Investigation of the peak action wavelength of light-activated gene transduction.

Light-activated gene transduction (LAGT) is an approach to localize gene therapy via preactivation of cells with UV light, which facilitates transduction by recombinant adeno-associated virus vectors. Previous studies demonstrated that UVC induces LAGT secondary to pyrimidine dimer formation, whereas UVA induces LAGT secondary to reactive-oxygen species (ROS) generation. However, the empirical UVB boundary of these UV effects is unknown. Thus, we aimed to define the action spectra for UV-induced LAGT independent of DNA damage and determine an optimal wavelength to maximize safety and efficacy. UV at 288, 311 and 320 nm produced significant dose-dependent LAGT effects, of which the maximum (800-fold) was observed with 4 kJ m⁻² at 311 nm. Consistent with its robust cytotoxicity, 288 nm produced significantly high levels of DNA damage at all doses tested, whereas 311, 320 and 330 nm did not generate pyrimidine dimers and produced low levels of DNA damage detected by comet assay. Although 288 nm failed to induce ROS, the other wavelengths were effective, with the maximum (10-fold) effect observed with 30 kJ m⁻² at 311 nm. An in vivo pilot study assessing 311 nm-induced LAGT of rabbit articular chondrocytes demonstrated a significant 6.6-fold (P<0.05) increase in transduction with insignificant cytotoxicity. In conclusion, 311 nm was found to be the optimal wavelength for LAGT on the basis of its superior efficacy at the peak dose and its broad safety range that is remarkably wider than the other UV wavelengths tested.

Development of adult bone marrow stem cells in H-2-compatible and -incompatible mouse fetuses.

Bone marrow of normal adult mice was found, after transplacental inoculation, to contain cells still able to seed the livers of early fetuses. The recipients' own hematopoietic stem cells, with a W-mutant defect, were at a selective disadvantage. Progression of donor strain cells to the bone marrow, long-term self-renewal, and differentiation into myeloid and lymphoid derivatives was consistent with the engraftment of totipotent hematopoietic stem cells (THSC) comparable to precursors previously identified (4) in normal fetal liver. More limited stem cells, specific for the myeloid or lymphoid cell lineages, were not detected in adult bone marrow. The bone marrow THSC, however, had a generally lower capacity for self-renewal than did fetal liver THSC. They had also embarked upon irreversible changes in gene expression, including partial histocompatibility restriction. While completely allogeneic fetal liver THSC were readily accepted by fetuses, H-2 incompatibility only occasionally resulted in engraftment of adult bone marrow cells and, in these cases, was often associated with sudden death at 3-5 mo. On the other hand, H-2 compatibility, even with histocompatibility differences at other loci, was sufficient to ensure long-term success as often as with fetal liver THSC.

Gene control of hematopoiesis. I. Erythrocyte mosaicism and permanent immunological tolerance in allophenic mice.

Erythropoietic cells of two unrelated strains, C3H (or C3Hf) and C57BL/6, can coexist throughout hematopoiesis in allophenic mice experimentally produced from aggregated, undifferentiated blastomeres of separate genotypes. The presence of two red cell genotypes in these circumstances signifies that the erythroid population must normally be multiclonal, i.e., derived mitotically from at least two genetically determined cells. The two strains were detected by hemagglutination and absorption tests of erythrocytes for the specific histocompatibility antigens dictated by the H-2(k) and H-2(b) alleles. Of 34 C3H(f) <--> C57BL/6 allophenics tested, 16 had both red cell types; the remaining 18 showed only C3H or C57 red cells and included 12 mice with both cell strains present in some other tissues. All animals with evidence of two H-2 phenotypes among circulating erythrocytes were permanently immunologically tolerant of both antigenic types and remained free of runt disease. They lived a full lifespan, up to 2 yr 7(1/2) months of age. The data suggest a possible specific selective advantage of C57BL/6 over C3H erythropoietic tissue. There is considerable individual variability, not only in proportions of antigenically distinct erythrocytes, but also in strain composition of other tissues in the same animals. A broad spectrum of distinctive situations is found, in which parameters are varied within or outside of the circulatory system. Allophenic mice can therefore serve as investigative tools for entirely new kinds of experimental studies of gene control mechanisms and blood physiology in normal hematopoiesis and in a number of hereditary blood diseases.

"Intrinsic" immunological tolerance in allophenic mice.

Mice experimentally derived from pairs of conjoined, undifferentiated, cleavage-stage embryos of different histocompatibility genotypes can retain cells of each strain, which still produce their characteristic antigenic products. The animals are permanently tolerant of cells of both original types, remain free of runt disease, and display a normal and specific immune response to introduction of a foreign antigen. Absence of autoimmunity in development of ordinary animals is explainable by the "intrinsic" kind of tolerance found here.

Human beta-globin gene sequences injected into mouse eggs, retained in adults, and transmitted to progeny.

Foreign gene sequences were retained in two adult mice (out of 62 analyzed) from fertilized eggs injected with a recombinant plasmid containing the human beta-globin genomic region and the herpes simplex viral thymidine kinase gene. The intact human and viral genes were found in DNA of one of the animals and, in the other, at least part of the human globin gene was present. The latter individual transmitted these sequences to its progeny in a Mendelian ration. Thus, human DNA may be incorporated into the germ line of mice for in vivo studies of regulation of gene expression in development, genetic diseases, and malignancy.

Seminal vesicle formation and specific male protein secretion by female cells in allophenic mice.

The relation of cellular sex genotype to phenotype was examined in seminal vesicles of adult allophenic mice with cellular sex chromosome mosaicism. Each animal originated from conjoined blastomeres of an embryo of female (XX) and one of male (XY) constitution, from different inbred strains. Cells of both sexes were detected in bone marrow and certain other somatic tissues; cellular sex of seminal vesicles was deduced from strain-associated electrophoretic variants of proteins coded for at autosomal loci. Seminal vesicles composed partly or entirely of female cells were found in male and pseudohermaphrodite individuals. In a pseudohermaphrodite, both allelic variants of the tissue-specific normal male seminal vesicle protein (Svp-locus) were present, signifying that female as well as male cells were synthesizing the protein. Male-determining factors on the Y chromosome are thus not required in cells that differentiate into functional seminal vesicles.

Transfer of nonselectable genes into mouse teratocarcinoma cells and transcription of the transferred human beta-globin gene.

Teratocarcinoma (TCC) stem cells can function as vehicles for the introduction of specific recombinant genes into mice. Because most genes do not code for a selectable marker, we investigated the transformation efficiency of vectors with a linked selectable gene. In one series, TCC cells first selected for thymidine kinase deficiency were treated with DNA from the plasmid vector PtkH beta 1 containing the human genomic beta-globin gene and the thymidine kinase gene of herpes simplex virus. A high transformation frequency was obtained after selection in hypoxanthine-aminopterin-thymidine medium. Hybridization tests revealed that the majority of transformants had intact copies of the human gene among three to six total copies per cell. These were associated with cellular DNA sequences as judged from the presence of additional new restriction fragments and from stability of the sequences in tumors produced by injecting the cells subcutaneously. Total polyadenylate-containing RNA from cell cultures of two out of four transformants examined showed hybridization to the human gene probe: one RNA species resembled mature human beta-globin mRNA transcripts; the others were of larger size. In differentiating tumors, various tissues, including hematopoietic cells of TCC provenance could be found. In a second model set of experiments, wild-type TCC cells were used to test a dominant-selection scheme with pSV-gpt vectors. Numerous transformants were isolated, and their transfected DNA was apparently stably integrated. Thus, any gene of choice can be transferred into TCC stem cells even without mutagenesis of the cells, and selected cell clones can be characterized. Cells of interest may then be introduced into early embryos to produce new mouse strains with predetermined genetic changes.

Cellular DNA rearrangements and early developmental arrest caused by DNA insertion in transgenic mouse embryos.

Insertional mutagenesis was investigated in a transgenic mouse strain (HUGH/4) derived from a fertilized egg injected with plasmid DNA containing the human growth hormone gene. Lethality occurred in homozygous embryos and was traced to the egg cylinder stage on days 4 to 5 of gestation, shortly after implantation. The mutation is on chromosome 12 and is distinct in location and integration pattern from another mutation also leading to lethality of homozygotes in the egg cylinder stage. Based on this and other evidence, relatively many genes may be recruited to activity near the time of implantation and may therefore present a large target of vulnerability to mutagenesis. The single insert in HUGH/4, consisting of approximately three tandem copies of plasmid sequences, is flanked by mouse cellular sequences that have undergone rearrangements, including a probable deletion. The data suggest the hypothesis that DNA rearrangements, which appear to be commonplace in transgenic mice, may arise because the initial insertional complex is unstable; stepwise changes may then be generated until a more stable conformation is achieved.

Metastatic cutaneous melanoma promoted by ultraviolet radiation in mice with transgene-initiated low melanoma susceptibility.

An inbred-strain (C57BL/6) transgenic (Tyr-SV40E) mouse model of ultraviolet radiation (UVR)-induced metastatic cutaneous melanoma was produced without the use of chemical carcinogens and without resulting in other skin malignancies. Expression of this transgene occurs specifically in melanocytic-lineage cells. In untreated hemizygous mice of transgenic line 12 there are no skin melanomas, and the oncogenic sequence, which is expressed at a very low level, functions solely as a weak initiating stimulus. UVR [including 65% ultraviolet B (280-320 nm wavelength)] supplied the necessary promoting stimulus leading to melanomas. Of various trial protocols, eight were successful and involved exposure of 112 mice for a limited time on each of 3-10 days starting at 2-3 days of age and totalling 1.1-3.7 J/cm2 UVR. Fourteen of these animals developed a total of 15 invasive skin melanomas on the head and body, arising between 37-115 weeks of age and, therefore, often after a relatively long latency. The tumors were melanotic and in five of the mice they yielded macrometastases in regional and distant sites. The single most favorable protocol (1.9 J/cm2 total UVR, at 0.38 J/cm2/day for 5 days starting at 3 days of age) led to the highest incidence of melanoma (5 of 19 mice) and one of the lowest mortality rates (2 of 19). No melanomas occurred in UVR-treated nontransgenic C57BL/6 controls. Benign skin keratoacanthomas arose and often regressed in treated transgenic as well as nontransgenic mice. This new transgenic mouse model introduces many novel possibilities for experimental analysis of the melanoma-promoting mechanisms of UVR and also of the ability of specific genetic changes to impede or facilitate the UVR effect.

Accelerated growth of melanomas after specific immune destruction of tumor stroma in a mouse model.

Destruction of the entire stroma in a tumor could provide a stringent test of the prospects for tumor eradication in a single treatment. This possibility was investigated by experimental immune destruction of the stroma in a mouse melanoma model. Melanomas were first produced by grafting skin from transgenic C57BL/6 females of high-melanoma susceptibility to low-susceptibility transgenic males so that the malignant cells would be genetically female and the stromal cells genetically male. Subcutaneous transplant lines were then established from the melanotic and the amelanotic zones of such a melanoma and were carried in transgenic male hosts to ensure the male composition of the stroma. Thus, the male-specific H-Y histocompatibility antigen, which is ubiquitously expressed on male cells, could provide the target for an immune attack against the stroma. The transplant lines were next passaged once in transgenic females preimmunized against the H-Y antigen by having received and rejected a graft of C57BL/6 nontransgenic male skin. The antistromal immune response of these hosts did not prevent recovery of the tumors, which required a substantially prolonged latency. However, after retransplantation to nonimmunized males and females, the latency was markedly shortened from the original level. Thus, the treatment had indirectly selected for more rapidly growing tumor cells and hastened malignant progression.

Differences in latency and inducibility of mouse skin melanomas depending on the age and anatomic site of the skin.

To determine whether the occurrence of skin melanoma is influenced by the age or the anatomic source of the skin in melanoma-susceptible transgenic mouse models, skin was grafted from donors of different ages or from different anatomic sites to a standard (lateral trunk) site in adult recipients of the same transgenic strain. In 27 grafts of neonatal body skin, melanomas arose with a significantly shorter latency than in 37 grafts of older body skin. The difference may reflect not only the larger number of extrafollicular melanocytes in a given area of neonatal skin but also their unusually high mitotic activity shortly after birth and the influence of other growing skin cells nearby. Each of these body-skin grafts usually developed a single tumor situated near the graft edge. Because maximal wound healing occurs at the edge of such full-thickness skin grafts, melanocytes near the edge would receive the highest exposure to growth factors and cytokines associated with wound healing. In contrast to these results, grafts of snout skin yielded many melanomas, each originating from melanocytes within a vibrissa follicle rather than at the graft edge. The relatively strong local tumorigenic stimulus may be attributable to intrafollicular growth factors normally involved in whisker growth. The above-described experiments support the conclusion that agents in the immediate skin environment of the melanocyte, in addition to the state of the melanocyte itself, contribute to melanoma formation.

Ultraviolet radiation-induced malignant skin melanoma in melanoma-susceptible transgenic mice.

A new mouse model of UV radiation-induced melanoma is described. Unlike previous models, melanoma induction requires only short-term irradiation, does not require the application of chemical carcinogens, and does not cause any other tumors. The model takes advantage of the fact that Tyr-SV40E (C57BL/6 strain) transgenic mice are all melanoma-susceptible, and that different inbred lines are susceptible to different extents. Four-day-old mice of a moderately susceptible line (line 9 transgenic homozygotes) were exposed for 20 min/day to 328 mJ/cm2 to UVB (280-320 nm wavelength, comprising 70% of the total irradiance) for up to 4 consecutive days. Melanocytic lesions resembling macules, nevi, or early melanomas gradually appeared in the irradiated mice that were not seen in unirradiated transgenic controls of similar age. To afford ongoing observation beyond the short life span of line 9 homozygous mice, skin samples containing a total of 26 selected lesions were grafted at 20 weeks after UV radiation to longer-lived unirradiated hosts of transgenic line 12, in which melanoma susceptibility is low. Ten lesions in the grafts became melanomas; all melanomas had ulcerated and two had metastasized. At the stages examined, all the tumors were deeply melanotic. The remaining 16 lesions were still indolent when the experiment was terminated at 57 weeks post-UV radiation. The present protocol lends itself to variations in the choice of transgenic line, the age of the treated mice, and the intensity and duration of ultraviolet light; appropriate combinations of these variables would be expected to yield melanomas in relatively long-lived transgenic mice without skin grafting. The new model provides an opportunity to determine the melanoma action spectrum, to characterize at the molecular level the melanomas induced by ultraviolet light in comparison with those of other origins, and to investigate in vivo the photoprotective role of melanin.

Expression of plasminogen activators and plasminogen activator inhibitors in cutaneous melanomas of transgenic melanoma-susceptible mice.

The Tyr-SV40E transgenic mouse model of malignant skin melanoma has been used here to generate melanomas in genetically identical (C57BL/6) mice for analysis of the plasminogen activator (PA) system during tumor development and progression. Twenty-two melanocytic lesions were examined by in situ zymography for PA activity and by immunohistochemistry for concomitant visualization of PA proteins; these lesions encompassed 3 nevi and 19 primary melanomas ranging from melanotic through mixed tumors to amelanotic tumors. Although urokinase-type plasminogen activator (u-Pa) activity was not detected at premalignant stages, it began to appear early in tumorigenesis and became more prominent in later stages of a majority of the tumors. The activity was largely attributable to the endothelium of sprouting capillaries and to a lesser degree to granulocytes, fibroblastic cells, and occasional melanoma cells within tumors. Tissue-type plasminogen activator (t-PA) was undetectable or low in all cases. Of the inhibitors (PAI), PAI-1 was seen in endothelial and fibroblastic cells and in the extracellular matrix, whereas PAI-2 occurred in only one case and was melanoma cell associated. Eleven additional melanomas were analyzed by reverse transcription-PCR for PA expression in RNA extracts from relatively large tumor samples. These were obtained from eight primary melanomas and three metastases, again spanning melanotic, mixed, and amelanotic cases. From four of the mixed primary tumors with distinct melanotic and amelanotic zones, the respective components were propagated separately in transgenic hosts as s.c. transplants to obtain data for clearly identifiable melanotic versus amelanotic parts. u-PA and PAI-1 mRNAs were expressed in all. t-PA expression varied greatly and was notably high in several amelanotic tumors or tumor components, possibly as a result of large blood vessels, as such vessels were seen to be t-PA positive in normal tissue. The u-PA activity in sprouting capillaries may indicate a role in neoangiogenesis. Therefore, according to these mouse models, u-PA may indirectly be a potential therapeutic target against melanoma progression.

Comparative decreases in tyrosinase, TRP-1, TRP-2, and Pmel 17/silver antigenic proteins from melanotic to amelanotic stages of syngeneic mouse cutaneous melanomas and metastases.

Malignant cutaneous melanomas and metastases were taken directly from in situ lesions of genetically identical (C57BL/6 strain) Tyr-SV40E transgenic mice, and samples were analyzed by Western immunoblotting with antisera specific for the COOH terminus of each of four melanocytic proteins. These were tyrosinase, TRP-1, TRP-2, and Pmel 17/silver. Of the 13 melanomas examined, there were 5 melanotic primary tumors, 5 amelanotic primary tumors, and 3 amelanotic metastases. The melanotic tumors expressed all of the markers to some extent. In contrast, the amelanotic tumors lacked detectable levels of one, two, or three of the proteins, except for an apparently amelanotic tumor sample in which all were expressed, but in which some melanotic cells were likely to have been present. Thus, despite some variability, there is clearly a downward trend in the presence of these proteins as the tumors become amelanotic, a pigmentary change associated with ongoing malignant progression. In the amelanotic tumors, tyrosinase was most often deficient, whereas TRP-2 was most often persistently expressed. These results, obtained from melanomas of syngeneic origin, indicate that tumors in the relatively early stages of malignancy might be more responsive than later-stage tumors to immunotherapy involving an ensemble of antigenic peptides of the tested gene products. Moreover, TRP-2 peptides may be especially useful for therapeutic intervention at the later stages.

Melanocyte culture lines from Tyr-SV40E transgenic mice: models for the molecular genetic evolution of malignant melanoma.

Transgenic Tyr-SV40E mice previously produced on the C57BL/6 inbred-strain background, with SV40 oncogenic sequences specifically expressed in pigment cells, are predisposed to melanoma [Bradl, M., Klein-Szanto, A., Porter, S. & Mintz, B. (1991). Proc. Natl. Acad. Sci. USA, 88, 164-168]. Separate lines of these animals differ genetically only in the number of copies and chromosomal site of integration of the transgene. Skin melanocytes from young mice with no apparent skin lesions were established in continuous culture from hemizygous donors with low, medium and high numbers of transgene copies, and from a homozygous offspring of the low-copy mouse line. The standard culture conditions enable C57BL/6 wild-type melanocytes to become stably immortalized without transformation. The transgenic cell lines all changed over time in an orderly progression. However, with greater numbers of transgene copies, the cells more rapidly displayed shorter doubling times, increased anchorage independence, reduced serum and growth factor requirements, decreased tyrosinase expression and melanin content, increased oncogene expression, and capacity to form malignant melanomas when tested by grafting. Melanocytes with the lowest number of transgene copies were of special interest. They grew more rapidly than the wild-type cells from the outset, but did not become tumorigenic until an apparently small number of still-unknown genetic changes had spontaneously occurred, or until the number of transgene copies was increased slightly by homozygosity. In contrast to the hemizygous low-copy cells, the homozygous counterparts underwent striking and rapid transformational changes and early conversion to malignancy. Thus such low-copy transgenic melanocyte lines afford an exceptional opportunity for molecular analysis of somatic genetic evolution toward malignant melanoma.

Neonatal W-mutant mice are favorable hosts for tracking development of marked hematopoietic stem cells.

Neonatal unirradiated mice of W-mutant genotypes, with a hematopoietic stem cell defect and anemia, were injected i.v. with normal fetal liver hematopoietic cells. Efficient, long-term engraftment occurred as a result of the competitive advantage to the donor stem cells. The frequency of engraftment and rate of repopulation characteristically diminish in the series W/Wv, Wf/Wf, and Wv/+, in which the severity of the endogenous defect is progressively less. H-2 compatibility is required in the inbred strain combinations examined; other histocompatibility loci play a minor role in some strain combinations. Engraftment is due to self-renewing hematopoietic stem cells ancestral to myeloid and lymphoid lineages. The more mildly defective mutants display much greater variability in the kinetics of repopulation--a result consistent with seeding by single, or very few, stem cells that form developing clones. Engraftment efficiency is reduced by prolonged culture of fetal liver cells during experimental infection by recombinant retroviruses; nevertheless, after 24 h in vitro to achieve retroviral marking, stem cells retain their ability to repopulate and develop in W/Wv neonates.

Multiple alternatively spliced transcripts of the mouse tyrosinase-encoding gene.

We have isolated and characterized tyrosinase-specific cDNAs from wild-type mouse skin, to provide a basis for the structural and functional analysis of mutations at the mouse tyrosinase-encoding (Tyr) locus. The cDNAs were synthesized by the polymerase chain reaction. At least twelve alternatively spliced transcripts of the Tyr gene were found, including nine not previously described. Of 51 clones obtained, most (59%) correspond to the full-length cDNA encoding active tyrosinase. The others are shorter and apparently arose by alternative splicing. They are attributable to exon skipping, usage of alternative 5' and/or 3' splice sites, and (in one case) retention of an intronic sequence. Patterns of alternative splicing also occur in other pigmented tissues.

Intracranial metastases in the initial staging of bronchogenic carcinoma.

We evaluated the effectiveness of neurologic examination, electroencephalography (EEG), and computed tomography (CT) in the initial staging of patients with nonsmall cell carcinoma. Eight of 66 patients had evidence of intracranial metastases. Three of these had no other metastases and would otherwise have been surgical candidates. Thus, thorough investigation for evidence of intracranial metastases is warranted at the time of initial staging. The CT proved to be more effective than clinical evaluation or EEG, alone or in combination, in detecting intracranial metastases. The CT screening of patients prior to curative resection should increase the success rate for such procedures by eliminating patients with preexisting metastases.

The role of open lung biopsy in the management and outcome of patients with diffuse lung disease.

BACKGROUND: Open lung biopsy (OLB) has long been considered the gold standard for the diagnosis of parenchymal lung disease. With recent advances in computed tomographic imaging and diagnostic techniques (eg, bronchoscopy), we thought it necessary to reevaluate the role of OLB in the management of patients with interstitial lung disease. METHODS: We carried out a retrospective analysis of 103 OLBs performed at Hadassah University Hospital, Jerusalem, and Carmel Medical Center, Haifa, between 1980 and 1994. Data gathered included demographic information, underlying condition, indications for biopsy, diagnosis before biopsy, final diagnosis, change in therapy, and mortality. "Benefit" was defined as a change in therapy resulting in survival. RESULTS: There were 45 immunocompetent patients (group 1), 39 immunocompromised patients (group 2), and 26 children (group 3), 7 of whom were included in group 2 for analysis. Overall, a diagnosis was reached after OLB in 85% of patients. An unexpected diagnosis was reached in 52%, and a change in therapy was instituted in 46%. The overall mortality rate was 20%. In group 1, the mortality rate was 13%, and "benefit" from OLB was reached in only 18%. In group 2, the mortality rate was 39%, and "benefit" was achieved in 46%, and in group 3, the mortality rate was 12% and "benefit", 50%. CONCLUSIONS: Open lung biopsy is an excellent diagnostic technique. In immunocompetent patients, the "benefit" is relatively low, as therapy (corticosteroids) is frequently used after biopsy. In immunocompromised patients, therapy changes substantially after OLB, but mortality is high. Therefore, OLB should be reserved for patients in whom the diagnosis is likely to lead to a change in therapy and in patients in whom the underlying condition has a reasonable prognosis according to the clinical impression by the attending physician.

Decline in MHC class I expression with increasing thickness of cutaneous melanomas in standard-strain transgenic mouse models.

Major histocompatibility complex (MHC) class I proteins are required for the formation of complexes with antigenic peptides that enable cytotoxic T lymphocytes (CTLs) to recognize and lyse target cells. The frequent loss of MHC class I expression reported in human melanomas and melanoma cell lines may therefore be an obstacle to CTL-based immunotherapy. We have investigated the expression of MHC class I proteins in the cutaneous melanomas of Tyr-SV40E (C57BL/6 strain) transgenic mice in order to evaluate their potential as experimental models for immunotherapy. The SV40 large T (LT) oncoprotein, which is expressed exclusively in the melanocytic lineages of these mice, was used as a marker for flow cytometric analysis of the parenchymal (potential target) cells of 35 freshly dissociated samples from 28 primary tumours. All the tumours were ulcerated and exceeded the Breslow thickness indicative of a poor clinical prognosis in human melanoma. Using antibodies against H-2D(b) and H-2K(b) class I proteins, the LT antigen-positive cells were found to have high levels of both these MHC class I molecules in the thinnest tumours (2 mm), whereas the levels tended to decline with increasing tumour thickness. Among the tumours > 4 mm thick, five had no detectable MHC class I expression. Unexpectedly, the apparent loss of H-2D(b) and H-2K(b) proteins was observed not only in LT-positive cells but also in LT-negative cell populations. Expression of both H-2D(b) and H-2K(b) was restored in tumours derived from a class I-low melanoma cell line by treatment of the hosts with interferon-gamma. These results implicate a regulatory defect as a principal cause of the loss of MHC class I antigens, as noted by others in some human tumours, and they demonstrate that this loss is remediable, even in advanced stages of melanomas.