A mechanistic approach to anticancer therapy: targeting the cell cycle with histone deacetylase inhibitors.

The activity of genes encoded by the highly-condensed DNA in cellular nuclei must be precisely regulated. Regulation of the accessibility of gene promoters to transcription complexes is one level of gene regulation and is influenced by histone tail modifications such as acetylation, methylation, and phosphorylation. Acetylation is a reversible modification catalyzed by histone acetyl transferase (HAT) and histone deacetyltransferase (HDAC) enzymes. Histone deacetylation is associated with transcriptional repression of genes, as the removal of acetyl groups from lysine residues allows for tighter electrostatic interactions between DNA and histones, limiting accessibility of the DNA for transcription. Inhibition of HDAC activity permits histones to remain in an acetylated state, and through the resulting alterations in gene regulation, inhibits cell cycle progression, inhibits differentiation, and in some cases induces apoptosis. Inhibition of proliferation by HDAC inhibitors is characterized by arrest at the G1 or G2/M phases of the cell cycle. Many types of tumor cells then undergo programmed cell death. Exposure to HDAC inhibitors may also allow reactivation of tumor suppressor genes which had been silenced by hypoacetylation during tumorigenesis. HDAC inhibitors from a number of chemical classes have shown promise as anti-cancer agents in animal studies and early clinical trials. The development of HDAC inhibitors which specifically target HDAC isozymes, and more detailed understanding of their anti-neoplastic actions, heralds a new epigenetic antitumor therapeutic strategy.

Retinoic acid receptor-independent mechanism of apoptosis of melanoma cells by the retinoid CD437 (AHPN).

Retinoic acid (RA) induces differentiation of S91 melanoma cells through activation of RA receptor (RAR)gamma without affecting cell viability. The novel RARgamma-agonist CD437 (AHPN), however, also induces concomitant apoptosis through an unknown mechanism which was investigated here. By utilizing DNA microarray analysis, five apoptosis-associated, CD437-induced transcripts (CITs) were identified. Interestingly, all CITs are also regulated by p53 in a DNA damage response, and consistent with this interpretation, CD437 was found to cause DNA adduct-formation. However, p53 is not required for CD437-dependent regulation of CITs. Among this set of genes, induction of p21(WAF1/CIP1) is likely to be responsible for early S-phase growth-arrest of CD437-treated cells, whereas ei24 is a critical mediator of CD437-induced apoptosis in S91 cells. These data suggest an RAR-independent mechanism in which CD437 causes DNA adduct-formation, resulting in induction of a p53-independent DNA damage response, and subsequent growth-arrest and apoptosis. CD437-mediated DNA adduct-formation may also explain its apoptotic effects in other cell types.

Redox control of retinoic acid receptor activity: a novel mechanism for retinoic acid resistance in melanoma cells.

Retinoic acid (RA) slows growth and induces differentiation of tumor cells through activation of RA receptors (RARs). However, melanoma cell lines display highly variable responsiveness to RA, which is a poorly understood phenomenon. By using Northern and Western blot analyses, we show that RA-resistant A375 and RA-responsive S91 melanoma cells express comparable levels of major components of RAR-signaling pathways. However, A375 cells have substantially higher intracellular reactive oxygen species (ROS) levels than S91 cells. Lowering ROS levels in A375 cells through hypoxic culture conditions restores RAR-dependent trans-activity, which could be further enhanced by addition of the antioxidant N-acetyl-cysteine. Hypoxia also enhances RAR activity in the moderately RA-responsive C32 cells, which have intermediate ROS levels. Conversely, increasing oxidative stress in highly RA-responsive S91 and B16 cells, which have low ROS levels, by treatment with H(2)O(2) impairs RAR activity. Consistent with these observations, RA more potently inhibited the proliferation of hypoxic A375 cells than that of normoxic cells. Oxidative states diminish, whereas reducing conditions enhance, DNA binding of retinoid X receptor/RAR heterodimers in vitro, providing a molecular basis for the observed inverse correlation between RAR activity and ROS levels. The redox state of melanoma cells provides a novel, epigenetic control mechanism of RAR activity and RA resistance.

Effects of retinoic acid and sodium butyrate on gene expression, histone acetylation and inhibition of proliferation of melanoma cells.

Retinoic acid (RA) induces growth-arrest of many tumor cell lines but it is an ineffective therapeutic against melanoma. We investigated whether the histone deacetylase (HDAC)-inhibitor sodium butyrate (BUT) can restore or potentiate the RA-response of RA-resistant human A375, and RA-responsive S91 murine melanoma cells. BUT induced expression of RARbeta and p21(waf1/cip1) mRNA in A375 cells but in S91 cells only p21(waf1/cip1) was induced. RA and BUT synergistically activated transcription of an RA-dependent reporter gene in S91, but not A375 cells. BUT increased histone H4-acetylation in both cell types. RA potentiated BUT-mediated inhibition of S91 cell proliferation, whereas A375 cells remained largely resistant to both compounds. HDAC-inhibitors may enhance the activity of RA on RA-responsive melanoma cells.

B7.1 costimulation increases T-cell proliferation and cytotoxicity via selective expansion of specific variable alpha and beta genes of the T-cell receptor.

BACKGROUND: Optimal T-cell activation requires not only ligation of the T-cell receptor (TcR) but also delivery of costimulatory signals by various accessory molecules. The interaction of the costimulatory molecule B7.1 (CD80) with its receptor CD28 provides a strong positive signal to T cells. METHODS: The B7.1 gene was transduced into cultured human ovarian, breast, and pancreatic tumor cells by using a retroviral vector. Autologous as well as allogeneic naive T-cells were stimulated with either wild-type or B7.1-transduced tumor cells in a mixed lymphocyte tumor cell culture (MLTC). In addition to cytolytic activity, T-cell proliferation, T-cell subset composition, and the frequencies of TcR variable (V) alpha and beta genes were compared in T cells from both types of MLTC. RESULTS: Introduction of the B7.1 gene into tumor cells was successful in all tumors to a varying degree. Those tumors expressing high levels of B7.1 induced significantly higher levels of T-cell proliferation than wild-type tumor cells. T-cell subset composition did not markedly differ between T cells stimulated with wild-type tumor cells or B7.1-expressing tumor cells. However, T cells stimulated with B7.1-expressing tumor cells showed a significantly increased cytolytic potential. The increased cytotoxic T lymphocyte activity was associated with a higher frequency of specific TcR V alpha and V beta genes. In addition, B7.1 costimulation promoted oligoclonality among the responding T cells. CONCLUSIONS: These data suggest that costimulation through B7.1 promotes T-cell proliferation and cytotoxic activity through clonal expansions of T cells bearing antigen-specific TcR V alpha and V beta genes and through promotion of oligoclonality. The data also suggest that promoting B7.1-mediated costimulation is an important aspect of immune therapies.

Soluble MUC1 secreted by human epithelial cancer cells mediates immune suppression by blocking T-cell activation.

Solid tumors may secrete factors that mediate immune suppression in patients. We investigated the effect of supernatants from 25 human tumor cell lines on T-lymphocytes from healthy donors. A profound inhibition of proliferation, cytokine secretion and cytotoxic activity was seen when T-cells were cultured in concentrated tumor supernatants from 6 cell lines fractionated into high (>100 kDa) m.w. molecules. Interestingly, the inhibitory effects were reversed when the tumor supernatant was removed. Cell cycle studies of inhibited T-cells showed most of them were growth arrested in the G(0)/G(1) phase similar to naïve T-cells. In addition, these T-cells did not express IL2-receptors and expression of CD54 (ICAM-1) and CD58 (LFA-3) resembled that of resting T-cells. Protein gel electrophoresis of the tumor supernatants and western blot analysis demonstrated the presence of soluble MUC1 in the inhibitory tumor supernatants but not in control supernatant. Most importantly, depletion of soluble MUC1 by immunoprecipitation from the tumor supernatants neutralized the inhibitory effects on T-lymphocytes. Therefore, our results show that MUC1 shed by cultured epithelial tumor cells mediates inhibition of T-cell proliferation and function by inducing cell growth arrest.

Pancreatic cancer cells can evade immune surveillance via nonfunctional Fas (APO-1/CD95) receptors and aberrant expression of functional Fas ligand.

BACKGROUND: The Fas (APO-1/CD95) receptor/Fas ligand (FasR/FasL) system plays a key role in immune surveillance. We investigated the possibility of a tumor escape mechanism involving the FasR/FasL system in pancreatic cancer cells. METHODS: Fourteen pancreatic cancer cell lines and 3 pancreatic cancer surgical specimens were studied for their expression of FasR and FasL by flow cytometry, immunoblotting, and immunohistochemistry, FasR function was tested with an anti-FasR antibody. FasL function was assessed by coculture assays using pancreatic cancer cells and FasR-sensitive Jurkat T-cells. RESULTS: FasR was expressed in normal pancreas, in 14 of 14 pancreatic cancer cell lines, and in 3 of 3 surgical specimens. However, only 1 of 14 cancer cell lines expressed functional FasR when grown in monolayer, although 3 additional cell lines displayed functional FasR when cultured in suspension. Normal pancreas did not express FasL, whereas 14 of 14 cancer cell lines and 3 of 3 surgical specimens expressed FasL. FasL expressed by pancreatic cancer cells mediated killing of Jurkat T-cells in coculture assays (P < .005). CONCLUSIONS: These data suggest that pancreatic cancer cells have 2 potential mechanisms of evading Fas-mediated immune surveillance. A nonfunctional FasR renders them resistant to Fas-mediated apoptosis. The aberrant expression of functional FasL allows them to "counterattack" activated Fas-sensitive T-cells. Alone or in unison, these tumor escape mechanisms may contribute to the malignant and often rapid course of pancreatic cancer disease.

T cells mediate treatment of six-day-old cytokine-gene-transfected mouse mammary tumor.

We have previously shown that immunologically different mouse mammary cancer cell lines induce antitumor responses after IL2 or IL4 gene transfection. We now report the ability of cytokine-transfected tumors to induce eradication of established wild-type tumor. Animals with 6-day-old tumor treated with IL2-transfected cells also had significantly smaller tumors 2.8 and 1.7 cm2 (EMT6 and 410.4). Findings were similar for IL4-transfected cells. Tumor infiltrating lymphocytes or cells from draining lymph nodes demonstrated tumor-specific in vitro cytotoxicity. Immunohistochemical studies revealed T cell infiltrates in transfected tumors.

Ligand-inducible retinoid X receptor-mediated protein: DNA interactions in the retinoic acid receptor beta2 gene promoter in vivo.

Retinoid X receptors (RXRs) are recently characterized transcription factors that are members of the nuclear hormone receptor superfamily. However, it is not known whether the endogenous RXR complex requires its ligand for access to its hormone response element (HRE) of a target gene in vivo. Hence, dimethyl sulfate-based genomic footprinting was carried out to examine occupancy of HREs in the retinoic acid (RA) receptor beta2 (RARbeta2) gene promoter in the murine melanoma cell line S91 cultured in the absence or presence of T3, all-trans-RA (atRA), or CD2624, an RXR-selective retinoid. No footprint was observed at the RA-response element (betaRARE) in the absence of ligands. However, a footprint was detected at the betaRARE and other cis-acting elements after a 6 h incubation with CD2624 and atRA. Interestingly, only the betaRARE was footprinted after 60 min incubation with CD2624. These results suggest that the endogenous RXR complex can interact with an HRE of a target gene in the presence of ligand, and subsequently may initiate additional interactions between DNA and other transcription factors.

Clone 10d/BM28 (CDCL1), an early S-phase protein, is an important growth regulator of melanoma.

Retinoic acid (RA) induces growth arrest and differentiation of many different tumor cells. RA activates RA receptors, which function as ligand-dependent transcriptional modulators. S91 murine melanoma cells stop proliferating and then reversibly differentiate into a melanocytic cell type after the administration of RA. The genetic changes that take place during this process serve as an excellent model for the etiology of melanoma. The use of subtractive hybridization techniques yielded several differentially expressed cDNAs that are associated with RA-induced growth arrest. One clone, cyclin D1, is repressed and is probably a differentiation marker. Two other cDNAs represent novel, RA-inducible genes. Expression of another cDNA, clone 10d, is strongly down-regulated. It is the homologue of the human gene BM28 (CDCL1) that is indispensable for entry into S phase and cell division. S91 cells that are permanently transfected with a plasmid that constitutively expresses clone 10d become significantly more resistant to RA, suggesting that repression of this gene is a critical event in RA-induced growth arrest. The use of reverse transcription-PCR for the detection of expression in human melanoma in vitro was performed to study the potential role of clone 10d/BM28 in this disease. It is expressed in 80% of melanoma cell lines but is virtually undetectable in primary melanocytes. The expression of BM28 is not regulated by RA in human, RA-resistant melanoma cells. These results suggest that clone 10d/BM28 functions as an important tumor cell growth promoter. The regulation of clone 10d may be directly mediated by RA receptors, and escape from negative regulation may, thus, contribute to the etiology of melanoma.

Specific activation of retinoic acid receptors (RARs) and retinoid X receptors reveals a unique role for RARgamma in induction of differentiation and apoptosis of S91 melanoma cells.

Retinoic acid (RA) and 9-cis-RA induce growth arrest and differentiation of S91 melanoma cells. RA activates retinoic acid receptors (RARs), whereas 9-cis-RA activates both RARs and retinoid X receptors (RXRs). Both classes of receptors function as ligand-dependent transcription factors. S91 melanoma cells contain mRNA for RXRalpha, RXRbeta, RARalpha, RARgamma, and RARbeta in low levels. Among these, only RARbeta gene transcription is induced by retinoids. However, at present the individual role(s) for each RXR and RAR isoform in these processes is unclear. We assessed the function of all isoforms in the S91 melanoma model by using RXR and RAR isoform-specific retinoids to study their effects on cell growth, RARbeta expression, and differentiation. Activation of each of the endogenous RXR or RAR isoforms induces RARbeta gene expression, and blocks cellular proliferation. However, only the RARgamma-ligands cause additional differentiation toward a melanocytic phenotype, which coincides with substantial apoptosis well before morphological changes are apparent. Apoptosis is completely dependent on de novo protein synthesis but cannot be induced by changes in activities of AP-1, protein kinase C, and protein kinase A, nor can it be blocked by the presence of the antioxidant glutathione. These results argue against a specific role for RARbeta, but suggest that RARgamma has a critical role in a genetic switch between melanocytes and melanoma, and induction of ligand-dependent apoptosis.

Repression of glucocorticoid receptor-mediated transcriptional activation by unliganded thyroid hormone receptor (TR) is TR isoform-specific.

Two genes, c-ErbA alpha and c-ErbA beta, generate at least three functional T3 receptor (TR) isoforms in the rat: TR alpha-1, TR beta-1, and TR beta-2. The latter is an N-terminal splice variant of TR beta-1 whose expression is high in the pituitary gland, whereas the other isoforms are more widely expressed. It is believed that TR beta-2 might play an important role in the pituitary, but no specific biological activities have been defined. Using in vitro translated receptors, we were unable to detect striking isoform-specific differences in T3 binding, DNA binding, homodimerization and heterodimerization activities with retinoid X receptor in the electrophoretic mobility assay, or T3-dependent repression and activation of basal transcription in transfection assays. The N-terminus of TR beta is completely dispensible for these activities. However, we found that unliganded TR alpha-1 and TR beta-1, but not TR beta-2, can repress glucocorticoid receptor-mediated transcriptional activation if the reporter gene promoter contains binding sites for both receptor species. TR beta-2 also synergizes most efficiently with the glucocorticoid receptor in the presence for the of T3. The TR beta-2-specific N-terminus is required for these effects. This result indicates that the relative abundance of specific TR isoforms may determine the with the glucocorticoid receptor in the presence of T3. The TR beta-2 specific N-terminus is required for these effects. This result indicates that the relative abundance of specific TR isoforms may determine the quantitative response of a gene that is regulated by T3 and glucocorticoids. In particular, pituitary TR beta-2 may be an important mediator of GH gene expression, which is regulated by both of these hormones.

Evidence that retinoid X receptors mediate retinoid-dependent transcriptional activation of the retinoic acid receptor beta gene in S91 melanoma cells.

S91 melanoma cells are growth arrested and differentiate when treated with retinoids. These processes correlate with expression of the retinoic acid receptor (RAR) beta gene, which is induced through a retinoic acid response element (beta RARE). We wished to determine which endogenous retinoid receptors (RARs and retinoid X receptors, RXRs) mediate induction of the RAR beta gene. We show that RXR alpha and RXR beta are constitutively expressed. Electrophoretic mobility shift assays with nuclear extracts show specific binding to the beta RARE (Complex I) in untreated cells, which can be supershifted by antibodies against RXRs but not by anti-RAR antibodies. After 48 h of treatment with retinoic acid, Complex I is replaced by a faster migrating Complex II, which can be supershifted by anti-RAR beta and anti-RXR alpha antibodies. This suggests that induction of the RAR beta gene is largely mediated by RXRs only. Accordingly, we also find that 9-cis RA, which activates both RAR and RXR, is a more potent inducer of the RAR beta gene than RA, which only activates RAR. After 48 h, all RXRs appear to be titrated by the newly synthesized RAR beta into an RAR beta.RXR heterodimer complex. Thus, it appears that the beta RARE is sequentially occupied by RXR dimers and RAR-RXR heterodimers.

Half-site arrangement of hybrid glucocorticoid and thyroid hormone response elements specifies thyroid hormone receptor complex binding to DNA and transcriptional activity.

Thyroid hormone receptors bind to thyroid hormone response elements (TREs) as heterodimers with 3,5,3'-L-triiodothyronine (T3) receptor auxiliary protein (TRAP) and retinoid X receptors (RXRs). Currently, it is not known whether TR/TRAP or TR/RXR heterodimers need to bind to both TRE half-sites and whether there is a preferred orientation for TR/RXR heterodimer binding to TREs or transcriptional activation. Accordingly, we created a mutant TR alpha (TR-P box) by changing 3 amino acids in the P box region of the first zinc finger of the DNA-binding domain to that of the glucocorticoid receptor (GR), and we examined wild-type TR alpha and TR-P box complex binding to hybrid response elements containing TRE and glucocorticoid receptor element (GRE) half-sites arranged as a direct repeat with a four-nucleotide gap. TR-P box/RXR heterodimers selectively bound to the hybrid response elements in which GRE half-site was the downstream half-site, whereas TR alpha/RXR bound to hybrid response elements in which GREs were in either position. Additionally, TR/TRAP or TR/RXR heterodimer required two half-sites for binding to DNA, with strong binding to at least one of the half-sites. Last, co-transfection assays and methylation interference studies using the hybrid response elements suggest that the sequential arrangement of strong and weak half-sites in the TRE may be a critical determinant of TR/RXR heterodimer binding and transcriptional activation.

Rat Rev-erbA alpha, an orphan receptor related to thyroid hormone receptor, binds to specific thyroid hormone response elements.

Rat Rev-erbA alpha (rRev), which is related to thyroid hormone receptor (TR), is a conserved member of the nuclear hormone receptor superfamily whose physiological roles are unknown ("orphan" receptor). We studied DNA binding of rRev in vitro by electrophoretic mobility shift assay. A fusion protein was constructed, called NGR.Rev, containing part of the N terminus of the glucocorticoid receptor fused to nearly full-length rRev. Inasmuch as rRev and TR share homology in their DNA-binding domains, we tested binding to three different thyroid hormone response elements (TREs) in which the half-sites are arranged in different orientations. NGR.Rev bound direct repeats (DR4), but not palindromic (TREpal) or inverted palindromic (F2H) repeats. Also, transfection of CV1 cells with a reporter gene containing the luciferase gene under control of the inducible thymidine kinase promoter resulted in an increase in luciferase activity when NGR.Rev was cotransfected and when the thymidine kinase promoter contained DR4. In addition, a series of deletions in the ligand-binding domain of NGR.Rev revealed regions that can modulate DNA binding. Finally, we studied DNA binding of bacterially produced fusion proteins that contain the DNA-binding domains of rRev or rTR alpha fused to glutathione S-transferase, to a panel of natural TREs. Our results indicate that Rev binds DNA with a different specificity than TR alpha-1 and might be involved in the regulation of a subset of thyroid hormone-regulated genes.

Region-specific anti-thyroid hormone receptor (TR) antibodies detect changes in TR structure due to ligand-binding and dimerization.

There are multiple factors that potentially can induce structural changes in DNA-bound thyroid hormone receptors (TRs) including protein-protein interactions, ligand-binding to TRs, and the thyroid hormone response element (TRE) sequence. We used a battery of anti-TR antibodies that recognize the amino-terminal, hinge, or carboxy-terminal regions of TRs to study changes in the epitope regions of in vitro translated TRs in electrophoretic mobility shift assays. We found that the carboxy-terminal and hinge region antibodies recognized TR homodimers but not TR/T3-receptor auxiliary protein or TR/retinoid X receptor heterodimers. The amino-terminal antibodies detected conformational changes due to ligand binding. In contrast, each antibody recognized TR complexes bound to TREs containing half-sites arranged in three different orientations. These results suggest that dimerization with nuclear proteins and ligand-binding, rather than the orientation of TRE half-sites, cause changes in several TR subregions.

Reconstitution of ligand-mediated glucocorticoid receptor activity by trans-acting functional domains.

Glucocorticoid receptors (GRs) are ligand-inducible transcription factors that contain several functional domains. We tested whether GR activity can be reconstituted using domains expressed in separate molecules. Hence, we developed a general approach in which proteins can be individually expressed but interact specifically through the leucine zippers of c-Jun and c-Fos fused to each protein. The GR was divided into two different fragments, one encoding the N-terminal trans-activation and DNA-binding domains and conferring constitutive activity to a glucocorticoid-responsive reporter gene, and one containing the C-terminal, ligand-binding domain. Coexpression of the trans-activation-DNA-binding domain and the ligand-binding domain fragments leads to reconstituted ligand-regulated GR activity that is completely dependent on the presence of compatible zippers. These results suggest that, in GRs and perhaps other members of the steroid/thyroid hormone receptor superfamily, ligand-mediated function does not require that these domains be present in cis, but that they can also function in trans. This, together with the absence of interdomain dimerization signals, also suggests that these domains possibly evolved from separate genes.

Ligand-binding and heterodimerization activities of a conserved region in the ligand-binding domain of the thyroid hormone receptor.

The ligand-binding domain of the thyroid hormone (3,5,3'-triiodothyronine) receptor (TR) contains poorly characterized subdomains involved with ligand binding, transactivation, and protein-protein interactions. The region between residues 288-331 of rat TR alpha-1 was analyzed by modeling and site-directed mutagenesis. Our results suggest that part of this sequence adopts an amphipathic alpha-helical conformation. The integrity of the putative helix is important for 3,5,3'-triiodothyronine binding but not necessarily for heterodimerization with nuclear factor(s). Mutants defective for both activities were found clustered in a region overlapping the C-terminal portion of the helix and further downstream. The sequence conservation of this particular region among the entire superfamily suggests a similar role in dimerization in other receptors.

Frameshift suppression at tandem AGA and AGG codons by cloned tRNA genes: assigning a codon to argU tRNA and T4 tRNA(Arg).

Arginine is coded for by CGN (N = G, A, U, C), AGA and AGG. In Escherichia coli there is little tRNA for AGA and AGG and the use of these codons is strongly avoided in virtually all genes. Recently, we demonstrated that the presence of tandem AGA or AGG codons in mRNA causes frameshifts with high frequency. Here, we show that phaseshifts can be suppressed when cells are transformed with the gene for tRNA(T4Arg) or E. coli tRNA(argU,Arg) demonstrating that such errors are the result of tRNA depletion. Bacteriophage T4 encoded tRNA(Arg) (anticodon UCU) corrects shifts at AGA-AGA but not at AGG-AGG, suggesting that this tRNA can only read AGA. Similarly, comparison of the translational efficiencies in an argU (Ts) mutant and in its isogenic wild type parent indicates that argU tRNA (anticodon UCU) reads AGA but not AGG. An argU (Ts) mutant barely reads through AGA-AGA at 42 degrees C but translation of AGG-AGG is hardly, if at all, affected. Overexpression of argU+ relaxes the codon specificity. The thermosensitive mutant in argU, previously called dnaY because it is defective in DNA replication, can be complemented for growth by the gene for tRNA(T4Arg). This implies that the sole function of the argU gene product is to sustain protein synthesis and that its role in replication is probably indirect.

Expression of the rat interferon-alpha 1 gene in Escherichia coli controlled by the secondary structure of the translation-initiation region.

A synthetic ribosome-binding site (RBS) containing a 7-nucleotide-long Shine-Dalgarno (SD) sequence was placed ahead of the rat interferon (IFN)-alpha 1 coding region. The translational efficiency of this construct was extremely low. Structural probing of transcripts with RNases T1 and U2 combined with computer predictions revealed the presence of a stable hairpin in which the SD region was base-paired to codons 3, 4 and 5 of the IFN mRNA. Each mutation in this stem changing an A-U to an A.C or a G-C a G.U pair increased translational efficiency about fourfold and this effect could be reversed by a compensating stabilizing substitution in the other strand of the stem. We conclude that the strength of an RBS is to a major degree determined by its involvement in secondary structure. We also show that the negative effect of secondary structure on the efficiency of an RBS can be overcome by allowing upstream translation to terminate within the base-paired region. In our clones, termination-dependent restarts occur at a frequency comparable to that taking place in constructs containing destabilized hairpins.