Apoptotic cell death induced by a mouse-human anti-APO-1 chimeric antibody leads to tumor regression

The murine anti-APO-1 antibody (gamma 3, kappa) induces programmed cell death (apoptosis) following binding to the APO-1 antigen (m.w., 48 kDa) expressed, e.g., on activated or malignant lymphocytes. APO-1 expression on malignant cell lines and tissues suggested potential clinical utility supported by anti-APO-1-mediated tumor regression in a nude mouse model. A mouse-human anti-APO-1 chimeric antibody (gamma 3, kappa) with an affinity similar to that of the murine antibody was produced. Chimeric anti-APO-1 showed the same potential to inhibit growth of the SKW6.4 B-lymphoblastoid cell line as murine anti-APO-1. In addition, both the chimeric and murine anti-APO-1 antibodies were equally capable of mediating complete macroscopic tumor regression of a SKW6.4 xenotransplant in SCID mice by induction of apoptosis. Induction of apoptosis was the only mechanism for tumor regression because neither murine nor chimeric anti-APO-1 showed anti-tumor activity against solid H53 tumor (APO-1 antigen-positive, anti-APO-1-resistant) xenotransplants. Our results indicate that the chimeric anti-APO-1 antibody effectively induces apoptosis and suggest that chimeric anti-APO-1 should be evaluated for the treatment of malignant cells expressing the APO-1 antigen. However, chimeric anti-APO-I might only be used therapeutically when the antibody can be targeted specifically to tumor cells.

Regulation of apoptosis in the immune system

Apoptosis in T and B lymphocytes is involved in all fundamental processes in the immune system. It is a mechanism to regulate the course of an immune response and to establish immunological memory as well as central and peripheral tolerance. Apoptosis in lymphocytes is regulated by gene products that induce or block this process. Elucidating the molecular basis for sensitivity and resistance towards induction of apoptosis is the key to the understanding of the development of the immune system, basic immune reactions and the pathogenesis of autoimmune diseases, AIDS and cancer.

Differential expression of APO-1 on human thymocytes: implications for negative selection?

Negative selection during T cell ontogeny involves selective induction of apoptosis in thymocytes. In peripheral lymphoid cells, apoptosis may be mediated via the APO-1 pathway. Here we report that APO-1 is constitutively expressed on the vast majority of human thymocytes but down-regulated at a mature stage of thymocyte development (TCR(hi)). This stage of development is characterized by CD28hi, CD44hi, CD69hi and up-regulation of Bcl-2 protein. We define a new thymocyte subpopulation that expresses high levels of APO-1 and intermediate levels of T cell receptor alpha/beta (TCR(im)/APO-1hi). The TCR(im)/APO-1hi population contains a large fraction of dead cells, suggesting that the APO-1 pathway may be involved in negative selection of at least a fraction of thymocytes after intrathymic activation.

Circulating haematopoietic progenitors during treatment of renal anaemia with recombinant human erythropoietin

The effect of recombinant human erythropoietin (rhEPO) and interleukin 3 (IL3) on circulating haematopoietic progenitors consisting mainly of immature burst-forming-units-erythrocytes (BFU-E), was investigated in ten paediatric patients treated by regular haemodialysis. During a 30-week study rhEPO treatment resulted in a rise of median haemoglobin levels from 6.7 g/dl to > 10 g/dl in all patients. Before initiating rhEPO treatment the number of circulating BFU-E in chronic renal failure patients responded to grading doses of rhEPO in vitro similar to that in control children; however, the dose-response curves were not predictive for the in vivo response to rhEPO. After an initial rise in five patients BFU-E numbers declined by week 30 of rhEPO treatment. BFU-E numbers decreased to 35% of pretreatment values. The number of granulocyte-macrophage colony forming cells (GM-CFC) also decreased during rhEPO treatment. Addition of IL3 to the culture medium containing saturating concentrations of granulocyte-macrophage colony stimulating factor did not stimulate BFU-E numbers of patients before rhEPO treatment or those of controls. However, 2 weeks after start of rhEPO treatment IL3 increased the growth of patient's BFU-E in vitro to 220% of pretreatment levels, followed by a gradual decrease of stimulation until the end of observation. These findings indicate that: (1) long-term recruitment of circulating haematopoietic progenitors during rhEPO treatment is low in children with renal anaemia; (2) rhEPO sensitivity of circulating BFU-E is not predictive for the in vivo response; (3) rhEPO treatment results in enhanced sensitivity of BFU-E to IL3.

Constitutive and induced expression of APO-1, a new member of the nerve growth factor/tumor necrosis factor receptor superfamily, in normal and neoplastic cells

BACKGROUND

APO-1 is a 48 kilodalton transmembrane, cysteine-rich glycoprotein identical with the Fas antigen which belongs to the nerve growth factor/tumor necrosis factor receptor superfamily. Cross-linking of APO-1 induces apoptotic cell death in sensitive cells.

EXPERIMENTAL DESIGN

As suggested by our preliminary results, APO-1 expression is not restricted to cells of the hematopoietic lineage. We therefore investigated APO-1 expression in normal human tissues and in various epithelial and nonepithelial tumors.

RESULTS

We show by immunohistochemistry that APO-1 is a non-lineage antigen constitutively expressed in a variety of epithelial cells. This includes the basal layers of various squamous epithelia, transitional epithelium and columnar epithelium of the biliary tract and intestine. Among the epithelial cell types of the reproductive system of both genders, APO-1 expression is complex. Except the satellite cells of autonomic ganglia, all cells of the nervous tissue are APO-1-negative. Among mesenchymal cells, constitutive APO-1 expression is rare but detectable in various kinds of activated cells, e.g. fibroblasts, osteoblasts, and subpopulations of endothelial cells. Within the immune system, APO-1 is broadly distributed among histiocytic cells but restricted to minor subpopulations of peripheral T and B cells. Immature T cells, i.e., thymocytes, do not express detectable APO-1-antigen. Expression of APO-1 was induced in phytohemagglutinin activated T cells and in a mammary carcinoma cell line by interferon-gamma alone and in combination with tumor necrosis factor alpha. Consistently, there was an in situ induction of APO-1 in several types of glandular epithelium in microtopographic association with lymphohistiocytic infiltrates. This inflammation-associated APO-1 induction went along with increased expression of this molecule within the lymphocytic compartment of the lesion. In tumors. APO-1 expression was heterogeneous. In comparison to their normal counterparts, some tumors showed abnormal hypo-expression or loss of APO-1. However, abnormal neo-expression was also found.

CONCLUSIONS

Tissue distribution, in vitro expression, and reaction upon cytokine-induced activation suggest that APO-1 might not only transmit apoptotic signals but might play a more general role in growth control.

Activation interferes with the APO-1 pathway in mature human T cells

One of the mechanisms to terminate a specific immune response may involve elimination of antigen activated T cells by programmed cell death, apoptosis. Apoptosis in activated T cells may be induced via the TCR-CD3 complex or/and cell surface molecules like the APO-1 (Fas) antigen, a new member of the nerve growth factor/tumor necrosis factor receptor superfamily. To investigate apoptosis in activated T cells we studied expression of APO-1 and sensitivity to APO-1 mediated apoptosis in human peripheral T lymphocytes. APO-1 is not expressed on cord blood and the majority of resting T cells, but on activated T cells. One day activated T cells in culture showed activation induced resistance to apoptosis (ARA). However, after prolonged in vitro culture, 6 day activated T cells acquired sensitivity to activation induced sensitivity to apoptosis (ASA). Restimulation of the ASA+ activated T cells by triggering TCR-CD3 or CD2 induced proliferation and apoptosis in a fraction of the cells. In the surviving fraction of ASA+ activated T cells, however, this treatment reinduced a transient ARA+ phenotype. Thus, activation of resting mature T cells or restimulation of activated T cells may induce a transient resistance to apoptotic signals. Activation signals may interfere with the APO-1 pathway and may prevent elimination of activated T cells in the periphery (peripheral selection).

APO-1-induced apoptosis of leukemia cells from patients with adult T-cell leukemia

The 48-Kd cell-surface protein APO-1 is a new member of the nerve growth factor (NGF)/tumor necrosis factor (TNF) receptor superfamily. APO-1 is expressed on various cells, including activated T and B cells and some lymphoid and nonlymphoid cell lines. Triggering of APO-1 by the monoclonal antibody anti-APO-1 induces programmed cell death (apoptosis) in APO-1-expressing cells. APO-1 is also present on T-cell lines derived from patients with adult T-cell leukemia (ATL). Therefore, we investigated APO-1 expression and APO-1-mediated induction of apoptosis ex vivo in cells from patients with ATL. Fresh leukemic cells from nine patients with ATL were assayed for APO-1 expression by two-color immunofluorescence. The leukemic cells from all patients strongly expressed APO-1. Incubation of ATL cells with anti-APO-1 in vitro inhibited spontaneous and cytokine-mediated DNA synthesis. Furthermore, DNA isolated from cells treated with anti-APO-1 exhibited polynucleosomal DNA fragmentation (DNA ladder) characteristic for apoptotic cell death. The analysis of APO-1-mediated apoptosis may represent a new approach to the study of growth control in lymphoid malignancies. In addition, induction of apoptosis by administration of anti-APO-1 may represent a new therapeutic approach for aggressive T-cell malignancies such as ATL.

[Intermediate results of a multicenter study in long-term management of HIV exposed and infected children]

AIM OF THE STUDY

In most cases (about 80%) the HIV-infection of children is acquired during pregnancy or birth. Therefore peculiarities for diagnostic procedures and in the natural course of the disease can be expected if compared to AIDS acquired at adult age. Further interesting questions are the frequency of the (vertical) transmission from the infected mother to the child, the influence of mother- and/or child-related factors as well as the method of delivery on this transmission frequency.

METHODS

In order to answer these questions the children of HIV-infected mothers since April 1, 1988 have been examined virologically, immunologically and clinically since birth by 6 study centers according to a standard protocol. The pregnancy data of the mothers have been documented, according to uniform criteria. Pre- or perinatally infected children not known to be antibody-positive at birth have been separately analysed.

RESULTS

The frequency of pre/perinatal transmission from the mother to the child is 15.3%, the elective caesarean section seems to lower the transmission rate. Mother and/or infant related cofactors of the transmission could not be defined. In comparison to the control groups of the intrauterine HIV-exposed but not infected children, at birth the HIV-infected children were clinically and immunologically not ill. During their first months some of the infected children had a significant increase of serum IgG. Oral candidiasis, chronic recurrent diarrhoea and bacterial infections are indicator symptoms of beginning HIV-disease. The Minimum-AIDS-Prevalence in the child's first year is 24%. The relatively late AIDS manifestation (3.7 years) and the average value of 28 months of survival in retrospective observed children were a remarkable result.

CONCLUSION

In general, the course of an HIV-infection in children of HIV-positive women in Germany seems to be comparable to results in other countries. The frequency of transmission from mother to child is relatively low (15.3%) and is an important result for maternity care. The caesarean section should be considered. The course of pregnancy must be examined more subtly in order to find out the still unclear causes of transmission. The Minimum-AIDS-Prevalence of 24% and the frequency of HIV-related signs and symptoms of 29% in the child's first year make it necessary to observe these children rather closely during this time.

Expression of the APO-1 antigen in Burkitt lymphoma cell lines correlates with a shift towards a lymphoblastoid phenotype

APO-1 is a cell surface molecule that induces apoptosis when ligated with the monoclonal antibody anti-APO-1. Expression of APO-1 and response to anti-APO-1 was investigated in a number of Epstein-Barr virus (EBV)-positive and -negative Burkitt lymphoma (BL) cell lines, in EBV-immortalized lymphoblastoid cell lines, and in cells from fresh BL biopsies. APO-1 was not expressed in EBV-negative cell lines and in EBV-positive BL cell lines with a phenotype corresponding to BL tumor biopsy cells (CD10+, CD21-, CD23-, CD30-, CD39-, CDw70-, CD77+). Accordingly, fresh BL cells obtained from three BL biopsies were APO-1 negative. EBV-positive BL cell lines that had acquired a lymphoblastoid phenotype (CD10-, CD21+, CD23+, CD30+, CD39+, CDw70+, CD77-) upon prolonged in vitro cultivation, as well as normal B-lymphoblastoid cell lines, expressed a high density of APO-1. APO-1 may, therefore, be regarded as a B-cell activation marker. APO-1 expression is not the only prerequisite for anti-APO-1-induced apoptosis because 6 of 7 APO-1-expressing EBV-positive BL cell lines were not sensitive to anti-APO-1, whereas all lymphoblastoid cell lines were killed by anti-APO-1. The sensitivity of lymphoblastoid cell lines to anti-APO-1-mediated apoptosis may open a new therapeutic approach for the treatment of EBV-induced lymphoproliferative lesions in immunocompromised individuals, because these are composed of cells with a lymphoblastoid phenotype.

Minimal requirements for the diagnosis, classification, and evaluation of the treatment of childhood acute lymphoblastic leukemia (ALL) in the "BFM Family" Cooperative Group

Minimal requirements and their rationale for the diagnosis and the response to treatment in childhood acute lymphoblastic leukemia (ALL) were defined in the recently instituted "BFM-Family"-Group, in which the German, Austrian, Dutch, Italian, Belgian, French and Hungarian childhood leukemia study groups cooperate. ALL is defined as > or = 25% lymphoblasts in the bone marrow; for confirmation of the diagnosis and classification the criteria of the French-American-British (FAB) criteria are retained. For determination of the extent of the disease at diagnosis or relapse the criteria by the Rome Workshop [1986] are recommended: An obligatory panel of monoclonal antibodies for immunophenotyping was defined, as well as criteria for precursor B-ALL and T-ALL. Cytogenetic studies may support the diagnosis and subtyping, and are essential to identify certain patients with a high risk of treatment failure (f.i. t(9;22), t(4;11)). The role of molecular genetics for the diagnosis and the characterization of leukemia and the value of its clinical application needs further elucidation. Relapse was defined as recurrence of evident leukemia in the blood, bone marrow (> or = 25% lymphoblasts) or at any other site (to be confirmed by histological examination). Bone marrow involvement combined with extramedullary relapse was defined as > or = 5% lymphoblasts in the bone marrow.

[Symptomatic HIV infection. Approaches to rational therapy]

The symptomatic HIV-infection is characterized by the involvement of multiple organs with a predominance of infectious complications. Treatment of these complications usually follows established therapeutic regimens for the resulting diseases of affected organs. Successful reconstitution of the immunodeficiency resulting from HIV-infection has been unsuccessful so far. Recently, promising results have been obtained by the treatment of patients with antiviral substances which inhibit viral replication. The prototype substance, Zidovudine (AZT) has been proven to be beneficial for patients in advanced stages of the disease and more recently has been shown to delay the occurrence of full-blown AIDS in patients with AIDS-related complex. In consequence, substances with less side effects such as Dideoxycytidine or Dideoxyinosine are currently under investigation.

Monoclonal-antibody-mediated apoptosis in adult T-cell leukaemia

The monoclonal antibody anti-APO-1 recognises a 52 kD cell membrane protein (APO-1) on some lymphoid tumour cell lines and on activated T cells. Binding of anti-APO-1 to cells expressing APO-1 results in programmed cell death, apoptosis, the most common form of death in eukaryotic cells. Expression of the antigen and sensitivity to the induction of cell death by anti-APO-1 were studied in human T-cell lines transformed by human leukaemia virus type 1 (HTLV-I) and in cultured cells from patients with adult T-cell leukaemia (ATL). APO-1 was strongly expressed on both types of cells and incubation of the cells with anti-APO-1 resulted in inhibition of proliferation and apoptosis. Induction of apoptosis may therefore be a possible therapeutic tool in HTLV-I-associated malignant disorders.

Autocrine growth factors secreted by the malignant human B-cell-line BJAB are distinct from other known cytokines

BJAB, a EBV-negative Burkitt-like lymphoma, did not grow under suboptimal culture conditions in low concentrations of serum unless appropriate cytokines were added. A subclone of BJAB, Clone 13, however, could be kept in long-term culture under such conditions without added cytokines. This suggested that growth of BJAB-Clone 13 was supported by autocrine growth factors (AGF). In fact, the supernatant of Clone 13 stimulated growth of the parental BJAB line and showed IL-1-like activity. Of several cytokines tested only AGF and IL-1 stimulated growth of BJAB. IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, GM-CSF, TNF-alpha, LT, IFN-gamma and TGF beta did not have this effect. The IL-1-like activity was completely neutralized by anti-IL-1 alpha antibodies. In contrast, AGF-activity was not affected by anti-IL-1 alpha. Rabbit antibodies produced against fractions enriched for AGF inhibited growth of BJAB. This inhibition was overcome by Clone 13-AGF, but not by IL-1 alpha. These data suggest that Clone 13-AGF is distinct from IL-1 alpha and might be a new cytokine.

[Multicenter study of long-term management of HIV exposed and HIV infected children. Study organization and aims]

In order ot improve and standardize medical and psychosocial care for infants born to HIV-positive mothers, a prospective multicenter study has been initiated at 6 university hospitals in West-Germany in 1987. Aims of the study are: 1) determination of the rate of vertical HIV-transmission, relevant maternal risk factors and the clinical symptomatology of pre-or perinatally acquired HIV-infections, 2) identification of immunological and virological laboratory markers important for the verification and staging of HIV-infections in infants, 3) elaboration of an interdisciplinary approach to the care of HIV-infected children and their families, integrating medical care, social and psychological attendance.

Lack of interleukin-2 (IL-2) dependent growth of TAC positive T-ALL/NHL cells is due to the expression of only low affinity receptors for IL-2

Binding of interleukin-2 (IL-2) to high affinity receptors on activated normal T cells was shown to be the essential step in induction of proliferation of such cells. The finding of abundant IL-2 receptors on malignant T cells in adult T cell leukemia suggested a deregulation of the IL-2/IL-2 receptor system and was assumed to account for aberrant growth in malignant disorders of T cells. In this study we use malignant T cells from nine patients with the clinical diagnosis of T-ALL or T-NHL and did not detect IL-2 dependent growth under conditions in which normal T cells responded to IL-2. IL-2 receptors comparable in numbers to activated T cells were found on T-ALL/T-NHL cells stimulated with PHA and PMA. However, binding studies using radiolabeled IL-2 indicated that the receptors present on malignant T cells were not able to bind to IL-2 with high affinity. Therefore, if IL-2 is involved in the proliferation of malignant T cells, its mechanism of growth regulation may be different from the one for normal T cells. Alternatively, IL-2 may not play a role in the regulation of growth of malignant T cells in vitro.

Monoclonal antibody-mediated tumor regression by induction of apoptosis

To characterize cell surface molecules involved in control of growth of malignant lymphocytes, monoclonal antibodies were raised against the human B lymphoblast cell line SKW6.4. One monoclonal antibody, anti-APO-1, reacted with a 52-kilodalton antigen (APO-1) on a set of activated human lymphocytes, on malignant human lymphocyte lines, and on some patient-derived leukemic cells. Nanogram quantities of anti-APO-1 completely blocked proliferation of cells bearing APO-1 in vitro in a manner characteristic of a process called programmed cell death or apoptosis. Cell death was preceded by changes in cell morphology and fragmentation of DNA. This process was distinct from antibody- and complement-dependent cell lysis and was mediated by the antibody alone. A single intravenous injection of anti-APO-1 into nu/nu mice carrying a xenotransplant of a human B cell tumor induced regression of this tumor within a few days. Histological thin sections of the regressing tumor showed that anti-APO-1 was able to induce apoptosis in vivo. Thus, induction of apoptosis as a consequence of a signal mediated through cell surface molecules like APO-1 may be a useful therapeutic approach in treatment of malignancy.

[Local recurrence of mediastinal non-Hodgkin lymphomas in childhood]

Local relapses in mediastinal non-Hodgkin's lymphomas in childhood are rare. Of the 15 children treated in our hospital during the last 11 years, 2 had early local or pleural relapses. At presentation both had extensive pleural effusions, and after completion of induction therapy still some fluid in the pleural space was present. A local irradiation was not performed initially. Be review of the literature for risks and benefit of local irradiation in mediastinal non-Hodgkin's lymphoma, local irradiation in children receiving aggressive chemotherapy should only be considered in few patients; preferentially in those, with incomplete disappearance of the mediastinal tumor or, as we may conclude from our patients, incomplete disappearance of pleural effusion.

Activation of macrophages by lymphokines from T-cell clones: evidence for different macrophage-activating factors

The data reported in this paper demonstrate that macrophage-activating factors (MAFs) are a heterogeneous group of T-cell-derived lymphokines. Two long-term T-cell clones, Cl 96 and PK 7.1.2E8, were potent sources of MAFs (MAF96 and MAF7.1.2E8). These MAFs could be distinguished by differential activation of macrophages. Activation of resident murine macrophages with MAF7.1.2E8 enhanced RNA and glycoprotein synthesis, hexosemonophosphate shunt (HMPS) activity, release of oxygen metabolites (O-2 and H2O2), pinocytosis and tumor cytostasis, whereas no effect on schistosomula killing and tumor cytolysis could be observed. In contrast, MAF96 enhanced glycoprotein synthesis, HMPS activity, release of oxygen metabolites and prostaglandin E, schistosomula killing, and tumor cytostasis and cytolysis, while RNA synthesis and pinocytosis were decreased. These findings show that MAFs from both T-cell clones share some properties but markedly differ in others. In addition, the macrophage-activating properties of MAF96 but not of MAF7.1.2E8 could selectively be inhibited by a rabbit anti-lymphokine antiserum. This demonstrates a serological difference between MAF activities from both clones. Although at optimal concns both MAFs were active in the absence of lipopolysaccharide (LPS), the activity of suboptimal doses of MAF96 but not of MAF7.1.2E8 could be enhanced by LPS. These findings show that different MAFs from T-cell clones may be useful to clarify molecular mechanisms of macrophage activation.

Macrophage-activating factors from different T cell clones induce distinct macrophage functions

The data reported in this paper are the first demonstration that different T cell clones (PC-AKR-CI 96, clone 96; PK 7.1.2 E8, clone 7.1.2 E8) secrete different macrophage-activating factors (MAF) that induce distinct macrophage activities. Incubation of resident murine macrophages with MAF 7.1.2 E8 increased RNA, protein, and glycoprotein synthesis, hexosemonophosphate shunt (HMPS) activity, release of oxygen metabolites (O-2, H2O2), pinocytosis, phagocytosis, and tumor cytostasis, whereas no effect on prostaglandin E (PGE) release, schistosomula killing, and tumor cytolysis could be observed. In contrast, MAF 96 increased glycoprotein synthesis, HMPS activity, release of oxygen metabolites and PGE, schistosomula killing, and tumor cytostasis and cytolysis, whereas RNA and protein synthesis and pinocytosis were decreased and phagocytosis remained unaffected. Thus, MAF from both T cell clones share some macrophage-activating properties but differ in others. Most importantly, both MAF could be differentiated serologically by a rabbit anti-lymphokine antiserum that selectively inhibited MAF 96 but not MAF 7.1.2 E8 activity. At optimal concentrations, MAF 96 and 7.1.2 E8 were active in the absence of lipopolysaccharide (LPS) whereas LPS enhanced the activity of suboptimal doses of MAF 96 but not of MAF 7.1.2 E8. These data are discussed with respect to the possibility that the functional dichotomy of T cell clones might reflect different activities of normal T cell subpopulations.

Macrophage-activating factors from different T cell clones induce distinct macrophage functions

The data reported in this paper are the first demonstration that different T cell clones (PC-AKR-CI 96, clone 96; PK 7.1.2 E8, clone 7.1.2 E8) secrete different macrophage-activating factors (MAF) that induce distinct macrophage activities. Incubation of resident murine macrophages with MAF 7.1.2 E8 increased RNA, protein, and glycoprotein synthesis, hexosemonophosphate shunt (HMPS) activity, release of oxygen metabolites (O-2, H2O2), pinocytosis, phagocytosis, and tumor cytostasis, whereas no effect on prostaglandin E (PGE) release, schistosomula killing, and tumor cytolysis could be observed. In contrast, MAF 96 increased glycoprotein synthesis, HMPS activity, release of oxygen metabolites and PGE, schistosomula killing, and tumor cytostasis and cytolysis, whereas RNA and protein synthesis and pinocytosis were decreased and phagocytosis remained unaffected. Thus, MAF from both T cell clones share some macrophage-activating properties but differ in others. Most importantly, both MAF could be differentiated serologically by a rabbit anti-lymphokine antiserum that selectively inhibited MAF 96 but not MAF 7.1.2 E8 activity. At optimal concentrations, MAF 96 and 7.1.2 E8 were active in the absence of lipopolysaccharide (LPS) whereas LPS enhanced the activity of suboptimal doses of MAF 96 but not of MAF 7.1.2 E8. These data are discussed with respect to the possibility that the functional dichotomy of T cell clones might reflect different activities of normal T cell subpopulations.