GFI1 as a novel prognostic and therapeutic factor for AML/MDS.

Genetic and epigenetic aberrations contribute to the initiation and progression of acute myeloid leukemia (AML). GFI1, a zinc-finger transcriptional repressor, exerts its function by recruiting histone deacetylases to target genes. We present data that low expression of GFI1 is associated with an inferior prognosis of AML patients. To elucidate the mechanism behind this, we generated a humanized mouse strain with reduced GFI1 expression (GFI1-KD). Here we show that AML development induced by onco-fusion proteins such as MLL-AF9 or NUP98-HOXD13 is accelerated in mice with low human GFI1 expression. Leukemic cells from animals that express low levels of GFI1 show increased H3K9 acetylation compared to leukemic cells from mice with normal human GFI1 expression, resulting in the upregulation of genes involved in leukemogenesis. We investigated a new epigenetic therapy approach for this subgroup of AML patients. We could show that AML blasts from GFI1-KD mice and from AML patients with low GFI1 levels were more sensitive to treatment with histone acetyltransferase inhibitors than cells with normal GFI1 expression levels. We suggest therefore that GFI1 has a dose-dependent role in AML progression and development. GFI1 levels are involved in epigenetic regulation, which could open new therapeutic approaches for AML patients.

Comprehensive polymorphism analysis of ABO using allele-specific separation by bead technology and subsequent sequencing.

The ABO system is the most important blood group system in transfusion and transplantation medicine. Over the last decade ABO genotyping was introduced to complement serological analysis but because of the enormous diversity at the ABO locus, DNA-based typing is highly sophisticated. In some genotype combinations and especially when dealing with hybrid genes, cis/trans linkage analysis is necessary. Therefore, we developed a simple and reliable approach to isolate ABO haplotypes. The described method physically separates diploid genomic DNA using probe-dependent allele hybridization followed by magnetic bead separation. After this haplotype-specific extraction (HSE), sequencing based typing (SBT) was performed to analyze coding and non-coding respectively regulatory regions of ABO.

HCF164 encodes a thioredoxin-like protein involved in the biogenesis of the cytochrome b(6)f complex in Arabidopsis.

To understand the biogenesis of the plastid cytochrome b(6)f complex and to identify the underlying auxiliary factors, we have characterized the nuclear mutant hcf164 of Arabidopsis and isolated the affected gene. The mutant shows a high chlorophyll fluorescence phenotype and is severely deficient in the accumulation of the cytochrome b(6)f complex subunits. In vivo protein labeling experiments indicated that the mutation acts post-translationally by interfering with the assembly of the complex. Because of its T-DNA tag, the corresponding gene was cloned and its identity confirmed by complementation of homozygous mutant plants. HCF164 encodes a thioredoxin-like protein that possesses disulfide reductase activity. The protein was found in the chloroplast, where it is anchored to the thylakoid membrane at its lumenal side. HCF164 is closely related to the thioredoxin-like protein TxlA of Synechocystis sp PCC6803, most probably reflecting its evolutionary origin. The protein also shows a limited similarity to the eubacterial CcsX and CcmG proteins, which are required for the maturation of periplasmic c-type cytochromes. The putative roles of HCF164 for the assembly of the cytochrome b(6)f complex are discussed.

Simultaneous measurement of cellular P-glycoprotein content and function by multiparametric flow-cytometry.

OBJECTIVE: A multiparametric approach was applied to simultaneously determine expression and function of the drug efflux pump P-glycoprotein (PGP) in multidrug-resistant (MDR) human leukemic lymphoblast cell lines and isolated leukemic blasts using flow-cytometry in a patient with acute myeloid leukemia (AML). METHODS: The antigen was measured by staining PGP using the monoclonal antibody 4e3 which does not inhibit the function of PGP. The 4e3 antibody binds to an external epitope of PGP and can therefore be used for staining living cells. Drug transport, mediated by PGP, was determined simultaneously by measuring rhodamine 123 (rho123) efflux. The MDR cell lines, CEM/VLB10-2 and CEM/VBL100 are 10-fold and 270-fold resistant to vinblastine (VBL), respectively, compared to the human PGP-negative parent cell line CEM/WT and they express different amounts of PGP. Initially, living cells were stained using the 4e3 antibody and a secondary antibody labeled with 7-amino-4-methylcoumarin-3-acetic acid (AMCA). Cells were then incubated for 60 min with rho123 (10 microM) and analyzed for rhodamine and AMCA-derived fluorescence. The decrease in rho123 fluorescence was determined after a further period of 30 min. RESULTS: CEM/VLB100 cells expressed larger amounts of PGP, and rho123 fluorescence after 30 min was 85% lower than the parent cell line. PGP expression and rho123 efflux were also detected in CEM/VLB10-2 cells which display a low degree of resistance, thus reflecting the high sensitivity of this method. PGP-expressing blasts and moderate rho123 efflux were also observed in a specimen derived from a patient with clinically resistant acute myeloid leukemia (AML). CONCLUSION: A multiparametric approach using flow-cytometry allows the reliable and sensitive measurement of both PGP expression and function simultaneously in single cells.

Determinants of carcinogen-induced malignant conversion in rat brain: proliferative properties of neural precursor cell subpopulations analyzed by multi-parameter flow cytometry.

The induction of neuroectodermal tumors in BDIX rats by N-ethyl-N-nitrosourea (EtNU) is a model system for the analysis of transformation risk as a function of target cell properties. The yield of neural tumors induced by EtNU varies with the developmental window chosen for the carcinogen pulse; i.e. with the relative proportions of different neural precursor cells exposed to EtNU at distinct developmental stages. Different subsets of fetal brain cells have been characterized previously with respect to their relative risk of malignant transformation using monoclonal antibodies. As DNA replication of target cells is considered to be a prerequisite for malignant conversion, we analyzed the cell cycle distributions, using flow-cytometry, of 4 subsets of neural precursor cells considered to be at high or low risk, respectively, of malignant conversion by EtNU in vivo. Cell populations associated with an elevated risk of transformation exhibited higher proportions of cells in S-phase. One of the 2 putative low-risk populations exhibited a significantly lower fraction of S-phase cells, while the value of the second one exceeded those obtained for the 2 high-risk subpopulations. Therefore, a higher than average fraction of cells in S-phase appears to be positively correlated with the cellular risk of malignant transformation by EtNU, but does not represent a dominant risk determinant per se.

Frequency of clonal B lymphocytes in chronic myelogenous leukemia evaluated by fluorescence in situ hybridization.

The demonstration of the Philadelphia (Ph) chromosome in B lymphocytes from patients with chronic myelogenous leukemia (CML) has provided evidence that the disorder originates in a pluripotent progenitor cell. Divergent results, however, exist as to the degree of contribution of clonally derived cells to the B-cell compartment. To address this issue, B lymphocytes were selected from the blood of seven patients in the chronic phase of Ph-positive CML and were examined with dual-color fluoresence in situ hybridization for the presence of the Ph translocation. The purity of the B-cell preparations ranged from 88% to 97% (mean 93%). The Ph translocation was detected in 22-34% (mean, 27%) of the sorted B cells. There was no evidence that the duration of the disease affects the ratio of Ph-positive and -negative B cells. In summary, clonally derived circulating B lymphocytes were present in all patients studied but made only minor contribution to this compartment.

Enrichment of residual tumor cells in bone marrow or peripheral blood cells in a neuroectodermal tumor model.

To enrich a small population of malignant cells contaminating human bone marrow or peripheral blood, a method of immunomagnetic depletion of the normal nucleated cells was developed. In a model using neuroectodermal cell lines, contaminations between 0.1% and 1.4% of tumor cells could be increased by a factor of 7.5 (median; range, 3.8 to 18.0; n = 10). Because incubation with antibodies is restricted to the magnetic beads in this method, the cell population after removal of the beads has not been incubated with antibodies. The risk of unspecific staining of the tumor cells during the enrichment procedure is minimal. This simple method is therefore well suited to the study of the characteristics of a minimal residual disease population.

Surface antigens of cell subpopulations in prenatal rat brain are expressed in a characteristic non-random pattern on their ethylnitrosourea-induced malignant counterparts.

Selective induction of neural tumors in the rat by single-dose exposure of the immature nervous system to ethylnitrosourea (EtNU) is a model for the study of cell lineage-, differentiation stage-, and carcinogen-dependent mechanisms in neuro-oncogenesis. Overall yields and relative frequencies of different types of neural tumors vary with the developmental window chosen for the EtNU-pulse. Precursor cells belonging to different neural lineages and targeted by the carcinogen at distinct developmental stages may thus bear a differential risk of malignant conversion. To specify subpopulations of neural precursors in fetal (prenatal day 18) BDIX-rat brain, four monoclonal antibodies (mAbs) recognizing cell surface differentiation antigens were used: mAb RB13-2 directed against O-acetylated gangliosides and binding to approximately 36% of fetal brain cells (FBC); mAb RB13-6 recognizing a 130 kDa glycoprotein (expressed by approximately 8% of FBC); and mAbs RB21-7 and RB21-15 which bind, respectively, to embryonal neural cell adhesion molecules (N-CAM) and a 24 kDa protein (expressed by approximately 55% and 12% of FBC). Antigen expression profiles were compared with those of 14 primary brain tumors and 16 malignant neural cell lines, all of which had been induced by EtNU on prenatal day 18 in vivo. Monoclonal antibodies RB13-2 and RB21-7 did not bind to any of the tumors or cell lines. In contrast, mAbs RB13-6 and RB21-15 both reacted with 14/14 tumors, and with 16/16 and 10/16 cell lines, respectively. Expression of the latter antigens might thus specify lineage-specific stages of FBC development/differentiation particularly susceptible to EtNU-induced malignant transformation. Two-color fluorescence analyses revealed three subsets of FBC binding mAb RB13-6 (RB13-2+/RB13-6+/RB21-15-; RB13-2-/RB13-6+/RB21-15-; and RB13-2-/RB13-6+/RB21-15+), representing successive stages of differentiation.

[Myogenic differentiation in experimental malignant schwannomas (an immunohistochemical and molecular genetic study of the "triton" tumor)]. / Miogennaia differentsirovka v éksperimental'nykh zlokachestvennykh shvannomakh (immunogistokhimicheskoe i molekuliarno-geneticheskoe issledovanie opukholi "triton".

Spontaneous myogenic differentiation was observed in 2 out of 15 monolayer cultures from schwannomas induced in BD1X rats by transplacental exposure to N-enthyl-N-nitrosourea (ENU). Cells were sorted following fluorescence-activating method with monoclonal antibody (Mab) 217c. Myotubes and numerous mononucleated cells no longer expressed the Schwann cell antigens 217c and S-100 protein, but rather revealed the presence of desmin, the alpha-sarcomeric form (alpha-sr) of actin, and the cell surface antigen specified by Mab RB21-7, a glycoprotein sharing an epitope with the neural cell adhesion molecule (N-CAM). Subcutaneous reimplantation of such cells into syngeneic animals resulted in the appearance of tumours composed of both S-100 positive Schwann cells and desmin and alpha-sr-actin positive rhabdomyoblasts, thus closely resembling the human "triton" tumour. With the use of the polymerase chain reaction and allele-specific oligonucleotide hybridization, DNA isolated from individual myotubes was analysed for the presence of a T-->A transversion mutation at nucleotide 2007 of the neu gene, which is diagnostic of ENU-induced rat schwannomas. All of the amplified DNA isolated contained the mutant neu allele, thus providing direct genetic proof for the capacity of mammalian neuroectodermal cells for myogenic differentiation.

Effect of H-2 compatibility in autoimmune destruction of islet allografts from B10 congenic lines to nonobese diabetic mice.

Autoimmune diabetes involves multiple antigens, and both cellular and humoral immune responses. Using CBA (H-2k) C57BL/6 (H-2b), and BALB/c (H-2d) newborn mouse pancreata, we previously demonstrated that acute and strong destruction of islet allografts by anti-islet autoimmunity in the nonobese diabetic (NOD) mouse H-2Kd, Db) is under the influence of major histocompatibility complex (MHC) antigens. In the current study, we have attempted to confirm these results in the absence of minor alloantigenic differences using B10 congenic strains as pancreatic donors. Pancreata from B10.BR (H-2k), C57BL/10SnJ (H-2b), and B10.D2 (H-2d) were transplanted under the kidney capsule of NOD mice within 1 month of diabetes onset. These recipients were immunosuppressed with cyclosporine (CsA) in a dosage that effectively prevents rejection of skin allograft, but not islet isograft destruction that is mediated by anti-islet autoimmunity. On day 10, the grafts were harvested and examined histologically to assess viability. Pancreatic allografts from B10.D2, sharing the H-2Kd with the NOD mouse, showed the strongest lymphocytic infiltration, and neither islets nor beta cells were found in all seven grafts. C57BL/10SnJ grafts, sharing the same H-2Db, also showed severe lymphocytic infiltration, and no intact islets, and only a few beta cells were found, as single cells, in three of eight grafts. In contrast, B10.BR grafts, completely incompatible at the H-2, showed the least infiltration, and normal islets containing many beta cells were found in 10 of 11 grafts. These results again suggested the hypothesis that islet allograft destruction by diabetic NOD mice is MHC restricted.(ABSTRACT TRUNCATED AT 250 WORDS)

Improved flow-cytometric detection of low P-glycoprotein expression in leukaemic blasts by histogram subtraction analysis.

Expression of the drug efflux pump P-glycoprotein (PGP) was determined by flow cytometry in human lung cancer cell lines and in leukaemic blasts derived from 60 patients with acute myeloid leukaemia (AML). Cells from the PGP-negative parent cell line H69/P and the multidrug resistant (MDR)-variant H69/LX4 could be clearly distinguished by immunostaining with the anti-PGP monoclonal antibody MRK16. In leukaemic blasts, the differences in fluorescence intensities between samples incubated with the idiotypic nonspecific (control sample) and specific antibody (test sample) were small, resulting in nondisjunct distributions. Only in a few leukaemia specimens were PGP-expressing cells detectable by simple subtraction of histograms using a threshold. Therefore, an improved histogram subtraction analysis, based on curve fitting and a statistical test, was applied to distinguish antigen-positive from antigen-negative cells. Moreover, a multiparametric staining procedure employing propidium iodide (PI) and Hoechst 33342 was used to reduce staining artefacts. By this approach, leukaemic cells with low expression of PGP were detected in 39 out of 60 cases. Subpopulations with strong PGP expression, resulting in disjunct fluorescence distributions, were not observed. Only in 5 out of 60 specimens were PGP expressing cells detected by a conventional subtraction of histograms using a threshold. Comparison of data obtained with or without the multiparametric gating procedure indicated that the increase in sensitivity was mainly due to the application of the data analysis. However, exclusion of cell debris using PI and Hoechst staining properties reduced the deviation of data from mean values. No relation between PGP expression and cell cycle position was observed in either cell lines or in leukaemic blasts.(ABSTRACT TRUNCATED AT 250 WORDS)

A standardized protocol for flow cytometric analysis of cells isolated from cerebrospinal fluid.

Flow cytometry (FC) is an useful tool for the analysis of subpopulations in complex cell suspensions. When applying this method to the cerebrospinal fluid (CSF), some characteristic properties of this cell type must be taken into consideration: there are only few cells which decay rapidly in their native medium and during centrifugation. One aim of the immunostaining procedure preceding flow cytometric analysis must be to minimize cell loss in order to get an undistorted picture of 'true' CSF cell populations. Consequently, morphological flow cytometric plots of high resolution are an indispensable precondition for reliable determination of subpopulations defined by monoclonal antibody (Mab) binding. We describe a standardized protocol for the flow cytometric examination of CSF cells which minimizes undesired cell loss. By the use of a 'quality control' the extent of cell loss could be monitored. Examples of morphological flow cytometric plots are given. The subsequent determination of Mab binding subpopulations is critical when fluorescence intensities of antigen positive and negative cells are non-disjunct. A statistical test was developed for these cases often seen when cell surface determinants are expressed at low levels only.

Cytotoxicity of adriamycin, idarubicin, and vincristine in acute myeloid leukemia: chemosensitization by verapamil in relation to P-glycoprotein expression.

A 4-day colorimetric tetrazolium dye (MTT) assay was used to assess the cytotoxicity of adriamycin (ADM), vincristine (VCR), and idarubicin (IDA) in blasts isolated from 37 patients with newly diagnosed and pretreated acute myeloid leukemia (AML). The effect of verapamil (VRP) as a chemosensitizer was studied in relation to the expression of the membrane efflux pump P-glycoprotein (PGP) as determined by a semiquantitative flow-cytometric proceder. A slight positive correlation was found between the fraction of cells expressing PGP and the ID50 values for ADM and VCR, but not between cellular PGP content and sensitivity to IDA. The overall data showed no significant sensitization effect of VRP. However, in specimens with more than 10% cells expressing PGP, 2 microM VRP sensitized cells to ADM and VCR significantly. The median of sensitization ratios (SRs), i.e., the ratios of cytotoxic drug ID50 in the absence/presence of VRP, were 1.89 and 2.0, respectively. No sensitizing effect of VRP on the cytotoxicity of IDA was observed. Related to the clinical status, the median fraction of PGP-positive blasts was elevated fourfold in pretreated patients (n = 16) in comparison to patients with de novo AML (n = 19). No differences in ID50 values were observed between newly diagnosed and pretreated patients. However, SRs for ADM and VCR were higher in samples of pretreated patients compared with de novo AML. PGP-mediated cellular drug resistance may thus be circumvented in leukemic blasts by application of chemosensitizers or, potentially, alternative anthracyclines.

Determination of immunoreactive fraction and kinetic parameters of a radiolabeled monoclonal antibody in the absence of antigen excess.

Monoclonal antibodies (Mabs) directed against cell surface determinants and conjugated to fluochromes, radionuclids or drugs are of increasing importance in cell and tumor biology as well as in clinical oncology. Many of the applications of Mab require precise and quantitative information regarding the molecular interactions of labeled antibody with the respective antigen expressed on the cell surface. These interactions are characterized by the association rate constant (ka), the dissociation rate constant (kd) and the antibody affinity constant (K). The immunoreactive fraction (IRF) of the labeled antibody molecules directly influences these parameters. IRF is usually reduced below 100% by antibody purification and labeling procedures and, in case of radiolabeled antibodies, by radiation damage during antibody storage. Besides the calculation of kinetic parameters, IRF should, therefore, be determined for the quality control of any antibody preparation before experimental or clinical application. Commonly used methods for measuring IRF are based on radioimmunoassays (RIA) on intact cells performed under antigen excess. However, especially with Mabs directed against cell surface antigens expressed in small numbers per cell and for displaying low affinity constants, these assays often give unsatisfactory results. We have, therefore, established a method which permits us to determine IRF, ka, kd and K for an 125I-labeled Mab with precision even in the absence of antigen excess.

A simple method for counting small numbers of cells by flow cytometry.

We describe a flow cytometric method for the quantitation of small numbers of cells that facilitates proliferation studies in microcultures. A constant number of fluorescent latex microspheres/sample is added to single cell suspensions prepared from the cultures. By flow cytometric analysis, cells are easily distinguishable from the microspheres and can be quantitated on the basis of their light scattering and fluorescence properties is contour plots. As the number of microspheres/sample is known, the relative proportions of cells and microspheres, respectively, can be converted into absolute cell numbers. This quick method is useful for any type of studies where cell numbers less than 1 x 10(5) in a small volume are to be determined.

Rat model of the human "Triton" tumor: direct genetic evidence for the myogenic differentiation capacity of schwannoma cells using the mutant neu gene as a cell lineage marker.

Spontaneous myogenic differentiation was observed in 2 out of 15 cases when cells from schwannomas induced in the offspring of BDIX rats by transplacental exposure to N-ethyl-N-nitrosourea (EtNU) were grown in monolayer culture following fluorescence-activated cell sorting with monoclonal antibody (Mab) 217c. Myotubes and numerous mononucleated cells no longer expressed the Schwann cell antigens 217c and S-100 protein, but rather revealed the presence of desmin, the alpha-sarcomeric form (alpha-sr) of actin, and the cell surface antigen specified by Mab RB21-7, a 250 kD glycoprotein sharing an epitope with the neural cell adhesion molecule (N-CAM). Subcutaneous reimplantation of such cells into syngeneic animals led to the appearance of tumors composed of both S-100 positive Schwann cells and desmin and alpha-sr-actin positive rhabdomyoblasts, thus closely resembling the human "Triton" tumor. With the use of the polymerase chain reaction and allele-specific oligonucleotide hybridization, DNA isolated from individual myotubes was analyzed for the presence of a T----A transversion mutation at nucleotide 2012 of the neu gene, which is diagnostic of EtNU-induced rat schwannomas. All of the amplified DNA isolates contained the mutant neu allele, thus providing direct genetic proof for the capacity of mammalian neuroectodermal cells for myogenic differentiation.

Calibration of fluorescence intensities to quantify antibody binding surface determinants of cell subpopulations by flow cytometry.

Quantitative indirect immunofluorescence analysis by flow cytometry was used to determine the mean number of antibody binding sites per cell in a small subpopulation of rat brain cells expressing low levels of a cell surface differentiation antigen recognized by monoclonal antibody (Mab) RB13-6 (Kindler-Röhrborn et al.: Differentiation 30:53-60, 1985). For these non-disjunct distributions of fluorescence intensities, the cut-off border between antigen-positive and antigen-negative cells was defined by a statistical test. To eliminate the influence of accidental disturbances leading to incorrect statistical decisions, the curves for antigen-negative cells were fitted according to cell number and shape. The flow cytometer was calibrated with the use of a clonal cell line for which the average number of Mab RB13-6 binding sites per cell had previously been determined by radioimmunoassay and Scatchard-plot analysis. Using this analytical procedure, both the proportion of Mab binding brain cells and the mean number of Mab binding sites per Mab binding cell could be determined as a function of developmental stage.

Differentiation potential of a monoclonal antibody-defined neural progenitor cell population isolated from prenatal rat brain by fluorescence-activated cell sorting.

We have studied the differentiation potential in vitro of a subpopulation of neural progenitor cells from BDIX-rat brain. These cells transiently express a cell surface determinant (CSD) specified by monoclonal antibody (Mab) RB13-2 (Kindler-Röhrborn, A. et al., Differentiation 30 (1985) 53-60), and recently identified as a set of O-acetylated gangliosides (Reinhardt-Maelicke, S. et al., submitted) also recognized by Mab D1.1 (Levine, J.M. et al., J. Neurosci., 4 (1984) 826-831) and partly by Mab JONES (Schlosshauer, B. et al., J. Neurosci., 8 (1988) 580-592), respectively. As analyzed by immunofluorescence, Mab RB13-2 binding brain cells (prenatal days 11-22; postnatal days 7 and 89) were localized in different areas of the proliferative ventricular layer of the prenatal cerebrum and in the external granular layer of the early postnatal cerebellum. No Mab RB13-2 positive brain cells were found in adult brain. Following their isolation by fluorescence activated cell sorting on prenatal day 18, the differentiation potential of Mab RB13-2 binding brain cells was studied by double-immunofluorescence analysis under different conditions of monolayer culture. In the presence of 10% fetal calf serum (FCS), these cells differentiated into glial fibrillary acidic protein (GFAP) positive flat astrocytes, whereas neurons (neurofilament-positive) and a smaller number of stellate astrocytes (GFAP-positive) developed in a chemically defined medium containing 0.5% FCS. Neural progenitor cells binding Mab RB13-2 may thus either retain more than one option for differentiation into specific cell types, or the expression of the CSD specified by Mab RB13-2 may be common to more than one subset of neural progenitor cells (with or without predetermined unidirectional differentiation pathways) whose survival and/or proliferative behavior could be differentially affected by microenvironmental conditions.

The effect of H-2 compatibility on pancreatic beta cell survival in the nonobese diabetic mouse.

The effect of major histocompatibility (MHC) antigens on the survival of newborn pancreatic iso and allografts was assessed in the nonobese diabetic (NOD) mouse, which develops a spontaneous autoimmune diabetes. The NOD mouse (H-2Kd,Db) rejects skin allografts from CBA (H-2k) within a 12-day period, indicating normal immune function toward alloantigens. A pancreatic allograft into the NOD mouse represents a presumed first-set allogeneic response, as well as a possible second-set immune response to islets. To assess the effect of donor H-2 antigen and the influence of autoimmune disease on pancreatic graft survival, newborn pancreata from various strains of mice were transplanted into diabetic NOD mice treated with 40 mg/kg/day cyclosporine (CsA) that prevented skin allograft rejection. The grafts were then harvested at day 10 to histologically assess the graft viability. CBA pancreatic grafts, incompatible at all MHC loci, showed the least lymphocytic infiltration, and good donor beta cell survival. Furthermore, CBA newborn pancreata under appropriate conditions were able to cure or improve the diabetic condition in 3/6 NOD mice. In the graft sharing class I MHC antigens, lymphocytic infiltration was significantly increased, while the donor beta cell number clearly decreased. The intensity of the graft destruction was intermediate in C57BL/6 allografts sharing H-2Db antigen, and strongest in BALB/c allografts sharing H-2Kd and in NOD isografts. The results indicate that in diabetic NOD mice the CsA dose controlling allograft rejection is incapable of controlling antiislet immunity. This antiislet immunity appears to exert its effect in an H-2-restricted manner. These findings may have important implications for the transplantation of pancreatic tissue in treating type I diabetes in humans.