Choline kinase-alpha by regulating cell aggressiveness and drug sensitivity is a potential druggable target for ovarian cancer.

BACKGROUND: Aberrant choline metabolism has been proposed as a novel cancer hallmark. We recently showed that epithelial ovarian cancer (EOC) possesses an altered MRS-choline profile, characterised by increased phosphocholine (PCho) content to which mainly contribute over-expression and activation of choline kinase-alpha (ChoK-alpha). METHODS: To assess its biological relevance, ChoK-alpha expression was downmodulated by transient RNA interference in EOC in vitro models. Gene expression profiling by microarray analysis and functional analysis was performed to identify the pathway/functions perturbed in ChoK-alpha-silenced cells, then validated by in vitro experiments. RESULTS: In silenced cells, compared with control, we observed: (I) a significant reduction of both CHKA transcript and ChoK-alpha protein expression; (II) a dramatic, proportional drop in PCho content ranging from 60 to 71%, as revealed by (1)H-magnetic spectroscopy analysis; (III) a 35-36% of cell growth inhibition, with no evidences of apoptosis or modification of the main cellular survival signalling pathways; (IV) 476 differentially expressed genes, including genes related to lipid metabolism. Ingenuity pathway analysis identified cellular functions related to cell death and cellular proliferation and movement as the most perturbed. Accordingly, CHKA-silenced cells displayed a significant delay in wound repair, a reduced migration and invasion capability were also observed. Furthermore, although CHKA silencing did not directly induce cell death, a significant increase of sensitivity to platinum, paclitaxel and doxorubicin was observed even in a drug-resistant context. CONCLUSION: We showed for the first time in EOC that CHKA downregulation significantly decreased the aggressive EOC cell behaviour also affecting cells' sensitivity to drug treatment. These observations open the way to further analysis for ChoK-alpha validation as a new EOC therapeutic target to be used alone or in combination with conventional drugs.

Monitoring response to cytostatic cisplatin in a HER2(+) ovary cancer model by MRI and in vitro and in vivo MR spectroscopy.

BACKGROUND: Limited knowledge is available on alterations induced by cytostatic drugs on magnetic resonance spectroscopy (MRS) and imaging (MRI) parameters of human cancers, in absence of apoptosis or cytotoxicity. We here investigated the effects of a cytostatic cisplatin (CDDP) treatment on (1)H MRS and MRI of HER2-overexpressing epithelial ovarian cancer (EOC) cells and in vivo xenografts. METHODS: High-resolution MRS analyses were performed on in vivo passaged SKOV3.ip cells and cell/tissue extracts (16.4 or 9.4 T). In vivo MRI/MRS quantitative analyses (4.7 T) were conducted on xenografts obtained by subcutaneous implantation of SKOV3.ip cells in SCID mice. The apparent diffusion coefficient (ADC) and metabolite levels were measured. RESULTS: CDDP-induced cytostatic effects were associated with a metabolic shift of cancer cells towards accumulation of MRS-detected neutral lipids, whereas the total choline profile failed to be perturbed in both cultured cells and xenografts. In vivo MRI examinations showed delayed tumour growth in the CDDP-treated group, associated with early reduction of the ADC mean value. CONCLUSION: This study provides an integrated set of information on cancer metabolism and physiology for monitoring the response of an EOC model to a cytostatic chemotherapy, as a basis for improving the interpretation of non-invasive MR examinations of EOC patients.

Alternative academic approaches for testing homologous recombination deficiency in ovarian cancer in the MITO16A/MaNGO-OV2 trial.

BACKGROUND: The detection of homologous recombination deficiency (HRD) can identify patients who are more responsive to platinum and poly ADP ribose polymerase inhibitors (PARPi). MyChoice CDx (Myriad) is the most used HRD test in ovarian cancer (OC). However, some limitations of commercial tests exist, because of the high rate of inconclusive results, costs, and the impossibility of evaluating functional resistance mechanisms. PATIENTS AND METHODS: Two academic genomic tests and a functional assay, the RAD51 foci, were evaluated to detect HRD. One hundred patients with high-grade OC enrolled in the MITO16A/MaNGO-OV2 trial and treated with first-line therapy with carboplatin, paclitaxel, and bevacizumab were analyzed. RESULTS: The failure rate of the two genomic assays was 2%. The sensitivity in detecting HRD when compared with Myriad was 98.1% and 90.6%, respectively. The agreement rate with Myriad was 0.92 and 0.87, with a Cohen's κ coefficient corresponding to 0.84 and 0.74, respectively. For the RAD51 foci assay, the failure rate was 30%. When the test was successful, discordant results for deficient and proficient tumors were observed, and additional HRD patients were identified compared to Myriad; sensitivity was 82.9%, agreement rate was 0.65, and Cohen's κ coefficient was 0.18. The HRD detected by genomic assays and residual tumor at primary surgery and stage was correlated with progression-free survival at multivariate analysis. CONCLUSIONS: Results suggest the feasibility of academic tests for assessing HRD status that show robust concordance with Myriad and correlation with clinical outcome. The contribution of the functional information related to the RAD51 foci test to the genomic data needs further investigation.

Enhancing ovarian cancer conventional chemotherapy through the combination with cannabidiol loaded microparticles.

In this work, we evaluated, for the first time, the antitumor effect of cannabidiol (CBD) as monotherapy and in combination with conventional chemotherapeutics in ovarian cancer and developed PLGA-microparticles as CBD carriers to optimize its anticancer activity. Spherical microparticles, with a mean particle size around 25 µm and high entrapment efficiency were obtained. Microparticles elaborated with a CBD:polymer ratio of 10:100 were selected due to the most suitable release profile with a zero-order CBD release (14.13 ± 0.17 µg/day/10 mg Mps) for 40 days. The single administration of this formulation showed an in vitro extended antitumor activity for at least 10 days and an in ovo antitumor efficacy comparable to that of CBD in solution after daily topical administration (≈1.5-fold reduction in tumor growth vs control). The use of CBD in combination with paclitaxel (PTX) was really effective. The best treatment schedule was the pre + co-administration of CBD (10 µM) with PTX. Using this protocol, the single administration of microparticles was even more effective than the daily administration of CBD in solution, achieving a ≈10- and 8- fold reduction in PTX IC50 respectively. This protocol was also effective in ovo. While PTX conducted to a 1.5-fold tumor growth inhibition, its combination with both CBD in solution (daily administered) and 10-Mps (single administration) showed a 2-fold decrease. These results show the promising potential of CBD-Mps administered in combination with PTX for ovarian cancer treatment, since it would allow to reduce the administered dose of this antineoplastic drug maintaining the same efficacy and, as a consequence, reducing PTX adverse effects.

Bonding characteristics of newly developed all-in-one adhesives.

This study evaluated the microtensile bond strength and the interfacial morphology of newer adhesives. The occlusal surfaces of extracted teeth were ground flat for random allocation to four equal groups. Resin composite was bonded to each surface using either Clearfil SE Bond [SEB], Clearfil Protect Bond [PB], G-Bond [GB], or an experimental adhesive, SSB-200 [SSB]. After storage for 24 h in water at 37 degrees C, they were sectioned into beams (cross-sectional area 1 mm(2)) for microtensile bond strength testing (muTBS) at a crosshead speed of 1 mm/min. The load at failure of each was recorded; the data were analyzed by one-way ANOVA and Games Howell tests. The surfaces of the fractured specimens were observed using SEM. For the ultra-morphology of the interface, the occlusal surfaces of four more teeth were prepared as before and a thin layer of flowable resin composite was bonded to each surface using one of the four adhesives. The mean muTBS ranged from 39.68 MPa (GB) to 64.97 MPa (SEB). There were no statistical differences between SEB and SSB, or between PB and GB (p > 0.05). The muTBS of SEB and SSB were significantly greater than that of PB and GB (p < 0.05). SEMs of the fractured surfaces revealed a mixed (cohesive/interfacial) failure. TEM examination highlighted differences in the hybrid layer; SEB had a thicker layer than the others. In conclusion, the newer all-in-one adhesives produced a thin hybrid layer but varied in their bond strengths. The 2-step self-etching adhesives do not necessarily produce higher bond strengths than that of the all-in-one systems.

Ascites interferes with the activity of lurbinectedin and trabectedin: Potential role of their binding to alpha 1-acid glycoprotein.

Trabectedin and its analogue lurbinectedin are effective drugs used in the treatment of ovarian cancer. Since the presence of ascites is a frequent event in advanced ovarian cancer we asked the question whether ascites could modify the activity of these compounds against ovarian cancer cells. The cytotoxicity induced by trabectedin or lurbinectedin against A2780, OVCAR-5 cell lines or primary culture of human ovarian cancer cells was compared by performing treatment in regular medium or in ascites taken from either nude mice or ovarian cancer patients. Ascites completely abolished the activity of lurbinectedin at up to 10nM (in regular medium corresponds to the IC90), strongly reduced that of trabectedin, inhibited the cellular uptake of lurbinectedin and, to a lesser extent, that of trabectedin. Since α1-acid glycoprotein (AGP) is present in ascites at relatively high concentrations, we tested if the binding of the drugs to this protein could be responsible for the reduction of their activity. Adding AGP to the medium at concentration range of those found in ascites, we reproduced the anticytotoxic effect of ascites. Erythromycin partially restored the activity of the drugs, presumably by displacing them from AGP. Equilibrium dialysis experiments showed that both drugs bind AGP, but the affinity of binding of lurbinectedin was much greater than that of trabectedin. KD values are 8±1.7 and 87±14nM for lurbinectedin and trabectedin, respectively. The studies intimate the possibility that AGP present in ascites might reduce the activity of lurbinectedin and to a lesser extent of trabectedin against ovarian cancer cells present in ascites. AGP plasma levels could influence the distribution of these drugs and thus they should be monitored in patients receiving these compounds.

CD3-CD28 costimulation as a means to avoiding T cell preactivation in bispecific monoclonal antibody-based treatment of ovarian carcinoma.

One of the major limitations to the immunotherapy of ovarian carcinoma based on the use of anti-CD3/antitumor bispecific monoclonal antibodies (bi-mAb) is the need for preactivation of effector cells ex vivo, because cross-linking of the T cell receptor-CD3 complex per se may lead to T-cell unresponsiveness or even apoptosis. The bi-mAb OC/TR, which recognizes the folate-binding protein (FBP) overexpressed in 90% of ovarian carcinomas and the CD3 molecule on T cells, has demonstrated efficacy in a clinical setting. Here we investigated the possibility of delivering accessory signals to OC/TR-retargeted peripheral blood mononuclear cells (PBMCs) via an anti-CD28 mAb or an anti-FBP/anti-CD28 bi-mAb. Coculture of resting PBMCs from healthy donors with OC/TR, anti-FBP/anti-CD28 bi-mAb, and FBP+ tumor cell lines resulted in a highly activated phenotype of effector cells and in a dramatic in vitro growth inhibition of the target cells without an increase in OC/TR-redirected lysis. Whereas both the CD4 and CD8 T cell subsets were involved in the growth inhibition, only the CD8 subpopulation accounted for the cytotoxic activity. The in vitro tumor growth inhibition was mediated mainly by soluble factors, which were active on both FBP+ and FBP- ("bystander effect") cell lines. Activation and antitumor activity were also observed, albeit to a lesser extent, using OC/TR and monospecific bivalent anti-CD28 mAb. In vitro analysis demonstrated that cross-linking between tumor and effector cells for at least 24 h was needed to achieve T-cell activation and development of antitumor activities. Thus, ex vivo CD3-CD28 costimulation on resting PBMCs might be of therapeutic utility for local treatment of minimal residual disease.

Chimeric murine-human antibodies directed against folate binding receptor are efficient mediators of ovarian carcinoma cell killing.

The MOv18 (gamma 1, kappa) and MOv19 (gamma 2a, kappa) murine monoclonal antibodies (MAbs) recognize different epitopes on the human folate binding receptor which is overexpressed on 90% of nonmucinous epithelial ovarian tumors. A chimeric murine-human (human gamma 1, kappa) version of both antibodies was constructed and expressed. The genes encoding the murine heavy and light chain variable regions of the MOv18 and MOv19 MAbs were cloned from the parental hybridomas, fused with genes encoding the human heavy (gamma 1) and light (kappa) chain constant regions, respectively, and expressed in the SP2/0 murine myeloma cell line. Using human peripheral blood mononuclear cells as effector cells and conditions that provide for maximum lysis (effector target = 50:1, saturating antibody concentration), the murine MOv18 MAb (IgG1) mediated variable levels of specific cytolysis of the target ovarian cancer cell line IGROV1. In contrast, the chimeric MOv18 MAb mediated higher and more consistent lysis even at a 10-100-fold lower antibody concentration. The murine MOv19 MAb (IgG2a) mediated specific lysis of IGROV1 cells, and the chimeric version of this antibody mediated an amount of lysis at least equal to that mediated by its murine counterpart. A comparison of the ED50 values obtained for the murine MOv19 and chimeric MOv19 antibodies indicates that the chimeric MOv19 MAb was 3 to 10 times more potent than the murine MOv19 antibody. In addition, the ED50 values obtained for the chimeric MOv18 and chimeric MOv19 MAbs were similar, indicating that these MAbs are equally potent. The level of maximal lysis obtained was dependent on the number of target molecules/cell; the same high level of lysis mediated by cMOv18, MOv19, and cMOv19 was observed with both IGROV1 and OvCA432 target cells. However, only low levels of lysis were obtained when the SW626 cell line, which expresses 1 x 10(4) folate binding protein sites/cell, was used as a target. An equimolar mixture of the chimeric MOv18 and MOv19 MAbs was no more effective in the mediation of lysis than an equivalent amount of either chimeric MAb alone. These data suggest that the folate binding receptor is expressed on IGROV1 cells at a density sufficient to provide for optimal levels of antibody-mediated lysis using a single chimeric antibody directed at the folate binding receptor.

Panning phage antibody libraries on cells: isolation of human Fab fragments against ovarian carcinoma using guided selection.

The display of repertoires of human antibody (Ab) fragments on filamentous phages and selection by binding of the phage to antigen (Ag) have provided a ready means of deriving human Ab against purified Ag. However, it has been more difficult to obtain phage Ab against an individual Ag of a complex mixture, such as cell surface Ag. Using the technique of "guided selection," we generated human Ab against the high-affinity folate-binding protein (FBP), a cell surface Ag that is overexpressed in many human ovarian carcinomas. The guiding Ab template was provided by the light chain of mouse monoclonal Ab Mov19 (K[aff], 10[8] M[-1]) directed against FBP; this was paired with repertoires of human heavy chains displayed on phages, and the hybrid Ab fragments were selected by binding to an ovarian carcinoma cell line (OVCAR3). The selected human heavy chains were then paired with repertoires of human light chains. Further panning led to the isolation of a human Fab fragment, C4, with a binding affinity of 0.2 x 10(8) M(-1). This was highly specific for FBP, as demonstrated by ELISA and flow cytometry data and by immunoprecipitation of the relevant molecule from the cell surface of ovarian carcinoma cells. Moreover, C4 targeted the same or a closely related epitope of the Ag, as did the template rodent monoclonal Ab Mov19. These results suggest the usefulness of guided selection as a simple means to deriving human Ab against cell surface Ag for which a rodent Ab is available.

Human T-lymphocytes targeted against an established human ovarian carcinoma with a bispecific F(ab')2 antibody prolong host survival in a murine xenograft model.

A bispecific F(ab')2 fragment with anti-CD3 and antitumor specificity was used to target activated human peripheral blood lymphocytes (PBL) against OVCAR-3 human ovarian carcinoma cells growing i.p. in athymic mice. Mice were given injections of OVCAR-3 cells on day 0 and treated with i.p. injections of activated PBL coated with the [anti-CD3 (TR66) x antitumor (MOv18)] bispecific F(ab')2 on day 4, using an approximate effector:target ratio of 1:1. Treatment was evaluated for the ability either to block tumor growth at 15 days or to prolong survival of tumor-bearing mice. After 15 days, the incidence of mice with tumor growth was 20% among those given PBL coated with bispecific F(ab')2, whereas the incidence among mice untreated or treated with PBL alone or PBL with either parental antibody ranged from 80 to 94%. The mean survival time of tumor-bearing mice treated with PBL and bispecific F(ab')2 was 104 days, which was 3.5 times that of untreated mice and twice that of mice given PBL alone or PBL with either parental antibody. These results provide support for the concept that treatment of ovarian cancer patients with targeted T-cells could prove beneficial.

Transfer of chimeric receptor gene made of variable regions of tumor-specific antibody confers anticarbohydrate specificity on T cells.

The antitumor specificity of T cells can be induced by gene transfer using a recently developed therapeutic approach (T body). In this work, we genetically conferred anticarbohydrate specificity onto T cells using the variable regions of monoclonal antibody MLuC1, which binds the Lewis(Y) (LeY) tumor-associated antigen that is overexpressed on several human carcinomas. The variable regions of MLuC1, which are in a single-chain Fv (ScFv) configuration, were cloned and spliced in a eukaryotic expression vector with both the gene encoding the signal-transducing gamma-chain of the human Fcgamma receptor and a flexible hinge domain. The chimeric ScFv-gamma gene was expressed in a murine cytotoxic T-cell hybridoma. Transfectants receiving vector only served as a negative control (mock). Screening for functional transfectants was carried out using a tumor growth inhibition assay. The soluble form of MLuC1 ScFv was recovered from bacteria periplasm and tested for binding to LeY-expressing cells by the fluorescence-activated cell sorter analysis. Despite the low binding ability of the soluble MLuC1 ScFv, 7 of 13 genetically engineered cytotoxic T lymphocyte clones inhibited the growth of LeY-positive cells and did not affect growth of LeY-negative cells. None of the mock clones tested specifically inhibited tumor growth. These data indicate that, by chimeric MLuC1 ScFv-gamma gene transfer, it is possible to confer anticarbohydrate specificity onto T cells and extend the applicability of the T-body approach to tumor-associated antigens that are naturally not recognized by T cells.

Targeting of anti-tumor responses with bispecific antibodies.

T cells can be induced to specifically lyse tumor cells with bispecific antibodies containing anti-T cell receptor mAbs crosslinked to anti-tumor mAbs. Such "targeted cytolysis" requires that the target cell be bound directly to the cytotoxic cell. In addition, targeted T cells mediate a second activity, the secretion of factors that can block the growth of both tumor target cells and bystander tumor cells. When given to nude mice bearing intraperitoneal human ovarian carcinoma, targeted human T cells cause the rapid removal of most tumor cells from the peritoneum, and markedly prolong the times of survival of treated mice. The efficacy of targeted T cells for treating human cancer is currently being tested in clinical trials.

Preparation and biological characterization of conjugates consisting of ricin and a tumor-specific non-internalizing MAb.

MOv18, a non-internalizing monoclonal antibody (MAb) with restricted tumor specificity, was conjugated to ricin toxin (RT). According to their ability to bind to galactose residues of Sepharose 6B, the immunoconjugates were fractionated into Bound and Unbound MOv18-RT. The two conjugates could be distinguished by SDS-PAGE, in vivo toxicity and agglutination capability. When the binding activity of both fractions was compared by solid-phase RIA to that of native MAb, it proved to be similar on the relevant target cells but significantly increased on the non relevant cells. On the latter, galactose totally cancelled the binding of the Unbound immunoconjugate, whereas it could only partially reverse that of the Bound MOv18-RT. By in vitro cytotoxic activity, either in the presence or absence of galactose, only a slight selectivity for relevant versus non-relevant target cells was observed for both conjugates. It seems that in the presence of a MAb which is incapable of internalization, the conjugate cytotoxicity could only be attributed to RT, with a loss of the MAb's specificity.

The effect of human serum on the binding activity of radiolabelled monoclonal antibodies.

Three murine monoclonal antibodies (MoAbs), MBr1 and MOv2 of IgM isotype and MOv8 of IgG isotype, with restricted reactivity for breast or ovarian carcinomas, were labelled with 125I in the perspective of obtaining specific and stable radioimmunopharmaceutical reagents. The radiolabelled MoAbs were analyzed for their "in vitro" stability in human blood. They were incubated at 37 degrees C for various lengths of time in human or, as a control, in murine blood and their binding capacity was evaluated by solid-phase RIA and compared with that obtained after incubation with buffer. In human blood, serum and plasma, but not with other components such as erythrocytes, leukocytes, HSA and IgG, the MoAbs revealed a loss of binding reactivity which was marked and constant for the IgM MoAbs, and only occasional for the IgG MoAb. In murine serum the decrease was not so rapid. The same change in the binding capacity was observed when the MoAbs were labelled with 3H or 35S, excluding the involvement of dehalogenating mechanisms. In the perspective of using MoAbs for intracavity therapy the effect of ascitic or pleural fluids on their binding activity was also evaluated. The inhibition of the binding reactivity was not as evident and was not related to the MoAb isotype.

Change in binding reactivity of an anti-tumor monoclonal antibody after the introduction of 2-pyridyl disulphide groups.

The use of monoclonal antibodies for in vivo therapeutic approaches depends largely on their specificity. During the characterization of ricin A-chain-murine monoclonal antibody conjugates we found that the binding specificity of a monoclonal antibody raised against human ovarian carcinoma (MOv2) seemed altered. Therefore, the binding reactivity of the unmodified antibody (MOv2), the conjugation intermediate (MOv2-PDP) and the conjugate (MOv2-A chain) was titrated by solid-phase radioimmunoassay on 11 human tumor cell lines belonging to seven different histotypes. The three reagents bound with the two reference cell lines (SW626:ovary carcinoma and HT-29:colon carcinoma). The MOv2-PDP and the Mov2-A chain also reacted with seven other cell lines which were unreactive with the unmodified MOv2. In addition MOv2-PDP exhibited reactivity on all normal cells tested (peripheral blood lymphocytes and skin fibroblasts). To elucidate the significance of these findings the following experiments were performed: cross inhibitions between the unmodified and modified monoclonal antibodies; comparative absorption tests with different cell lines; and immunoblotting analysis of the target antigens. The results suggest that after chemical modification with SPDP the monoclonal antibody MOv2 increases its binding activity, so that even a low number of antigenic sites can be detected. This study underlines the need to redefine the specificity of a conjugate before considering therapeutic applications.

Retargeting of human lymphocytes against human ovarian carcinoma cells by bispecific antibodies: from laboratory to clinic.

We have selected a monoclonal antibody (MOv18) reactive with ovarian carcinoma, which exhibits a restricted tumor specificity, a high affinity constant and which recognizes a 38-kDa glycoprotein homogeneously expressed on the cell surface of 90% of human ovarian carcinomas. Localization studies with radiolabelled MOv18 showed that MOv18 could specifically reach ovarian carcinoma cells growing in the peritoneal cavity of nu/nu mice. MOv18 did not mediate antibody-dependent cellular cytotoxicity via Fc and could not be used as a carrier for toxins due to poor internalization of the antigen-antibody complex. Bispecific F(ab')2 antibodies made with MOv18 and an antibody reactive with CD3 were able to redirect the cytotoxicity of peripheral blood lymphocytes (PBL) against ovarian carcinoma cells both in vitro and in vivo in an animal model. The treatment of athymic mice bearing a human ovarian carcinoma intraperitoneally, with human PBL coated with the bispecific F(ab')2, significantly prolonged survival of the animals compared with tumour-bearing untreated and control mice treated with PBL alone. Four ovarian cancer patients have been treated with autologous lymphocytes coated with this bispecific F(ab')2 in a phase I clinical trial. No serious side-effects were observed but patients developed human anti-murine antibodies mainly directed against the idiotype of MOv18. We have now begun phase II clinical trials in ovarian cancer patients.

Possible targets on carcinoma for bMAb retargeting of lymphocyte or drug cytotoxicity.

Retargeting of drugs or lymphocyte cytotoxicity through bi-specific monoclonal antibodies (bMAbs) has been proven a therapeutic tool against human carcinoma both in pre-clinical in vitro and in vivo studies. Some of these reagents have already been introduced into clinical trials and preliminary results appear to be promising. However, improvement of the specificity of this approach could be achieved by selecting more suitable target molecules on tumor cells. The research focused on developing MAbs directed against molecules with tumor-restricted distribution and homogeneous expression. Cell-membrane receptors for nutrients or growth factors which operationally represent tumor-specific molecules due to their overexpression, could be considered appropriate targets. Several bMAb anti-nutrients (the folate-binding protein) or growth-factor receptor (c-erbB1, c-erbB2)/anti-triggering molecules have been generated and all were able to efficiently retarget the relevant population of lymphocytes on tumor cells. In order to design a more reliable and selective therapeutic tool, the following parameters were analyzed: correlation between cytotoxicity and antigen level, potency and possible modulation of the target molecule.

Reversion of transformed phenotype in ovarian cancer cells by intracellular expression of anti folate receptor antibodies.

The alpha-folate receptor (FR) is selectively overexpressed in 90% of nonmucinous ovarian carcinomas, whereas no expression is detectable in normal ovarian surface epithelium (OSE). Indirect evidence suggests that FR expression is associated with tumor progression and affects cell proliferation. To evaluate better the role of FR, we developed an approach based on intracellular expression of single-chain (sc) antibodies (intrabody) to downmodulate membrane expression of FR in ovary cancer cells. IGROV-1 and SKOV3 ovarian carcinoma cell lines were transfected with an anti-FR intrabody. Transfectants and parental cells were tested for FR, integrins and anti-FR intrabody expression by fluorescence-activated cell sorting (FACS), reverse transcription and polymerase chain reaction (RT-PCR) and/or immunoblotting. Cell growth characteristics and adhesion properties were evaluated in liquid, semisolid and organotypic cultures. The anti-FR scFv inhibited FR expression from 60 to 99%. At physiological concentrations of folate, proliferation varied directly as a function of FR expression. FR downmodulation was accompanied by reduced colony-forming ability in soft agar, morphological change of the cells, significant enhanced adhesion to laminin or Matrigel, a two- to three-fold increase in alpha6beta4 integrin expression, and a marked reduction in laminin production. In three-dimensional organotypic cultures, anti-FR intrabody-transfected IGROV1 cells grew as a single-ordered layer, reminiscent of normal OSE growth in vivo. In conclusion, the anti-FR intrabody reverses the transformed phenotype in ovary cancer cells and may provide an efficient means to inhibit selectively the growth of these cells.