Abstract
The wavelike pattern of fetal T cell neogenesis is largely determined by the intermittent generation and exportation of waves of prothymocytes by the hemopoietic tissues in coordination with their gated importation by the thymus. Having previously shown that the importation of prothymocytes by the adult mouse thymus is also gated and that thymocytopoiesis proceeds in discrete (albeit overlapping) waves, we now demonstrate that prothymocytes are periodically exported in saturating numbers from the adult mouse bone marrow. Experiments in normal, radioablated, and parabiotic mice document the cyclical accumulation (3–5 wk) of prothymocytes in both the steady state and regenerating bone marrow, followed by their release into the blood ∼1 wk before intrathymic gate opening. The results also show that circulating donor-origin thymocyte precursors can transiently (∼1 wk) establish high level chimerism in the bone marrow after the mobilization of endogenous prothymocytes, presumably by occupying vacated microenvironmental niches. Hence, by analogy with the fetal state, we posit the existence of a feedback loop whereby diffusible chemokines of thymic origin regulate the production and/or release of bone marrow prothymocytes during each period of thymic receptivity. Because each resulting wave of thymocytopoiesis is accompanied by a wave of intrathymic dendritic cell formation, these coordinated events may help to optimize thymocyte selection as well as production.
We have previously demonstrated that thymocytopoiesis in adult mice is maintained by the gated importation of hematogenous precursors (prothymocytes) at ∼4-wk intervals (1, 2). At the level of the thymus, prothymocyte importation appears to be regulated both by the receptivity of a finite number (∼100) of intrathymic microvascular (IMV)3 gates and by the availability of an equivalent number of specific intrathymic microenvironmental (IMN) niches (3). Importantly, IMV gate closure is log dose responsive and can be induced in individual thymus lobes; and binding of prothymocytes to the IMN niches obeys one-on-one receptor occupancy kinetics. The simplest interpretation of these results is that individual IMV gates are closed upon occupation of individual, anatomically associated, IMN niches by individual prothymocytes. Conversely, IMV gate opening appears to occur only after the vast majority of the IMN niches have been vacated, suggesting that it is triggered by a threshold-dependent signal downstream of the IMN niches.
We noted previously that, at the time of IMV gate opening, sufficient prothymocytes are present in the peripheral blood to compete on an equal basis with a saturating dose of prothymocytes in donor bone marrow (BM) (3). Yet, at most other time points, blood samples contain barely detectable prothymocyte activity, even when assessed by intrathymic (i.t.) injection into irradiated recipients (4, 5). Therefore, despite the general presumption that prothymocytes are continuously released from the BM in adult mice, these observations raised the possibility that the release of prothymocytes into the blood occurs periodically in coordination with the onset of IMV gate opening. They also provided a potential explanation for our heretofore unpublished observation that prothymocyte chimerism occurs periodically and transiently in the BM of parabiotic mice.
We now show that both the accumulation in and exportation from the BM of newly formed prothymocytes occur in discrete waves that are coordinated temporally with IMV gate opening. We further show that the ability of hematogenous precursors to transiently induce prothymocyte chimerism in the BM of normal and parabiotic mice correlates temporally with the periodic release of endogenous prothymocytes from a finite number of presumptive microenvironmental niches in the BM. These results in their aggregate suggest that, as in the fetus (6, 7, 8), a regulatory feedback loop exists in the adult that synchronizes the generation and/or release of prothymocytes from the BM with their gated importation by the thymus. A schematic diagram of these coordinated cycles of prothymocytopoiesis and thymocytopoiesis is presented, and the possible underlying regulatory mechanisms are discussed.
Materials and Methods
Animals
Inbred male and female mice of the Thy-1-congenic B10.S strain, bred in our laboratory (3), were used for parabiotic studies. Inbred female Ly-45-congenic C56BL/6 mice, obtained from National Cancer Institute (Charles River, Frederick, MD), were used for adoptive transfer studies. Animals were maintained on commercial mouse chow and water ad libitum in the Center for Laboratory Animal Care, University of Connecticut Health Center. Irradiated mice (and nonirradiated controls) were placed on acidified, chlorinated water for 10 days.
Parabiosis
Pairs of 5-wk-old, sex- and weight-matched, Thy-1 alloantigen-disparate B10.S mice were parabiosed, as previously described (1). At the time of sacrifice, the parabiotic mice were anesthetized with ketamine/acepromazine; the exchange of blood between the parabionts was interrupted by gentle compression of the anastomotic bridge with padded Bainbridge forceps; the animals were surgically separated; and the tissues were harvested.
Preparation of cell suspensions
BM cell suspensions were prepared by flushing the marrow from the femur with cold RPMI 1640 (Life Technologies, Gaithersburg, MD) supplemented with sodium bicarbonate (2 mg/ml) and 1% HEPES (1.5 M). Repeated gentle pipetting further dispersed the cells, which were then washed in cold medium and centrifuged at 4°C for 5 min at 1500 rpm. Thymocyte cell suspensions were prepared by gently pressing thymus lobes, stripped of attached lymph nodes, through a stainless steel screen (50 mesh), followed by washing in cold medium. Suspensions of nucleated peripheral blood cells (PBCs) were prepared by mixing 0.5 ml of heart blood with 10 ml Alsever’s solution, lysing the RBCs with 0.165 M NH4C1, and washing the nucleated cells in cold medium. Cell counts were performed on a calibrated Z1 Coulter Counter (Beckman Coulter, Hialeah, FL).
Intrathymic adoptive transfer assay for prothymocytes
Recipient mice received 6 Gy total body irradiation (0.97 Gy/min) from a 137Cs source (Gamma Cell 40 Irradiator; Atomic Energy of Canada, Ottawa, Canada) 2–4 h before BM or PBC injection. The thymus was surgically exposed, and the indicated numbers of cells were injected into the anterior superior portion of each lobe (10 μl/site) using a 1-ml syringe (with attached 28-gauge needle) mounted on a Tridek Stepper (Indicon, Brookfield Center, CT), as described (4). The skin incision was closed with Nexaband Liquid (Veterinary Products, Phoenix, AZ).
Intravenous adoptive transfer assay for prothymocytes
The indicated number of BM cells suspended in 0.5 ml of RPMI 1640 was injected through a 28-gauge needle into the lateral tail veins of irradiated (6 Gy) or nonirradiated recipient mice. Control mice were injected with RPMI 1640 alone.
Flow immunocytometric analysis
Thymocytes were harvested 28 days after BM or PBC transfer or at timed intervals after parabiosis, as indicated. The percentages of donor- and host-origin cells were determined by flow immunocytometric analysis (FACS Analyzer or FACScan; BD Biosciences, Sunnyvale, CA) after development for immunofluorescence. Fluorescein-labeled 30-H12 mAb against Thy-1.2 was obtained from BD Immunocytometry Systems (Mountain View, CA). The OX7 mAb against Thy-1.1 was obtained from Bioproducts for Science (Indianapolis, IN). Anti-CD45.1 and anti-CD45.2 mAbs were obtained from The Jackson Laboratory (Bar Harbor, ME). Dead cells and nonlymphoid cells were excluded by gating for forward and side angle light scatter, and 10,000 viable cells were collected in each file. Specificity and sensitivity of staining were controlled by checkerboard analysis against normal Thy-1.1 and Thy-1.2 or CD45.1 and CD45.2 thymocytes and purposeful mixtures thereof. The percentage of positive cells was calculated by using the intersection of the fluorescence histogram with its control profile to determine the cutoff point.
Results
Periodic establishment of transient prothymocyte chimerism in the BM of parabiotic mice
We previously studied the kinetics of the establishment of thymocyte chimerism in 105 pairs of Thy-1 congeneic B10.S parabiotic mice (1). The results showed that the level of chimerism (percentage of donor-origin thymocytes) increased progressively between wk 4 and 7 of parabiosis, decreased modestly between wk 7 and 11, and then returned to peak levels through wk 21. Although not reported at that time, we also assayed the BM from each of these parabionts for prothymocyte chimerism by transferring saturating numbers of BM cells (20 × 106) i.v. into sublethally irradiated (6 Gy) Thy-1.1 and Thy-1.2 recipients. The results, shown in Fig. 1⇓, A–C and E–G, demonstrate a cyclical and roughly synchronous pattern of transient (1–2 wk) prothymocyte chimerism in the BM of the parabiotic partners, as evidenced by the induction of thymocyte chimerism among the BM cell recipients. Thus, spikes of prothymocyte chimerism were detected in the BM of the Thy-1.1 parabionts at wk 5, 11–13, and 21 and of the Thy-1.2 parabionts at wk 4–5, 7–10, possibly 17, and 21 of parabiosis. Because BM samples were not obtained at wk 1, 12, 14–16, or 18–20 of parabiosis, it is possible that one or more additional spikes of prothymocyte chimerism occurred, but were overlooked. As indicated below, this is especially likely for wk 1.
Periodic establishment of transient prothymocyte chimerism in the BM of Thy-1 congenic parabiotic mice. BM cells (20 × 106), obtained from each partner at the indicated time points after the establishment of parabiosis, were injected i.v. into sublethally irradiated (6 Gy) recipients (Thy-1.1 BM into Thy-1.1 recipients; Thy-1.2 BM into Thy-1.2 recipients). Thymocytes were harvested from the recipients 28 days later, and the frequency of chimerism (percentage of mice with ≥5% Thy-1-disparate thymocytes) (A, E) was determined by FACS analysis. Of the positive mice, the percentage (B, F) and number (C, G) of Thy-1-disparate thymocytes per thymus were calculated (thickened baselines indicate <5% (B, F) and <10 × 106 (C, G) Thy-1-disparate thymocytes). In addition, the number of Thy-1-identical thymocytes per thymus (D, H), corrected for background numbers of thymocytes in sham-injected (medium only) recipients, was calculated for each time point. Results represent the mean values ± SD for four to eight pairs of parabiotic mice conjoined at 5 wk of age. Data were not available for wk 1, 12, 14–16, and 18–20 (broken baseline). ∗, p < 0.05 between the indicated value and the preceding peak value in D and H.
It should be noted that higher levels of thymocyte chimerism were induced by the BM of the Thy-1.2 parabionts than the Thy-1.1 parabionts (∼70% vs ∼30% donor-origin cells) (Fig. 1⇑, B vs F). This is presumably related to our observation in this same cohort of mice that Thy-1.1 prothymocytes have a ∼2-fold competitive advantage over an equivalent number of Thy-1.2 prothymocytes in establishing thymocyte chimerism (1). Hence, when normalized, the data suggest that the peak levels of prothymocyte chimerism approximated 50% in both parabiotic partners.
As shown in Fig. 1⇑, D and H, the initial two spikes of prothymocyte chimerism (in which the data are most complete) were preceded (and followed) by a wave of endogenous prothymocyte activity in the BM. Quantitatively, the numbers of thymocytes generated at the peaks of each wave of endogenous prothymocyte accumulation were ∼2-fold greater than those at the valleys. Hence, by reference to log dose-response curves in this assay system (4), it can be estimated that the number of prothymocytes present per unit number of BM cells varied 4- to 8-fold between the peaks and the valleys. Furthermore, when corrected for the competitive advantage of Thy-1.1 over Thy-1.2 prothymocytes in these mice (1), the total prothymocyte activity (endogenous plus chimeric) in the BM during this period was essentially constant. In addition, the marked increase in endogenous prothymocyte activity between wk 2 and 3 of parabiosis suggested that a spike of prothymocyte chimerism would likely have occurred during wk 1. Unfortunately, this could not be confirmed because blood exchange was not fully established in these mice until the end of the first week of parabiotic union (1).
The simplest explanation for these observations is that circulating prothymocytes establish transient chimerism in the BM of contralateral parabionts by occupying a set of niches that have been vacated by the periodic release of a wave of endogenous prothymocytes into the blood. This chimerism is terminated ∼1 wk later by the export of these chimeric prothymocytes and the in situ generation of a new wave of endogenous prothymocytes. Furthermore, as the apparent periodicity of prothymocyte chimerism in the BM of these parabionts approximates that of IMV gate opening in the steady state thymus (2), an attractive corollary is that the periodic release of prothymocytes from the BM is coordinated with their gated importation by the thymus. This notion is supported by the present evidence (Fig. 1⇑, D and H) for the exportation of prothymocytes from BM to blood during wk 1 of parabiosis, and our earlier evidence in these same mice (1) for the initial importation of prothymocytes from the blood to the thymus during wk 2.
Cyclical variation of prothymocyte activity in regenerating BM
Although the general wavelike pattern of prothymocyte accumulation in BM is clear from the preceding experiments, the kinetics are somewhat obscured by asynchrony within the cohort of mice and possibly by the parabiotic state itself. We therefore attempted to synchronize the generation of prothymocytes by sublethally irradiating (6 Gy) a cohort of nonparabiotic C56BL/6 mice. At weekly intervals thereafter for a period of 19 wk (6–24 wk of age), the femoral BM cells were harvested from a different group of mice and assayed for prothymocyte activity by i.v. transfer into sublethally irradiated Ly-45 congenic recipients. The results in Fig. 2⇓ show that, after a lag of ∼2 wk, prothymocyte activity reappeared in the regenerating BM and progressively increased to peak levels by wk 7–9. Thereafter, the prothymocyte activity varied cyclically (periodicity 3–4 wk), reaching peaks at wk 12, 15, and 19 and valleys at wk 10–11, 14, and 17 after irradiation. The difference between the levels of thymocyte chimerism at the peaks and valleys remained relatively constant between cycles (p > 0.05), approximating 2-fold when expressed as percentage of donor-origin thymocytes (Fig. 1⇑A) and 3-fold when expressed as numbers of donor-origin thymocytes (Fig. 2⇓B). Parallel control experiments using saturating numbers (20 × 106) of BM cells from normal (nonirradiated) weanling mice demonstrated that the total prothymocyte activity per femur in the postirradiation BM did not reach saturating levels, and that the sensitivity of the assay system (total number of donor-origin thymocytes generated) did not vary significantly over time (data not shown). Therefore, by reference to log dose-response curves (3), it can be estimated that the numbers of prothymocytes in the regenerated BM varied cyclically over a range of ∼5-fold. Again, the most satisfying explanation for these results is that the vast majority of newly formed prothymocytes are released into the blood by the end of each cycle, and that these are rapidly replaced by proliferating precursors whose progeny occupy a finite number of niches in the BM.
Cyclical variation of the relative prothymocyte activity in the regenerating BM of sublethally irradiated mice. A cohort of 5-wk-old CD45.2 C56BL/6 mice was sublethally irradiated (6 Gy) to synchronize the regeneration of prothymocytes in BM. BM cells were harvested from a different group of four mice at weekly intervals thereafter and tested for relative prothymocyte activity. For this purpose, the total BM cells in a single femur from each donor were injected i.v. into an irradiated (6 Gy) 5-wk-old CD45.1 recipient (one femoral equivalent per recipient), and the peak level of thymocyte chimerism for each time point was determined 28 days later. Results indicate the mean percentage ± SD (A) and mean number ± SD (B) of donor-origin thymocytes. Representative experiment (one of two with similar results). ∗, p < 0.01 between the indicated values and the preceding peak value.
Coordination of the exportation of prothymocytes by the BM with their gated importation by the thymus
To formally test our hypothesis of cyclically coordinated exportation and importation of prothymocytes by BM and thymus, we compared the kinetics of appearance of prothymocytes in the blood of normal (nonirradiated) adult mice with both the availability of niches for prothymocytes in BM and the status (open or closed) of the IMV gates for prothymocytes. In these experiments, a cohort of normal C56BL/6 (CD45.2) mice was divided into six groups and, at weekly intervals between 7 and 12 wk of age, a different group was injected i.v. with BM cells from weanling CD45.1 donors. In addition, some mice in each group were not injected. The receptivity of the thymus for prothymocytes (open IMV gates; empty BM niches) was determined by measuring the frequency and mean levels of thymocyte chimerism 28 days after cell transfer. The receptivity of the BM for prothymocytes (empty BM niches) was determined by measuring the ability of such BM, obtained 4 days after cell transfer, to establish thymocyte chimerism in irradiated CD45.2 recipients. The relative prothymocyte activity in the blood of the uninjected mice was determined by measuring the ability of the nucleated cells in 0.5 ml of heart blood to induce thymocyte chimerism after i.t. injection into irradiated CD45.1 recipients (one donor per recipient).
As shown in Fig. 3⇓, A and B, the thymus was receptive to the entrance of hematogenous prothymocytes at wk 9 only, when 50% of the normal recipients became chimeric. This is consistent with our previous observations (2). The BM in these same recipients (Fig. 3⇓, C and D) were receptive to the induction of prothymocyte chimerism at wk 7 and 8 (peak), and again at wk 12. Temporally, a wave of prothymocyte activity appeared in the blood of the uninjected mice between wk 7 and 11 (Fig. 3⇓, E and F). Of special importance, the highest levels of prothymocyte activity in the blood of the uninjected mice and of prothymocyte chimerism in the BM of the injected mice were observed ∼1 wk before the onset of thymic receptivity. Thereafter, the BM became refractory to colonization by hematogenous prothymocytes, and the prothymocyte activity in the blood progressively decreased ∼5-fold over a 4-wk period. Hence, in their aggregate, these results indicate that under steady state conditions prothymocytes are cyclically released from BM to blood, after which many are recruited in a gated fashion by the thymus. In addition, the phenomenon of transient chimerism of the BM with donor prothymocytes correlates closely with (and presumably is causally related to) the timed release of endogenous prothymocytes into the blood.
Coordinated exportation and importation of prothymocytes by the BM and thymus in normal mice. A cohort of 5-wk-old (±3 days) CD45.2 C56BL/6 mice was divided into six groups (25 mice each), and at weekly intervals over a 6-wk period the members of a different group were injected i.v. with a saturating dose (20 × 106) of CD45.1 BM cells (15 mice) or were uninjected (10 mice). Of the BM cell-injected subgroup, 10 mice were examined 28 days after cell transfer for the development of thymocyte chimerism (A, B), and 5 mice were examined 4 days after cell transfer for the occurrence of prothymocyte chimerism in BM (C, D). In the latter assay, 20 × 106 BM cells from the primary CD45.2 recipients were injected i.v. into irradiated (6 Gy) CD45.2 secondary recipients, and the development of CD45.1 thymocyte chimerism was determined 28 days later. Of the uninjected subgroup of mice, 10 mice were examined each week for relative prothymocyte activity in peripheral blood (E, F). In this assay, the total nucleated cells in 0.5 ml of blood were injected i.t. into irradiated (6 Gy) CD45.1 recipients (one donor per recipient), and CD45.2 thymocyte chimerism was evaluated 28 days later. The results of each of the preceding assays are expressed as the percentage of mice with detectable donor-origin (thymus, BM) or host-origin (blood) prothymocyte activity (≥5% thymocyte chimerism in irradiated recipients) (A, C, E), and as the mean levels (% ± SD) of donor-origin thymocytes in the chimeric recipients (B, D, F). Representative experiment (one of three with similar results). ∗, p < 0.05 between the levels of prothymocyte activity in the blood at the indicated ages and the peak level at wk 8 in F (insufficient data at wk 11).
Discussion
During fetal and neonatal life in birds and mice (as well as in larval and postmetamorphic frogs), waves of prothymocytes are exported from the primary hemopoietic tissues to the blood and imported in a gated manner by the thymus (6, 7, 8). This mechanism permits the sequential generation, selection, and exportation of developmentally and functionally discrete populations of thymocytes to the skin, liver, mucosa, and peripheral lymphoid tissues (9, 10). However, once this repertoire is established, heterogeneous populations of T cells rather than waves of specialized T cells appear to be produced. Nonetheless, the gated importation of saturating numbers of hematogenous prothymocytes by the thymus continues into adult life, even during thymic involution (2, 3).
These observations raised the possibility that the BM also continues to release waves (rather than constant numbers) of prothymocytes into the blood, possibly in coordination with their gated importation by the thymus. The present results strongly support this hypothesis. Thus, in a cohort of normal mice observed over a 6-wk period, a wave of prothymocyte activity, peaking ∼1 wk before IMV gate opening, appeared in the blood. This is consistent with our recent observation (5) that ∼25% of blood samples from a cohort of normal 7- to 12-wk-old mice contained readily detectable prothymocyte activity (33% of blood samples were positive in the present study). Furthermore, the difference between the mean numbers of prothymocytes in the positive and negative blood samples was at least 10-fold, consistent with the present data. These results appear to explain why blood, which ordinarily is a poor source of prothymocytes (4), contains sufficient prothymocytes to saturate the thymus at the time of IMV gate opening (3).
Two additional observations made in this work also point to the periodic release of prothymocytes from the BM of adult mice. The first concerns the cyclical pattern of endogenous prothymocyte activity present in the BM of both parabiotic and sublethally irradiated mice followed over a 19- to 21-wk period. The second concerns the receptivity of the BM to the induction of high levels of transient prothymocyte chimerism at the end of each cycle (and at the time of appearance of maximal prothymocyte activity in the blood). In a separate study, we have observed the periodic entrance of saturating numbers of prothymocytes into the thymus under circumstances in which the IMV gates fail to close (D. Foss and I. Goldschneider, manuscript in preparation). None of these phenomena would be expected if the release of prothymocytes from the BM occurred continuously and at relatively low levels.
We have previously demonstrated that the relatively low mean levels of thymocyte chimerism observed in parabiotic mice (16–25%) are due to the inefficient exchange of hematogenous prothymocytes between the parabiotic partners (1). Hence, it was surprising that the hematogenous thymocyte precursors that entered the BM of these same parabionts established much higher levels of chimerism (30–70%). Indeed, when the results in Fig. 1⇑, B and F, are corrected for the ∼2:1 competitive advantage of Thy-1.1 over Thy-1.2 prothymocytes for IMN niches (1), it appears that the population of thymocyte precursors that return to BM is randomly exchanged in the blood. We therefore infer that these cells differ from those that migrate to thymus, possibly representing the less mature subset of prothymocytes described by Spangrude et al. (11). It is unlikely that they are pluripotent hemopoietic stem cells, which appear to be exchanged inefficiently between parabiotic partners, establish much lower levels of long-term chimerism in the BM, and maintain a relatively stable frequency in the blood (1, 12). In addition, the results of BM transfer experiments (Fig. 3⇑) suggest that these cells gain the ability to migrate to the thymus within 4 days. This may help to explain why the chimeric prothymocytes only remain in the BM ∼1 wk before they are replaced by newly generated endogenous prothymocytes. It is of importance that the total prothymocyte activity in the chimeric BM of the parabiotic partners (Fig. 1⇑C), when corrected for the competitive advantage of Thy-1.1 over Thy-1.2 prothymocytes, approximated the peak levels of endogenous prothymocytes in nonchimeric BM. As in the thymus (3), these results suggest that a finite number of niches for prothymocytes also exists in the BM, and that most of these niches are periodically and synchronously vacated.
Conceptually, a series of regulatory feedback loops would be required to coordinate the production and release of prothymocytes from the BM with their gated importation by the thymus. As illustrated schematically in Fig. 4⇓, at least seven events must be integrated in a cyclical fashion: 1) mobilization of prothymocytes from the BM; 2) recruitment of mobilized prothymocytes to (and/or arrest by) specialized vascular endothelium in the thymus; 3) opening of the IMV gates; 4) entrance of prothymocytes into the IMN niches; 5) closing of the IMV gates; 6) generation of a wave of developing thymocytes (and dendritic cells) (5); 7) regeneration of the pool of prothymocytes in the BM. Although considered separately below, these are not necessarily exclusive processes. For example, mobilization and recruitment of prothymocytes may have a common mechanism.
Integrated scheme of the cyclical accumulation, mobilization, and gated importation of prothymocytes by the BM and thymus of normal adult mice. Four major phases of this cycle are arranged clockwise around a 4-wk time wheel, each phase lasting ∼1 wk. This time frame is based on the data from cohorts of normal age-matched mice (Figs. 1⇑ and 3⇑) (2 ), but may more closely approximate 3 wk in individual mice or groups of synchronized mice (Fig. 2⇑) (2 ,3 ). In phase 1 (upper panel), a wave of prothymocytes (PT; large •) is released from the BM into the blood, vacating a set of microenvironmental niches (open box in BM) that are then temporarily available for reoccupation by circulating immature prothymocytes (data not shown). At this time, the IMV gates for prothymocytes (filled bar in thymus) are closed, and the IMN niches for prothymocytes (open box in thymus) are vacant. In phase 2 (right panel), the IMV gates open, and hematogenous prothymocytes enter the IMN niches in saturating numbers. At this time, few prothymocytes remain in the BM niches, and excess prothymocytes in the blood are presumed to undergo apoptosis. In phase 3 (lower panel), the IMN niches in the inner thymic cortex are stably occupied by recently imported prothymocytes and their pro-T1 (CD25− CD44+) TN (CD3− CD4−, CD8−) progeny. At this time, the IMV gates close, few prothymocytes remain in the blood, and newly generated prothymocyte precursors (small •) begin to occupy the vacated niches in the BM. In phase 4 (left panel), the developing thymocytes begin to vacate the IMN niches, whereas the BM niches are filled with maturing prothymocytes. At this time, few prothymocytes are present in the blood, and the IMV gates are closed. These events initiate a wave of thymocytopoiesis that lasts ∼8 wk. Hence, by reference to our data on the kinetics of thymocytopoiesis after IMV gate opening (2 ) and to the scheme of Lind et al. (44 ) on early thymocyte migration, pro-T1 (TN1) thymocytes would appear to be generated in the INM niches in the inner thymic cortex during phase 3; pro-T2 (TN2) thymocytes (CD25+ CD44+) in the mid-cortex during phase 4; pro-T3 (TN3) thymocytes (CD25+ CD44−) and pro-T4 (TN4) thymocytes (CD25− CD44−) in the outer cortex during phase 1; and double-positive (DP; CD4+ CD8+) thymocytes in the mid-cortex during phase 2. The generation and exportation of negatively selected CD4 and CD8 single-positive thymocytes from the thymus medulla (not shown) would primarily occur during the next cycle of prothymocyte mobilization and importation, during which time a second wave of thymocytopoiesis would be initiated in the thymic cortex. Consequently, despite the gated entry of prothymocytes, the production of overlapping waves of thymocytopoiesis maintains relatively constant numbers and subsets of thymocytes under steady state conditions (2 ). Some of the possible mechanisms that coordinate these cyclical events are presented in Discussion.
The simplest, but certainly not the only, explanation for the mobilization/recruitment of BM prothymocytes would be the release of a diffusible chemokine by the thymus ∼1 wk before the opening of the IMV gates. This is not a new idea, as many authors have provided evidence that chemotactic factors secreted by thymic epithelial cells can attract migrant precursors from the primary hemopoietic tissues (see Anderson et al. (13) for review). However, most of these studies, our own included (14), were conducted in the nonvascularized thymus (fetal or cultured), and the responsible factors were not identified. Of particular interest to the present study is the report of Wilkinson et al. (15), who found that fetal thymus lobes taken during the refractory period following initial lymphoid colonization attracted precursors much less efficiently than did alymphoid thymic lobes. The authors argued that the attraction of T cell precursors by the avascular fetal thymus is regulated by its lymphoid content, and speculated that this is due either to decreased chemoattractant synthesis/secretion or increased chemoattractant consumption consequent to thymocyte-epithelial cell interaction. However, given the overlapping wavelike generation of thymocytes in the vascularized thymus (2), it is possible that similar thymocyte-stromal cell interactions may signal the periodic mobilization of prothymocytes from the adult BM.
Inasmuch as the mobilization of prothymocytes from the BM occurs when the IMN niches in the deep cortex have been vacated (2, 3) and the resulting wave of newly generated triple-negative (TN; CD3−, CD4−, CD8−) thymocytes is migrating toward the outer cortex (16), one could envision a number of sites at which thymic stromal cells might be induced to generate a diffusible chemotactic signal for prothymocyte recruitment. Of the candidate chemokines tested, both stromal cell-derived factor-1 (CXCL12) and thymus-expressed chemokine (CCL25) have been implicated in possible prothymocyte mobilization/recruitment (17, 18, 19, 20). β2-microglobulin is another thymus stromal cell-derived molecule (albeit not a chemokine) that has been postulated to be involved in prothymocyte mobilization/recruitment (21, 22, 23). In addition, the variant isoforms of the CD44 integrin (CD44v), which appear to be expressed preferentially by hemopoietic progenitor cells, have recently been implicated in the recruitment and initial interaction of prothymocytes with the thymic stroma (24, 25, 26). However, none of these factors appears to be essential for this process (15, 26, 27, 28, 29, 30). Important insights into the identity of the key factors responsible for prothymocyte mobilization/recruitment may be forthcoming from the intriguing observations that prothymocyte importation by lymph nodes can be induced in normal mice by treatment with oncostatin-M (30, 31, 32) and possibly leukemia-inhibitory factor (33).
Once mobilized, prothymocytes must be attracted to the thymus and arrested at the IMV gates, presumably by the periodic expression of tissue-specific patterns of chemokines and adhesion molecules analogous to those expressed by the high endothelial venules in the secondary lymphoid tissues (34, 35, 36, 37). In this regard, Horst et al. (38) and Iizuka et al. (39) have identified a subpopulation of venules in the corticomedullary region of the thymus that express the HECA-452 and mucosal addressin cell adhesion molecule-1 homing receptors characteristic of high endothelial venules. Transmigration of prothymocytes between endothelial cell junctions and localization in the associated IMN niches would then follow, presumably guided by chemokines and adhesion molecules from the underlying fibroblastic and epithelial elements (13, 40).
Because IMV gate opening does not occur until ∼1 wk after the release of prothymocytes from the BM, we infer that it is induced by a separate signal. Although the nature of this signal is not known, adoptive transfer studies suggest that it is not associated with the mobilized prothymocytes themselves (otherwise there would be no refractory period). Neither does it appear to be induced by a diffusible factor from the BM, as gate opening is not necessarily synchronized between the parabiotic partners (2, 5). Rather, because IMV gate opening occurs ∼1 wk after the IMN niches for prothymocytes empty, we postulate that it is associated with a downstream signal induced by the interaction of the most recent wave of developing thymocytes with the stroma in the outer thymic cortex (see Fig. 4⇑).
Whatever the mechanism for IMV gate opening, it appears that once a prothymocyte is stably bound to an IMN niche or begins to differentiate into pro-T1 (CD25− CD44+) cells, a signal is generated that closes the associated IMV gate (3). Thus, IMV gate closing occurs progressively during the first week of prothymocyte importation, is log dose dependent, and can be induced unilaterally by the i.t. injection of BM cells into a single thymus lobe. Indeed, in the presence of subsaturating numbers of prothymocytes, many IMV gates remain open and are available for further prothymocyte recruitment during the normally refractory period. We do not know whether the resulting asynchronicity of thymocytopoiesis (and associated thymic dendritic cell development) (5) has adverse consequences, but the postulated thymus-BM feedback loop seems designed to prevent this.
A potentially important insight into the regulation of IMV gate closing has recently been provided by H. Petrie et al. (Memorial Sloan-Kettering Cancer Center; personal communication), who observed that nonirradiated IL-7R−/− mice, but not equally thymocytopenic RAG−/− mice, routinely develop high levels of thymocyte chimerism after the i.v. transfer of wild-type BM cells. These results not only suggest that most of the IMN niches in IL-7R−/− mice are available for engraftment by IL-7R+/+ prothymocytes, but that the IMV gates remain open in the absence of IL-7R signaling. Given that IL-7R has been shown to activate integrin-mediated adhesion of early thymocytes (41), it is reasonable to speculate that IL-7 signaling may be involved in IMV gate closing. Alternatively, the effect of IL-7 may be indirect, if IL-7R−/− prothymocytes, like their Janus kinase 3−/− counterparts (42), are present in abnormally low numbers in BM or fail to efficiently bind to the IMN niches.
A final consideration concerns a possible role for diffusible thymic factors in regulating the repopulation of the BM with prothymocytes (43). The central question in this study is whether prothymocytopoiesis in BM normally proceeds at a constant rate or in periodic bursts. Although this remains to be determined, the rapid refilling of the vacated BM niches with newly formed prothymocytes after the release of their predecessors into the blood favors the latter possibility. If true, then the “clock” that determines the periodic release of prothymocytes from the BM may be set at the level of prothymocyte generation as well as (or instead of) prothymocyte mobilization.
In summary, the present results support the existence of an integrated regulatory circuit that coordinates the cyclical release (and possibly generation) of saturating numbers of prothymocytes from the BM with their gated importation by the adult thymus. Based on available evidence from the fetus, we postulate that the resulting waves of developing thymocytes activate this feedback loop by modulating the production, use, and/or release of diffusible chemokines by the stromal elements of the thymic cortex.
Acknowledgments
We thank Frances Tausche for expert technical assistance and Jennifer Wegh for excellent secretarial assistance.
Footnotes
-
↵1 This study was supported in part by National Institutes of Health Grants AI33741 and AI49882.
-
↵2 Address correspondence and reprint requests to Dr. Irving Goldschneider, Department of Pathology, School of Medicine, University of Connecticut Health Center, 263 Farmington Avenue, Farmington, CT 06030-3105. E-mail address: igoldsch{at}neuron.uchc.edu
-
↵3 Abbreviations used in this paper: IMV, intrathymic microvascular; BM, bone marrow; IMN, intrathymic microenvironmental; i.t., intrathymic; PBC, peripheral blood cell; TN, triple negative.
- Received May 27, 2003.
- Accepted July 31, 2003.
- Copyright © 2003 by The American Association of Immunologists
References
In this issue
