The goal of this study was the development of a system in which the cooperative interactions between CD4 and CD8 T cells specific for defined peptides from a single minor histocompatibility antigen could be studied. A transgenic mouse strain that expresses chicken ovalbumin (Act-mOVA) on the surface of all cells in the body was produced as a source of tissues containing such an antigen. Skin grafts from Act-mOVA donors were rapidly and completely rejected by wild-type recipients, but only when both CD4 and CD8 T cells were present. CD4 T cells by themselves caused an incomplete form of rejection characterized by rapid but partial contraction of Act-mOVA grafts. CD8 T cells alone caused complete rejection of Act-mOVA skin grafts but only after a long delay. Adoptively transferred ovalbumin-specific TCR-transgenic CD4 and CD8 T cells were stimulated by Act-mOVA graft antigens and CD8 T-cell accumulation in the grafts was enhanced by specific CD4 T cells. These findings, together with the fact that the ligand for ovalbumin peptide-specific CD8 T cells can be detected in Act-mOVA tissues with an MHC-restricted antibody, make this an ideal system for the study of cooperation between CD4 and CD8 T cells.
T cells are essential for allograft rejection (1). In many situations where the donor and recipient differ with respect to major or minor histocompatibility antigens, either a CD4 or CD8 T-cell subset alone is sufficient to cause rejection (2,3). Both subsets are required, however, in certain situations, usually where the donor and recipient differ with respect to only a few minor antigens (4,5). For example, grafts expressing antigens capable of stimulating only CD8 or CD4 T cells are poorly rejected unless an additional epitope capable of activating the other T-cell subset is included (6–9). These findings indicate that synergistic interactions between graft antigen-specific CD4 and CD8 T cells are required for graft rejection under conditions where antigenic disparity is limited. Such interactions may also explain the finding that CD4 T cells are required for CD8 T cells to efficiently eliminate certain viruses (10,11) or tumors (12).
Understanding the interactions between graft-antigen specific CD4 and CD8 T cells is important for developing better means to prevent transplant rejection. However, for most naturally occurring minor antigens where rejection depends on both CD4 and CD8 T cells, the nature of the relevant proteins and/or peptides is unknown (13). Thus, an in vivo system where the specific CD4 and CD8 T cells, as well as the antigenic targets of these cells, can be followed would be helpful. One approach to this problem has been the use of transplants from transgenic animals expressing a known foreign protein. While several studies of this sort have been performed, they have not shed light on the interactions of CD4 and CD8 T cells because either both T-cell populations were not required for rejection (3,14) or the antigen-specific T cells could not be identified in vivo (9). Although the latter problem has been addressed by the adoptive transfer of T cells of known specificity from TCR-transgenic mice into recipients of grafts expressing the relevant MHC I alloantigen (15,16), these studies did not focus on rejection of minor antigen–disparate transplants or interactions between T-cell subsets. Another recent approach has been the use of peptide-MHC tetramers to track minor antigen-specific T cells during rejection (17), but to date only the CD8 T-cell response has been studied. To begin to address these limitations, we describe here a novel system in which chicken ovalbumin-expressing skin grafts from a transgenic mouse line are rejected via a mechanism that depends on both CD4 and CD8 T cells, and which allows in vivo tracking of the graft antigen-specific T cells and the antigenic peptide-MHC I target.
Materials and Methods
Transgenic mice expressing membrane-bound chicken ovalbumin (Act-mOVA) were generated by the University of Minnesota Mouse Genetics Laboratory as described later. C57BL/6 (B6) mice used as graft donors and recipients were purchased from the National Cancer Institute (Frederick, MD). B6 MHC II-deficient mice were purchased from Taconic (Germantown, NY). B6.PL (Thy1.1 congenic), B6.SJL (CD45.1 congenic), B6 B cell-deficient, B6 αβ-T cell-deficient, and B6 CD8-deficient mice were purchased from The Jackson Laboratory (Bar Harbor, ME). OT-I (18) and OT-II (19) TCR-transgenic mice were kindly provided by Matthew Mescher (University of Minnesota) and Leo Lefrancois (University of Connecticut), respectively, and crossed to the congenic strains in our facility.
Generation of Act-mOVA transgenic mice
A PCR product containing the full-length chicken ovalbumin cDNA was generated using the pAc-neo-OVA plasmid (20) as a template and a forward primer that incorporated the leader sequence from the H-2Kb gene (21). A second PCR product, containing the short extracellular spacer and transmembrane region of H-2Db (21), was generated using pCMU-Db (kindly provided by Stephen Jameson, University of Minnesota) as a template (22). After cloning, these two PCR products were linked together via an XbaI restriction site to produce a construct containing the ovalbumin gene with the attached transmembrane anchor. EcoRI sites at either end of this plasmid allowed insertion into the pCAGGS vector (23) containing the CMV-IE enhancer, chicken β-actin promoter, and rabbit β-globin polyA tail to produce pODpCAGGS (Figure 1A). The purified, SalI-HindIII-digested transgene was used to generate transgenic B6 mice. Founders were identified by Southern blotting of tail DNA using the ovalbumin transgene as a probe. Two transgenic founders (Act-mOVA-I and Act-mOVA-II) were selected for further study and maintained as heterozygotes through matings with wild-type B6 animals.
Ovalbumin was detected in the tissues of transgenic mice by conventional immunohistochemistry. Frozen tissues embedded in OCT compound (Allegiance Health Care Corporation, Waukegan, IL) were cut into 10-μm sections, dehydrated for 10 min in acetone, fixed for 15 min in 1% formaldehyde, and treated to reduce nonspecific binding and background owing to endogenous peroxidase and biotin, as described previously (24). Sections were then incubated for 60 min with a polyclonal rabbit anti-chicken ovalbumin antibody (Calbiochem, La Jolla, CA) conjugated with Sulfo-NHS-biotin (Pierce, Rockford, IL). Detection of bound antibody was carried out by incubating the sections with ABC-Elite complex and 3,3-diaminobenzidine (Vector Laboratories, Burlinghame, CA) for 30 min and 5 min, respectively. Sections were counter-stained with hematoxylin. Images were acquired on a Zeiss Axioplan 2 microscope (Carl Zeiss Incorporated, Thornwood, NY) and captured with a SPOT camera (Diagnostic Instruments Incorporated, Sterling Heights, MI).
Binding of the 25-D1.16 antibody specific for ovalbumin 257–264-Kb complexes (kindly provided by Kristin Hogquist, University of Minnesota) was detected by immunofluorescence. Tissue sections were incubated with biotin-labeled 25-D1.16 for 30 min followed by sequential incubation in HRP-labeled streptavidin (30 min) and Cy3-labeled tyramide (6 min) from the TSA-Biotin kit (PerkinElmer Life Sciences, Boston, MA), followed by the nuclear stain 4′,6-diamidine-2′-phenylindole (DAPI; Vector Laboratories). Slides were mounted with Vectashield (Vector Laboratories) to preserve fluorescence. Two sets of digital images were acquired using a Bio-Rad MRC-1000 confocal microscope equipped with a krypton/argon laser and CoMOS vs. 7.0a software (Bio-Rad Life Sciences, Hercules, CA); one in the Cy3 channel and one in the DAPI channel. Photoshop 5.5 software (Adobe Systems, San Jose, CA) was used to assemble separate grayscale versions of the images, which were then pseudo-colored red and blue, respectively, and overlaid. On some sections, tissue autofluorescence in the FITC channel was collected and used for morphology in the place of DAPI.
Lymph node cells from B6 mice (2 × 105 per well), OT-I, or OT-II mice (5 × 104 per well) were cultured in triplicate in 96-well round bottom plates (Corning Incorporated, Corning, NY) with 5 × 105 irradiated splenocytes (3000 rads) from Act-mOVA or B6 mice. The cells were cultured in complete medium for 2–4 days at 37 °C, 5% CO2; the last 16 h in the presence of 3H-thymidine (1 μC/mL). Incorporation of 3H-thymidine into DNA was determined by liquid scintillation counting.
Full-thickness tail skin grafts (0.5–1 cm2) were placed in a graft bed on the right lateral thorax of recipient mice and secured with Nexaband surgical glue (Veterinary Products Laboratories, Phoenix, AZ). Antibiotic ointment, Vaseline gauze, and a Band-Aid were applied then secured with Elastikon tape (Johnson & Johnson, Arlington, TX). Bandages were removed after 10 days and rejection was scored as the day when the graft was no longer attached to the recipient. Graft surface area was measured with calipers at the time of surgery and at later times starting on the day the bandages were removed. Statistical significance between recipient groups was determined by the Wilcoxon rank-sum test (25) at α= 0.05.
Lymph node and spleen cells from OT-I or OT-II mice were treated with anti-CD4 or anti-CD8, antibodies, respectively, and then rabbit complement (CedarLane Laboratories, Ontario, Canada). Small samples were stained with fluorochrome-labeled anti-CD8 and anti-TCR Vα2 or anti-CD4 and anti-TCR Vα2 antibodies (BD PharMingen, San Diego, CA) and analyzed by flow cytometry. The percentages of CD8+, Vα2+ OT-I cells or CD4+, Vα2+ OT-II cells were then used to calculate the number of transgenic cells for adoptive transfer. Unirradiated recipient mice received 2 × 106 OT-I and/or OT-II cells via the tail vein. For immunohistochemical experiments, OT-I cells were purified by negative selection using magnetic beads (Miltenyi Biotec, Auburn, CA) coated with biotin-labeled anti-HSA, anti-B220, anti-CD11b, anti-CD4, anti-Gr-1, and anti-CD16/CD32 antibodies (BD PharMingen) to remove any CD45.1+ cells that were not CD8 T cells.
The number of OT-I or OT-II cells present in the graft-draining lymph nodes (pooled right axillary and brachial) of recipient mice at various times after adoptive transfer was determined by multiplying the percentage of CD45.1+, Vα2+, CD8+ cells or Thy1.1+, Vα2+, CD4+ cells determined as described earlier by the total number of viable lymphocytes. The number of OT-I or OT-II cells present in Act-mOVA skin grafts was determined by immunohistology. Graft tissue including a surrounding cuff of recipient skin and the underlying muscle was sectioned, treated and stained with biotin-labeled anti-Thy 1.1 (for detection of OT-II) antibodies, HRP-labeled streptavidin, and Cy3-labeled tyramide as described earlier. Residual HRP and free biotin sites from this step were then blocked by a 60-min incubation in 1% H2O2, 0.1% sodium azide, and then avidin and biotin (Vector Laboratories). Biotin-labeled anti-CD45.1 (for detection of OT-I) was then added, followed by HRP-labeled streptavidin, biotin-labeled tyramide (PerkinElmer Life Sciences), and Cy5-labeled streptavidin (Caltag, South San Francisco, CA). Separate Cy3, Cy5, and FITC (to capture autofluorescence for tissue morphology) images covering the length of the graft were acquired using a Bio-Rad MRC-1000 confocal microscope as described earlier. The number of OT-I or OT-II cells present in each section was then determined as described by Reinhardt et al. (24), then multiplied by the calculated number of sections in the graft (10-μm section-thickness multiplied by the graft length at the time of excision). The average number of total cells calculated from two sections through each graft, each at least 100 sections apart, was reported as the total number of cells in that graft. Multiple sections through different grafts at various time points revealed that cells were evenly distributed throughout the width of an individual section.
Act-mOVA transgenic mice express membrane ovalbumin, which functions as a minor histocompatibility antigen
A transgenic mouse displaying chicken ovalbumin on the surface of most cells of the body was produced to serve as a donor of tissue expressing a single, defined minor histocompatibility antigen. Ovalbumin was chosen because it contains peptides that are recognized by CD8 and CD4 T cells. Peptide 254–267 (SIINFEKL) binds to Kb molecules (26) and is recognized by CD8 T cells, including those from the OT-I TCR-transgenic line (18). Peptide 323–339 (ISQAVHAAHAEINEAGR) binds to I-Ab and is recognized by CD4 T cells, including those from the OT-II TCR-transgenic line (19). A construct was produced in which the leader sequence from the H-2Kb gene (21), the full-length ovalbumin coding sequence (27), and the H-2Db transmembrane sequence (21) were placed under the control of the chicken β-actin promoter (23) (Figure 1A). This construct was injected into fertilized B6 (H-2b) eggs, and two transgenic lines, Act-mOVA-I and Act-mOVA-II, which transmitted the transgene were produced (Figure 1B). Similar experimental results were obtained with either line, and for convenience both lines will be referred to hereafter as Act-mOVA mice.
Ovalbumin expression was detected by immunohistology in all organs from Act-mOVA mice (not shown). In skin, ovalbumin was heavily expressed in the epidermis, including the epithelial cells forming the hair follicles, and in the dendritic cells scattered throughout the dermis of Act-mOVA (Figure 1C), but not wild-type B6 (Figure 1D) mice. The 25-D1.16 antibody (28) was used to assess which of these ovalbumin-expressing cells were producing ovalbumin 257–264-Kb complexes. Despite the fact that many cell types in Act-mOVA skin expressed ovalbumin, only cells with dendritic morphology (probably bone marrow-derived dendritic cells) were stained with the 25-D1.16 antibody, indicating that these cells displayed the highest amounts of ovalbumin 257–264-Kb complexes (Figure 1E). This staining was specific, as evidenced by the finding that no staining was detected in the skin of wild-type B6 mice (Figure 1F).
In vitro stimulation of specific T cells was used to determine whether or not APC from the spleens of Act-mOVA mice constitutively process ovalbumin and display ovalbumin peptide-MHC complexes. Splenocytes from Act-mOVA mice but not wild-type B6 mice stimulated the proliferation of naive T cells from the OT-I and OT-II lines, demonstrating that Act-mOVA splenocytes display 254–267-Kb and 323–339-I-Ab complexes (Figure 1G). Act-mOVA splenocytes failed to stimulate significant proliferation of naive B6 T cells (Figure 1G), consistent with previous reports that the frequency of minor histocompatibility antigen-specific T cells in naive individuals is very low (29,30).
Act-mOVA skin grafts are rejected in a T cell-dependent manner
The constitutive display of ovalbumin peptide-MHC complexes by cells from Act-mOVA mice raised the possibility that wild-type B6 mice would reject Act-mOVA skin grafts. Full-thickness tail skin grafts from Act-mOVA mice were rejected by B6 recipients [Figure 2A; mean survival time (MST) ± SEM = 33.9 ± 4.3 days, n = 22], whereas control B6 isografts were accepted indefinitely (MST > 200 days, n = 10). Rejection was likely immune-mediated because mice that had rejected an Act-mOVA skin graft showed accelerated rejection of a second Act-mOVA skin graft (MST = 18 ± 1.7 days, n = 4; not shown).
The immunological basis for the rejection of Act-mOVA skin grafts was assessed using gene-targeted mice as recipients. Although B6 recipients of Act-mOVA skin grafts produced serum antibodies specific for ovalbumin, antibodies were not required for rejection, as four B cell-deficient μMT mice rejected their Act-mOVA skin grafts in less than 30 days (not shown). In contrast, three αβ-T cell-deficient TCRα–/– mice accepted their Act-mOVA skin grafts for greater than 200 days, indicating a critical role for αβ-T cells (not shown). As depicted in Figure 2(A), recipients that lacked CD4 T cells (MHC II-deficient mice) or CD8 T cells (CD8-deficient mice) were greatly impaired in their ability to reject Act-mOVA grafts. These results showed that both CD4 and CD8 αβ-T-cell subsets, but not B cells, were essential for complete and rapid rejection of Act-mOVA skin grafts.
Although neither CD4 nor CD8 T cells were capable of mediating complete and rapid skin graft rejection on their own, measurement of graft size revealed that subtle forms of rejection were in fact occurring in mice containing only one T-cell subset. B6 isografts shrank approximately 20% during the first few days after the bandages were removed on day 10 (Figure 2B–D, dotted lines), regardless of the recipient genotype, presumably as a nonimmunological consequence of wound healing. Thereafter, the isografts remained constant in size and healthy in appearance for the remainder of the observation period of 200 days. As expected, all Act-mOVA skin grafts shrunk progressively until finally sloughing off of wild-type B6 recipients containing both CD4 and CD8 T cells (Figure 2B, solid lines; note that the final two grafts rejected on days 84 and 128). In contrast, Act-mOVA skin grafts placed on recipients lacking CD4 T cells were indistinguishable from healthy isografts for ∼40 days, after which they contracted progressively (Figure 2C, solid lines) such that 25% were lost completely after 115 days (data not shown). Act-mOVA skin grafts placed on most recipients lacking CD8 T cells shrank rapidly to 10–30% of their original size but were then retained indefinitely in an apparently healthy condition (Figure 2D, solid lines). These results suggested that CD4 T cells cause rapid damage to Act-mOVA skin grafts, but in the absence of CD8 T cells cannot cause complete elimination of graft tissue. In contrast, CD8 T cells are capable of complete elimination of graft tissue in the absence of CD4 T cells, but do so slowly and inefficiently. Together, these findings supported the possibility that although the individual T-cell populations could mediate some aspects of rejection, additive or synergistic effects of both T-cell subsets are required for maximal and rapid rejection.
Tracking ovalbumin-specific T cells after adoptive transfer
As a first step toward understanding the roles for CD8 and CD4 T cells, a small number of cells from the OT-I and/or OT-II TCR-transgenic lines were transferred into wild-type B6 recipients 1 day before they received Act-mOVA or control B6 skin grafts. This adoptive transfer approach was used to track the T-cell response because endogenous ovalbumin peptide-MHC-specific T cells are difficult to detect in naive mice (31–33).
As shown in Figure 3(A), the B6 mice that received either OT-I cells or OT-II cells alone rejected Act-mOVA skin grafts faster than the B6 mice that did not receive TCR-transgenic T cells (MST = 21.7 ± 1.1 days for recipients of OT-I cells and 17.0 ± 0.4 days for the OT-II recipients compared with 32.3 ± 2.7 days for the untransferred recipients). The rate of rejection was accelerated even further in the B6 mice that received OT-I and OT-II cells together (MST = 14.2 ± 0.4 days). Therefore, both TCR-transgenic T-cell populations participated in the rejection of Act-mOVA skin grafts and maximal rejection was achieved with collaboration between the two cell types. These results validated the Act-mOVA system as a model for studying the rejection of minor-mismatched skin and supported using the OT-I and OT-II cells after adoptive transfer as indicators of the endogenous antigen-specific T-cell response during graft rejection.
Flow cytometric analysis of single-cell suspensions showed that small numbers of OT-I (CD45.1+, Vα2+, CD8+) and OT-II cells (Thy1.1+, Vα2+, CD4+) were present in the lymph nodes of B6 mice (which express CD45.2 and Thy 1.2) 1 day after adoptive transfer, but were not detectable in the lymph nodes of untransferred mice (Figure 3B, middle and left panels). As seen in the in vitro assay, both OT-I and OT-II cells in the transferred mice proliferated in the lymph nodes draining the site of an Act-mOVA skin graft, indicating that the T cells were stimulated by their cognate peptide-MHC complexes in vivo (Figure 3B, right panels). By 9 days after transplantation of an Act-mOVA graft, OT-I cells increased dramatically in the lymph nodes of mice that received only OT-I cells, and to a slightly greater extent in mice that also received OT-II cells (Figure 4A). OT-II cells also accumulated to large numbers in the lymph nodes 9 days after placement of Act-mOVA skin grafts but did not benefit from the presence of OT-I cells (Figure 4A). Therefore, ovalbumin-specific CD8 and CD4 T cells accumulated in the lymph nodes that drained the Act-mOVA skin grafts.
Neither OT-I nor OT-II cells were detected in day 3 Act-mOVA skin grafts on mice that received both populations of transgenic T cells (Figures 4B,C). By day 9, however, many OT-I and OT-II cells had entered the Act-mOVA skin grafts on mice that received either OT-I or OT-II cells alone (Figure 4C). Interestingly, the number of OT-I cells that infiltrated the Act-mOVA skin grafts was increased by approximately threefold in mice that also contained OT-II cells. As in the lymph nodes, the number of OT-II cells that entered the Act-mOVA skin grafts was similar whether or not OT-I cells were present. Therefore, ovalbumin-specific CD4 and CD8 T cells accumulated in Act-mOVA skin grafts, and the CD8 T cells did so to a greater extent in the presence of ovalbumin-specific CD4 T cells.
To test the possibility that ovalbumin 257–264-Kb-specific CD8 T-cell entry into the grafts could contribute to rejection by cytolytic destruction of graft parenchymal cells, the display of ovalbumin 257–264-Kb complexes within the grafts was measured using the 25-D1.16 antibody. Three days after transplantation, ovalbumin 257–264-Kb complexes were detected primarily on dermal dendritic cells (Figure 4D, left panel) as seen in the intact donor skin from Act-mOVA mice (Figure 1E). The 25-D1.16 antibody only stained cells within the graft; underlying and adjacent recipient cells were not stained (Figure 4D). Remarkably, by day 9 after transplantation ovalbumin 257–264-Kb complexes were also present on epithelial cells in the hair follicles and epidermis of Act-mOVA grafts (Figure 4D, right panel). These results showed that ovalbumin 257–264-Kb complexes, which are normally displayed primarily on dendritic cells, are induced on parenchymal cells during the course of rejection of Act-mOVA skin grafts. Additionally, infiltrating ovalbumin 257–264-Kb-specific CD8 T cells would have access to cytolytic targets in the graft and, potentially, greater numbers of CD8 T cells would be in proximity to targeted cells if ovalbumin-specific CD4 T-cell help was involved in the response.
Act-mOVA skin grafts underwent the relatively slow course of rejection observed for skin grafts expressing other natural or artificial minor histocompatibility antigens (3,34–37). In addition, like several other minor antigens (6–9), complete rejection of Act-mOVA grafts depended on both CD4 and CD8 T cells. Both T-cell subsets may be required for rapid and complete rejection in this system because they contribute different but essential mechanisms of graft damage. The rapid but incomplete loss of graft tissue caused by CD4 T cells may be related to the fact that most cells in Act-mOVA grafts do not express MHC II molecules. Thus, although CD4 T cells may recognize ovalbumin peptide-MHC II complexes on dendritic cells and macrophages in Act-mOVA grafts, causing the production of toxic mediators that damage the graft in a bystander fashion, CD4 T cells can never directly kill parenchymal cells in the graft. Once donor ovalbumin-expressing APCs disappear, the bystander activation and related graft damage may subside because ovalbumin is not efficiently captured from ovalbumin-expressing parenchymal cells by recipient APCs that repopulate the graft. Conversely, although CD8 T cells have direct cytotoxic potential, their quantity and/or cytotoxic function may only become sufficient for rejection in the presence of CD4 T cells. The Act-mOVA system will provide the unique opportunity to directly measure the relevant peptide-MHC I target of the effector CD8 T cells, via the 25-D1.16 antibody.
The necessity for CD4 and CD8 T cells for maximal rejection of Act-mOVA grafts is consistent with the possibility that CD8 T cells depend on helper signals from CD4 T cells as reported for several other immune responses (38–40). Our adoptive transfer experiments indicated that graft antigen-specific CD4 T cells increase the number of graft antigen-specific CD8 T cells that accumulated in the grafts. It is also possible that CD4 T cells enhance the proliferation of CD8 T cells by producing growth factors such as IL-2 (41,42). Alternatively, CD4 T cells may enhance the survival of CD8 T cells (40), perhaps by activating CD40-expressing APCs (43–45). The adoptive transfer of TCR-transgenic T cells described here should make the Act-mOVA system useful for future studies aimed at distinguishing between these possibilities in skin grafts, vascularized organ transplants, and the setting of graft vs. host disease.
The authors acknowledge Jennifer Walter for expert technical assistance, Drs Stephen McSorley, Matthew Mescher, Stephen Jameson and Timothy Behrens for assistance and discussion, and Micki Dyer and Sandra Horne from the University of Minnesota Cancer Center ES Cell/SPF Animal Facility for production of the transgenic mice.
The work was supported by grants from the National Institutes of Health (AI35296 and AI50162 to M.K.J and AI07313 to B.D.E). Additional funding for B.D.E. was provided by a Medical Scientist Training grant from the National Institutes of Health (T32 G08244). Support for E.I was through the Minnesota Medical Foundation and the Vikings Children's Fund.