin-Elmer TaqMan EZ RT-PCR kit (Perkin-Elmer, Foster City, CA) with gene-specific sense and antisense primers and a probe fluorescently labeled at the 5' end with 6-carboxy-fluorescein (28, 29). Primers and fluorescently labeled probes were then generated by using Primer Express software (Perkin-Elmer) and were synthesized by Perkin-Elmer. To avoid amplification of contaminating genomic DNA, primer pairs were selected that crossed intron/exon borders whenever possible. Samples were reverse transcribed for 30 min at 60°C and then subjected to 40 rounds of amplification for 15 s at 95°C and 1 min at 60°C with the ABI Prism 7700 sequence detection system as described by the manufacturer (Perkin-Elmer; Ref. 29). Sequence-specific amplification was detected as an increased fluorescent signal of 6-carboxy-fluorescein during the amplification cycle. Quantitation of gene-specific message levels was based on a comparison of the fluorescence intensity in the unknown mRNA sample to the fluorescence intensity from a standard curve of known mRNA levels. Amplification of the genes for mouse GAPDH or cyclophilin were performed on all samples tested to control for variations in amounts of RNA. Levels of gene-specific messages were graphed as normalized message units as determined from the standard curve. The data are expressed as a ratio of the transcript values obtained from long-standing skin grafts compared with their normal skin counterpart, or as TaqMan units that are relative values of transcript levels for a given gene, which have been normalized to a housekeeping gene (GAPDH). A no-template control was included in each amplification reaction to control for contaminating templates. For valid sample analysis, the fluorescence intensity in the no template control was required to be zero.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In our first series of experiments, we asked whether a healed
graft carrying a single, weak, minor histocompatability Ag, would
induce tolerance, be ignored, or induce immunity in T cell populations
that were new thymic emigrants. We assessed the response of newly
generated T cells by giving female B6 nude mice a nude male skin graft
and 9 wk later, a female thymus. These thymuses were from normal B6
donors (Fig. 1
A), or from
B10-RAG donors (Fig. 1B) that have no functional T cells and
must be repopulated by stem cells from the nude host before any T cell
development can occur. The expectation was that newly maturing T cells
from the female thymus should slowly seed the periphery in low numbers,
encounter the H-Y Ag either in the male skin graft or in its draining
lymph node (25, 30), and become tolerant. As a control to
ensure that central thymic tolerance was functioning properly in these
animals, we also grafted one cohort of nude females with a male B10-RAG
thymus.
|
, A and B, shows that newly generated T
cell populations rejected the long-standing male grafts, though the
grafts had healed for 9 wk (or 14 wk, data not shown) and appeared as
healthy as the host’s own skin. The rejection did not require the
transfer of mature T cells with the thymus graft because the B10-RAG
thymus grafts, like the normal B6 thymuses, lead to rejection.
Recipients reconstituted with a male thymus did not reject the male
skin (Fig. 1B) showing that central tolerance mechanisms
were functionally intact and able to induce tolerance in the newly
developing T cells, but peripheral tolerance to the grafts did not
occur. Thus it appeared that mature naive T cells and, more
surprisingly, newly developing T cells, were immunized rather than
tolerized by the peripheral Ag on the healed grafts, despite the low
frequency of responding cells to H-Y and the lack of cross-reactive
environmental Ags to drive the response (7, 24).
In the experiments above, the bone-marrow-derived precursors were
from adult marrow, and the thymuses in which they developed were from
neonatal animals. It remained possible that peripheral tolerance
depends on some property intrinsic to fetal but not adult precursor
cells or thymuses. It has also been reported that nude mice have a
small number of mature T cells, which may have contributed to the
rejection process. Therefore, we reperformed the grafting experiments
in B10-RAG recipients by reconstituting with male or female fetal liver
cells. In addition, to test whether the maturity of the thymus itself
has an influence, we thymectomized a cohort of female B10-RAG mice and
reconstituted them with a day-15 fetal B6 thymus plus an injection of
day-15 fetal liver cells (Fig. 1, C and D). Once
again, in both groups, we found that the female mice reconstituted with
female fetal tissues rejected the long-standing male grafts. Thus, T
cells generated from fetal precursors, in a fetal thymus, in hosts that
(unlike nudes) have no preexisting T cells, went on to reject a healed
preexisting male skin graft, a surprising result that was not predicted
by any current model of immunological tolerance. The mice did not
reject grafts of female skin, showing that some aspects of peripheral
tolerance were intact, and those given male thymuses did not reject
male skin, showing that central tolerance was also intact. Rejection of
preexisting grafts was not peculiar to B6 or B10 mice, as C.B-17 SCID
mice reconstituted with day-12 BALB/c fetal liver cells also rejected a
preexisting B10.D2 minor-histocompatibility different skin graft (not
shown and 59).
Lack of autoimmune responses in reconstituted mice
The finding that the long-standing male grafts were rejected by newly generated recent thymic emigrants counter the expectations of most time-based models because these models hold that a preexisting Ag should induce tolerance. They counter the ignorance model because the graft Ags are peripheral Ags that should be ignored. They counter the Danger model because the grafts had healed. Healthy tissues should not immunize.
However, one recently elaborated time-based model did offer a possible
explanation for the rejection. This model suggests that tolerance to
peripheral Ags cannot be established in adult mice. It was proposed
that establishment of tolerance to skin, for example, requires that
naive T cells traffic into the skin, which occurs only in the neonatal
period (1). If this were the case, we would expect to find
general signs of peripheral autoimmunity, as has sometimes been seen in
other systems (31, 32). In addition, because we used adult
mice, there is no possibility for traffic of T cells into neonatal skin
for the establishment of tolerance. Thus, we would expect to see either
heavy lymphocytic infiltration or rejection of the syngeneic skin
grafts based on the data/conclusions reported by Alferink and
colleagues in their transgenic system (1). Although the
acceptance of female skin grafts, the long life span of the mice after
reconstitution, and an absence of splenomegaly or lymphadenopathy (not
shown) suggested that the mice were not autoimmune, we nevertheless
took a further look at a potential generalized lack of self-tolerance.
Nine to 21 mo after reconstitution, we tested nine different organs for
signs of autoimmunity such as lymphocytic infiltration and/or tissue
destruction (14 reconstituted and 6 age-matched normal mice were
analyzed). Ovaries, thyroid/parathyroids, and skin appeared normal in
all mice and there was no extensive lymphocytic infiltration of
parenchymal tissue or damage to tissues that are highly sensitive to
autoimmune destruction, such as glomeruli and pancreatic islets (Fig. 2, A–D). We did
find some weak lymphocytic infiltration of the intestinal lamina
propria in a few mice, and some focal perivascular nodules of
lymphocytic infiltration in one or more organs, including the kidney,
pancreas, heart, lung, and liver of all mice. However, these types of
focal infiltrates were also found in all similarly aged normal
mice.
|
E) and no infiltration of the graft was
evident on histological examination, which strongly suggests that these
mice were solidly tolerant of their own (female) adult skin, though
they had rejected the male skin. The simplest explanation for our results would be that tolerance to skin Ags is determined centrally. Although several Ags (e.g., insulin, myelin basic protein, and ocular S-Ag), the expression of which was thought to be restricted to a peripheral tissue, were found on further examination to be expressed at low levels in the thymus (33, 34, 35), there is some evidence for a skin Ag that is not on thymic epithelium- (36) or bone marrow-derived cells (37, 38). Thus, in the face of such potential skin-specific Ags, the newly arising T cells in adult mice appeared to become tolerant of female skin, but rejected the male skin.
To test for self tolerance to a second tissue, we challenged a set of
B6 nude female mice that had been reconstituted with a B6 neonatal
female thymus by grafting with a syngeneic ovary. These mice showed the
same strong tolerance for a peripheral organ, as there were no signs of
infiltration or damage to the graft (Fig. 2F) or to host
ovaries. Taken together, these data establish that peripheral tolerance
does not require a neonatal tolerance window. Tolerance was established
despite the fact that the newly arising T cells developed in an adult
body. The only reactivity we saw was to the long-standing, male skin
graft.
Constitutive presentation of skin graft Ag in draining lymph nodes by host APC
"Ignorance" models postulate that peripheral tissue Ags go
undetected by T cells because APCs do not capture them and they
consequently do not make it into lymphoid tissue (Ags in lymphoid
tissue being immunogenic). However, our healed grafts did not go
undetected but were rejected. To test whether the graft Ags were
carried into lymphoid tissue, perhaps by a mechanism unanticipated by
ignorance models, we used a sensitive in vitro test for the presence of
H-Y. Five months after grafting male skin onto immunoincompetent hosts,
we tested the cells from nodes draining the skin grafts as APCs in
titrated in vitro tests, asking whether Rachel, a
CD4+ T cell clone specific for
H-Y/I-Ab, would respond. Further, to test whether
such presentation was attributable to the migration of any remaining
donor APC or whether host APCs were mostly involved, we also tested the
APCs from nodes draining an MHC class II-negative skin graft. We found
that the graft Ag was indeed presented in the draining nodes and that
host, rather than donor, APCs accounted for most of this presentation.
Fig. 3 shows that Rachel responded well
to lymph node cells draining a male graft, whether or not the graft
expressed MHC class II, but not to lymph node cells from ungrafted
female control mice (Fig. 3, A and B). Thus, host
APCs are able to present graft Ags as long as 5 mo after grafting. By
comparison with titrated male lymph node cells, where every APC is
expected to express H-Y, the APCs in nodes draining the male skin
grafts were 60 fold less efficient, cell for cell. If we assume that
the non-APCs in the nodes have little positive or negative effect in
our assay, this suggests that about one in every 60 host APCs presented
enough H-Y Ag to stimulate the T cell clone.
|
C shows that this was not the
case. The cross-presentation by host APC was immunogenic not only in
vitro, but also in vivo, as even well-healed, MHC-II-negative male skin
grafts, where only host APCs can present to Th cells, were rejected on
reconstitution with naive female T cells. Thus, the H-Y Ag was present
in the draining nodes, it was present on host APCs, and it was
immunostimulatory. The rate of rejection of healed grafts was
significantly slower than that of the fresh grafts
(p = 0.03, logrank analysis), which may
indicate that the fresh surgical damage and consequent tissue necrosis
(40) provided a greater stimulus for APC activation and
rejection.
It has been demonstrated that dendritic cells in nodes can
cross-present Ags captured from other dendritic cells. Because the
lifespan of these cells is not known, there was a remote
possibility that the immunogenic H-Y Ag in the draining nodes had been
captured from dendritic cells migrating from the grafts early after
grafting and that this Ag was still being presented 5 mo later. To test
whether the presence of the Ag in the nodes required the continued
presence of the graft, we gave nude female mice a male skin graft and
later removed the male graft (and replaced it with female skin), or
not, and then injected female lymphocytes to see whether they could
become primed to the graft Ag in the absence of the graft. Fig. 3D shows that female T cells were primed to make
anti-H-Y CTL when the graft was left on or when it was removed 1
day before injection of the female lymphocytes. However, they were not
primed if the graft was removed 6 wk before female lymphocyte
injection, indicating that the functional lifespan of activated APC in
the draining lymph node is relatively short. Similar results were
obtained in RAG-KO female hosts when a RAG-KO male skin graft was
removed before T cell injection (not shown and 59). Thus, it
appeared that host APCs continue to capture and present graft Ags on
both MHC class I and II for months after transplantation.
Long-standing skin grafts are not completely normal
Though the constitutive cross-presentation shown here is not predicted by the ingnorance model, it is hardly surprising, as Kurts and colleagues have shown a similar phenomenon for OVA expressed transgenically on islets and kidney cells (25). However, the transgenically expressed Ag in their study led to a cross-presentation that was tolerogenic, whereas constitutive cross-presentation from a long-established graft in our study was instead immunogenic. One possibility for the difference comes from the danger model, which postulates that signals from stressed cells should activate local APCs, such that they should present captured Ags along with costimulatory signals to passing T cells (23, 40, 41, 42). This impelled us to have another look at the assumptions under which we were operating. Up to this point, we had assumed that a long-standing graft is as healthy as normal tissue. For time-based models or ignorance models, the health of a grafted tissue is not terribly relevant to whether it can induce tolerance. However, to some of the context-based models, the health of the graft is crucial. Therefore, we decided to have a closer look to determine whether the skin grafts were indeed as quiescent as normal skin or whether there may be some differences between the graft and normal skin that could explain rejection.
To minimize any potential positive or negative effects attributable to
an adaptive immune response, we took samples from graft recipients that
had not been reconstituted with T cells, thus ensuring that there was
no interference by T cells capable of responding to the graft.
Furthermore, because normal mouse skin contains 
TCR dendritic
epithelial T cell (DEC) populations that may contribute to the normal
healing process by producing epithelial cell growth factor
(43), we used TCR-KO female recipients and male donors,
which lack mainstream TCR
T cells but contain the DEC cells
and therefore should heal normally. Fig. 4 shows that after reconstitution with
naive female lymphocytes, all of the grafts were rejected, illustrating
that long-healed male skin is rejected even in the presence of the
DECs.
|
,
IL-6, TNF-, GM-CSF) were similar between normal and long-standing
grafted skin when analyzed by conventional (nonquantitative) PCR.
However, a more sensitive method may be necessary to detect relevant
differences between grafted and normal skin. When we used real-time
quantitative PCR to look at the mRNA levels of several molecules known
to be expressed in damaged or recovering cells, or known to be involved
in inflammation and immunity, we initially saw some striking
differences.
Normalizing mRNA levels to GAPDH, we compared long-healed grafts (4 mo
after transplantation) with freshly grafted skin (4 days after
transplantation, a positive control for surgery-induced damage and
inflammation); and with normal skin from the grafted recipients. Fig. 5A shows that long-healed
grafts had higher levels than normal skin for 6 of the 10 messages. To
verify these increases in stress/inflammation molecules, we analyzed
more healed grafts and normalized the mRNA in relation to both GAPDH
and cyclophilin expression. Fig. 5B shows that when
normalized to GAPDH, the increases in mRNA for stress/inflammation
molecules were very reproducible. However, these increases were far
less apparent when the mRNA was normalized to cyclophilin expression.
The level of cyclophilin mRNA expression was approximately the same in
normal skin and the long-standing graft, whereas GAPDH was generally
lower in the long-standing graft. Thus, there appears to be a decrease
in GAPDH expression in the healed skin grafts when compared with
normal skin.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
These are not the first results to challenge the time-based models of tolerance. For example, Lambert’s group showed that neonatally tolerant mice are tolerant at the level of CTL but not Th cells (45), Coutinho’s group showed that they contain large numbers of polyclonally activated cells (46), and Mahana et al. (3) found that normal neonates are able to make Abs against self Ags. Furthermore, it was recently shown that neonates can make normal CTL and Th responses if given the appropriate Ag doses, adjuvants, or APCs (4, 5, 6). In response to these results, proponents of time-based models have argued that neonates are not appropriate test animals because the tolerogenic period is already over in neonatal animals, and that the Ag must therefore be introduced even earlier if it is to be obligatorily tolerogenic (9, 17, 18). However, this view also has been challenged in several ways. As early as 1979, it was shown that skin grafts that had been grafted onto nude mice before reconstituting the recipients with a thymus graft were rejected by the newly generated T cells (11), and similar results were later obtained in SCID and RAG-KO mice (10, 15). Furthermore, grafts given to chicken or sheep embryos before the development of immunocompetence also were rejected when the animals developed immune competence (12, 13, 14). However, in all of these studies, the grafts differed from the hosts at MHC and/or multiple minor loci, and therefore the frequency of anti-graft T cells was most likely greater than the frequency of T cells to any peripheral tissue-specific Ag. Therefore, it could be argued that these studies were not valid tests of those time-based models in which tolerance is thought to occur as a result of a low frequency of Th cells.
H-Y is a single, weak minor histocompatibility Ag. In unprimed mice, the frequency of responding T cells is undetectably low. It lacks environmental mimics (7, 24), and the response requires cellular collaboration between T cells (47, 48, 49, 60). In these ways, it is quite likely to mimic a conventional peripheral self-Ag, and for these reasons, both Bretscher (9) and Bingaman et al. (10) suggested that the ultimate test of peripheral tolerance models should be done with H-Y. Nonetheless, by using this antigenic system, we found that both mature naive, and, most surprisingly, newly arising T cells, rejected the long-standing male skin grafts, while at the same time, these T cells became tolerant of their own tissues.
None of the current time-based models are easily able to incorporate our data. Most time-based models would have predicted that newly arising T cells would have accepted the long-standing male skin grafts. Some of the models (9, 16, 17) might argue that the male grafts were too small to serve as an adequate source of Ag during the early tolerizable-only period in which the frequency of T cells is low. Though we cannot completely rule this out, it seems very unlikely, given that graft Ags were readily detected in the draining lymph nodes and the male grafts were quite large (up to 2 cm in diameter) and that other, much smaller organs, such as ovaries and islets, were tolerated. A more recent time-based model, which suggests that establishment of tolerance to skin occurs only in the neonate because neonatal skin allows traffic of T cells whereas adult skin does not (1) would have suggested that the newly arising T cells in an adult body would have rejected both the male and female adult grafts. Our data show, to the contrary, that tolerance to skin can be established to adult skin in an adult animal and does not require any property unique to neonatal tissues.
Our data are also difficult for the ignorance model of Zinkernagel (21), which would have predicted that the newly arising T cells would have "ignored" both the male and female grafts. This model proposes that peripheral Ags are not captured by host APCs and presented on MHC class I molecules unless the Ag is directly inserted into an APC, for example, by virus infection. Clearly, however, like many other minors that have been shown previously to be cross-presented in vivo (50, 51), the H-Y Ag reached the draining nodes, was presented by host APCs, and was able to activate CTL there. Thus, the ignorance model will need to be provided with additional or amended assumptions to explain these data.
Although apparently ruling out existing time-based models, our data also do not fit easily with the Danger model, which suggests that tolerance will be established to an Ag unless APC costimulation is induced by signals from stressed or damaged tissues (22, 23). Though the longstanding grafts were grossly normal macroscopically, histologically, and by conventional PCR (Ref. 10 and our unpublished data), when analyzed with a sensitive quantitative PCR assay, the level of expression of GAPDH was decreased in comparison with normal skin. It remains possible that abnormal expression of other molecules not yet examined may provide a better explanation for the immunogenic cross-presentation leading to graft rejection that we have observed. Rejection was not the result of long-lived APC that were stimulated to cross-present graft Ags at the time of grafting. Instead, we found that host APCs had a short functional lifespan (<6 wk), but the graft Ags were continuously presented in the draining lymph nodes for at least 5 mo after grafting. Thus, there is likely to be continuous migration of activated APCs that have captured skin graft Ags and, unlike APCs migrating from normal internal tissues such as kidney and pancreas (25, 52), these APCs are immunogenic.
Finally, the finding that the female graft was accepted whereas the male was rejected although both tissues potentially continue to generate activated APCs, suggests that tolerance can be established in the face of costimulation. The most likely explanation for this paradox lies in the size (antigenic load) of the two different tissues. Consistent with this view, it has been shown that increasing the size of a skin graft decreases the likelihood of rejection (53, 54). In our experiments, the H-Y Ag expression is restricted to the male skin graft while any skin-specific Ags on the syngeneic female graft are present not only on the graft but also on the rest of the host skin, a very large tissue. Should autoreactive T cells be activated by the syngeneic graft or by a few activated APC draining normal skin tissue, they would kill a few target tissue cells, as well as the activated APCs that stimulated them. However, because CTL killing induces apoptosis and such apoptotic death does not activate resting APCs, the self Ags in the dying tissues would be presented by nonactivated APCs (40, 55, 56), which would be tolerogenic. Therefore, the response would end and tolerance would eventually result. This would not occur for the male graft because the tissue expressing the Ag is much smaller, allowing the graft to be eliminated before the Ag presentation becomes tolerogenic.
Altogether, these studies demonstrate that it is not the timing of Ag exposure that determines the induction of tolerance vs reactivity. We are left with the context of Ag presentation as the determining factor. Dissecting the difference between tolerogenic and immunogenic cross-presentation by APC in lymphoid tissue is a critical next question. By characterizing the differences, we may find ways to manipulate Ag presentation to produce tolerance vs immunity at our discretion.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 Abbreviations used in this paper: B6, C57BL/6; KO, knockout; B10-RAG, B10-RAG-2-KO; DEC, dendritic epithelial T cell.
Received for publication May 15, 2000. Accepted for publication January 4, 2001.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  C. Schroder, R. N. Pierson III, B.-N. H. Nguyen, D. W. Kawka, L. B. Peterson, G. Wu, T. Zhang, M. S. Springer, S. J. Siciliano, S. Iliff, et al. CCR5 Blockade Modulates Inflammation and Alloimmunity in Primates J. Immunol., August 15, 2007; 179(4): 2289 - 2299. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. F. LaRosa, A. H. Rahman, and L. A. Turka The Innate Immune System in Allograft Rejection and Tolerance J. Immunol., June 15, 2007; 178(12): 7503 - 7509. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. F. Nawar, S. Arce, M. W. Russell, and T. D. Connell Mutants of Type II Heat-Labile Enterotoxin LT-IIa with Altered Ganglioside-Binding Activities and Diminished Toxicity Are Potent Mucosal Adjuvants Infect. Immun., February 1, 2007; 75(2): 621 - 633. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Wang, S. S. Lee, C. Dell'Agnello, V. Tchipashvili, J. D'Avilla, E. Czismadia, B. Y. Chin, and F. H. Bach Bilirubin Can Induce Tolerance to Islet Allografts Endocrinology, February 1, 2006; 147(2): 762 - 768. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Sho, K. Kishimoto, H. Harada, M. Livak, A. Sanchez-Fueyo, A. Yamada, X. X. Zheng, T. B. Strom, G. P. Basadonna, M. H. Sayegh, et al. Requirements for induction and maintenance of peripheral tolerance in stringent allograft models PNAS, September 13, 2005; 102(37): 13230 - 13235. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. J. Bell, B. Min, R. K. Gregg, H.-H. Lee, and H. Zaghouani Break of Neonatal Th1 Tolerance and Exacerbation of Experimental Allergic Encephalomyelitis by Interference with B7 Costimulation J. Immunol., August 15, 2003; 171(4): 1801 - 1808. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. E. Phillips, T. G. Markees, J. P. Mordes, D. L. Greiner, and A. A. Rossini Blockade of CD40-Mediated Signaling Is Sufficient for Inducing Islet But Not Skin Transplantation Tolerance J. Immunol., March 15, 2003; 170(6): 3015 - 3023. [Abstract] [Full Text] [PDF]
|
|
; female grafts)
later the grafts were removed and RNA extracted for real-time
quantitative RT-PCR analysis of the indicated mRNA by using the
Perkin-Elmer TaqMan EZ RT-PCR kit and normalization of mRNA amounts
based on GAPDH expression (see Materials and Methods).
For comparison, normal host skin, adjacent to the graft was also
analyzed (
). Results for individual mice are depicted and are
expressed in message units as determined from comparison with standard
curves specific for each gene. Because different genes had different
scales of amplitude, due to the specific efficiency of their primers
and probes, we multiplied or divided their values by 10 so that the
results of the normal skin samples were around 1 to be able to present
them in the same figure. B, Female immunodeficient
recipient mice (B6-TCR
, •) expression (this panel also
includes values for the three mice analyzed in A), and
in the bottom panel, cyclophilin (
,