Gene therapy: a brief overview of the past, present, and future.

Gene therapy has only recently begun to make serious progress, beginning with two approved gene therapy trials in the United States in late 1990. The death of an 18-year-old man participating in a gene therapy trial delivered a major setback in terms of public concerns, but the resulting improvements in scrutiny of trial design and ethical standards will benefit the field in the long run. The three main issues for the coming decade will be public perceptions, scale-up and manufacturing, and commercial considerations. Focusing on single-gene applications, which tend to be rarer diseases, will produce successful results sooner than the current focus on the commoner, yet more complex, cancer and heart disease.

Long-term multilineage expression in peripheral blood from a Moloney murine leukemia virus vector after serial transplantation of transduced bone marrow cells.

Using a mouse bone marrow transplantation model, the authors evaluated a Moloney murine leukemia virus (MMLV)-based vector encoding 2 anti-human immunodeficiency virus genes for long-term expression in blood cells. The vector also encoded the human nerve growth factor receptor (NGFR) to serve as a cell-surface marker for in vivo tracking of transduced cells. NGFR(+) cells were detected in blood leukocytes of all mice (n=16; range 16%-45%) 4 to 5 weeks after transplantation and were repeatedly detected in blood erythrocytes, platelets, monocytes, granulocytes, T cells, and B cells of all mice for up to 8 months. Transgene expression in individual mice was not blocked in the various cell lineages of the peripheral blood and spleen, in several stages of T-cell maturation in the thymus, or in the Lin(-/lo)Sca-1(+) and c-kit(+)Sca-1(+) subsets of bone marrow cells highly enriched for long-term multilineage-reconstituting activity. Serial transplantation of purified NGFR(+)c-kit(+)Sca-1(+) bone marrow cells resulted in the reconstitution of multilineage hematopoiesis by donor type NGFR(+) cells in all engrafted mice. The authors concluded that MMLV-based vectors were capable of efficient and sustained transgene expression in multiple lineages of peripheral blood cells and hematopoietic organs and in hematopoietic stem cell (HSC) populations. Differentiation of engrafting HSC to peripheral blood cells is not necessarily associated with dramatic suppression of retroviral gene expression. In light of earlier studies showing that vector elements other than the long-terminal repeat enhancer, promoter, and primer binding site can have an impact on long-term transgene expression, these findings accentuate the importance of empirically testing retroviral vectors to determine lasting in vivo expression.

NKT cells are phenotypically and functionally diverse.

NK1.1(+)alpha betaTCR(+) (NKT) cells have several important roles including tumor rejection and prevention of autoimmune disease. Although both CD4(+) and CD4(-)CD8(-) double-negative (DN) subsets of NKT cells have been identified, they are usually described as one population. Here, we show that NKT cells are phenotypically, functionally and developmentally heterogeneous, and that three distinct subsets (CD4(+), DN and CD8(+)) are differentially distributed in a tissue-specific fashion. CD8(+) NKT cells are present in all tissues but the thymus, and are highly enriched for CD8alpha(+)beta(-) cells. These subsets differ in their expression of a range of cell surface molecules (Vbeta8, DX5, CD69, CD45RB, Ly6C) and in their ability to produce IL-4 and IFN-gamma, with splenic NKT cell subsets producing lower levels than thymic NKT cells. Developmentally, most CD4(+) and DN NKT cells are thymus dependent, in contrast to CD8(+) NKT cells, and are also present amongst recent thymic emigrants in spleen and liver. TCR Jalpha281-deficient mice show a dramatic deficiency in thymic NKT cells, whereas a significant NKT cell population (enriched for the DN and CD8(+) subsets) is still present in the periphery. Taken together, this study reveals a far greater level of complexity within the NKT cell population than previously recognized.

Development of T lymphocytes at extrathymic sites.

T lymphocytes expressing both CD4 and CD8 are the predominant cell type in the thymic cortex but are extremely rare outside the thymus of normal mice. In this article, we show that if precursor thymocytes (CD4-CD8-) from fetal or adult donors are injected i.v. into irradiated recipients, some of these cells will lodge in lymph nodes and develop into both CD4+CD8+ (double-positive) and CD4+ or CD8+ (single-positive) cells. This phenomenon also occurred in thymectomized recipients, strongly suggesting it is genuine extrathymic development. Prethymic precursors (e.g., fetal liver), were unable to use the lymph node for T cell development, without thymic processing. The data suggest that given unusual circumstances (irradiation or thymectomy and availability of appropriate precursors), the lymph nodes can support T cell development.

CD34+ cells from mobilized peripheral blood retain fetal bone marrow repopulating capacity within the Thy-1+ subset following cell division ex vivo.

Ex vivo cell cycling of hematopoietic stem cells (HSC), a subset of primitive hematopoietic progenitors (PHP) with engrafting capacity, is required for transduction with retroviral vectors and to increase transplantable HSC numbers. However, induction of division of HSC ex vivo also may lead to differentiation and loss of in vivo marrow repopulating potential. We evaluated mobilized peripheral blood (MPB) PHP for maintenance of stem cell function after ex vivo culture under conditions that we show can induce cycling of a majority of PHP with minimal differentiation. The following methods were combined: cell labeling with the division tracking dye carboxyfluorescein-diacetate succinimidylester (CFSE), analysis of primitive cell surface marker expression, an ex vivo PHP assay, and an in vivo marrow repopulating assay. MPB-purified CD34+ Thy-1+ cells were labeled with CFSE dye and cultured for 112 hours in serum-deprived medium in the presence of the cytokine combinations of thrombopoietin (TPO), flt3 ligand (FL), and c-kit ligand (KL), or TPO, FL, and interleukin 6 (IL-6). Both cytokine combinations supported division of greater than 95% of cells within 112 hours with an average 2.1-fold (TPO, FL, KL) or 1.3-fold (TPO, FL, IL-6) increase in total cell numbers. An average of 21.6% (TPO, FL, KL) and 27.4% (TPO, FL, IL-6) of the divided cells still expressed the Thy-1 marker after 112 hours. Functional assays were performed to compare cultured and uncultured cells. CD34+ Thy-1+ CFSElo (post division) cells showed maintenance of cobblestone area-forming cell (CAFC) frequency (a mean of 1/9.0) relative to the starting population of uncultured CD34+ Thy-1+ cells (a mean of 1/8.4). In contrast, CD34+ cells that had lost Thy-1 expression during culture (CD34+ Thy-1 CFSElo) showed a mean 5.8-fold reduction in CAFC frequency (a mean of 1/52.5). Only the Thy-1-expressing fraction of cells post culture could engraft in vivo in the SCID-hu bone assay. Because the majority of HSC functional activity post culture was found in the CD34+ Thy-1+ fraction, we focused on this fraction for subsequent analysis. CFSE labeling allows segregation and purification by flow cytometry of cells having undergone discrete numbers of divisions during culture. Very few cells that divided more than four times in culture still expressed Thy-1. Cells that retained expression of Thy-1 during culture retained CAFC activity relative to fresh CD34+ Thy-1+ cells, after undergoing at least two divisions. CAFC frequency decreased after four divisions in culture with TPO, FL, and KL or after three divisions in TPO, FL, and IL-6. We then compared populations of Thy-1+ cells that had undergone sequential numbers of divisions in culture for their ability to engraft in the SCID-hu bone assay. Engrafting ability was retained throughout four divisions in both cytokine combinations. These data demonstrate that primitive MPB CD34+ cells maintain HSC function coincident with Thy-1 expression while undergoing two to four divisions under these culture conditions. Essentially all CD34+ Thy-1+ cells divided under the conditions tested, promoting susceptibility to retroviral transduction.

Thrombopoietin, flt3, and kit ligands together suppress apoptosis of human mobilized CD34+ cells and recruit primitive CD34+ Thy-1+ cells into rapid division.

Various combinations of cytokines have profoundly different effects on inhibition of apoptosis and stimulation of self-renewal division of hematopoietic stem cells (HSC) in short-term, ex vivo culture. Our goal was to quantitate expansion of cells with a primitive CD34+ Thy-1+ phenotype, as well as cell cycling, division history, differentiation, and apoptosis of CD34+ cells enriched from normal donor mobilized peripheral blood (MPB) cells. The balance of these parameters determines the net number of transplantable HSC produced in ex vivo cultures. Comparing several different combinations of cytokines added to 90-hour cultures of MPB CD34 cells, thrombopoietin (TPO), flt3 ligand (FL), and c-kit ligand (KL) gave the best result, with the lowest percentage of apoptotic cells and a mean 1.2-fold increase in the number of CD34+ Thy-1+ cells. A combination of interleukin 3 (IL-3), interleukin 6 (IL-6), and leukemia inhibitory factor (LIF) gave the worst outcome, including a decrease of CD34+ Thy-1+ cell number to a mean of 30% of the starting cell number. Cell division history was tracked using the dye 5-(and 6-) carboxyfluorescein diacetate succinimidyl ester (CFSE). Division of CD34+ Thy-1+ cells was faster and more synchronous in TPO, FL, and KL than in IL-3, IL-6, and LIF, which left a significant proportion of CD34+ cells undivided. Such detailed analyses of short-term, ex vivo cultures generated "replication scores," which allowed prediction of a sixfold improvement of the efficiency of gene transduction of primitive hematopoietic progenitors from MPB, using TPO, FL, and KL to replace IL-3, IL-6, and LIF. Analysis of retroviral transduction efficiency confirmed the increase of transgene expression from MPB primitive hematopoietic progenitors assayed after stromal culture was fivefold, validating the usefulness of multiparameter analysis of short-term cultures for survival and replication of CD34+ Thy-1+ cells.

Quantitation of the proliferative potential of highly enriched human primitive hematopoietic progenitor cells using a stroma-free limiting dilution assay with automated scoring.

BACKGROUND: Increasing use of phenotypically-enriched stem cell populations for clinical hematopoietic transplants has led to an urgent demand for a reliable, rapid and simple functional assay which would provide an estimation of the reconstituting potential of cells prior to transplantation. METHODS: We have developed a 2-week quantitative, stroma-free assay to measure the frequency of primitive progenitors within hematopoietic cell samples. This relatively short-term assay provides frequency information which correlates with that measured by a 5-week stroma-dependent CAFC assay. Cells with the phenotype CD34+Thy-1+ were purified by fluorescence-activated cell sorting from peripheral blood apheresis products of multiple myeloma patients mobilized with cytoxan and GM-CSF. CD34+Thy-1+ cells were plated at limiting dilution into microtiter wells and cultured in an Iscove's based serum-deprived culture medium, supplemented with the cytokines, interleukin (IL)-3, IL-6, G-CSF, Flk2/Flt3 ligand (FL) and Kit ligand (KL). After 2 weeks, cell proliferation in individual wells was quantified by microscopy and bright-field imaging, or by using a fluorescent nucleic acid-binding dye and fluorimetry. Poisson statistics were used to calculate the frequency of wells containing cells with high proliferative potential (wells containing > or = 500 cells). RESULTS: Progenitor cell frequencies generated using this assay were compared by linear regression analysis to those generated from 32 parallel CAFC and CFU-C assays performed on the same patient samples. Correlations were r = 0.80, r2 = 0.65, and r = 0.76, r2 = 0.58, respectively; these correlations were highly significant (p < 10(-7)). DISCUSSION: This limiting dilution assay should more directly quantitate the potential of primitive hematopoietic cells than a CFU-C assay. It also has advantages over both the CAFC and the CFU-C assay, in that scoring has been automated, making it simple, rapid, and objective compared with manual cobblestone area or colony counting. The described limiting dilution assay may provide a useful alternative to assays currently used to evaluate the viability and proliferative potential of purified hematopoietic cells intended for transplant.

HIV, but not murine leukemia virus, vectors mediate high efficiency gene transfer into freshly isolated G0/G1 human hematopoietic stem cells.

Recent studies have opened the possibility that quiescent, G0/G1 hematopoietic stem cells (HSC) can be gene transduced; lentiviruses (such as HIV type 1, HIV) encode proteins that permit transport of the viral genome into the nucleus of nondividing cells. We and others have recently demonstrated efficient transduction by using an HIV-1-based vector gene delivery system into various human cell types including human CD34(+) cells or terminally differentiated neurons. Here we compare the transduction efficiency of two vectors, HIV-based and murine leukemia virus (MuLV)-based vectors, on untreated and highly purified human HSC subsets that are virtually all in G0/G1. The HIV vector, but not MuLV vector supernatants, transduced freshly isolated G0/G1 HSC from mobilized peripheral blood. Single-step transduction using replication-defective HIV resulted in HSC that expressed the green fluorescent protein (GFP) transgene while retaining their stem cell phenotype; clonal outgrowths of these GFP+ HSC on bone marrow stromal cells fully retained GFP expression for at least 5 weeks. MuLV-based vectors did not transduce resting HSC, as measured by transgene expression, but did so readily when the HSC were actively cycling after culture in vitro for 3 days in a cytokine cocktail. These results suggest that resting HSC may be transduced by lentiviral-based, but not MuLV, vectors and maintain their primitive phenotype, pluripotentiality, and at least in vitro, transgene expression.

Sustained gene expression in retrovirally transduced, engrafting human hematopoietic stem cells and their lympho-myeloid progeny.

Inefficient retroviral-mediated gene transfer to human hematopoietic stem cells (HSC) and insufficient gene expression in progeny cells derived from transduced HSC are two major problems associated with HSC-based gene therapy. In this study we evaluated the ability of a murine stem cell virus (MSCV)-based retroviral vector carrying the low-affinity human nerve growth factor receptor (NGFR) gene as reporter to maintain gene expression in transduced human hematopoietic cells. CD34(+) cells lacking lineage differentiation markers (CD34(+)Lin-) isolated from human bone marrow and mobilized peripheral blood were transduced using an optimized clinically applicable protocol. Under the conditions used, greater than 75% of the CD34(+) cell population retained the Lin- phenotype after 4 days in culture and at least 30% of these expressed a high level of NGFR (NGFR+) as assessed by fluorescence-activated cell sorter analysis. When these CD34(+)Lin-NGFR+ cells sorted 2 days posttransduction were assayed in vitro in clonogenic and long-term stromal cultures, sustained reporter expression was observed in differentiated erythroid and myeloid cells derived from transduced progenitors, and in differentiated B-lineage cells after 6 weeks. Moreover, when these transduced CD34(+)Lin-NGFR+ cells were used to repopulate human bone grafts implanted in severe combined immunodeficient mice, MSCV-directed NGFR expression could be detected on 37% +/- 6% (n = 5) of the donor-type human cells recovered 9 weeks postinjection. These findings suggest potential utility of the MSCV retroviral vector in the development of effective therapies involving gene-modified HSC.

High doses of purified stem cells cause early hematopoietic recovery in syngeneic and allogeneic hosts.

In humans, autologous transplants derived from bone marrow (BM) usually engraft more slowly than transplants derived from mobilized peripheral blood. Allogeneic BM transplants show a further delay in engraftment and have an apparent requirement for donor T cells to facilitate engraftment. In mice, Thy-1.1(lo)Lin-/loSca-1+ hematopoietic stem cells (HSCs) are the principal population in BM which is responsible for engraftment in syngeneic hosts at radioprotective doses, and higher doses of HSCs can radioprotect an allogeneic host in the absence of donor T cells. Using the mouse as a preclinical model, we wished to test to what extent engraftment kinetics was a function of HSC content, and whether at high doses of c-Kit+Thy-1.1(lo)Lin-/loSca-1+ (KTLS) cells rapid allogeneic engraftment could also be achieved. Here we demonstrate that engraftment kinetics varied greatly over the range of KTLS doses tested (100-10,000 cells), with the most rapid engraftment being obtained with a dose of 5,000 or more syngeneic cells. Mobilized splenic KTLS cells and the rhodamine 123(lo) subset of KTLS cells were also able to engraft rapidly. Higher doses of allogeneic cells were needed to produce equivalent engraftment kinetics. This suggests that in mice even fully allogeneic barriers can be traversed with high doses of HSCs, and that in humans it may be possible to obtain rapid engraftment in an allogeneic context with clinically achievable doses of purified HSCs.

Thrombopoietin, kit ligand, and flk2/flt3 ligand together induce increased numbers of primitive hematopoietic progenitors from human CD34+Thy-1+Lin- cells with preserved ability to engraft SCID-hu bone.

CD34(+)Thy-1(+)Lin- cells are enriched for primitive hematopoietic progenitor cells (PHP), as defined by the cobblestone area-forming cell (CAFC) assay, and for bone marrow (BM) repopulating hematopoietic stem cells (HSC), as defined by the in vivo SCID-hu bone assay. We evaluated the effects of different cytokine combinations on BM-derived PKH26-labeled CD34(+)Thy-1(+)Lin- cells in 6-day stroma-free cultures. Nearly all (>95%) of the CD34(+)Thy-1(+)Lin- cells divided by day 6 when cultured in thrombopoietin (TPO), c-kit ligand (KL), and flk2/flt3 ligand (FL). The resulting CD34(hi) PKHlo (postdivision) cell population retained a high CAFC frequency, a mean 3.2-fold increase of CAFC numbers, as well as a capacity for in vivo marrow repopulation similar to freshly isolated CD34(+)Thy-1(+)Lin- cells. Initial cell division of the majority of cells occurred between day 2 and day 4, with minimal loss of CD34 and Thy-1 expression. In contrast, cultures containing interleukin-3 (IL-3), IL-6, and leukemia inhibitory factor contained a mean of 75% of undivided cells at day 6. These CD34(hi) PKHhi cells retained a high frequency of CAFC, whereas the small population of CD34(hi) PKHlo postdivision cells contained a decreased frequency of CAFC. These data suggest that use of a combination of TPO, KL, and FL for short-term culture of CD34(+)Thy-1(+)Lin- cells increases the number of postdivision PHP, measured as CAFC, while preserving the capacity for in vivo engraftment.

Stem cell antigen 2 expression in adult and developing mice.

Stem cell antigen 2 (Sca-2) expression can distinguish the most immature T-lymphocyte precursors in the thymus from the hemopoietic stem cells. Sequence analysis of the Sca-2 protein showed that Sca-2 is a glycosylphosphatidylinositol (GPI) anchored molecule that shares some characteristics with the members of the Ly-6 multigene family, and that it is the same as the thymic shared antigen-1 (TSA-1). Here we extend these studies and critically reassess the expression of the Sca-2/TSA-1 antigen in hematopoietic tissues of adult and developing mice. With more sensitive methods we show that the distribution of Sca-2/TSA-1 differs from existing reports. We find especially high expression of Sca-2/TSA1 at day 14 of fetal development.

Thymic emigration: conveyor belts or lucky dips?

The thymic medulla has always seemed a rather uncomplicated compartment, simply storing mature thymocytes until they are exported to the peripheral lymphoid organs. However, as discussed here by Roland Scollay and Dale Godfrey, a careful look at recent data suggests that events in the medulla may be more complex and protracted than previously thought.

Perturbed development of T and B cells in mice expressing an Hlx homeobox transgene.

Within the lymphoid compartment of mice, the Hlx homeobox gene is expressed only at early stages of B-lymphoid differentiation. To determine whether Hlx influences lymphopoiesis, transgenic mice were developed to enforce Hlx expression throughout the B and T cell lineages. The strain with the highest transgene expression in both cell types (Hlx-94) exhibited marked perturbations in both B and T cell development. In these mice, the thymus lacked almost all mature CD4+8- and CD8+4- cells and the medulla was greatly shrunken, whereas nearly one-half the T cells in the periphery were CD4+8+, a cell type normally confined to the thymus. The peripheral CD4+8+ cells had some features of mature T cells, including responsiveness to mitogens. Presumably these cells had emigrated prematurely from the thymus and generated mature T cells in the periphery. Bone marrow transplantation experiments indicated that the defects was intrinsic to the Hlx-94 hematopoietic cells rather than support cells. Although thymocyte development in Hlx-94 mice was blocked at the stage when selection normally occurs, analysis of lymphocyte populations in the progeny of crosses with mice transgenic for an anti-HY T cell receptor indicated that neither positive nor negative selection of T cells was markedly affected. In addition to T cell defects, Hlx-94 mice had subnormal numbers of B lymphoid cells in the bone marrow and spleen, and their surface phenotype suggested that B cell development after the pro-B stage was impeded. Furthermore, the B cell response to stimulation with LPS was impaired. These striking developmental defects suggest that the Hlx gene may help to govern lymphoid maturation.

The gamma delta T lymphocytes of the perinatal murine thymus.

We have previously shown that the adult thymus contains three subsets of gamma delta T cells that can be defined by the expression of THY-1 and heat-stable antigen (HSA). In this study, the number of cells in each of these thymic gamma delta populations was investigated at different stages throughout life. In adult mice, the populations stay relatively constant, however, in contrast, there were major variations in them early in development. It was shown that only two of the gamma delta populations were present in the prenatal thymus, a major population of Thy-1+ HSA - cells, and a smaller population of Thy-1+ HSA+ cells. However, after birth, most of the Thy-1+ HSA - cells appear to loose the Thy-1 antigen, forming the third population of HSA - Thy-1 - cells. The adult configuration of populations appeared to be established within the first week after birth. Therefore, whereas the gamma delta populations stayed relatively constant from this time point onwards, there were major variations early in development. Throughout life, most gamma delta thymocytes are CD4- CD8-, however, in the neonatal thymus, there are some CD+ and CD+ gamma delta thymocytes, and these are contained in the Thy-1+ HSA - population.

Thymic stem cells in mouse bone marrow.

There is still controversy concerning the nature of the stem cells from bone marrow that colonize the thymus during embryogenesis and continually throughout life. To identify the bone marrow stem cells that home to and populate the thymus, we screened murine bone marrow cells for the presence of a population of surface phenotype similar to the earliest known intrathymic precursor. We have identified a population characterized by expression of an intermediate level of heat stable antigen, a very low level of Thy-1, and high levels of CD44 and class I major histocompatibility complex antigens. It is negative for B-cell, granulocyte, macrophage, and erythrocyte markers (B220, Gr-1, Mac-1, and TER 119). All these markers are common to the intrathymic precursors and bone marrow stem cells. However, this new bone marrow population is Sca-2+, similar to the intrathymic precursor, which makes a clear distinction from the Sca-2- bone marrow hematopoietic stem cells previously characterized. The frequency of the new population in the normal mouse bone marrow is about 0.25%. When transferred intrathymically or intravenously to lethally irradiated mice, it has a higher expansion potential (2 x 10(5)) than has been found for the intrathymic precursors (10(3)), but less than was found for the Sca-2- multipotent stem cell (10(7)). These transfer studies also showed that it was pluripotent, in that its precursor activity was not restricted to the production of T or B lymphocytes. However, it gave a reduced spleen colony number and smaller colonies (day-12 colony-forming unit spleen) when compared with multipotent stem cells. Thus, the cell we have identified appears to be the latest pluripotent cells so far identified in bone marrow and is therefore a good candidate for a bone marrow prothymocyte, but it appears not to be T-cell-committed.

Intrathymic lymphoid precursor cells during fetal thymus development.

Our previous studies have demonstrated the presence, in the adult mouse thymus, of a population of early precursor cells able to give rise to T and B lymphocytes but not myeloid cells. This population of cells expresses low levels of CD4 and has been termed the "low CD4 precursors." All these precursors were found to be c-kit positive, and they precede the better known CD4-CD8- precursor stage. In this study, embryonic and neonatal thymuses were examined to see whether a similar low CD4 precursor was part of the pathway of T cell development during ontogeny. A population with the phenotypic characteristics of the adult low CD4 precursor was found from day 15 of embryonic development, although the expression of low levels of CD4 was apparent only from embryonic day 17. Functional tests of these putative precursors showed they had no thymus-reconstituting ability when isolated from thymuses at any time during embryonic life, and very low reconstituting ability even 24 days after birth. These results raise questions about the adult low CD4 precursor as an obligatory stage in the development of T cells in the thymus.

TCR beta and TCR alpha gene enhancers confer tissue- and stage-specificity on V(D)J recombination events.

We describe transgenic mice carrying germline variable gene segments associated with either the T cell receptor (TCR) beta or alpha gene enhancers (E beta or E alpha). Transgenic constructs underwent high rates of site-specific rearrangements predominantly in T cells from independent mice. Rearrangements of the E beta-containing transgenes began at different stages of T cell differentiation in embryonic and adult thymus than did the E alpha-containing ones, with a pattern superimposable upon the patterns of TCR beta or TCR alpha gene expression, respectively. We demonstrate that sequences within the TCR beta and TCR alpha gene enhancers confer tissue- and stage-specificity upon the V(D)J recombination events affecting adjacent gene segments. The patterns of transgene expression also gave information on developmental events and lineage relationships (gamma delta versus alpha beta) during T cell development.

Development of gamma delta T cells in the adult murine thymus.

The development of T cells belonging to the gamma delta lineage is not well understood. We have analyzed the cells in the adult murine thymus which express the gamma delta TcR on the surface in order to learn more about this process. Our data demonstrate a number of clear subpopulations of gamma delta expressing cells in the thymus based on the expression of Thy-1 and HSA (heat-stable antigen). Only one of these subpopulations, the one expressing both Thy-1 and HSA, contains dividing cells or has a significant rate of turnover. Together with the fact that emigrant gamma delta cells are HSA+Thy-1+, this suggests that this thymic subpopulation is the sole, or major, source of exported cells. However, the turnover of cells from this population is 5 x 10(4) - 10 x 10(4) cells per day, while previous estimates of the rate of export of gamma delta cells are in the order of 10(4) cells per day. Furthermore the V gamma profile of recent gamma delta+ emigrants differs from that of the thymic HSA+Thy-1+ cells. This raises the possibility that only a selected subpopulation of the thymic gamma delta+HSA+Thy-1+ population is exported, and that some gamma delta cells may die in situ in the thymus. The function of the other gamma delta thymic subpopulations, which are turning over very slowly or not at all, (i.e. the HSA-Thy-1- and HSA-Thy-1+ subpopulations) remains unclear.