Antigen-stimulated CD4 T-cell expansion is inversely and log-linearly related to precursor number

Abstract

Antigen-driven expansion of specific CD4 T cells diminishes, on a per cell basis, as infused cell number increases. There is a linear relation between log precursor number and log factor of expansion (FE), with a slope of ∼−0.5 over a range from 3 to 30,000 precursors. Cell number dependence of FE is observed at low precursor number, implying that the underlying process physiologically regulates antigen-driven T-cell expansion. FE of small numbers of transgenic precursors is not significantly affected by concomitant responses of large numbers of cells specific for different antigens. Increasing antigen amount or exogenous IL-2, IL-7, or IL-15 does not significantly affect FE, nor does FE depend on Fas, TNF-α receptor, cytotoxic T-lymphocyte antigen-4, IL-2, or IFN-γ. Small numbers of Foxp3-deficient T-cell receptor transgenic cells expand to a greater extent than do large numbers, implying that this effect is not mediated by regulatory T cells. Increasing dendritic cell number does result in larger FE, but the quantitative relation between FE and precursor number is not abrogated. Although not excluding competition for peptide/MHC complexes as an explanation, fall in FE with increasing precursor number could be explained by a negative feedback in which increasing numbers of responding cells in a cluster inhibit the expansion of cells of the same specificity within that cluster.

T-cell responses to antigens are characterized by rapid expansion in numbers of specific cells, followed by a steep contraction phase and then a relative stabilization at frequencies above those in the naive cell population (1–5). We show here and others have previously reported that the factor of expansion (FE), the ratio of the number of antigen-specific cells at 7 d after immunization to the number before immunization, is dependent on the number of cells transferred (3–6). Here, we report that over a range of precursor frequencies ranging from 3 to 30,000 among lymph node T cells, there is a log-linear relation between precursor number and FE, with a slope of ∼−0.5. Estimates of the physiological frequency of antigen-specific precursors range from 20 to 3,000 per animal, corresponding to frequencies ranging from 0.2 to 30 per 100,000 naive CD4 T cells (7–10). Thus, the fall in FE described here occurs within the physiological range of precursor number and must reflect a fundamental property of antigen-mediated immune responses.

Results

FE Depends on Number of Precursors.

To study response by small numbers of CD4 T cells, we used a real-time PCR analysis capable of detecting less than 1 T-cell receptor (TCR) transgenic T cell in 100,000 nontransgenic T cells, utilizing a primer-probe set capable of specifically amplifying the CDR3 region of the TCR Vβ chain of 5C.C7 cells (Fig. S1A), whose TCR is specific for a cytochrome C peptide in association with I-E^k (11–13).

We tested the efficiency of homing of transferred cells to lymph nodes. As shown in Fig. S1B, transfer efficiency was ∼6% when evaluated 5 or 14 d after transfer.

We next determined whether we could detect responses from single cells that had homed to lymph nodes. Twelve mice received transfers of 10 5C.C7 cells each, diluted in 500,000 B10.A cells. At a homing efficiency of 6%, each mouse would receive, on average, 0.6 cells in its lymph nodes, which means that some mice receive no cells and others receive 1 or possibly 2 cells. Seven recipients had no detectable response, whereas 5 had numbers of 5C.C7 cells ranging from 317 to 6,732, with a mean FE of 2,159, assuming all the responders received only one precursor (Fig. 1A). Thus, we could detect responses of a single precursor in all the lymph nodes. We also examined responses in mice that received an average of 3 cells. Their mean FE was 1,479, which was not statistically different from FE from single cells.

Fig. 1.

(A) Twelve mice were injected with 10 5C.C7 cells mixed with 500,000 B10.A cells, and 5 mice were injected with 50 5C.C7 cells mixed with 500,000 B10.A cells. Both groups were immunized with pigeon cytochrome C (PCC) protein and LPS 1 d after transfer. Seven days after immunization, lymph nodes were harvested, genomic DNA was prepared, and real-time PCR was performed to evaluate the number of 5C.C7 cells. (B) Twenty mice received 3 5C.C7 cells, and 10 mice received 30 5C.C7 cells. Experimental design, including immunization and processing, was as described in A. (C) Mice receiving either 300 or 30,000 5C.C7 cells were immunized and analyzed as in A. The data are derived from a total of 44 experiments for the group that received 300 5CC7 cells and 18 experiments for the group that received 30,000 5C.C7 cells. (D) Mice received from 600 to 54,000 5C.C7 cells and were immunized with PCC and LPS; after 7 d, lymph nodes were processed as before for real-time PCR.

Twenty mice received 3 cells each, and 10 received 30 cells each. FE was 1,999 for recipients of 3 cells and 484 for recipients of 30, with a P value of 0.0029 for comparison of the groups (Fig. 1B), indicating that an increase in precursor number within the physiological range (i.e., from ∼0.05/100,000 to ∼0.5/100,000) results in a fall in FE.

Comparing FE in recipients of 300 and 30,000 precursors shows a greater fall in FE with this larger increase in precursor number, diminishing from 136 to 12.5, with a P value of 5.0 × 10⁻⁷(Fig. 1C). A comparison of log FE against log of numbers of precursors over a range from 600 to 54,000 precursors in a single experiment reveals a slope of −0.42 with an r² of 0.79 (Fig. 1D). Pooling results from individual experiments comparing 3, 30, 300, and 30,000 precursors reveals that FE falls by almost 0.5 log for every log increase in precursor number.

Diminution in FE with Increasing Precursor Number Is Antigen-Specific.

FE falls as precursor number increases in four other TCR transgenic systems: the HY-specific, I-A^b–restricted Marilyn (14); the HY-specific, I-E^k–restricted A1(M) (15); the ovalbumin (OVA)-specific, I-A^b–restricted OT-II (16); and the OVA-specific, I-A^d–restricted DO11.10 (17). We wished to determine whether a response by a large number of CD4 T cells from one transgenic donor would diminish FE of small numbers of precursors derived from a distinct transgenic donor. B6 male mice received either 900 or 90,000 Marilyn or OT-II cells, a mixture of 900 Marilyn and 90,000 OT-II cells, or a mixture of 900 OT-II and 90,000 Marilyn cells and were immunized with HY Dby peptide and OVA (Fig. 2A). FE was 19.5 in recipients of 900 Marilyn cells alone but only 3.4 in recipients of 90,000 Marilyn cells alone (P value = 1.2 × 10⁻⁵), whereas for Marilyn cells, FE was 34.2 in recipients of 900 Marilyn cells and 90,000 OT-II cells. Similarly, recipients of 900 OT-II cells alone had an FE of 85, whereas recipients of 90,000 cells had an FE of 5. For recipients of 900 OT-II and 90,000 Marilyn cells, FE of the OT-II cells was 40. Although this is lower than in recipients of 900 OT-II cells alone, the difference is not statistically significant. More importantly, the difference in OT-II FE between recipients of 90,000 OT-II cells and a mixture of 900 OT-II and 90,000 Marilyn cells is highly significant (P = 7 × 10⁻⁵). Similar results were obtained when 5C.C7 and A1(M) cells were cotransferred (Fig. 2B). Thus, large numbers of cells of one specificity, if they block the expansion of small numbers of cells of another specificity at all, do so only inefficiently.

Fig. 2.

(A) C57BL/6 CD45.1 mice received 900 or 90,000 cells from either Marilyn or OT-II transgenic mice, both CD45.2. Two additional groups of C57BL/6 CD45.1 mice received a mixture of 900 OT-II cells plus 90,000 Marilyn cells or 900 Marilyn cells plus 90,000 OT-II cells. One day after transfer, mice were immunized with HY Dby peptide and OVA plus LPS. Seven days later, cell numbers were evaluated for both transgenic TCRs via FACS analysis using Vβ6 TCR staining for Marilyn cells and Vβ5 TCR staining for OT-II cells among the CD45.2-positive cells. (B) B10.A CD45.1 mice received 900 or 90,000 cells from either 5C.C7 or A1(M) transgenic mice, both expressing CD45.2. Two additional groups of B10.A CD45.1 mice received a mixture of 900 5C.C7 cells plus 90,000 A1(M) cells or 900 A1(M) cells plus 90,000 5C.C7 cells and were immunized 1 d later with HY peptide and PCC plus LPS. Seven days after immunization, cell numbers were evaluated via FACS analysis from lymph node single-cell suspensions. The 5C.C7 cells were identified as Vβ3-positive CD45.2 cells, whereas the A1(M) cells were identified as Vβ3-negative CD45.2 cells.

When Are Changes in FE Observed?

Mice that had received 300 5C.C7 cells had ∼2,800 cells 2 d after immunization, for an FE of ∼9. Mice that had received 30,000 cells had 307,000 cells on day 2 for an FE of ∼10 (Fig. S2). When analyzed on day 5, mice that had received 300 cells showed a continued expansion to almost 35,000 cells or an FE of 116, whereas those that had received 30,000 cells showed only a modest further increase, with FE rising to ∼11. The similar degrees of expansion in the early phase of the response occur before the release of responding cells from the lymph nodes that is seen by 62 h but not at 38 h (Fig. 3). This raises the possibility that the advantage in FE displayed by mice that received small numbers of precursors might reflect the behavior of cells after they have been freed from their initial sequestration, as judged by their capacity to enter the recirculating pool of primed cells. Injecting anti-CD62L 2 d after immunization, which should inhibit reentry of central memory cells from the blood into lymph nodes, had little or no effect on the fall of FE with increase in precursor number (Fig. S3). Thus, differences in FE, although delayed in time, can occur in the lymph nodes that were the original sites of the immune response.

Fig. 3.

B10.A CD45.2 mice received 30,000 5C.C7 CD45.1 cells. One day later, they were immunized with PCC plus LPS. Blood and lymph nodes were harvested, and the number of CD4-positive cells and CD45.1-positive cells was evaluated via FACS analysis. The bars represent the ratio between the fraction of CD45.1 cells among CD4-positive cells in the blood and the fraction of CD45.1 cells among CD4-positive cells in the lymph nodes.

Changes in Proliferation Rates Can Partially Account for Changes in FE with Increasing Precursor Number.

We immunized recipients of 300 and 30,000 cells, treated them with BrdU, and killed them 6 h later, at 3.5, 4.5, 5.5, 6.5, and 7.5 d after immunization. At each time point, a greater proportion of 5C.C7 cells in recipients of 300 cells took up BrdU than that of 5C.C7 cells in recipients of 30,000 cells (Fig. 4). The differences in proliferation rate would account for an ∼8.5-fold difference in the numbers of 5C.C7 cells at day 7.5, which is somewhat less than the ∼10-fold difference in FE in these groups. Although this analysis does not take into account the difference in the actual time at which the responses reached their peak and in the balance between cell proliferation and death around this time, it indicates that differences in the proliferation rates of the responding cell populations can account for a substantial portion of the differences in FE. We next asked what mechanisms were responsible for inhibiting cell division at higher cell densities.

Fig. 4.

B10.A mice received 300 or 30,000 CD45.1 5C.C7 cells. One day later, mice were immunized with PCC plus LPS. Mice received a BrdU “pulse” and were killed 6 h later on days 3.5, 4.5, 5.5, 6.5, and 7.5. Lymph nodes were harvested, and the proportion of BrdU-positive and BrdU-negative cells among CD45.1 cells was evaluated via FACS analysis.

Increasing Antigen Dose or Dendritic Cell Numbers Does Not Alter the Fall in FE with Increasing Precursor Number.

Increasing the antigen dose from 100 μg to 1 mg resulted in a modest increase in FE, but this increase was similar in magnitude in recipients of 300 and 30,000 cells, such that the ratios of FE at the two precursor frequencies hardly changed (Fig. 5A). Treating mice with 10 μg of human FLT3 ligand per day for 10 d (18–22) resulted in a sixfold increase in CD11c⁺ cells in lymph nodes. FLT3 ligand treatment resulted in a considerable increase in FE, but the increase was similar in magnitude in recipients of 300 and 30,000 cells (Fig. 5B), strongly arguing that the precursor-dependent differences in FE could not be accounted for by limitations in dendritic cell (DC) number in the high precursor cell number group, whereas there were adequate numbers of DCs in the low cell number group. Rather, DCs are “limiting” at both high and low precursor frequency.

Fig. 5.

(A) B10.A mice that received 300 or 30,000 5C.C7 cells were immunized 1 d later with 100 or 1,000 μg of PCC plus LPS. Seven days after immunization, lymph nodes were harvested and genomic DNA was prepared for real-time PCR. (B) B10.A mice were treated daily for 10 d with human FLT3 ligand or PBS. On day 9, the mice received 300 or 30,000 5C.C7 cells and were immunized on day 10 with PCC plus LPS. Seven days after immunization, lymph nodes were harvested and genomic DNA was prepared for the evaluation of the number of 5C.C7 cells via real-time PCR. (C) B10.A mice received 300 or 30,000 5C.C7 cells. The day after transfer, they received a 7-d miniosmotic pump with 10 μg of IL-2, IL-15, IL-7, or PBS and were immunized with PCC plus LPS. Seven days later, lymph nodes were harvested and genomic DNA was prepared for 5C.C7 cell evaluation via real-time PCR.

Fas, TNF-α Receptor, Cytotoxic T-Lymphocyte Antigen-4, IFN-γ, IL-2 IL-7, or IL-15 Has Little or No Effect on FE.

Implantation of miniosmotic pumps that delivered PBS or 10 μg of IL-2, IL-7, or IL-15 over a 7-d period failed to affect the FE of the recipients of either 300 or 30,000 5C.C7 cells (Fig. 5C).

To test whether expression of inhibitory molecules on the surface of responding cells or secretion of inhibitors might be responsible for differences in FE, we used 5C.C7 KO mice as donors and compared FEs of 300 or 30,000 of their cells with those of WT 5C.C7 cells. 5C.C7 cells deficient in Fas (5C.C7 lpr/lpr mice) show no differences from WT 5C.C7 cells (Fig. 6A). Similar experiments were carried out with 5C.C7 TNF-α receptor (TNF-αR)–deficient cells and cytotoxic T-lymphocyte antigen-4 (CTLA-4)–deficient cells. Recipients of TNF-αR–deficient cells displayed slightly poorer FEs at both 300 and 30,000 cells than recipients of WT cells (Fig. 6E), whereas recipients of CTLA-4–deficient cells showed a modest increases in FE (Fig. 6B). In neither case was the FE ratio altered. We tested the difference in FE of recipients of 300 or 30,000 cells from 5C.C7 mice deficient in IFN-γ or IL-2. No differences were observed (Fig. 6 C and D).

Fig. 6.

(A) B10.A mice received 300 or 30,000 cells from either WT or lpr 5C.C7 mice. One day later, animals were immunized with PCC plus LPS. Seven days after immunization, lymph nodes were harvested for genomic DNA preparation, followed by real-time PCR determination of cell number. (B) B10.A mice received 300 or 30,000 cells of either CTLA-4 KO 5C.C7 or WT 5C.C7 cells. The experimental design was the same as in A. (C) B10.A mice received 300 or 30,000 IFNγ^−/− 5C.C7 cells or WT 5C.C7 cells. The experimental design was the same as in A. (D) B10.A mice received 300 or 30,000 cells of either IL-2 KO 5C.C7 or WT 5C.C7 cells. The experimental design was the same as in A. (E) B10.A mice received 300 or 30,000 TNF-αR KO 5C.C7 cells or WT 5C.C7 cells. The experimental design was as the same as in A. (F) C57BL/6 CD 45.1 mice received 600 or 60,000 Scurfy OT-II or WT OT-II cells, both CD45.2. The following day, mice were immunized with OVA plus LPS. Seven days after immunization, Scurfy and WT OT-II cell number were evaluated via FACS analysis.

Regulatory T Cells in the Responding Population Cannot Account for Differences in FE.

Six and 10 d after priming, there were <1% detectable Foxp3⁺ cells among the 5C.C7 cells and no differences between their frequency in the 300 and 30,000 cell groups. The frequency of Foxp3⁺ endogenous CD4 T cells was also similar in mice that had received 300 and 30,000 5C.C7 cells (Fig. S4). We then compared FEs in CD45.1 C57BL/6 recipients of 60,000 or 600 OT-II cells from Rag2^−/− WT or scurfy donors that were immunized with OVA. There were no statistically significant differences between FEs of WT or scurfy responding cells at high cell numbers or at low cell numbers (Fig. 6F). Thus, natural or induced regulatory T cells (Tregs) among the responding cell population are not responsible for the delayed inhibition of responding cell expansion in our system and cannot account for differences in FE at low and high precursor frequencies.

Discussion

Physiological responses of naive specific T cells occur when those cells are present at low precursor frequencies, on the order of 1 in 10⁵ cells or even less (8, 9). Nonetheless, it is usual to study responses by transferring sufficiently large numbers of TCR transgenic CD4 T cells to syngeneic recipients, such that the responding cells can be confidently identified and characterized. These studies assume that responses at high cell frequency faithfully reflect responses that occur at physiological numbers of precursors. Others have already demonstrated that the magnitude of response of transgenic cells is related to responding cell frequency (3–5), however, and recent technological advances have allowed the demonstration that endogenous antigen-specific cells respond similarly to transgenic cells at low precursor frequency (9, 23).

Using a method that allowed us to track responses when the frequency of naive antigen-specific cells was as low as one cell in all the recipient's lymph nodes (i.e., ∼1 in 10⁷ cells), we found that over several orders of magnitude of precursor frequency, the relation between the change in FE in response to immunization and the change in precursor number was log-linear; log FE fell by approximately half a log for each log increase in precursor number. Fall in FE with increasing precursor numbers occurred even when responding cell numbers were within the physiological range.

FE superiority of low precursor numbers has been explained by competition for antigen when precursors were present at high frequency, whereas at low precursor frequency, such competition did not occur (3, 4). Hataye et al. (3) reported that when precursor frequencies were high, responding cells proliferated somewhat less than when frequencies were low. Our results were consistent. The proportion of cells in S phase at low precursor frequency over the period between 3.5 and 7.5 d was sufficiently greater than that of cells at high precursor frequency to account for most of the differences in FE.

The decrease in FE with increase in precursor number is antigen-specific. Responses of large numbers of precursors of one specificity have a limited effect on the FE of small numbers of precursors of a different specificity despite the fact that small and large numbers of precursors of the same specificity have very different FEs. This result leads to the postulate that even when different antigens are administered together, different antigen-presenting cells (APCs) predominantly present the different antigens.

Can our results be explained by competition for DCs or for antigen? Increasing the dose of antigen had a very limited effect on FE. Although increasing DC number did increase FE, it did not change the degree of advantage that precursors at low frequency displayed over precursors at high frequency. These findings suggest that the numbers of DCs are limiting for the expansion of precursors but are “equally limiting” for cells at high and low precursor frequency. Limitation in effective APC number implies that as responding cell number increases, there will be more antigen-specific T cells per cluster surrounding an effective APC.

Our data indicate that the effects of precursor number on FE do not begin to occur until 3 or more days after immunization. Thus, the early (up to day 2.5) FE is no different when 300 and 30,000 precursors are compared, but a difference in FE is clear at day 5. Differences in expansion coincide with the release of responding cells into the blood. Although differences in FE do not require recirculation, this suggests that the differences in FE occur when precursors are freed of their initial relatively stable interactions with APCs. Because CD4 T cells need continued exposure to antigen to continue their proliferation (24, 25), these expanded responding cells are presumably seeking new APCs. At that time, after there has been substantial expansion in specific CD4 T cells and the amount of antigen in the system is diminishing, the number of effective APCs may be limiting even at low precursor frequencies. Indeed, we observed that FE rose to the same extent at both high and low cell density with increasing DC number, implying that DCs were limiting at both precursor frequencies. The observed differences in response may reflect an increase in the numbers of specific CD4 T cells clustering about individual APCs as the number of responding cells increases.

Among possible mechanisms determining the poorer expansion seen at higher precursor frequency would be competition of the responding cells for access to peptide/MHC complexes and/or for costimulatory molecules expressed on the APC (11, 26, 27), competition for locally produced growth-promoting factors (28–30), or inhibition by the action of cells within the microenvironment of the cluster of responding cells (11).

Increasing antigen dose only modestly increases FE, making it unlikely that the amount of antigen per APC is limiting, although it is certainly possible that antigen processing is limiting and does not increase with increase in dose. Indeed, because DCs are limiting, many T cells clustering about a single APC might each receive a less robust TCR-mediated stimulation than the relatively few T cells clustering around the same DC (11, 26).

Consumption of growth-promoting factors is also a possibility. Because the effect of cell number on FE is antigen-specific, such competition would have to be in the localized microenvironment of the responding cell. If clusters were larger in the presence of more precursors, there would be more local cells to consume the growth factor(s), thus limiting the overall expansion of individual precursors. Administering IL-2, IL-7, or IL-15 did not alter the FE of either small or large numbers of precursors, suggesting that competition for growth-promoting factors is an unlikely cause of the fall in FE with increasing numbers of precursors, although one cannot rule out that a factor other than those tested was the critical one.

An alternate possibility is that cells that are expanding in an individual cluster, under the stimulus of a single APC, will limit one another's expansion by actively suppressing each other's growth. Obvious mechanisms would be by Fas- (31–33) or TNF-α– (34, 35) mediated cell death; by inhibitory signaling, such as that provided through engagement of CTLA-4 (36, 37); by secretion of inhibitory cytokines; by in situ development of induced Tregs; or by recruitment of natural Tregs (38). Our experiments do not provide support for any of these possibilities.

We suggest an alternate explanation for the inverse, log-linear relation between precursor number and factor of expansion, in which the most highly differentiated effectors (or memory cells) in a local environment limit the growth of less differentiated effectors (Fig. S5) (39). This may reflect the action of the most differentiated cells to increase the rate of differentiation, and thus diminish overall division of the less differentiated cells, controlling expansion as the number of differentiated cells in a given cluster increases. This model has the advantage that the same mechanism that regulates the typical expansion and contraction kinetics of the immune response also explains the fall in FE by ∼0.5 log for each log increase in precursor number over a range of 3–30,000 precursors per mouse (SI Text). A mathematical analysis of this model is presented in the companion paper (40).

Materials and Methods

Mice.

Animals were 6–12 wk old, gender- and age-matched in each experiment, and maintained in the pathogen-free National Institute of Allergy and Infectious Diseases, National Institutes of Health animal facility. The care and handling of the animals used in our studies were in accordance with the guidelines of the National Institutes of Health Animal Care and Use Committee. Details as to the mice used in this work can be found in SI Materials and Methods.

Adoptive Cell Transfer and Immunization.

Varying numbers of TCR transgenic lymph node cells were mixed with syngeneic lymph node cells to give a total of 1 million transferred cells per recipient mouse. Cells were allowed to home for a period of at least 24 h. Animals were immunized s.c. by a single dose of 100 μg of cytochrome C protein or 10 μg of pigeon cytochrome C (PCC) peptide in PBS unless otherwise indicated. LPS (25 μg per animal) was used as adjuvant when indicated.

In Vivo Treatment with Cytokine, Ligands, and Antibodies.

Cytokines were suspended in PBS or sterile water as indicated. Cytokines (10 μg) were delivered via an s.c. miniosmotic pump (model 2001; Alzet) at a rate of 1 μL/h for 7 d. Human FLT3 ligand was administered i.p. at 10 μg/d in PBS, with 1 mg/mL mouse serum albumen, for 10 d. Anti-CD62L was given i.p. in a single dose of 250 μg.

In Vivo Proliferation and Division Rate.

In vivo proliferation rates and BrdU uptake were measured as described in SI Materials and Methods.

Flow Cytometry.

Cells were stained with the antibodies and analyzed as described in SI Materials and Methods.

Evaluation of FoxP3 Expression.

Cells were adoptively transferred and immunized as described above. Six or 10 d later, lymph nodes were harvested and single cell suspensions were prepared. Lymph nodes were then processed using the eBioscience FoxP3 staining buffer set (catalog no. 5523). Following treatment, the cells were stained with anti–FoxP3-phycoerythrin (PE) and evaluated via flow cytometry.

Enumeration of Cells Using Real-Time PCR.

Enumeration of cells via real-time PCR was carried out as described in SI Materials and Methods.