- Split View
-
Views
-
Cite
Cite
Soo Jung Yang, Sejin Ahn, Chan Sik Park, Kevin L Holmes, Jenifer Westrup, Cheong Hee Chang, Moon G Kim, The quantitative assessment of MHC II on thymic epithelium: implications in cortical thymocyte development, International Immunology, Volume 18, Issue 5, May 2006, Pages 729–739, https://doi.org/10.1093/intimm/dxl010
- Share Icon Share
Abstract
The dynamics of MHC II expression in various thymic stromal compartments was investigated. By including MHC II in flow cytometry in addition to the cortical CDR1, medullary UEA-1 and pan-epithelial G8.8 markers, thymic stromal compartments were subdivided into at least six different populations. The total level of surface and cytoplasmic MHC II from fresh cortical thymic epithelial cells (cTECs) of normal mouse was as high as MHC II levels in medullary thymic epithelial cells (mTECs). MHC II levels as well as the percentages and cycling status of thymic epithelial cell populations expressing MHC II were not static during post-natal development, suggesting quantitative flexibility in presenting signals to the developing thymocytes. Although there was no evidence found for regulation of surface MHC II levels by TCR or by IFN-γ, the absence of class II transactivator reduced both the level of MHC II expression and the number of MHC II+ cells. Surprisingly, MHC II molecules were found to form distinct focal aggregates on the surface of cTEC but not mTEC using high-resolution analysis by confocal microscopy. Moreover, these aggregates were formed independent of TCR or TCR-bearing cells in the thymus. These aggregates could potentially generate a functional unit containing a much higher local MHC II concentration to yield a higher avidity interaction. We discuss possible mechanisms for positive selection by weak interactions in the presence of such preformed MHC II aggregate units in cTEC.
Abbreviations
- AAALAC
Association for Accreditation of Laboratory Animal Care
- CIITA
Class II transactivator
- cTEC
cortical thymic epithelial cell
- DC
dendritic cell
- DP
double positive
- KO
knockout
- MFA
mean fluorescence amplitude
- mTEC
medullary thymic epithelial cell
- NIAID
National Institute of Allergy and Infectious Diseases
- NIH
National Institutes of Health
- SP
single positive
- TEC
thymic epithelial cell
- TNC
thymic nurse cell
Introduction
Thymic stromal cells provide the proper microenvironment for developing thymocytes to undergo all steps in differentiation such as lineage commitment and positive and negative selection (1–4). It was suggested that different compartments of the thymic microenvironment are specialized to support various differentiation steps: the cortical region is for positive selection and the medullary region is for negative selection (5, 6). Among thymic stromal cells, epithelial cells play a major role in thymocyte selection by presenting ligands on self-MHC molecules. These complexes initiate TCR signals, which are quantitatively and qualitatively different for each developing thymocyte. The intensity of TCR signaling is believed to be critical for thymocyte fate (1, 3, 4, 7). Only the proper amount and the right type of signaling at the right moment, through ligand-loaded MHC and TCR interaction, along with signaling through other accessory proteins on the cell surface, will allow each thymocyte to mature properly. Cortical double-positive (DP) thymocytes undergoing positive selection have lower TCR levels, while medullary single-positive (SP) mature thymocytes have higher TCR levels on their surface. This difference in surface TCR expression has been used to support an avidity model for thymic selection events (7). However, the distinct TCR patches found in contact sites between DP thymocytes and cortical thymic epithelial cells (cTECs) raised an unresolved question about the specific mechanism in the biological functions of high avidity forming structural units (8).
For epithelial cells, many histologic studies have shown weaker MHC staining in the cortical area than in the medulla (1, 9–11). It was thought that this difference in staining intensity for MHC II is also consistent with the idea that a weak TCR–MHC II interaction is essential for positive selection in the cortex, while a stronger interaction is required for negative selection in the medulla (1).
The thymic microenvironment supports thymocyte development throughout life. Hematopoietic precursor T cells are first seeded to the organ when the thymic rudiment is formed by non-hematopoietic cells around fetal day 11 in the mouse. In the initial rudiment, the epithelial cell compartments are not clearly delineated (12). As fetal thymocytes develop, however, the microenvironment organizes into more structured compartments. The medullary structure is formed last, when SP cells are present. Its formation is influenced by the presence of more mature thymocytes, a phenomenon referred to as cross-talk (13). Once full stromal compartments are established, the thymus continues to generate and export functional T cells throughout life, although its output declines slowly after puberty. It is still not clear how different epithelial compartments of the microenvironment are established or controlled at a cellular or molecular level. In this study, we investigated the profiles of MHC II in developing thymic stromal cells expressing ligands for CDR1 (cTEC) and for UEA-1 [medullary thymic epithelial cell (mTEC)] during the post-natal period. Unexpectedly, we found high surface expression in the cTEC, the ongoing compartmentalization process showed dynamic changes in the cell numbers and levels of MHC II, CDR1 and UEA-1, as well as in cell-cycle status. In addition, little or no effect on thymic epithelial cell (TEC) MHC II protein expression was observed for potential modifying factors such as TCR and IFN-γ, in contrast to class II transactivator (CIITA). Furthermore, we found MHC II focal aggregates in cortical epithelium, even in the absence of TCR-bearing cells. We discuss the implications of these aspects of MHC II expression for thymic selection.
Methods
Mice
All mice were handled according to American Association for Accreditation of Laboratory Animal Care (AAALAC) regulations. C57BL/6 mice were purchased from DCT/Charles River (Frederick, MD, USA). Genetically modified mouse lines were bred at the National Institute of Allergy and Infectious Diseases (NIAID) Taconic Farms, Inc. barrier in Germantown, NY, USA. They are the TCRα−/− line on a B6 background: C57BL/6-[KO] [knockout (KO)] TCRα; the RAG1−/− line on a B6 background: C57BL/6J[KO]RAG1 N10; the 5C.C7 TCR transgenic line on a non-selecting B10 background: C57BL/10-[Tg]TCR-Cyt-5CC7-I-[KO]RAG2; and the AND-TCR transgenic line on a selecting B10 background: C57BL/10-[Tg]-TCR-Cyt-AND-[KO]RAG2. The CIITA−/−mouse line [22] was provided by Indiana University and bred in an AAALAC-approved NIAID facility at National Institutes of Health (NIH).
Thymic stromal cell isolation
Thymuses were finely minced in cold PBS and washed several times, and thymocytes were removed by allowing the suspension to stand for 1 min. The remaining fragments were subjected to digestion with 0.25% trypsin (Sigma, St Louis, MO, USA) and 1 mg ml−1 deoxyribonuclease I (Sigma) for 30 min at 37°C. After adding 10% Fetal Bovine Serum (FBS), the resulting suspension was filtered and then centrifuged at 6500 r.p.m. for 20 s. The pellet was incubated with anti-Fc mAb 2.4G2 and anti-mouse CD45 microbeads (Milteny Biotec, Sunnyvale, CA, USA) for 20 min at 4°C and CD45−cells were collected using magnetic bead cell sorting (MACS, Milteny Biotec).
Flow cytometric analysis
Cells were washed in cold FACS buffer (PBS + 1% BSA), subsequently stained on ice with 2.4G2 antibody, the primary antibody, and then a secondary antibody, and analyzed on a FACSCalibur (BD Biosciences, San José, CA, USA) with two lasers in the presence of 1–2 mg ml−1 of propidium iodide (PI). Analyses were done using CellQuest (BD Biosciences) or FlowJo 6.0 (http://flowjo.com). For cell-cycle analysis, purified cells were re-suspended at 1 × 106 cells ml−1 in DMEM containing 10% FBS; then Hoechst 33342 (Molecular Probe) was added to a final concentration of 20 mg ml−1. After incubation for 20 min at 37°C, cells were spun and re-suspended in ice-cold FACS buffer before staining for flow cytometry. Cells were analyzed on an LSR II flow cytometer (BD Biosciences), equipped with argon, HeNe and HeCad lasers and acquired using FACSDiVa software. Doublets were excluded by Hoechst fluorescence pulse width versus area gating. Data were analyzed in the cell-cycle platform (Watson pragmatic model) of FlowJo.
mAb and peptides
The following mAbs were used: anti-CD45 (30-F11, rat IgG2b) PE–Texas Red conjugate, anti-CD45 (30-F11.1, rat IgG2b) conjugated with FITC, anti-MHC II (I-Ab) antibody conjugated with PE (M5/114.15.2, rat IgG2a) or with FITC (AF6-120, mouse IgG2a), anti-aminopeptidase A (BP-1/6C3, rat IgG2a) conjugated with FITC and anti-TCR (H57-597, hamster IgG2). The anti-CD45 antibody was purchased from Caltag (Burlingame, CA, USA) and all others from BD PharMingen (San Diego, CA, USA). Clones secreting anti-EpCAM (G8.8) (8) and anti-aminopeptidase A (CDR1, IgG2a) (14) were purchased from American Type Culture Collection (Rockville, MD, USA) and were prepared by Larry Lantz (Custom Antibody Services Facility, NIAID, NIH). Biotinylated UEA-1 was purchased from Vector Laboratories (Burlingame, CT, USA).
Cytoplasmic staining of MHC II
Purified cells were first incubated with 2.4G2 and then with about a 1000-fold excess amount of unlabeled anti-I-A (M5/114) to block the cell-surface I-Ab along with anti-CD45 FITC and biotinylated CDR1 or UEA-1 to stain for surface molecules. After 20 min at 4°C, the cells were washed with FACS buffer and incubated for 15 min at 4°C with streptavidin-Allophycoerythrin (Caltag) as a secondary antibody. Following washing, Fix and Perm Medium A (Caltag) was added and the cells were incubated at room temperature for 15 min. Cells were subsequently washed, and Fix and Perm Medium B was added with BSA (New England Biolabs, Beverly, MA, USA) and anti-MHC II antibody conjugated with PE (M5/114). Cells were incubated for 15 min and analyzed.
Immunostaining and microscopy
Serial frozen sections (4 mm) were air dried and fixed in cold acetone (−20°C) for 5 min. They were washed, hydrated, incubated with blocking solution (X-0909, Dako, Carpenteria, CA, USA) and then stained with FITC-conjugated anti-I-Ab antibody together with biotinylated other antibodies along with streptavidin Texas Red (Jackson ImmunoResearch, West Grove, PA, USA). All the antibodies used were cleared from precipitates by spinning at 13 000 r.p.m. for 1 h at 4°C. For quantitation of the mean fluorescence amplitude (MFA) from confocal microscopy, various methods were examined: line selection for profile, ellipse selection for 10 randomly placed small areas and poly selection for one whole selection using Leica confocal software version 2.5 (Leica Microsystems, Exton, PA, USA). MFA data from the line selection method are shown in Fig. 4. Cytospin (Shandon Lipshaw, Pittsburg, PA, USA) smears were prepared by centrifugation using sorted TEC. Cells (103–104) in 100 ml were spun at 1000 r.p.m. for 5 min and allowed to air dry. Images were collected on a TCS-NT/SP confocal microscope or Leica DM RX (Leica Microsystems). Images were processed using the softwares indicated in the figure legends.
Results
Surface MHC II profiles in different thymic stromal compartments
To study how each TEC establishes a different compartment at the single-cell level, we first assigned adult thymic stromal cells to either a cTEC or an mTEC phenotype using cell-surface markers in a flow cytometric analysis. The reagents used for specific compartments were the CDR1 mAb (2, 14–16) for cTEC and the UEA-1 lectin, which is reported to bind to a sub-population of mTEC (17, 18). As shown in Fig. 1(A), both cell populations were positive for the pan-epithelial marker EpCAM (G8.8 mAb) (16). We attempted to use MHC II as an additional marker to differentiate compartments because cortical epithelium showed a lower level of staining than medullary epithelium in histologic sections (9, 12, 18). MHC II expression levels were examined in the different gates that appeared separable based on the contour levels for CDR1 and UEA-1 binding (Fig. 1B and C). Comparing the levels of I-Ab-specific MHC II staining in the same gates to that of I-Ak cells as a negative control, it is clear that the relative levels of MHC II vary in different compartments. CDR1hi cTECs (gate I) were mostly MHC IIhi. The mean fluorescence intensity was even higher than that of CD45+CD11c+ thymic dendritic cells (DCs). The histogram for UEA-1+ mTEC (gate II) revealed two distinct populations that were either MHC IIhi or MHC IIlo. The MHC IIhi peak in UEA-1+ mTEC was equivalent in surface MHC II expression to that of the CDR1+ cTEC, while the MHC IIlo peak is very close to negative; however, this MHC IIlo population has a distinctive level that is higher than the negative control (easily seen in Fig. 1C). Also, two levels of UEA-1 binding can be seen: the UEA-1lo cells are MHC IIlo, while the UEA-1hi cells are MHC IIhi (see below). The population in gate III is CDR1lo as well as UEA-1+. It expresses almost exclusively a high level of MHC II, similar to that of most CDR1+ cTECs. The small population in gate IV contains CDR1lo and UEA-1− cells. They appeared different from any other population in their heterogeneous MHC II level, being mostly MHC IIlo or MHC IImed. Finally, the cells in gate V are most likely mesenchymal cells that are MHC II−, CDR1− and UEA-1−. Thus, cTECs are mainly homogeneous for CDR1+MHC IIhi in this type of analysis, while UEA-1+ mTECs can be divided into two distinct populations: UEA-1hi/medMHC IIhi and UEA-1med/loMHC IIlo; however, there is also a clearly independent subset that is UEAmed/hiCDRloMHC IIhi (gate III).
MHC II expression during development of TEC compartments
Next we followed changes in the MHC II expression profile in the two major epithelial cell compartments from birth to 8 weeks (Fig. 2A). As shown in the histograms, the profile of MHC II surface expression on thymic stromal cell preparations revealed at least four distinct peaks. In newborn, a MHC IIhi peak was most evident, a MHC IImed peak less so and a MHC IIlo peak the least prominent. Newborn stroma also contained a large number of MHC II− cells. This negative peak slowly declines to a small percentage as the animals age. In contrast, in the adult, the MHC IIlo peak (also UEA-1lo) becomes much more prominent. Cells in the MHC II− peak were also negative for a pan-epithelial marker, EpCAM (G8.8), at all ages (16, data not shown).
When MHC II levels were analyzed on newborn mTECs, most of the UEA-1+ cells were MHC IIhi or MHC IImed. The MHC IIhi cells persisted into the adult stage (8 weeks). In contrast, the MHC IImed cells were clearly present only up to 2 weeks of age, after which time they were replaced by a large MHC IIlo population. By 8 weeks of age, MHC IIlo cells were more abundant than MHC IIhi cells. Thus, newborn mTECs consist of MHC IIhi and MHC IImed cells, while mature mTECs are a mixture of MHC IIhi and MHC IIlo cells. When the MHC II profile is displayed against CDR1, it is clear that the highest CDR1+ cells are also high for MHC II. This is in contrast to our original assumption that the MHC II level on cTEC would be lower than that on mTEC. In the neonatal stage, the percentage of CDR1−UEA-1+ medullary epithelial cells (7.1%) was much lower than that of CDR1+UEA-1− cortical cells (46%), although the level of CDR1 was lower than in the older mice. The levels of CDR1 on fetal day 15 were even lower (data not shown). The cTEC population defined by CDR1+UEA-1−MHC IIhi cells contributed a greater percentage of cells in the newborn and decreased gradually with age (newborn, 46%; 1 week, 42%; 2 weeks, 15%; 6 weeks, 10.5%; 8 weeks, 9%); however, the absolute number of cTECs in our preparations actually appeared to increase with age (data not shown). The percentage of medullary cells gradually increased with age (newborn, 12%; 1 week, 34%; 2 weeks, 52%; 6 weeks, 53%; 8 weeks, 55% in Fig. 2A). Also, there is a small population that is positive for both CDR1 and UEA-1 (gate III in Fig. 1B) that is clearly visible in the neonatal period (newborn to 2 weeks), but decreases at later time points to a barely detectable level.
In Fig. 2B, we show the spatial organization of the thymus compartments in histologic sections by confocal microscopy. We found that the architectural organization of the post-natal thymus changes with age as predicted from previous studies performed at different stages of development (9, 19). The small isolated medullary islets stained with UEA-1 are seen in the newborn thymus, but grow larger and fuse with other islets to form a single medulla when the organ is mature at 8 weeks. At all ages, high MHC II expression co-localized with UEA-1 staining in the medullary areas. The subcapsular region also stained more intensely for MHC II than did the cortex. In addition, we noticed that two different levels of UEA-1 staining and MHC II staining could be clearly observed in the medullary area at the older ages (2, 6 and 8 weeks), consistent with the fact that UEA-1lo cells are present in the medulla.
MHC II expression and cell-cycle profile in the TEC compartments
To understand the mechanistic basis for the various number of cells expressing different levels of MHC II from the newborn period to the larger adult state, we looked at the cell-cycle distribution using the lipophilic dye Hoechst 33324 in live cells after staining with CDR1, UEA-1 and MHC II. As shown in Table 1, all compartments expressing MHC II at the newborn stage revealed a greater percentage (12.5% for MHC IIhiCDR1+, 19.8% for MHC IIhiUEA-1hi, 12.1% for MHC IIloUEA-1lo) of cells in S phase than non-epithelial stromal cells (7.1% in MHC II−). Of interest, UEA-1hi cells manifested the most cells in S phase (close to 20%) throughout the entire 8-week period, while MHC IIhiCDR1+ and MHC IIloUEA-1lo cells showed decreased percentages during maturation. From these results, we conclude that MHC II-expressing TECs possess relatively more dividing potential compared with that of non-MHC II-expressing cells at the newborn stage. This is consistent with the decreasing percentage of MHC II− cells in older animals (Fig. 2A).
Populationa | Newborn | 1 week of age | 8 weeks of age | |||
S | G2 | S | G2 | S | G2 | |
MHC IIhiCDR1+ | 12.5 ± 0.4b | 4.2 ± 0.4 | 8.0 ± 0.4 | 2.6 ± 0.1 | 7.8 ± 0.6 | 4.4 ± 2.1 |
MHC IIhiUEA-1+ | 19.8 ± 3.6 | 6.4 ± 1.4 | 19.3 ± 0.3 | 5.2 ± 0.3 | 18.4 ± 1.6 | 6.7 ± 0.4 |
MHC IIloUEA-1− | 12.1 ± 0.4 | 3.9 ± 0.3 | 11.6 ± 0.3 | 5.9 ± 0.8 | 10.1 ± 7.5 | 1.6 ± 1.9 |
MHC II− | 7.1 ± 1.8 | 1.3 ± 0.4 | 9.0 ± 0.2 | 1.8 ± 0.0 | 8.0 ± 0.1 | 2.0 ± 0.1 |
Populationa | Newborn | 1 week of age | 8 weeks of age | |||
S | G2 | S | G2 | S | G2 | |
MHC IIhiCDR1+ | 12.5 ± 0.4b | 4.2 ± 0.4 | 8.0 ± 0.4 | 2.6 ± 0.1 | 7.8 ± 0.6 | 4.4 ± 2.1 |
MHC IIhiUEA-1+ | 19.8 ± 3.6 | 6.4 ± 1.4 | 19.3 ± 0.3 | 5.2 ± 0.3 | 18.4 ± 1.6 | 6.7 ± 0.4 |
MHC IIloUEA-1− | 12.1 ± 0.4 | 3.9 ± 0.3 | 11.6 ± 0.3 | 5.9 ± 0.8 | 10.1 ± 7.5 | 1.6 ± 1.9 |
MHC II− | 7.1 ± 1.8 | 1.3 ± 0.4 | 9.0 ± 0.2 | 1.8 ± 0.0 | 8.0 ± 0.1 | 2.0 ± 0.1 |
Thymic stromal cell populations prepared from newborn, 1 week and 8 weeks of age mice were separated by the indicated markers. Isolated TEC populations were gated for CDR1+, UEA-1 high and low and MHC II level.
Mean ± SE of the percentages in S and G2 phase. Analysis was performed in the FlowJo cell-cycle platform using the Watson pragmatic model to calculate the S peak. Data are from two to three independent measurements.
Populationa | Newborn | 1 week of age | 8 weeks of age | |||
S | G2 | S | G2 | S | G2 | |
MHC IIhiCDR1+ | 12.5 ± 0.4b | 4.2 ± 0.4 | 8.0 ± 0.4 | 2.6 ± 0.1 | 7.8 ± 0.6 | 4.4 ± 2.1 |
MHC IIhiUEA-1+ | 19.8 ± 3.6 | 6.4 ± 1.4 | 19.3 ± 0.3 | 5.2 ± 0.3 | 18.4 ± 1.6 | 6.7 ± 0.4 |
MHC IIloUEA-1− | 12.1 ± 0.4 | 3.9 ± 0.3 | 11.6 ± 0.3 | 5.9 ± 0.8 | 10.1 ± 7.5 | 1.6 ± 1.9 |
MHC II− | 7.1 ± 1.8 | 1.3 ± 0.4 | 9.0 ± 0.2 | 1.8 ± 0.0 | 8.0 ± 0.1 | 2.0 ± 0.1 |
Populationa | Newborn | 1 week of age | 8 weeks of age | |||
S | G2 | S | G2 | S | G2 | |
MHC IIhiCDR1+ | 12.5 ± 0.4b | 4.2 ± 0.4 | 8.0 ± 0.4 | 2.6 ± 0.1 | 7.8 ± 0.6 | 4.4 ± 2.1 |
MHC IIhiUEA-1+ | 19.8 ± 3.6 | 6.4 ± 1.4 | 19.3 ± 0.3 | 5.2 ± 0.3 | 18.4 ± 1.6 | 6.7 ± 0.4 |
MHC IIloUEA-1− | 12.1 ± 0.4 | 3.9 ± 0.3 | 11.6 ± 0.3 | 5.9 ± 0.8 | 10.1 ± 7.5 | 1.6 ± 1.9 |
MHC II− | 7.1 ± 1.8 | 1.3 ± 0.4 | 9.0 ± 0.2 | 1.8 ± 0.0 | 8.0 ± 0.1 | 2.0 ± 0.1 |
Thymic stromal cell populations prepared from newborn, 1 week and 8 weeks of age mice were separated by the indicated markers. Isolated TEC populations were gated for CDR1+, UEA-1 high and low and MHC II level.
Mean ± SE of the percentages in S and G2 phase. Analysis was performed in the FlowJo cell-cycle platform using the Watson pragmatic model to calculate the S peak. Data are from two to three independent measurements.
Regulation of MHC II expression levels
To test the possible effects of TCR engagement on MHC II surface expression levels, we examined cells from mutant thymuses deficient in TCR or in mature thymocytes. Neither RAG1 nor TCRα deficiency resulted in a higher level of surface expression of MHC II when compared with wild-type B6 levels by flow cytometry (Fig. 3A). In addition, we examined cells from mice expressing transgenic TCR specific for pigeon cytochrome c on either a selecting (20) or a non-selecting (21) background (AND-B10 and 5C.C7-B10, respectively). The percentages of cTECs and mTECs were similar, and the cTECs expressed high levels of MHC II. Neither sample showed any difference in surface MHC II levels. From these results, we conclude that at least the steady state level of MHC II in isolated TECs is not affected by either TCR expression or the presence of mature thymocytes. We also investigated whether IFN-γ, a well-known regulatory factor for MHC II gene expression, showed any effect on MHC II protein expression. As displayed in Fig. 3(B), IFN-γ-deficient TECs showed no differences in the levels of MHC I or MHC II in a total CD45− stromal cell analysis compared with the levels in a wild-type mouse. CDR1+- and UEA-1+-specific compartments were also not different (data not shown). These results indicate that IFN-γ is not required to maintain high levels of MHC II expression on the surface of isolated TECs. Finally, we examined the role of CIITA, an MHC II transactivator (22, 23). As shown in Fig. 3(C), the levels of MHC II as well as the number of cells expressing MHC II were significantly reduced in both the cortical and medullary epithelial cells of CIITA-deficient thymus. Therefore, MHC II levels can be regulated by an essential transcription factor in IFN-γ in independent fashion.
In contrast to our observations by flow cytometry on isolated TECs, the levels of MHC II staining in histologic samples of various thymuses were quite different. In sections of adult thymus, MHC II staining of the cortex was quite low compared with that of the medulla (Fig. 4A). Mutations in either TCRα or RAG1 revealed a higher level of staining in cortical histologic sections. Quantitation of MFA by confocal microscopy gave 36.3 ± 0.8 for RAG1−/− and 12.5 ± 1.3 for TCRα−/−thymus compared with 7.0 ± 0.3 for B6 thymus (Fig. 4A). Furthermore, the results for the non-selecting background (5C.C7-B10) were similar to that for the TCRα−/−sections (13.0 ± 0.9), and the results for the selecting background (AND-B10) (8.4 + 0.6) were closer to that of the wild-type B6 sections. Overall, the results for all the mice are consistent with the notion that a thymus filled with T cells stains less brightly in its cortex.
Levels of MHC II in flow cytometry versus confocal microscopy
We further investigated several parameters to resolve the apparent differences between the two methods. Because confocal microscopy detects both intracellular and cell-surface molecules, we tested the possibility that mTECs expressed more intracellular molecules. As shown in Fig. 4(B), CDR1+ cells actually stained more brightly for intracellular MHC II than did UEA-1+ cells. When pulse height was compared with pulse area for TEC surface MHC II fluorescence in order to exclude the possibility of artifact presentation due to clustering of the signals during the flow cytometry analysis, the levels for both cTECs and mTECs were similar (data not shown). Thus, the total amount of MHC II (cell surface and intracellular) appears to be higher in cTECs than in mTECs and, therefore, cannot account for the discrepancy.
Another factor considered was the size of the cells as this could affect the density of MHC II staining. As shown in Fig. 4(C), both subsets showed a broad distribution for the forward side scatter, but the CDR1+ population possessed a greater frequency of cells with high forward scatter, suggesting that they might be larger in size. They had more forward light scatter than both UEA-1+ mTECs and CD11c+ DCs. This observation was confirmed by direct observation of isolated permeablized and stained epithelial cells placed on slides (Fig. 4D). CDR1+ cells were heterogeneous in size, but most of the cells were larger than UEA-1+ cells. It was thus possible that the same number of MHC II molecules in the smaller volume of the mTECs would appear brighter in the sections if the cells contained equivalent amounts of protein; however, as shown in Fig. 4(D), whole MHC II staining of isolated permeablized CDR1+ cells was actually brighter than that for the UEA-1+ cells, perhaps because they possessed a greater amount of intracellular MHC II molecules (Fig. 4B).
Focal aggregate formation by MHC II molecules in cortical epithelium
While examining cytospin preparations of purified thymic stromal cells, we noticed that some CDR1+ cells contained smaller CD45+ cells and appeared to have typical thymic nurse cell (TNC) morphology (24). As shown in Fig. 5(A), one TNC contains >10 thymocytes. In higher resolution, MHC II staining of permeablized TNCs showed a distinctive punctate pattern that was especially dense in the focal contact points, suggesting that peptide–MHC II complexes might be expressed in a small compact area rather than homogeneously spread out over the epithelial cell surface. MHC II aggregates in a thymocyte-free area (Fig. 5A, middle panel) raised the possibility that they might be formed independently of TCR engagement. We therefore compared the surface staining patterns of TNCs from B6 and TCRα-deficient thymuses after staining with anti-TCRβ and anti-MHC II mAb without permeabilization or fixation (Fig. 5B and C). TCRα−/− TNCs appeared normal in morphology although smaller in size, and the pattern of MHC II aggregates in isolated TNCs were indistinguishable from that in normal B6 thymus. In B6 TNCs, only occasional double-stained spots were observed (yellow in the figure), indicating that most MHC II aggregates were not stably engaged with TCR. We conclude that MHC II aggregate formation can occur without TCR contact.
Next, we investigated the surface MHC II staining pattern in unpermeablized TECs other than TNCs. As shown in Fig. 5D, most CDR1+ (or UEA-1−) cells revealed a pattern of MHC II aggregates on their surface, while CDR1− (or UEA-1+, data not shown) cells revealed more diffuse staining without any clear evidence of dense aggregates (Fig. 5D and data not shown). The aggregates on CDR1+ cells were smaller than staining artifacts (data not shown) and distinctively round, indicating that they are not formed by non-specific antibody aggregation. We also sought evidence for cortical MHC II aggregates in thymic sections to exclude the possibility of their being generated during the TEC purification process. As shown in Fig. 6(A and B) (captured at the maximum magnification of the confocal microscope), MHC II aggregates were clearly visible in the normal B6 thymic cortex but not in the medulla. The shapes and sizes of the MHC II aggregates could be clearly represented in a software-mediated, stylized gray scale presentation (embossing in Photoshop) in the second panel. When the stronger medullary staining was reduced through imaging software to the levels of cortical staining, we still did not find any similar aggregates (data not shown), suggesting that the failure to detect aggregates in the medulla is not due to the intense signal from the staining. Only occasionally were MHC II aggregates co-stained with TCR (shown as arrows in Fig. 6A, right panel). In addition, TCRα and RAG1-deficient thymuses also showed clear MHC II aggregates (Fig. 6C and D), strongly supporting the hypothesis that they are formed independently of TCR engagement.
Discussion
In this paper, the focus was on the significance of measurable MHC II profiles in the different TEC compartments for the mechanism of thymocyte selection.
cTECs express a relatively high level of MHC II on their surface
Positive selection, the process for DP thymocytes to interact with self-peptide–MHC complexes, is thought to occur mostly, if not entirely, in the cortex. The key to this fatal step is determined by the moderate strength of TCR signaling transduced from the TECs through peptide-loaded MHC. In this regard, a quantitative analysis of MHC II levels on surface of cTECs is an important aspect for understanding the mechanistic aspects of this event. By carefully looking at MHC II expression with the techniques currently available, we conclude that the total surface MHC II level on the cTECs is as high as that on UEA-1 binding, MHC IIhi mTECs and higher than that of UEA-1+MHC IIlo mTECs or thymic DCs (Figs 1 and 2). In addition, the cytoplasmic pool of MHC II is as high as that of the UEA-1+ medullary cells, if not higher (Fig. 4B and D) this internal pool could supply sufficient MHC II molecules to maintain the high level on the surface. Therefore, the total amount of MHC II present in cTECs is greater than we previously assumed based solely on the results obtained from histologic studies (11, 18).
A number of hypotheses were examined to explain the duller staining of the cortex, relative to that of the medulla in histologic sections. One potentially significant variable was the number of thymocytes in the sections. Comparison of cTEC MHC II staining in RAG−/− and TCRα−/−sections showed a significantly brighter staining of the RAG− thymus (Fig. 4). This might be because the TCRα−/− has numerous DP thymocytes in its cortex, whereas the RAG1−/− does not. The higher density of DP thymocytes can maximize the surface interacting with cTECs. In addition, they could mask the detection of MHC II signals scattered over the large stretched cTEC dendritic surface either by making the antibody inaccessible for staining or merely by covering the existing MHC II complexes. This interpretation is consistent with at least one other example in the literature. When anti-EpCAM antibody G8.8 was originally described, its histologic staining appeared specific for the medulla (17); however, examination by flow cytometry showed that it binds to all epithelial cells (16, Fig. 1A). Thus, we conclude that the low-intensity staining in cortical sections, relative to the high intensity observed in the medulla, is due to thymocyte masking of MHC II on the smaller number of cTECs in the cortex, contrasted with the larger number of mTECs and DCs in the medulla, where thymocyte density is lower.
cTECs express MHC II as an aggregated form on the surface
Of interest, MHC II molecules were distributed as clusters in TNCs and other cTECs as revealed by both high-resolution cytologic and histologic analyses (Figs 5 and 6). This MHC II aggregate formation did not completely depend on the presence of a functional TCR. Furthermore, most MHC II focal aggregates were not involved in interactions with thymocytes, as only a few were co-localized with TCRs (Figs 5 and 6). Although the detailed structure of the focal MHC II aggregates is unclear, we have ruled out the possibility that they are internalized vesicular compartments or artifacts by the following experiments. First, surface staining was performed without permeabilization at a cold temperature in the presence of NaN3, which should block energy-dependent membrane internalization. Second, the same kind of focal aggregates were observed in the cortical area of thymic sections (Fig. 6). In fact, patchy (9), focal (12) and punctuated (24) MHC II staining patterns have been previously described in electron microscopic studies in situ, although these results were not carefully examined. In addition, similar focal aggregate structures were described in the TCR distribution on cortical DP thymocytes (8). One specialized type of cTEC, the TNC, was originally defined in vitro as a multicellular complex of an epithelial cell and thymocytes (for review, see 25, 26). Although it is not easy to identify intact TNCs by conventional microscopy in sections, it is clear by electron microscopy that there are multiple types of TNCs present (27). Mouse TNCs are known to express cortical markers (28). Although there are conflicting reports on the developmental stage of the thymocytes in TNC clones (29–32), the TNC has been considered a site for positive selection (33). As seen in Fig. 5(B and C), MHC II aggregates were clearly visible all over the surface of the TNC; however, only a fraction of TCRβ molecules were co-localized with MHC II aggregates of normal TNCs. Similar observations were made with TNCs from TCRα-deficient thymuses, although the number of TCRβ molecules appearing to co-localize with the MHC II aggregates were much less. The DN1–DN3 stages of double-negative cells were not capable of forming TNC because we could not detect them in the thymus from RAG1−/− mice (data not shown).
Maintaining the weak interaction between TCR and peptide-loaded MHC II aggregates in cTECs
The MHC II aggregates revealed in this study raise the intriguing possibility that cTECs are equipped to generate a high local avidity with the TCR when it is engaged, even higher than that provided by mTECs. Thus, MHC II concentration should not be a limiting factor for positive selection. This is in agreement with a recent report that a thymus bearing two different cells carrying related TCR transgenes responding to the same antigen did not compete for MHC II. Instead, they competed for other environmental factors (34). On the other hand, one can ask how the thymocytes avoid being negatively selected by these MHC II aggregates.
One possible mechanism to achieve a moderate TCR interaction with a cortical epithelium displaying an abundant amount of aggregated MHC is to shorten the kinetic parameters of interaction. However, recent two photon-imaging results (35, 36) are not consistent with this view. TCR transgenic thymocytes either move rapidly around stromal cells or make stable contact for many hours with single stromal cells in the presence of selecting MHC molecules. It was also shown in a fetal thymic organ culture system that positive selection involves sustained interaction between DP cells and the microenvironment (37). An alternative hypothesis is that the avidity of each clonotypic TCR for self-peptide–MHC complexes could be lowered by non-homogeneous peptide loading in an MHC II focal aggregate. Because there are so many peptides to be loaded under normal conditions (38), a random mechanism would generate a diverse repertoire of endogenous peptide–MHC complexes on the cTEC surface. If focal aggregate formation of MHC molecules is also random, the TCRs would be unlikely to find many MHC molecules loaded with the same peptide in a given cluster. This is consistent with the idea that low-abundance peptides play a major role in selecting a large repertoire of T cells (39).
Complexity of MHC II expression in TEC compartments—time and location make a difference
The heterogeneous nature of mTECs was previously described in histologic studies (18, 40, 41). mTEC cells are categorized as non-classical I-O+UEA-1− or conventional MHC II+UEA-1+. Using flow cytometry, we have divided UEA-1+MHC II+ cells into MHC IIhi and MHC IIlo in adult thymus and MHC IIhi and MHC IImed in neonates. In our study, we were unable to identify the I-O-expressing population because there is no mAb-staining reagent available to detect these cytoplasmic, non-classical, MHC II-bearing cells. While it is possible that the UEA-1−CDR1−MHC II− population contains such cells, we suspect that the UEA-1loMHC IIlo population is in fact the I-O-expressing population. This latter interpretation is more consistent with our observation that the UEA-1loMHC IIlo population expresses G8.8 while the UEA-1−CDR1−MHC II− population does not (Fig. 1A and data not shown). In addition, the MHC IIlo population expresses a lower level of UEA-1-binding ligand on its surface (Figs 1C and 2C). Therefore, the UEA-1−MHC II− population identified in histologic studies may correspond to the UEA-1loMHC IIlo population seen with flow cytometry. It is unlikely that the low levels of MHC II on UEA-1lo cells is due to the acquisition of MHC II from other MHC II-expressing cells since MHC II was not transferred from MHC-expressing hematopoietic cells to MHC II-negative TECs (42).
Finally, MHC II+UEA-1+ mTECs are thought to be the major players in thymic tolerance induction (in addition to DCs) by promiscuously/ectopically expressing peripheral proteins, at least in the adult (6, 43, 44). Whether this task is carried out only by the MHC IIhi sub-population of UEA-1+ cells or by both MHC II subsets remains to be determined.
Although the neonatal thymus contains fully mature SP thymocytes and mTECs, it is still undergoing structural re-organization (Fig. 2). Post-natal development decreases the proportion of cTECs and increases the proportion of mTECs (Fig. 2). The isolated medullary islets throughout the newborn thymus fuse by enlarging until eventually they form a single central medulla in the adult. Growth of the thymus in the neonatal stage is most likely achieved by expansion of MHC II-expressing cells, as shown by the greater percentage in S phase at the neonatal stage (Table 1). MHC II levels in CDR1+ cTECs are high in the newborn and remain high, while those in UEA-1+ mTECs change (Fig. 2). In any case, the ever-changing numbers of cells and levels of MHC II might not provide a consistent fixed quantitative background for the selection events that occur in this compartment.
MHC II expression in minor populations
In adult thymus, there are two minor populations that appear distinct: CDR1loUEA-1+ (gate III) and CDR1loUEA-1− (Gate V). The former population showed high MHC II expression while the latter showed intermediate levels. Recent experiments have provided evidence for a single embryonic origin of TECs (19, 45, 46). These authors also suggested that medulla–cortex compartmentalization is established from a single progenitor during the fetal stage. Furthermore, an epithelial precursor population has been identified using keratin (K5 and K8) expression (47) and functionally characterized using surface markers such as MTS 24 and/or MTS 20 (48, 49). So far, this precursor population is not homogeneously defined for surface markers. It will be of interest to determine whether either of our observed minor populations contains such precursor cells using functional analysis as well as by further characterization with other markers (e.g. K5 and K8).
Regulation of surface MHC II expression in TEC compartments
It is well known that the MHC II genes are expressed constitutively in all APCs and that their expression can be further induced by IFN-γ (50, 51). It was previously suggested using histologic sections that IFN-γ can up-regulate MHC II expression in cortical thymic epithelium (11, 24). Here, we used the IFN-γ-deficient mouse and flow cytometry to show that MHC I and MHC II expression on the surface of TECs is not different in any compartment. This indicates that differentiation of TECs and their MHC expression do not depend on IFN-γ. This result is also consistent with the fact that IFN-γ-deficient mice undergo normal thymic development (52). In contrast, deletion of the transcription factor CIITA resulted in a significant reduction in the number of MHC II+ cells and in the level of MHC II per cell in both cTECs and mTECs. CIITA is known to control the specific expression of MHC II molecules in peripheral antigen presenting cells and to be responsible for their constituitive and inducible MHC II expression. In thymic epithelium, however, deletion only partially affected MHC II expression. Therefore, the MHC II gene in this tissue must be regulated by additional factors.
We thank Ronald Schwartz, B. J. Fowlkes, Ronald N. Germain, Polly Matzinger, Wendy Shores and Mario Roederer for their discussions and helpful suggestions. We also thank Larry Lantz in the NIAID FACS facility for help with mAb production and conjugation. We express special thanks to Owen M. Schwartz and Juraj Kabat in the NIAID Biological Imaging Facility for their help with confocal microscopy and image analysis. This work was partially supported by NIH intramural program and Inha University research Grant.
References
Author notes
Transmitting editor: A. Singer
These authors contributed equally to this study.