An integrated benefit-risk assessment of cobalt-containing alloys used in medical devices: Implications for regulatory requirements in the European Union.

In 2017, the European Union (EU) Committee for Risk Assessment (RAC) recommended the classification of metallic cobalt (Co) as Category 1B with respect to its carcinogenic and reproductive hazard potential and Category 2 for mutagenicity but did not evaluate the relevance of these classifications for patients exposed to Co-containing alloys (CoCA) used in medical devices. CoCA are inherently different materials from Co metal from a toxicological perspective and thus require a separate assessment. CoCA are biocompatible materials with a unique combination of properties including strength, durability, and a long history of safe use that make them uniquely suited for use in a wide-range of medical devices. Assessments were performed on relevant preclinical and clinical carcinogenicity and reproductive toxicity data for Co and CoCA to meet the requirements under the EU Medical Device Regulation triggered by the ECHA re-classification (adopted in October 2019 under the 14th Adaptation to Technical Progress to CLP) and to address their relevance to patient safety. The objective of this review is to present an integrated overview of these assessments, a benefit-risk assessment and an examination of potential alternative materials. The data support the conclusion that the exposure to CoCA in medical devices via clinically relevant routes does not represent a hazard for carcinogenicity or reproductive toxicity. Additionally, the risk for the adverse effects that are known to occur with elevated Co concentrations (e.g., cardiomyopathy) are very low for CoCA implant devices (infrequent reports often reflecting a unique catastrophic failure event out of millions of patients) and negligible for CoCA non-implant devices (not measurable/no case reports). In conclusion, the favorable benefit-risk profile also in relation to possible alternatives presented herein strongly support continued use of CoCA in medical devices.

Carcinogenic hazard assessment of cobalt-containing alloys in medical devices: Review of in vivo studies.

Cobalt (Co) alloys have been used for over seven decades in a wide range of medical devices, including, but not limited to, hip and knee implants, surgical tools, and vascular stents, due to their favorable biocompatibility, durability, and mechanical properties. A recent regulatory hazard classification review by the European Chemicals Agency (ECHA) resulted in the classification of metallic Co as a Class 1B Carcinogen (presumed to have carcinogenic potential for humans), primarily based on inhalation rodent carcinogenicity studies with pure metallic Co. The ECHA review did not specifically consider the carcinogenicity hazard potential of forms or routes of Co that are relevant for medical devices. The purpose of this review is to present a comprehensive assessment of the available in vivo preclinical data on the carcinogenic hazard potential of exposure to Co-containing alloys (CoCA) in medical devices by relevant routes. In vivo data were reviewed from 33 preclinical studies that examined the impact of Co exposure on local and systemic tumor incidence in rats, mice, guinea pigs, and hamsters. Across these studies, there was no significant increase of local or systemic tumors in studies relevant for medical devices. Taken together, the relevant in vivo data led to the conclusion that CoCA in medical devices are not a carcinogenic hazard in available in vivo models. While specific patient and implant factors cannot be fully replicated using in vivo models, the available in vivo preclinical data support that CoCA in medical devices are unlikely a carcinogenic hazard to patients.

Acute Alpha-Naphthylisothiocyanate-induced Liver Toxicity in Germfree and Conventional Male Rats.

Differences in the responses of conventional and germfree male Sprague-Dawley rats to acute injury induced by alpha-naphthylisothiocyanate (ANIT), a well-characterized biliary epithelial toxicant, were evaluated. Conventional and germfree rats were dosed once orally with 50 mg/kg of ANIT or corn oil alone and serially sacrificed daily for the next 3 days. Germfree rats treated with ANIT tended to have greater increases in virtually all liver and biliary-related analytes compared with conventional rats treated with ANIT; however, significant differences were found only in a few of these analytes including increased bile acids on day 3, total bilirubin on day 4, glutamate dehydrogenase (GLDH) on day 3, and reduced paraoxonase 1 (PON1) on days 2 and 3. Histologic differences between the conventional and germfree rats were modest, but most pronounced on day 2 (24-hr post dosing). Based on subjective scoring, biliary necrosis, neutrophilic cholangitis, and portal tract edema were more severe in germfree rats at 24 hr post dosing compared with conventional rats. Biliary epithelial replication did not differ between treated groups, however. Overall, germfree rats had a modestly greater level of biliary tract injury based on subjective histologic scoring and clinical chemistry measurements following an acute exposure to the well-characterized biliary toxin, ANIT; however, the difference between the ANIT-treated germfree and conventional groups was modest and most evident only within the first day following exposure. These findings suggest that the microbiome did not significantly affect ANIT-induced acute biliary tract injury in the conditions of this study.

Integrated approach to early detection of cardiovascular toxicity induced by a ghrelin receptor agonist.

Cardiovascular (CV) safety concerns are among the leading causes of compound attrition in drug development. This work describes a strategy of applying novel end points to a 7-day rodent study to increase the opportunity to detect and characterize CV injury observed in a longer term (ie, 28 days) study. Using a ghrelin receptor agonist (GSK894281), a compound that produces myocardial degeneration/necrosis in rats after 28 days at doses of 0.3, 1, 10, or 60 mg/kg/d, we dosed rats across a range of similar doses (0, 0.3, 60, or 150 mg/kg/d) for 7 days to determine whether CV toxicity could be detected in a shorter study. End points included light and electron microscopies of the heart; heart weight; serum concentrations of fatty acid-binding protein 3 (FABP3), cardiac troponin I (cTnI), cardiac troponin T (cTnT), and N-terminal proatrial natriuretic peptide (NT-proANP); and a targeted transcriptional assessment of heart tissue. Histologic evaluation revealed a minimal increase in the incidence and/or severity of cardiac necrosis in animals administered 150 mg/kg/d. Ultrastructurally, mitochondrial membrane whorls and mitochondrial degeneration were observed in rats given 60 or 150 mg/kg/d. The FABP3 was elevated in rats given 150 mg/kg/d. Cardiac transcriptomics revealed evidence of mitochondrial dysfunction coincident with histologic lesions in the heart, and along with the ultrastructural results support a mechanism of mitochondrial injury. There were no changes in cTnI, cTnT, NT-proANP, or heart weight. In summary, enhancing a study design with novel end points provides a more integrated evaluation in short-term repeat dose studies, potentially leading to earlier nonclinical detection of structural CV toxicity.

Effects of Solutol (Kolliphor) and cremophor in polyethylene glycol 400 vehicle formulations in Sprague-Dawley rats and beagle dogs.

When conventional vehicles (eg, methylcellulose and water) impart inadequate physical, chemical, and/or biological properties for proper toxicological assessment of test article formulations, nonconventional vehicles may be considered. Often toxicity data for nonconventional vehicle formulations are limited. Studies were conducted to collect toxicity data from a rodent and a non-rodent species given 2 nonconventional vehicles, Solutol HS15/polyethylene glycol (PEG) 400 and Cremophor RH40/PEG 400, with differing formulations and dose volumes (10 mL/kg for rats; 2 or 5 mL/kg for dogs). In rats, both vehicles caused increase in kidney weights (males only) and decrease in thymic weights (males only) without concurrent microscopic findings; altered urine electrolytes, minimally decreased serum electrolytes (males only), and increased serum total cholesterol (females only) were also present. The Cremophor formulation was also associated with increased serum urea (males only) and urine phosphorus: creatinine. For rats given the Solutol formulation, both genders had decreased urine glucose parameters and males had increased urine volume. In dogs, loose/watery feces and emesis were present given either vehicle, and mucus-cell hyperplasia of the ileum was present given the Solutol formulation. Increased red blood cell mass and decreased urine volume in dogs given 30% Solutol/70% PEG 400 (5 mL/kg/d) were likely due to subclinical dehydration and hemoconcentration. For the Cremophor formulations, dose volume-dependent increased incidence of minimal subepithelial gastric hemorrhage was noted in dogs, and dogs given 5 mL/kg/d showed increased serum urea nitrogen. Overall, regardless of the formulation or dose volume, neither vehicle produced overt toxicity in either species, but the Solutol formulation produced fewer effects in rats. Generally, lower dose volumes minimized the severity and/or incidence of findings.

Effects of Kupffer cell depletion on acute alpha-naphthylisothiocyanate-induced liver toxicity in male mice.

Depletion of Kupffer cells, known to modulate chemical-induced hepatocellular injury, has not been studied with regard to biliary epithelial injury. Here, the authors investigated the effect of Kupffer cell depletion by clodronate on the toxicity of alpha-naphthylisothiocyanate (ANIT), known to injure biliary epithelium as well as hepatocytes. Up to 99% depletion of Kupffer cells occurred in ANIT and liposome-encapsulated clodronate-treated mice. The effect of Kupffer cell depletion was most evident one day following ANIT treatment. Histologically, there was a modest increase in neutrophil infiltration of the bile ducts, hepatocytic necrosis, and microvesicular vacuolization in the ANIT and clodronate-treated mice, but differences between other groups did not persist. Clinical pathology analytes related to the biliary or hepatocellular injury were significantly elevated in ANIT and clodronate-treated mice compared to mice given clodronate only. This was also true for mice given ANIT and empty liposomes in the case of the biliary analytes. However, group means were typically higher for the ANIT and clodronate-treated group than others on the first 2 days following ANIT injection. These findings suggest that Kupffer cell reduction increases hepatobiliary damage due to ANIT treatment.

Acute ANIT toxicity in male IL-10 knockout and wild-type mice.

The innate immune response is known to modify hepatocellular injury induced by toxicants. To assess the role of IL-10, a component of the innate immune response, in toxicant-induced injury of biliary epithelium, wild-type (WT) and IL-10 knockout mice (KO) were given a single toxic dose (50 mg/kg) of alpha-napthylisothiocyanate (ANIT) and assessed at twenty-four-hour intervals for four days following treatment. Clinical signs of toxicity were greater in WT mice. Unexpectedly, over the course of the study, there was a consistent tendency for ANIT-treated IL-10 KO mice to have less hepatocellular injury than WT mice. However, changes in the biliary epithelium differed in that there was more histologic evidence of inflammation and necrosis on days 2 and 3, respectively, in ANIT-treated IL-10 KO mice compared with WT mice. Proliferation of biliary epithelium and hepatocytes was greater and/or occurred earlier in the ANIT-treated IL-10 KO mice compared with the ANIT-treated WT mice, suggesting a greater reparative response was needed for recovery after toxicant injury in the IL-10 KO mice. Overall, our data suggest that IL-10 KO mice have less hepatocellular injury than WT mice following a toxic dose of ANIT and that biliary epithelial injury is accentuated in the KO mice.

Time course gene expression using laser capture microscopy-extracted bile ducts, but not hepatic parenchyma, reveals acute alpha-naphthylisothiocyanate toxicity.

Acute toxic responses to a 50-mg/kg oral dose of 1-naphthylisothiocyanate (ANIT) were evaluated by microarray analysis of laser capture-microdissected rat biliary epithelium or hepatic parenchyma obtained 2 and 6 hours postdose. Distinct differences in gene expression patterns between biliary epithelium and hepatic parenchyma were noted at the 2-hour postdose time point, where 375 genes were altered in biliary epithelium but only 38 genes were altered in hepatic parenchyma. Endoplasmic reticulum stress genes were uniquely expressed in biliary epithelial cells at 2 hours postdose. By 6 hours postdose, 620 genes were altered in biliary epithelium, but only 32 genes were altered in hepatic parenchyma. In biliary epithelium, expression of genes involved in the unfolded protein response had decreased compared with the 2-hour time point, while expression of genes involved in protein degradation such as proteasome-ubquination pathways and cell death pathways had increased. At this same time, hepatic parenchymal gene expression changed little. Within 6 hours following oral exposure to ANIT, prior to morphologic changes, specific biliary epithelial gene expression changes, indicative of a vigorous unfolded protein response with protein destruction and cell death pathway activation were noted, in contrast to minor changes in the hepatic parenchyma.

PPAR alpha, more than PPAR delta, mediates the hepatic and skeletal muscle alterations induced by the PPAR agonist GW0742.

Therapeutic use of certain peroxisome proliferator-activated receptor (PPAR) alpha agonists (fibrates) for the treatment of dyslipidemia has infrequently been associated with the untoward side effect of myopathy. With interest in PPAR-delta as a therapeutic target, this study assessed whether a PPAR-delta agonist induced similar hepatic and skeletal muscle alterations as noted with some fibrates. PPAR-alpha null (KO) and corresponding wild-type (WT) mice were administered toxicological dosages of a potent PPAR-delta agonist tool ligand (GW0742; which also has weak PPAR-alpha agonist activity) or a potent PPAR-alpha agonist (WY-14,643) for 10 days. Increases in liver weights and clinical chemistry indicators of skeletal muscle damage and/or liver injury were more pronounced in WT mice compared with KO mice administered the PPAR-delta agonist. Likewise, the incidence and severity of skeletal myopathy were greater in WT mice given GW0742 compared with KO mice. Ultrastructural and immunohistochemical analyses revealed significant peroxisome proliferation in muscle and liver of WT mice treated with each agonist; however, KO animals showed little or no evidence of hepatic and muscle peroxisome proliferation. PMP-70 protein expression in liver was consistent with these results. The hepatomegaly, hepatic and skeletal muscle peroxisome proliferation, and skeletal myopathy induced by this PPAR-delta ligand was predominantly mediated by its cross-activation of PPAR-alpha, though PPAR-delta agonism contributed slightly to these effects.

Use of genetically modified mouse models to assess pathways of benzene-induced bone marrow cytotoxicity and genotoxicity.

Benzene induces bone marrow cytotoxicity and chromosomal breaks as a primary mode of action for the induction of bone marrow toxicity. Our research group has used genetically modified mouse models to examine metabolic and genomic response pathways involved in benzene induced cytotoxicity and genotoxicity in bone marrow and in hematopoietic stem cells (HSC). We review our studies using NQO1-/- mice and mEH-/- mice to examine the roles of these enzymes, NAD(P)H:quinone oxidoreductase-1 (NQO1) and microsomal epoxide hydrolase (mEH) in mediating benzene-induced toxicity. NQO1 catalyzes the detoxication of benzene quinone metabolites and mEH catalyzes the hydrolysis of benzene oxide. Our studies using gene expression profiling of bone marrow and enriched HSC populations isolated from the bone marrow of benzene-exposed mice demonstrate differential gene expression responses of key genes induced by inhaled benzene. These studies show that benzene toxicity is regulated by a number of genetic pathways that affect the production of reactive metabolites and DNA damage response pathways in a target tissue.

Exposure of hematopoietic stem cells to benzene or 1,4-benzoquinone induces gender-specific gene expression.

Chronic exposure to benzene results in progressive decline of hematopoietic function and may lead to the onset of various disorders, including aplastic anemia, myelodysplastic syndrome, and leukemia. Damage to macromolecules resulting from benzene metabolites and misrepair of DNA lesions may lead to changes in hematopoietic stem cells (HSCs) that give rise to leukemic clones. We have shown previously that male mice exposed to benzene by inhalation were significantly more susceptible to benzene-induced toxicities than females. Because HSCs are targets for benzene-induced cytotoxicity and genotoxicity, we investigated DNA damage responses in HSC from both genders of 129/SvJ mice after exposure to 1,4-benzoquinone (BQ) in vitro or benzene in vivo. 1,4-BQ is a highly reactive metabolite of benzene that can cause cellular damage by forming protein and DNA adducts and producing reactive oxygen species. HSCs cultured in the presence of 1,4-BQ for 24 hours showed a gender-independent, dose-dependent cytotoxic response. RNA isolated from 1,4-BQ-treated HSCs and HSCs from mice exposed to 100 ppm benzene by inhalation showed altered expression of apoptosis, DNA repair, cell cycle, and growth control genes compared with unexposed HSCs. Rad51, xpc, and mdm-2 transcript levels were increased in male but not female HSCs exposed to 1,4-BQ. Males exposed to benzene exhibited higher mRNA levels for xpc, ku80, ccng, and wig1. These gene expression differences may partially explain the gender disparity in benzene susceptibility. HSC culture systems such as the one used here will be useful for testing the hematotoxicity of various substances, including other benzene metabolites.

Gene expression profile in bone marrow and hematopoietic stem cells in mice exposed to inhaled benzene.

Acute myeloid leukemia and chronic lymphocytic leukemia are associated with benzene exposure. In mice, benzene induces chromosomal breaks as a primary mode of genotoxicity in the bone marrow (BM). Benzene-induced DNA lesions can lead to changes in hematopoietic stem cells (HSC) that give rise to leukemic clones. To gain insight into the mechanism of benzene-induced leukemia, we investigated the DNA damage repair and response pathways in total bone marrow and bone marrow fractions enriched for HSC from male 129/SvJ mice exposed to benzene by inhalation. Mice exposed to 100 ppm benzene for 6h per day, 5 days per week for 2 week showed significant hematotoxicity and genotoxicity compared to air-exposed control mice. Benzene exposure did not alter the level of apoptosis in BM or the percentage of HSC in BM. RNA isolated from total BM cells and the enriched HSC fractions from benzene-exposed and air-exposed mice was used for microarray analysis and quantitative real-time RT-PCR. Interestingly, mRNA levels of DNA repair genes representing distinct repair pathways were largely unaffected by benzene exposure, whereas altered mRNA expression of various apoptosis, cell cycle, and growth control genes was observed in samples from benzene-exposed mice. Differences in gene expression profiles were observed between total BM and HSC. Notably, p21 mRNA was highly induced in BM but was not altered in HSC following benzene exposure. The gene expression pattern suggests that HSC isolated immediately following a 2 weeks exposure to 100 ppm benzene were not actively proliferating. Understanding the toxicogenomic profile of the specific target cell population involved in the development of benzene-associated diseases may lead to a better understanding of the mechanism of benzene-induced leukemia and may identify important interindividual and tissue susceptibility factors.

Human CD34+ hematopoietic progenitor cells are sensitive targets for toxicity induced by 1,4-benzoquinone.

Chronic human exposure to benzene has been linked to several hematopoietic disorders, including leukemia and lymphomas. Certain benzene metabolites, including benzoquinone (BQ), are genotoxic and mutagenic. Bone marrow stem cells are targets for benzene-induced cytotoxicity and DNA damage that could result in changes to the genome of these progenitor cells, thereby leading to hematopoietic disorders and cancers. Human bone marrow CD34(+) hematopoietic progenitor cells (HPC) were exposed in vitro to 1,4-BQ to assess cytotoxicity, genotoxicity, and DNA damage responses and the molecular mechanisms associated with these events. CD34(+) HPC from 10 men and 10 women were exposed to 0, 1, 5, 10, 15, or 20 microM of 1,4-BQ and analyzed 72 h later. Apoptosis and cytotoxicity were dose-dependent, with exposure to 10 microM 1,4-BQ resulting in approximately 60% cytotoxicity relative to untreated controls. A significant increase in the percentage of micronucleated CD34(+) cells was detected in cultures treated with 1,4-BQ. In addition, the p21 mRNA level was elevated in 1,4-BQ-treated cells, suggesting that human CD34(+) cells utilize the p53 pathway in response to 1,4-BQ-induced DNA damage. However, there were no significant changes in mRNA levels of the DNA repair genes ku80, rad51, xpa, xpc, and ape1 as well as p53 following treatment with 1,4-BQ. Although interindividual variations were evident in the cellular response to 1,4-BQ, there was no gender difference in the response overall. These results show that human CD34(+) cells are sensitive targets for 1,4-BQ toxicity that use the p53 DNA damage response pathway in response to genotoxic stress. Human CD34(+) HPC will be useful for testing the toxicity of other benzene metabolites and various hematotoxic chemicals.

Variations in Prkdc and susceptibility to benzene-induced toxicity in mice.

Benzene, a carcinogen that induces chromosomal breaks, is strongly associated with leukemias in humans. Possible genetic determinants of benzene susceptibility include proteins involved in repair of benzene-induced DNA damage. The catalytic subunit of DNA-dependent protein kinase (DNA-PKcs), encoded by Prkdc, is one such protein. DNA-PKcs is involved in the nonhomologous end-joining (NHEJ) pathway of DNA double-strand break (DSB) repair. Here we compared the toxic effects of benzene on mice (C57BL/6 and 129/Sv) homozygous for the wild-type Prkdc allele and mice (129/SvJ) homozygous for a Prkdc functional polymorphism that leads to diminished DNA-PK activity and enhanced apoptosis in response to radiation-induced damage. Male and female mice were exposed to 0, 10, 50, or 100 ppm benzene for 6 h/d, 5 d/week for 2 weeks. Male mice were more susceptible to benzene toxicity compared with females. Hematotoxicity was evident in all male mice but was not seen in female mice. We observed similar, large increases in both micronucleated erythrocyte populations in all male mice. Female mice had smaller but significant increases in micronucleated cells. The p53-dependent response was induced in all strains and genders of mice following benzene exposure, as indicated by an increase in p21 mRNA levels in bone marrow that frequently corresponded with cell cycle arrest in G2/M. Prkdc does not appear to be a significant genetic susceptibility factor for acute benzene toxicity. Moreover, the role of NHEJ, mediated by DNA-PK, in restoring genomic integrity following benzene-induced DSB remains equivocal.

Genetic susceptibility to benzene-induced toxicity: role of NADPH: quinone oxidoreductase-1.

Enzymes that activate and detoxify benzene are likely genetic determinants of benzene-induced toxicity.NAD(P)H: quinone oxidoreductase-1 (NQO1) detoxifies benzoquinones, proposed toxic metabolites of benzene. NQO1 deficiency in humans is associated with an increased risk of leukemia, specifically acute myelogenous leukemia, and benzene poisoning. We examined the importance of NQO1 in benzene-induced toxicity by hypothesizing that NQO1-deficient (NQO1-/-) mice are more sensitive to benzene than mice with wild-type NQO1 (NQO1+/+; 129/Sv background strain). Male and female NQO1-/- and NQO1+/+ mice were exposed to inhaled benzene (0, 10, 50, or 100 ppm) for 2 weeks, 6 h/day, 5 days/week. Micronucleated peripheral blood cells were counted to assess genotoxicity. Peripheral blood counts and bone marrow histology were used to assess hematotoxicity and myelotoxicity. p21 mRNA levels in bone marrow cells were used as determinants of DNA damage response. Female NQO1-/- mice were more sensitive (6-fold) to benzene-induced genotoxicity than the female NQO1+/+ mice. Female NQO1-/- mice had a 9-fold increase (100 versus 0 ppm) in micronucleated reticulocytes compared with a 3-fold increase in the female NQO1+/+ mice. However, the induced genotoxic response in male mice was similar between the two genotypes (> or = 10-fold increase at 100 ppm versus 0 ppm). Male and female NQO1-/- mice exhibited greater hematotoxicity than NQO1+/+ mice. p21 mRNA levels were induced significantly in male mice (>10-fold) from both strains and female NQO1-/- mice (> 8-fold), which indicates an activated DNA damage response. These results indicate that NQO1 deficiency results in substantially greater benzene-induced toxicity. However, the specific patterns of toxicity differed between the male and female mice.

Male mice deficient in microsomal epoxide hydrolase are not susceptible to benzene-induced toxicity.

Enzymes involved in benzene metabolism are likely genetic determinants of benzene-induced toxicity. Polymorphisms in human microsomal epoxide hydrolase (mEH) are associated with an increased risk of developing leukemia, specifically those associated with benzene. This study was designed to investigate the importance of mEH in benzene-induced toxicity. Male and female mEH-deficient (mEH-/-) mice and background mice (129/Sv) were exposed to inhaled benzene (0, 10, 50, or 100 ppm) 5 days/week, 6 h/day, for a two-week duration. Total white blood cell counts and bone marrow cell counts were used to assess hematotoxicity and myelotoxicity. Micronucleated peripheral blood cells were counted to assess genotoxicity, and the p21 mRNA level in bone marrow cells was used as a determinant of the p53-regulated DNA damage response. Male mEH-/- mice did not have any significant hematotoxicity or myelotoxicity at the highest benzene exposure compared to the male 129/Sv mice. Significant hematotoxicity or myelotoxicity did not occur in the female mEH-/- or 129/Sv mice. Male mEH-/- mice were also unresponsive to benzene-induced genotoxicity compared to a significant induction in the male 129/Sv mice. The female mEH-/- and 129/Sv mice were virtually unresponsive to benzene-induced genotoxicity. While p21 mRNA expression was highly induced in male 129/Sv mice after exposure to 100-ppm benzene, no significant alteration was observed in male mEH-/- mice. Likewise, p21 mRNA expression in female mEH-/- mice was not significantly induced upon benzene exposure whereas a significant induction was observed in female 129/Sv mice. Thus mEH appears to be critical in benzene-induced toxicity in male, but not female, mice.

Influence of CD4 T cells and the source of major histocompatibility complex class II-restricted peptides on cytotoxic T-cell priming by dendritic cells.

We have previously reported that bone marrow derived dendritic cells (DC) pulsed with major histocompatibility complex (MHC) class I-restricted peptide efficiently prime a cytotoxic T lymphocyte (CTL) response in vivo. Here we assess the involvement of CD4(+) T cells in the induction of CD8(+) CTL by DC by testing the ability of class II-deficient (C2D) DC, class II mutant (Alpha beta mut) DC and autologous serum generated DC (AS DC) to present class I-restricted antigens in vitro and in vivo. DC generated from the bone marrow of class II knockout mice and transgenic mice expressing a mutant class II that can not bind CD4 were phenotypically similar to wild type (wt) DC, except with regard to MHC class II expression. The C2D and Alpha beta mut DC, though fully capable of presenting the class I-restricted ovalbumin (OVA) peptide to a T-cell hybridoma in vitro, failed to prime a CTL response in vivo. Restoration of class II expression on C2D DC allowed priming of a CTL response; thus, the defect in CTL priming was indeed caused by the absence of class II expression. Likewise, DC generated in autologous serum were unable to prime a CTL response as these DC only express 'self' class II epitopes and therefore would not activate syngeneic CD4(+) T cells. Addition of exogenous class II epitopes rescued the ability of AS DC to prime a CTL response. These observations provide convincing evidence that efficient CTL induction by DC in vivo requires concomitant presentation of class II epitopes for CD4(+) T-cell induction.