In vivo tracking of ‘color-coded’ effector, natural and induced regulatory T cells in the allograft response

Abstract

Here we present methods to longitudinally track islet allograft—infiltrating T cells in live mice by endoscopic confocal microscopy and to analyze circulating T cells by in vivo flow cytometry. We developed a new reporter mouse whose T cell subsets express distinct, ‘color-coded’ proteins enabling in vivo detection and identification of effector T cells (T_eff cells) and discrimination between natural and induced regulatory T cells (nT_reg and iT_reg cells). Using these tools, we observed marked differences in the T cell response in recipients receiving tolerance-inducing therapy (CD154-specific monoclonal antibody plus rapamycin) compared to untreated controls. These results establish real-time cell tracking as a powerful means to probe the dynamic cellular interplay mediating immunologic rejection or transplant tolerance.

Understanding the cellular immune response to transplanted tissue has a key role in developing strategies to improve transplant outcomes. Tissue biopsy is the standard method for accessing immune cell infiltration in the graft, but the method is both invasive and inadequate when temporal information is needed to characterize an immune reaction that progresses dynamically over time. Advances in molecular imaging techniques that combine cell labeling with the use of whole-body imaging modalities such as positron emission tomography, magnetic resonance imaging and bioluminescence imaging have led to promising approaches for tracking immune cells noninvasively in vivo^1–4, but these methods lack the sensitivity and spatial resolution to quantify events at the single-cell level. In addition, there is need to track multiple cell populations using labels that are not diluted by cell division.

Here we describe a real-time optical method for simultaneous tracking of multiple T cell subsets that are color-coded with distinct fluorescent reporters in a mouse model in which pancreatic islet transplants are placed beneath the renal capsule. As rejection occurs at 2 weeks in control untreated hosts, we track the T cell response for a period of 2 weeks after transplantation in both the circulatory compartment (by in vivo flow cytometry)⁵ and within the allograft (by endoscopic confocal microscopy)⁶. In vivo flow cytometry allows noninvasive, continuous detection and quantification of fluorescently labeled cells in the circulation without the need to draw blood samples⁵. Endoscopic confocal microscopy enables minimally invasive imaging of internal organs with cellular definition by inserting a narrow-diameter endomicroscope through a small incision in the skin⁶. We show that repeated imaging of the islet allograft just beneath the renal capsule can be accomplished in the same mouse over the two-week period.

Islet transplantation is a promising clinical approach to restore insulin production and glucose regulation in patients with type 1 diabetes. The immune response to allogeneic islet transplants is CD4⁺ T cell dependent^7–9, and includes both donor reactive, tissue-destructive T_eff cells and tissue-protective T_reg cells. The acquisition of transplant tolerance, a state in which the transplant is not rejected despite the cessation of immunosuppressive therapy, is associated with an alteration in the functional balance of T_eff and T_reg cells, as deduced in passive lymphocyte transfer experiments^10–12. In addition, the pool of T_reg cells includes both nT_reg and iT_reg populations that arise during intrathymic T cell maturation or in the periphery when naive CD4⁺ T cells are activated by antigen in the presence of transforming growth factor-β (TGF-β) and in the absence of interleukin-6 (IL-6) and IL-21, respectively^13,14. The relative importance of iT_reg and nT_reg cells in the induction and maintenance of transplant tolerance is unclear because it has not been possible to readily distinguish these two T_reg subsets in vivo. Using our cell tracking technology, together with the color-coding scheme, we are now able to monitor not only T_eff and T_reg cells but also distinguish nT_reg from iT_reg cells in live mice to serially analyze the immune response to rejecting or tolerized major histocompatibility complex–mismatched islet transplants. As part of this study, we have tested the hypotheses that tolerance-inducing therapy will accelerate the rate and magnitude of conversion or accumulation of iT_reg cells from naive alloreactive T cells and lead to a profound increase in the number of allograft-infiltrating T_reg cells.

Results

Imaging and quantification of graft-infiltrating T cells

For islet transplantation, we first prepared C57BL/6 Rag1^−/− recipient mice that lack lymphocytes by adoptive transfer of 1 × 10⁶ nT_reg cells (DsRed⁻CD4⁺GFP⁺) purified from C57BL/6 Foxp3-eGFP regulatory T cell reporter mice¹³ together with 9 × 10⁶ T_eff cells (DsRed⁺CD4⁺GFP⁻) purified from C57BL/6 DsRed–knock-in mice (Fig. 1a). The proportion and number of transferred nT_reg and T_eff cells mimicked those present in normal wild-type mice (Supplementary Fig. 1a). On the next day, we placed a DBA/2 allogeneic islet graft underneath the left renal capsule (Supplementary Fig. 1a,b)⁸. One group of recipients received a tolerance-inducing protocol consisting of a 14-d course of CD154-specific monoclonal antibody (mAb) plus rapamycin. This regimen promotes conversion of antigen-activated naive T cells to iT_reg cells and donor-specific allograft tolerance¹⁵. In untreated hosts, all islet allografts were rejected by 14 d after transplantation (mean graft survival time, 12 d), whereas allografts in hosts receiving CD154-specific mAb plus rapamycin tolerizing treatment survived indefinitely (Fig. 1b).

Figure 1.

In vivo imaging of color-coded T cells. (a) FACS sorting of DsRed⁺CD4⁺GFP⁻ red T_eff cells from DsRed–knock-in mice and CD4+GFP+ green nT_reg cells from the original knock-in mice. (b) Graft survival curves of mice treated with CD154-specific mAb plus rapamycin and untreated rejecting controls. The difference in the survival curves is significant, as calculated by either log-rank (Mantel-Cox) (P = 0.0004) or Gehan-Breslow-Wilcoxon (P = 0.0012) tests. (c) Representative image of allograft-infiltrating nT_reg (green), T_eff (red) and iT_reg cells (yellow) acquired by intravital microscopy. Scale bar, 50 μm.

We took a two-step approach to imaging the islet allograft. First, we verified our ability to identify and enumerate various T cell subsets at this location by intravital microscopy. Subsequently, we developed a minimally invasive method to accomplish these tasks through an endomicroscope.

Under appropriate conditions, CD4⁺Foxp3⁻ T_eff cells can convert into a Foxp3⁺ phenotype, a characteristic of iT_reg cells^16,17. To validate our color-coded system, we monitored in vitro conversion of T_eff to iT_reg cells by culturing purified T_eff cells collected from Ds-Red–knock-in mice (DsRed⁺CD4⁺GFP⁻) with DBA/2-derived B220⁺ splenic B cells in complete medium containing recombinant mouse TGF-β, IL-2 plus IL-4-specific and interferon-γ-specific antibodies^13,18. Approximately 85% of T_eff cells cultured in these conditions acquired eGFP expression within 4 d (Supplementary Fig. 2), indicating their conversion to iT_reg cells.

Similarly, in our in vivo model, some DsRed⁺CD4⁺GFP⁻ T_eff cells converted to Foxp3⁺GFP⁺ iT_reg cells after transplantation and become yellow (Fig. 1c). These yellow iT_reg cells (DsRed⁺CD4⁺GFP⁺) could be readily distinguished from the green nT_reg cells (DsRed⁻CD4⁺GFP⁺) that were originally transferred from the knock-in mice. Thus, we created a color-coded system in which T_eff cells were red, nT_reg cells were green and iT_reg cells were yellow (Fig. 1c). To verify that yellow cells were true double-positives and not an artifact created by overlapping red (T_eff) and green (nT_reg) cells within the allograft, we acquired Z-stack images in 1- to 2-μm steps (Supplementary Video 1) and generated reconstructed orthogonal slices (xz and yz planes) for analysis (Supplementary Fig. 3). Only red and green double-positive T cells that were negative in the third autofluorescence channel were identified as iT_reg cells (Supplementary Fig. 4).

Using intravital microscopy on days 1 and 4 after transplantation, very few T cells were identified within the islet allografts in either untreated hosts or hosts treated with CD154-specific mAb plus rapamycin (data not shown). On weeks 1 and 2 after transplantation, the densities of allograft-infiltrating T cells were much greater in untreated hosts than in treated hosts, with a predominant pattern of allograft infiltration by red T_eff cells (Fig. 2a). We analyzed only those images taken from the islet allograft sites, which can be readily distinguished from the surrounding kidney structure by their distinctive autofluorescence pattern (Supplementary Fig. 5). To our surprise, when we determined the absolute cell density of the three subsets of infiltrating T cells, the densities of allograft-infiltrating nT_reg and iT_reg cells were comparable in untreated and treated hosts at weeks 1 and 2 after transplantation (Fig. 2b,c). The prominent difference was the density of infiltrating T_eff cells, which was consistently and significantly higher in untreated hosts than in treated hosts at weeks 1 and 2 (Fig. 2d). As a result, the ratio of infiltrating nT_reg and iT_reg cells to infiltrating T_eff cells in treated hosts was much higher than in untreated hosts (Fig. 2e,f). Infiltration of the allograft by nT_reg cells was more prominent than infiltration by iT_reg cells.

Figure 2.

Analysis of infiltrating T cells within islet allografts. (a) Representative intravital microscopy images showing T cell infiltration within islet allografts in untreated hosts and hosts treated with CD154-specific mAb plus rapamycin on week 1 and week 2 after transplantation. Scale bar, 100 μm. (b–d) Summary of cell density of islet allograft–infiltrating nT_reg (b), iT_reg (c) and T_eff (d) cells, as detected by intravital imaging. (e,f) Summary of the ratios of islet allograft–infiltrating nT_reg to T_eff (e) and iT_reg to T_eff (f) cells, as detected by intravital imaging. Error bars represent means ± s.d.

To follow the time course of the T cell response in individual animals receiving islet transplants, we performed minimally invasive imaging with a 1.24-mm diameter endomicroscope inserted through a small incision in the skin⁶. This instrument allowed repeated imaging of the allograft site with minimal surgical manipulation and was able to resolve individual allograft infiltrating T cells (Fig. 3). By serially imaging the same mice at days 3, 5, 7, 10, 12 and 14 after islet transplantation, we obtained the kinetics of T cell infiltration in both the hosts treated with CD154-specific mAb plus rapamycin treated hosts and the untreated controls. Consistent with the intravital microscopy data, we observed a pronounced difference between the two hosts starting at day 7, dominated by the markedly increased number of red (T_eff) cells in the untreated allograft (Fig. 3).

Figure 3.

Serial endomicroscopy of infiltrating T cells within islet allografts. Representative endomicroscopy images within islet allografts on days 3, 5, 7, 10, 12 and 14 after transplantation in untreated hosts and hosts treated with CD154-specific mAb plus rapamycin. Each row of images is from the same mouse at the given time points. Infiltrating nT_reg (green), T_eff (red) and iT_reg cells (green + red) accumulate in the allograft over time. Scale bar, 50 μm.

Flow cytometry analysis of color-coded T cells

To test the accuracy of our color-coded system, we reconstituted C57BL/6 CD45.2⁺ Rag1^−/− mice with 1 × 10⁶ CD45.2⁺DsRed⁻CD4⁺ GFP⁺ nT_reg cells from knock-in mice and 9 × 10⁶ CD45.1⁺ DsRed⁺CD4⁺GFP⁻ _eff cells from CD45.1⁺ CD45.2⁺ (F₁ generation) DsRed–knock-in mice. Islet allografts were performed exactly as described previously⁸. We analyzed spleen samples from both untreated recipients and recipients treated with CD154-specific mAb plus rapamycin 2 weeks after transplantation (Table 1). In untreated hosts, 98.9% of all CD4⁺DsRed⁺ T cells were CD4⁺CD45.1⁺, whereas in treated hosts the percentage was 96.2%. Similarly, 97.5% and 88.5% of CD4⁺CD45.1⁺ cells were CD4⁺DsRed⁺ in untreated and treated hosts, respectively. To identify false positives, we determined the percentages of CD4⁺CD45.1⁺ cells in the CD4⁺DsRed⁻ population and CD4⁺DsRed⁺ cells in CD4⁺CD45.1⁻ population. In the untreated hosts, the percentages were 2.4% and 1.1%, respectively, whereas the percentages in the treated hosts were 8.3% and 2.6%, respectively. Thus, the detection of cell subsets via analysis of cell surface CD45 isoform expression as an independent marker validated the accuracy of the color-coded system.

Table 1. External validation of DsRed with the congenic marker CD45.1 by ex vivo flow cytometry.

CD45.1+ in DsRed+	Untreated	98.9%
population	CD154-specific mAb plus rapamycin	96.2%
DsRed+ in CD45.1+	Untreated	97.5%
population	CD154-specific mAb plus rapamycin	88.5%
CD45.1+ in DsRed−	Untreated	2.4%
population	CD154-specific mAb plus rapamycin	8.3%
DsRed+ in CD45.1−	Untreated	1.1%
population	CD154-specific mAb plus rapamycin	2.6%

The correlation of the CD45.1 congenic marker with DsRed was calculated from spleen samples from the untreated mice and mice treated with CD154-specific mAb plus rapamycin on week 2 after transplantation by ex vivo flow cytometry on the basis of the percentage of CD4⁺CD45.1⁺ cells in the CD4⁺DsRed⁺ population and on the percentage of CD4⁺DsRed⁺ cells in the CD4⁺CD45.1⁺ population. False positives were calculated as the percentages of CD4⁺CD45.1⁺ cells in the CD4⁺DsRed⁻ population and DsRed⁺ cells in the CD4⁺CD45.1⁻ population.

We used ex vivo flow cytometry analysis to determine the number, distribution and ratios of red T_eff green nT_reg and yellow iT_reg cells in the allograft-draining lymph node at 2 weeks after islet transplantation from both untreated hosts and hosts treated with CD154-specific mAb plus rapamycin (Supplementary Table 1). The ratio of iT_reg to T_eff cells in the allograft was consistently elevated in treated hosts at 1 week and 2 weeks after transplantation and in the draining lymph node at 2 weeks after transplantation (Supplementary Table 1). The ratio of nT_reg to T_eff cells in the allograft was consistently elevated in treated hosts compared to untreated hosts at 1 week and 2 weeks after transplantation, but, in the draining lymph node at 2 weeks, the difference between treated and untreated hosts disappeared (Supplementary Table 1). As might be anticipated, the ratios were far more variable in spleen and nonallograft-draining peripheral lymph node, sites that are distant from the transplant (data not shown). Nonetheless, the ratios of iT_reg cells to T_eff cells and nT_reg cells to T_eff cells were consistently higher in hosts treated with CD154-specific mAb plus rapamycin compared to untreated hosts in spleen and peripheral nondraining lymph node (data not shown).

Monitoring circulating T cells by in vivo flow cytometry

To determine whether these T cell subtypes can be detected in the peripheral blood, we serially monitored the number of fluorescent cells flowing through an ear artery by in vivo flow cytometry. In both hosts treated with CD154-specific mAb plus rapamycin and untreated hosts, we detected circulating T_eff and nT_reg cells by 2 d after transplantation, and we detected iT_reg cells by 4 d after transplantation (Fig. 4a). The number of circulating T cells increased from day 2 to week 1 after transplantation, with T_eff cells greatly outnumbering both T_reg cell populations (Fig. 4b), but we did not identify any statistically significant difference between the two groups. Circulating T_eff cells were significantly more abundant in untreated hosts than in hosts treated with CD154-specific mAb plus rapamycin at week 2, at which time the allograft was undergoing rejection in the untreated control mice (Fig. 1b). In contrast, there was still no statistical difference in the numbers of T_reg cells identified in these two groups. The ratio of T_reg to T_eff cells in the circulation was significantly higher in hosts treated with CD154-specific mAb plus rapamycin as compared to untreated hosts at weeks 1 and 2 after transplantation (Fig. 4c).

Figure 4.

Detection of nT_reg, T_eff and iT_reg cells by in vivo flow cytometry in the peripheral blood. (a) A representative in vivo flow cytometry trace showing the identification of single positive nT_reg (green box), T_eff (red boxes) and double-positive iT_reg (yellow box) cells. The second peak in the DsRed channel occurred about 45 ms before the second peak in the GFP channel. As this time difference was greater than the uncertainty of the measurements, these two peaks were distinguished as separate cells and not a double-positive iT_reg cell. (b) In vivo flow cytometry showing T_eff (red), nT_reg (green) and iT_reg cells (yellow) in the peripheral blood. There is a ten-fold difference in scale between mice treated with CD154-specific mAb plus rapamycin and untreated mice. Each curve represents serial analysis of the same blood vessel of the same animal. (c) Summary of the ratio of circulating T_reg to T_eff cells, as detected by in vivo flow cytometry. Error bars represent means ± s.d.

Discussion

Analysis of the T cell-dependent immune response to allogeneic tissues has been hampered by the inability to serially identify iT_reg and nT_reg cell subsets in a living host. The recent development of T_reg cell reporter GFP knock-in mice has provided a powerful means to identify the Foxp3⁺ T_reg cell population, but Foxp3 promoter–driven GFP is expressed by both nT_reg and iT_reg cells^13,19.

We have now devised a means to serially analyze the allograft response through application of a unique tricolor-coded reporter system that enables discrimination not only between T_eff and T_reg cells but also between nT_reg and iT_reg cells. By placing islet transplants beneath the renal capsule, allograft-infiltrating T cells can be serially analyzed by endoscopic microscopy over time in the same host. Concurrently, subsets of circulating T cells can be quantified in a peripheral artery of the same living host by in vivo flow cytometry, without the need to draw blood samples. Compared to previous studies using traditional immunohistochemistry or standard flow cytometry that provide static ‘snapshots’ of T cell infiltration during allograft response^20,21, our technology allows a more complete, serial characterization of the immune response by dynamic tracking of various T cell populations both in the circulatory compartment and at the graft site of the same living host over a course of days to weeks.

With these tools, we have analyzed the T cell response to islet allografts in untreated rejecting recipients and in recipients given a transplant tolerance–inducing regimen^22,23. The fate of the allograft, either rejection or tolerance, depends upon the functional balance of alloreactive graft–protecting T_reg cells to alloreactive graft–destroying T_eff cells^11,24–26. In the absence of a favorable change in balance of T_reg to T_eff cells, the former cells are unable to restrain the latter cells from rejecting the transplant. Although it is known that tolerance cannot readily be induced in major histocompatibility complex–mismatched transplants in the absence of apoptosis of alloreactive T cells^22,23, the precise numeric change in the balance between T_eff and T_reg cells is not known. Moreover, it is not known whether nT_reg or iT_reg cells predominate in the graft-protective response noted in tolerized hosts. In fact, it is expected but not actually known whether these T_reg cell populations expand or accumulate more markedly in tolerized compared to rejecting recipients. Through the implementation of serial microendoscopy, we observed an expected increase in the of iT_reg to T_eff ratio among CD4⁺ T cells that infiltrate the transplant in tolerized hosts as compared to rejecting hosts. Although we anticipated a massive increase in conversion or accumulation of iT_reg cells in tolerized as compared to rejecting hosts, this hypothesis is not supported by our data, at least within the context of our passive transfer model. We did not observe a striking numeric increase in allograft-infiltrating iT_reg cells in tolerized hosts compared to untreated hosts. Also unanticipated was our observation that infiltration of the allograft by nT_reg cells predominated over infiltration by iT_reg cells. The most dramatic change in the response of tolerized versus rejecting recipients was the pronounced decrease in tempo and magnitude of allograft infiltration by T_eff cells, a change that overshadows any change in nT_reg or iT_reg populations in tolerized hosts. As a consequence, the proportion of allograft-infiltrating T_reg to T_eff cells is very markedly enhanced in tolerized hosts. These data corroborate and give insight into the key role of reduction of the pool of donor-reactive effector T cells as a precondition for tolerance induction^22,23. A reduction in allograft infiltration by graft destructive effector, but not regulatory, T cells enables T_reg cells to restrain the ability of the diminished cohort of allograft-infiltrating T_eff cells to reject the transplant.

In keeping with the appearance of T_reg cells within the allograft, we detected both nT_reg and iT_reg cells in the peripheral blood by in vivo flow cytometry. Indeed, we detected all three T cell subtypes in the circulation of both rejecting and tolerized hosts, days before they were discerned infiltrating the allografts. In agreement with the imaging data, we did not detect massive conversion of iT_reg cells in tolerized hosts, even though the technology is clearly able to detect converted iT_reg cells in the circulation, thereby providing unequivocal evidence of the capacity of naive T cells to convert to the Foxp3⁺ phenotype in vivo. Detection of circulating iT_reg cells as early as day 4 after transplantation suggests that iT_reg conversion is an inevitable early event in the allograft response. Notably, the circulating T_reg to T_eff ratio is consistently and substantially higher in tolerized compared to rejecting hosts, raising the possibility that peripheral blood analysis can be used as an early diagnostic method, enabling more timely and effective therapy to improve transplantation outcomes.

The serial application of endoscopic confocal microscopy and in vivo flow cytometry to a color-coded T cell reporter system has enabled a clearer understanding of both quantitative and qualitative characteristics of the CD4⁺ T cell response to allografts in rejecting and tolerized hosts. Indeed, the hypothesis that we held concerning the ability of co-stimulation blockade–based transplant tolerizing therapy to enable massive expansion of allograft-infiltrating T_reg cells, especially iT_reg cells, was proven incorrect. Although we have studied the allograft response in a system based on transfer of lymphocytes into lymphopenic hosts, albeit in a manner that minimizes homeostatic proliferation, we believe application of these methods to other models will enable further progress in a meticulous dissection of the cellular and molecular basis of the allograft response. Our studies also point to a clear need for further refinement of the endoscopic imaging technique, so that serial monitoring of the cellular response in the allograft-draining lymph node with minimal invasion will become possible.

Methods

Methods and any associated references are available in the online version of the paper at http://www.nature.com/naturemedicine/.