|Home | About | Journals | Submit | Contact Us | Français|
The behavior of self-reactive T cells in the peripheral immune system has often been studied by following the fate of adoptively transferred antigen-specific T cells in antigen expressing mice. In most cases, after a period of expansion, such cells undergo a slow clonal deletion, accompanied by the onset of anergy and/or suppression in the remaining cells. Here, we demonstrate that at initial frequencies approaching those found in normal repertoires, it is possible to completely avoid deletion and still maintain peripheral tolerance. At starting numbers of <1000 T cells, stimulation by chronic self-antigens resulted in a period of robust clonal expansion, followed by a steady plateau phase extending beyond four months. Despite their stable persistence, the self-reactive T cells did not convert to a Foxp3+ fate. However, they displayed a considerable block in their ability to make IL-2, consistent with the onset of anergy – in a precursor frequency or deletion independent fashion.
In an adaptive immune repertoire, the frequency of T cells that are specific for any given pathogen is thought to be very low. Although the precise numbers are difficult to estimate, in the mouse, it is thought to range in frequency from 1/1,000 to 1/100,000 [1, 2] and numerically as low as 20 per mouse [3, 4]. The robust clonal expansion and differentiation that follows antigen recognition in vivo, is therefore geared to expanding these rare precursors to large numbers of potent effector cells, in a short amount of time. However, the same process can be lethal if the target epitope is derived from a self-antigen.
Therefore, the vertebrate has evolved several processes to curtail self-reactive T cells. After central tolerance deletes a large proportion of these, very few escape to the periphery. This makes it even more difficult to isolate and follow their behavior in unmanipulated animals (until an autoimmune process activates and expands them). Instead, we and others have routinely used adoptive transfer model systems that infuse a traceable population of self-reactive T cells into mice and follow their fate. Such studies have led to the description of three mechanisms that seem to operate – often in concert, to enforce peripheral tolerance. The first is clonal deletion. Although it can be very effective, when actually studied in the periphery it seems to take a very long time to eliminate the autoreactive population . In cases where the antigen is chronic, this presents a problem since the animal continues to suffer a risk of autoimmunity while the cells are being “slowly deleted”. Therefore, two other processes are thought to operate to keep the cells in check - a functional inactivation, originally termed anergy and the action of regulatory T cells [6, 7]. However a clear separation between the three processes in vivo and an understanding of the principles that lead to the choice of any one or a combination of them is still lacking. We have previously reported that adoptively transferring antigen specific T cells to mice expressing their target antigen resulted in the induction of anergy and “slow deletion”, but not of regulatory T cells .
Typically, these studies involved the infusion of 1–3 million TCR transgenic T cells to congenic hosts. About 10% of the injected cells effectively incorporate into the secondary lymphoid organs. Nevertheless, work from several labs (using acute immunization, not chronic or self-antigens) subsequently suggested that at such high frequencies, the T-cell responses were severely constrained by interference between the transferred T cells themselves [8–14]. This phenomenon – termed clonal competition, affects the robustness of the initial T-cell response, the subsequent survival of the activated T cells (“memory”) and even the extent of differentiation into different subsets [13, 15]. We therefore wondered if such a “precursor frequency effect” could also influence the behavior of self-reactive T cells.
Interestingly, we find that chronic antigen stimulation elicits a precursor frequency independent response pattern, compared to an acute challenge. In the latter case the expansion phase and to a much lesser extent, the onset of contraction was influenced by how many T cells participated in the original response. However, the self-reactive T cells were only minimally affected by precursor frequency during the initial expansion phase. Furthermore, in the later phase, recipients seeded with about a 100 self-reactive T cells showed no evidence of clonal deletion for over 4 months. But, even at lower frequency, the self-reactive T cells entered an anergic state marked by reduced recall cytokine production and no conversion to Foxp3 positivity. These data suggest that in the normal repertoire, T cells reactive to chronic self-antigens that escape thymic deletion can respond and persist in the periphery, albeit in an anergic state.
The impact of initial precursor frequency on the magnitude of the subsequent T-cell response was modeled using an adoptive transfer strategy wherein log dilutions of congenically marked naïve T cells were injected intravenously into recipient mice and challenged in vivo. We first confirmed that such a strategy can reliably introduce a decreasing number of T cells to the secondary lymphoid organs, by titrating injected T cells and enumerating the number of those recovered a day later.. As expected, FACS analysis showed a clear titration in the percentage of 5C.C7 (Vβ3+,CD4+) T cells seeding the recipients (Figure 1A).
In order to derive a reliable value for the number of T cells that populate the animal, we combined two such experiments (n= 6–7 mice) and calculated the recovery of 5C.C7 cells as a fraction of injected cell numbers (Supporting Information Figure 1A). After eliminating the outliers, we calculated the mean seeding efficiency for each dilution (Supporting Information Figure 1B). As shown in Figure 1B, the recovery is close to 20% of the input at all dilutions (linear regression coefficient of 0.9969) except the lowest. In the experiments that follow (Figures 2B & 2D), we use this calculated efficiency to normalize T-cell expansion (as a function of the actual initial frequency). So, an injection dose of 103, corresponds to an actual precursor frequency of 129 ± 33 5C.C7 T cells in the recipient while that of 105 amounts to 21,866 ± 1320 cells.
The presence of a large frequency of antigen-specific T cells at the beginning of the response has been shown to blunt the clonal expansion and accelerate the subsequent clonal contraction after an acute antigenic immunization . Similar to those studies, 5C.C7 T cells challenged acutely with PCC peptide (with LPS as an adjuvant) attained an expansion maximum that was inversely proportional to the initial precursor frequency (Figure 2A). This is most evident in Figure 2B where expansion is represented as the fold increase from the initial seeding frequency on day1. At the peak of their expansion (day 4), the 105 group increased in number by around 40 fold. However, lower frequencies resulted in a significantly greater burst - 175–456 fold for the 104 and 1367 to 3504 fold for the 103, albeit at a later time-point (Day 8). These data are consistent with the idea that T cells can clonally compete for antigen [8, 9].Each T-cell at lower frequencies can have more access to the antigen, resulting in stronger initial stimulation. The extended expansion could then be a programmed consequence of this initial signal . Alternately, since acute antigen can linger in vivo for over 3 days, the extended proliferation by the lower frequency groups could also be a result of continuing to receive stronger stimulation at these later times . Regardless, after this phase, the expanded cells begin classical clonal contraction. In this model, the contraction is not much influenced by the initial frequency and all groups decay similarly – even over longer time frames (Figure 2E).
In contrast, even the first phase of the response of 5C.C7 T cells to a chronic self-antigen (PCC expressed constitutively from an MHCI promoter) was less dependent on initial frequency (Figures 2C, D and F). The number of T cells after four days of expansion was quite proportional to the initial input (Figure 2C). As a result, the normalized fold expansions over a 100 fold input range was within 4 fold of each other (Figure 2Dd - (103 = 5–21 fold, 104 = 5–19 fold, 105 = 0.4–8 fold (potentially 6–8 fold excluding an outlier)). Therefore, unlike the responses to acute antigen, that to a chronic antigen seems to be largely precursor frequency independent. Furthermore, the extra burst of continued expansion seen in the acutely challenged lower precursor frequency groups between day 4 and 8 was also absent in the self-antigen stimulated T cells – contributing to a surprisingly synchronous dynamics at all precursor frequencies tested.
However, the most striking feature of the 5C.C7 response to chronic antigen, especially at lower frequencies is the absence of any obvious contraction after the initial expansion. At low frequencies (103 and 104 input), the expanded T cells reach a plateau phase that maintained the cell numbers reached at the peak (Figure 2C and F). This was not a sampling error since closer time points between day 4 and 12 also did not reveal any transient peak or crash (data not shown). The 105 group does go through a short deletion phase, but quickly reaches a plateau number of ~20,000 T cells, comparable to the lower frequency groups – suggesting that this density represents the number of post-expansion 5C.C7 T cells that can be supported over the long term in the presence of chronic antigen stimulation. The “crash phase” then, is likely to be a simple homeostatic correction of the overshoot of cell density in the higher frequency groups (105 shown here and 106 previously reported ) as they move towards this set point. This plateau is stable for as long as 135 days (Figure 2F). In the case of the acute antigen (Figure 2E), the cells do seem to crash below this set-point, suggesting that chronic antigen recognition plays an important role in this maintenance phase. Although absolute numbers of T cells show variability between experiments (comparing 2A vs 2E and 2C vs 2F), the profile and even the plateau reached by 103 groups, especially in the chronic antigen model was surprisingly coherent over 3 experiments. A similar behavior was also observed in a second model tracking male antigen (Dby) specific TCR-transgenic T cells (A1(M)) in male mice (Supporting Information Figure 2B). The variation in absolute numbers in the acute challenge model (see 2A vs 2E) makes it difficult to conclude if the precursor frequency does have an impact on the number of T cells recovered in this context at very late time points (30+ days). However in two additional experiments the number of T cells recovered very late after an acute challenge of high or low precursor infusions, was not statistically significant (Supporting Information Figure 2A).
Finally, in the model of 5C.C7 recognition of chronic PCC, an autoimmune arthritis can be clinically scored if the peripheral tolerance is incomplete . In none of the groups reported here were we able to score arthritis above the baseline, suggesting that peripheral tolerance is intact in all groups (data not shown). This further emphasizes the conclusion that clonal deletion is not a critical contributor to the development of such tolerance in the case of chronic peripheral self-antigen stimulation.
The absence of clonal deletion in the lower frequency group, prompted us to examine if the other major mechanisms of peripheral tolerance are intact in the model – namely anergy and conversion to a regulatory T-cell fate. We examined the latter by staining for the canonical marker Foxp3 and did not find significant conversion in the chronic hosts (Figure 3A, closed bars in in3B)3B) with only a minimal conversion in the acute hosts (Figure 3A, open bars in in3B).3B). While this argues against skewing of the autoreactive T-cell itself, it does not, of course, rule out the possibility that endogenous regulatory T cells participate in the peripheral tolerance process.
Finally we tested if the T cells that persist for such extended periods in the presence of chronic antigen, are in fact anergic. The in vivo parallel of anergy, known as adaptive tolerance, is typically marked by a severe blunting of the signaling cascades downstream of the TCR leading to a reduction in the ability of the T-cell to secrete cytokines such as IL-2 . Consistent with this, 5C.C7 T cells recovered 13 days later from 103 injected PCC-transgenic mice failed to make IL-2 (detected by capture assay as shown in Figure 3C). This contrasted with the robust IL-2 detected in similar cells that were acutely immunized with PCC in antigen deficient mice (open bar in Figure 3D). Therefore, in this model, at near physiological precursor frequencies, the induction of anergy seems to operate but without the accompaniment of clonal deletion or the conversion to a regulatory-Foxp3 lineage. These results are strikingly similar to the fate of the T cells in a lymphopenic model where we observe anergy but no deletion or suppression. In this context, however, it must be emphasized that the choice of a nondeletional tolerance mechanism is not simply restricted to anergy. In fact, in similar models, under lymphopenic conditions, T cells have been shown to develop anergy in concert with a suppressive phenotype . The variables that allow this phenotype to develop in specific models may relate to TCR affinity, antigen presentation etc., but are not well understood.
The mechanisms controlling T-cell numbers in vivo remains an enduring mystery. Recent work suggests that clonal competition regulates the pool of memory T cells generated after acute immunization. We suggest that it seems to be less of a factor in the case T cells responding to chronic, self-antigens. Interestingly, these T cells can also persist in vivo for extended periods with no evidence of clonal deletion or conversion to Treg cells. Instead they can adopt an anergic fate without triggering measurable immunopathology. This leads us to speculate that with tools of the appropriate sensitivity, one should be able to find a large number of autoreactive T cells, even in a normal repertoire, maintained in a tolerant state by non-deletional mechanisms.
Mice from the NIAID contract facility (Taconic Farms, Germantown, NY) were housed pathogen free. B10.A CD45.2 mice were also crossed to B6,CD45.1 mice to generate a B10.A,CD45.1 strain . To generate B10.A, mPCC(tg),CD45.1 mice, B10.A mPCC-transgenic, CD45.2 mice  were bred to B10.A,CD45.1.. The IEk restricted MCC/PCC specific TCR transgenic 5C.C7 mice on Rag2−/−, CD45.1+/+ and CD45.2+/+ backgrounds have been previously described . . A1(M) mice originally from Steve Cobbold  on a CBA/Ca background were backcrossed 11 times onto a B10.A,Rag2−/− background  and maintained by homozygous breeding. All animal protocols were as approved by the NIAID animal care and use committee (ACUC).
For adoptive cell transfers, cell suspensions from pooled lymph nodes of donor TCR-Tg Rag2−/− mice (>90% CD4+ T cells) were used without further enrichment and injected by the suborbital route. Acute antigen challenges were performed by intraperitoneal injections of 30 μg of antigenic peptide (DbY or PCC – Anaspec or Bachem, USA) mixed with 5 μg of LPS (Sigma, MI, USA).
T cells in transfer recipients were enumerated by isolating all lymph nodes and spleen, chopping them to approximately 1 mm cubes and digesting with 2 mg/ml collagenase-D (Roche, USA) solution containing 3 mM CaCl2 in 1xPBS, at 37°C for 45 min. Digested tissue was dissociated using gentleMACS dissociator and gentleMACS dissociator C tubes (Miltenyi biotec, Germany) with manufacturer's programmed settings m_Spleen 2.01 followed by m_Spleen 3.02 run serially on each sample. 500μl aliquots of the single cell suspensions were stained to obtain the percentage of CD4+ T cells and used to calculate the number of CD4+ T cells in each animal without any further manipulation.
However, in order to track exceedingly low numbers of transferred T cells, further enrichment was necessary. Following absolute counts as, as stated above, remaining cells were washed and centrifuged over Ficoll-Paque PLUS (GE healthcare Bioscience) followed by enrichment for T cells by negative selection. Briefly, a cocktail of mouse and rat antibodies to B220 (RA3-6B2), CD11b (M1/770), I-EK (14.4.4s), CD8 (53–6.7), and MHC II (M5.114) (BD bioscience) were used to label the cells and the bound fraction, pulled out using anti-mouse IgG and anti-rat IgG coated Dynabeads (Dynal Invitrogen). T cells were analyzed on a FACS Canto II cytometer (BD Immunocytometry) afer staining with appropriate fluorophore coupled antibodies (Biolegend, Ebioscience or BD).
We thank Eleanore Chuang for assistance with experiments, Carol Henry & Calvin Eigsti for Flow Sorting and Pascal Chappert for discussions. This research was supported by the Intramural Research Program of the NIH, NIAID.
Conflict of interest:
The authors declare no financial or commercial conflict of interest.