Recently discovered type III IFNs (IFN-λ) exert their antiviral and immunomodulatory activities through a unique receptor complex composed of IFN-λR1 and interleukin-10 receptor 2. To further study type III IFNs, we cloned and characterized mouse IFN-λ ligand-receptor system. We showed that, similar to their human orthologues, mIFN-λ2 and mIFN-λ3 signal through the IFN-λ receptor complex, activate IFN stimulated gene factor 3, and are capable of inducing antiviral protection and MHC class I antigen expression in several cell types including B16 melanoma cells. We then used the murine B16 melanoma model to investigate the potential antitumor activities of IFN-λs. We developed B16 cells constitutively expressing murine IFN-λ2 (B16.IFN-λ2 cells) and evaluated their tumorigenicity in syngeneic C57BL/6 mice. Although constitutive expression of mIFN-λ2 in melanoma cells did not affect their proliferation in vitro, the growth of B16.IFN-λ2 cells, when injected s.c. into mice, was either retarded or completely prevented. We found that rejection of the modified tumor cells correlated with their level of IFN-λ2 expression. We then developed IFN-λ-resistant B16.IFN-λ2 cells (B16.IFN-λ2Res cells) and showed that their tumorigenicity was also highly impaired or completely abolished similar to B16.IFN-λ2 cells, suggesting that IFN-λs engage host mechanisms to inhibit melanoma growth. These in vivo experiments show the antitumor activities of IFN-λs and suggest their strong therapeutic potential. (Cancer Res 2006; 66(8): 4468-77)
- Anticancer activity
- Antiviral cytokines
- Immune response
IFNs are the key cytokines in the establishment of a multifaceted antiviral response. Three distinct types of IFNs (type I, II, and III) are recognized based on their structural features, receptor usage, and biological activities. Although all IFNs are important mediators of antiviral protection, their roles in antiviral defense vary. Type I IFNs (IFN-α/β/ω/ε/κ in humans) possess strong intrinsic antiviral activity and are able to activate a potent antiviral state in a wide variety of cells ( 1). The essential role of the functional type I IFN system in the induction of antiviral resistance is clearly shown in type I IFN receptor knockout mice because such animals are highly susceptible to viral infections ( 2, 3). In contrast, studies with IFN-γ and IFN-γ receptor knockout mice ( 4– 6) and in people bearing inactive variant forms of the IFN-γ receptor ( 7, 8) revealed that the antiviral activity of IFN-γ is not its primary physiologic function. IFN-γ is a Th1 cytokine that stimulates cell-mediated immune responses that are critical for the development of host protection against intracellular parasites and is a part of antiviral and antitumor defenses ( 9– 11). Therefore, both type I and type II IFNs stimulate a variety of innate and adaptive immune mechanisms that contribute to eliminating viral infections ( 1, 9– 13).
IFNs are part of a larger family of proteins that also includes six interleukin-10 (IL-10)-related cytokines: IL-10, IL-19, IL-20, IL-22, IL-24, and IL-26. These diverse cytokines were first grouped into the same family because they all use receptors that share common motifs in their extracellular domains and, therefore, form the class II cytokine receptor family (CRF2). Subsequently, IFNs and IL-10-related cytokines are referred to as CRF2 cytokines. The most recent addition to the CRF2 family, type III IFNs or IFN-λs (also known as IL-28/29), show structural features of the family of IL-10-related cytokines but possess antiviral activity, which defines them as a new type of IFNs ( 1, 14– 16). Antiviral activity of IFN-λs was shown for several viruses in in vitro experiments and also in a mouse model of vaccinia virus infection ( 1, 14, 15, 17).
Three distinct human IFN-λ proteins, IFN-λ1, IFN-λ2, and IFN-λ3, have been characterized and found to bind and signal through a receptor complex composed of the unique IFN-λR1 chain and the IL-10 receptor 2 (IL-10R2) chain, which is shared with the IL-10, IL-22, and IL-26 receptor complexes ( 14, 15, 18– 22). In contrast, all type I IFNs exert their biological activities through a heterodimeric receptor complex composed of IFN-αR1 and IFN-αR2 chains, and IFN-γ engages IFN-γR1 and IFN-γR2 chains to assemble a functional receptor complex. IFNs activate primarily the Jak-signal transducers and activators of transcription (STAT) signal transduction pathway. IFN-γ causes the phosphorylation of Jak1 and Jak2, leading primarily to the mobilization of STAT1. Type I and type III IFNs, acting through distinct receptors, stimulate similar signaling events that include activation of Jak1 and Tyk2 kinases and several latent transcriptional factors of the STAT family including STAT1, STAT2, STAT3, STAT4, and STAT5 ( 15, 23). Activated STAT1, STAT3, and STAT5 form homodimers and heterodimers and bind to an IFN-γ activation site (GAS) in the promoters of IFN-inducible genes. IFN receptor engagement also leads to the activation of the IFN-stimulated gene factor 3 (ISGF3) transcription complex. ISGF3 is composed of STAT1 and STAT2 and IFN regulatory factor 9 (ISGF3 or p48). ISGF3 regulates gene transcription by binding to an IFN-stimulated response element (ISRE). Consequently, biological activities induced by either type I or type III IFNs are also similar and include induction of antiviral protection and up-regulation of MHC class I antigen expression in several cell types ( 15).
Sequences of the murine IFN-λs and their receptor were previously reported by us ( 15, 17). Here we present detailed analysis of the mouse IFN-λ antiviral system and its comparison with the human IFN-λ antiviral system. We show that similar to their human orthologues, mouse IFN-λs possess strong antiviral and immunomodulatory activities. Importantly, for the first time, we show antitumor activity of IFN-λs against B16 melanoma, suggesting that type III IFNs may have therapeutic potential in cancer treatment.
Materials and Methods
Plasmid construction. Based on the sequence of the mouse genome, primers were designed to amplify and clone the appropriate regions of the cDNAs encoding mouse ligands and their receptor into corresponding vectors. These primers contained sequences homologous to coding regions of the corresponding genes and sequences recognized by restriction endonucleases to facilitate cloning of these genes. Mouse IFN-λ2 and IFN-λ3 genes were amplified by PCR with primers 5′-CCGGTACCATGCTCCTCCTGCTGTTGCCTCTGC-3′ (mifnl-2F) and 5′-GAGAATTCCAGGTCAGACACACTGGTCTCC-3′ (mifnl-4R) and mouse genomic DNA from 129/Sv strain, and cloned into the pcDEF3 (pEF) vector ( 24) with the use of KpnI and EcoRI restriction endonucleases, generating plasmids pEF-mIFN-λ2gene and pEF-mIFN-λ3gene. After transfection of these plasmids into COS-1 cells, total RNA was isolated from the transfectants, converted to cDNA, and amplified with mifnl-4R and either mifnl-2F or 5′-CCGGATCCTGTCCCCAGGGCCACCAGGC-3′ (mifnl-6F) primers to obtain mIFN-λ2 and mIFN-λ3 cDNA fragments, which were cloned into either the KpnI and EcoRI restriction sites of the pEF vector or the BamHI and EcoRI restriction sites of the pEF-SPFL vector ( 24), respectively, resulting in plasmids pEF-mIFN-λ2, pEF-mIFN-λ3, pEF-FL-mIFN-λ2, and pEF-FL-mIFN-λ3. Because plasmid pEF-SPFL encodes the IFN-γR2 signal peptide followed by the FLAG epitope ( 24), this abutted the reading frames of the IFN-λs to the frame of the FLAG epitope. Therefore, these plasmids encode mIFN-λs tagged at their NH2 terminus with the FLAG epitope (FL-mIFN-λ2/3). The pEF-mIFN-λ plasmids encode intact mIFN-λs with their own signal peptides.
Genomic DNA from a feral mouse and various mouse strains (129/Sv, C57BL/6, FVB, and CD1) and primers 5′-CCGGTACCATGGCTACAGTGTGCCTGCTGGGT-3′ (mifnl-1F) and 5′-CCGAATTCAGACACACTGGTCTTCACTGGCC-3′ (mifnl-3R) were used for PCR to amplify mouse IFN-λ1 gene fragment. The resulting PCR products were either cloned into pCR2.1 vector (Invitrogen, Carlsbad, CA) and sequenced or subjected to direct sequencing with primers 5′-CCGGATCCTCCTTCCAAGCCCACCCCAACC-3′ (mifnl-5F) and 5′-CCAGAATTCTCCACTGATGAG-3′ (mifnl-10R). Feral mouse IFN-λ1 gene fragment was also amplified by PCR with primers mifnl-1F and 5′-TTCGCTAGCGACACACTGGTCTTCACTGGC-3′ (mifnl-15R) and cloned into the pEF2-X-FL vector ( 25) with the use of KpnI and NheI restriction endonucleases, generating plasmids pEF-mIFN-λ1gene-FL. In this plasmid, the stop codon of the mIFN-λ1 gene encoded in the last exon (exon 5) was deleted and replaced with the FLAG epitope (mIFN-λ1gene-FL).
Mouse IFN-λR1 cDNA was cloned with the use of nested PCR as follows. A first round of PCR was done with primers 5′-GCTGCATCTTCCTAGAGGCTCCAG-3′ (mR12-3F) and 5′-GAGAATGTGTAGATGGACCACCAG-3′ (mR12-8R) using cDNAs obtained from mouse bone marrow cells as template. The first-round PCR product was used as a template for a second round of PCR amplification, with primers 5′-AGCGCTAGCGGCAATGCCCTCACTCTTGCTTC-3′ (mR12-4F) and 5′-GCGAATTCACCTGACCAAGTAATCTCC-3′ (mR12-7R), to generate the extracellular domain of mIFN-λR1, which was subsequently cloned into plasmid pEF3-IL-10R1/γR1 ( 19), using NheI and EcoRI restriction endonucleases to excise a cDNA fragment encoding the IFN-γR1 intracellular domain and replace it with that for mIFN-λR1, resulting in plasmid pEF-IL-10R1/mIFN-λR1. Primers 5′-GCCGCCCTTGTCACCTGCCGC-3′ (mR12-1F), 5′-GGGATGTACATGTGGCGGGCCGACCGGTGG-3′ (mR12-2F), 5′-GCTATCAGTAGAAGCAAGAGTGAG-3′ (mR12-5R), and 5′-GCCTCTAGAGCTCTTTTGTCCCCTGGAGCCTC-3′ (mR12-6R) and the same pool of cDNAs were used for nested PCR (mR12-1F and mR12-5R primers for the first round and mR12-2F and mR12-6R primers for the second round) to amplify a cDNA fragment encoding the mIFN-λR1 extracellular domain. The resulting PCR product was digested with BsiWI and XbaI restriction endonucleases and cloned into KpnI and NheI restriction sites of either plasmid pEF-IL-10R1/mIFN-λR1 or plasmid pEF3-IL-10R1/γR1, resulting in plasmids pEF-mIFN-λR1/λR1 and pEF-mIFN-λR1/γR1, respectively.
The nucleotide sequences of the modified regions of all constructs were verified in their entirety by DNA sequencing. The assigned GenBank accession numbers were as follows: mouse IFN-λ2 and IFN-λ3 cDNAs from 129/Sv strain, AY869695 and AY869696, respectively ( 17); mouse IFN-λR1 cDNA, AY184375 ( 15); and mouse IFN-λ1 gene from 129/Sv and FVB strains, DQ340653 and DQ340654, respectively.
Cells, transfection, flow cytometry, and electrophoretic mobility shift assay. The 16-9 hamster-human somatic cell hybrid line and its derivative 16-9 cell line expressing the human chimeric IFN-λ receptor complex (hIFN-λR1/γR1 + hIL-10R2; ref. 15) were maintained in Ham's F12 medium with 10% fetal bovine serum (FBS). Colorectal adenocarcinoma HT29 cells were maintained in RPMI medium with 10% FBS. COS-1 cells, a SV40-transformed fibroblast-like simian CV-1 cell line, mouse L929 (s.c. connective tissue) cells, NIH 3T3 fibroblast-like cells, B16 melanoma cells, and derivatives were grown in DMEM with 10% FBS. Primary keratinocytes and bone marrow–derived macrophages were generated from C57BL/6 mice as described ( 26, 27). Splenocytes were isolated as follows: spleen was extracted, placed into RPMI medium with 5% FBS, and minced. After gravitational sedimentation of large tissue clumps, the remaining cells in suspension were harvested, pelleted by centrifugation, and used for experiments.
16-9 and COS-1 cells were transfected as described ( 15). COS cell supernatants were collected at 72 hours and used as a source of the expressed proteins. pEF-mIFN-λ2 or pEF-mIFN-λ3 plasmid was transfected into 0.5 × 106 to 1 × 106 B16 cells with the use of transIT-LT1 reagent (Mirus, Madison, WI) and G418-resistant colonies (800 μg/mL) were harvested and tested for the production of mIFN-λs.
To assess the growth (proliferation rate) of parental and transfected B16 cells, the cells were seeded at 0.5 × 105 per well in six-well plates in triplicates, grown at 37°C, and counted after 24, 48, 72, and 120 hours.
To detect changes in MHC class I antigen (H-2Kb) expression, cells were treated for 72 hours with cytokines or cell-conditioned medium, harvested, and incubated with mouse monoclonal antibody against H-2Kb (eBiosciences, San Diego CA), followed by incubation with FITC-goat antimouse immunoglobulin (Sigma, St. Louis, MO). Cell-surface staining was analyzed by flow cytometry.
To evaluate STAT activation, cells were treated for 15 minutes at 37°C with various cytokines and used for electrophoretic mobility shift assay (EMSA) experiments with GAS probe as described ( 15, 19). Human and murine mIFN-λs were from PeproTech, Inc. (Rocky Hill, NJ); mIFN-αA was from R&D Systems (Minneapolis, MN).
Virus infection and antiviral protection. Vesicular stomatitis virus (VSV; an RNA rhabdovirus, Indiana strain) infected or uninfected B16 cells were disrupted and the RNA was isolated as described ( 15) and used for nested reverse transcription-PCR (RT-PCR) with mIFN-λ-specific primers with the use of the GeneAmp RT-PCR kit (Perkin-Elmer, Boston, MA). The first round was done with primers mifnl-2 and mifnl-4 and the second round with mifnl-6 and mifnl-4. RNA samples were also used for RT-PCR with murine β-actin-specific primers. Antiviral assays were done essentially as described ( 28). After incubation with test IFNs for 24 hours, HT-29 or B16 cells were challenged with VSV and then further incubated until controls showed full killing by virus (1-2 days). Cells not killed were visualized by staining with crystal violet.
Mice, tumor transplantation, and histologic analysis of tumor tissue. Female C57BL/6 mice, 6 to 8 weeks old, were obtained from Charles River Laboratories, Inc. (Wilmington, MA). Mice were injected s.c. on the flank with 0.5 × 106 or 1 × 106 cells in 0.1 mL of PBS and tumors grew in the site of injection. Mice were checked for tumor growth by palpation of the injection site every 1 to 2 days. Mice were assigned tumor positive after first detection of tumors; animals were marked and kept for further analyses of tumors. Mice that rejected the primary tumor cells were challenged s.c. with parental B16 cells on the opposite flank. Tissue extraction for histology was done in mice with similar tumor size. Animals with massive tumor burden were euthanized. Extracted tumors were immediately fixed in 10% phosphate-buffered formalin. Paraffin-embedded sections were stained with H&E and analyzed by an expert pathologist blinded to treatment assignment.
The Kaplan-Meier estimator was used to calculate the median survival (tumor appearance) time and to derive tumor appearance (survival) curves. The log-rank test was used to test whether survival curves were identical. All statistical analyses were done with the use of the Survival package in program R ( 29).
Cloning and characterization of mouse IFN-λs and their receptor. There are three IFN-λ genes clustered on human chromosome 19 encoding highly homologous IFN-λ1, IFN-λ2, and IFN-λ3 proteins (ref. 15; Fig. 1 ). Further analysis of this genomic region suggested that, after the divergence of the IFN-λ1 and IFN-λ2 genes, a more recent duplication event occurred in which a fragment containing the IFN-λ1 and IFN-λ2 genes was copied and integrated back into the genome in a head-to-head orientation with the IFN-λ1-IFN-λ2 segment ( Fig. 1B). Divergence within this region created the IFN-λ3 gene, which is almost identical to the IFN-λ2 gene. However, in the duplicated fragment, the part which contained the IFN-λ1 gene was extensively mutated so that only separate pieces which do not encode a functional gene (IFN-λ4Ψ; Fig. 1B) can be found in this region.
Analysis of the mouse genome revealed that the region colinear with the human IFN-λ (hIFN-λ) gene cluster is located on chromosome 7A3 and has a similar organization with the hIFN-λ locus ( Fig. 1B). Two genes colinear with the hIFN-λ2 and hIFN-λ3 genes seemed to be intact and were predicted to encode functional proteins, which were designated mouse IFN-λ2 (mIFN-λ2) and mIFN-λ3 in accordance with the corresponding human genes. Mouse IFN-λ2 and IFN-λ3 also show higher sequence identity to human IFN-λ2 and IFN-λ3 than to hIFN-λ1 ( Fig. 1A). In contrast to the IFN-λ2 and IFN-λ3 genes, the mIFN-λ1 gene seemed to have lost the entire exon 2 and acquired a stop codon within exon 1; however, exons 3, 4, and 5 are intact. To investigate whether these mutations are strain specific, the IFN-λ1 genomic fragments from several mouse strains were cloned and sequenced. Although substantial sequence variations have been revealed in the region between exons 1 and 3 ( Fig. 1B), the stop codon within exon 1 was present in all the strains and exon 2 could not be predicted in any of the strains.
The mouse IFN-λ2 and IFN-λ3 genes were cloned into the pEF mammalian expression vector and transfected into COS-1 cells. The individual mIFN-λ cDNAs were synthesized from RNA isolated from the transfected COS cells and subsequently cloned into the pEF-SPFL plasmid in-frame with the FLAG epitope (referred to hereafter as FL-mIFN-λs). The immunoblots of conditioned media from transfected COS-1 cells revealed that secreted FL-mIFN-λ2 migrated on an SDS-PAGE gel as a broad band of ∼30 to 36 kDa whereas FL-mIFN-λ3 appeared as two distinct bands of ∼22 to 25 and 35 kDa, respectively ( Fig. 1C), suggesting possible glycosylation of mIFN-λs. Both mIFN-λ2 and mIFN-λ3 possess a potential site for N-linked glycosylation (Asn105-Met-Thr in mIFN-λ2 and Asn107-Asp-Ser in mIFN-λ3; Fig. 1A). Treatment with peptide N-glycosidase F (PNGase F) reduced the apparent molecular weights of these proteins, confirming that they are glycosylated (ref. 17 and data not shown).
Because there were several open reading frames in the region between exons 1 and 3 of the mIFN-λ1 gene, we investigated whether these frames could lead to the production of the modified IFN-λ1 protein. The entire mIFN-λ1 gene was cloned from the wild-type strain into the pEF-X-FL mammalian expression vector in the way that a potential IFN-λ1 protein would be tagged on its COOH terminus with the FLAG epitope. The plasmid was then transfected into COS-1 cells and the conditioned medium was analyzed by Western blotting for the presence of a FLAG-tagged protein. We failed to detect the presence of a FLAG-tagged protein ( Fig. 1C), showing that the mouse IFN-λ1 gene is a pseudogene.
The mouse IFN-λR1 (mIFN-λR1) chain was also cloned and it is ∼67% similar to its human counterpart ( Fig. 1D). The receptor is encoded on mouse chromosome 4D3. Although the mouse and human IFN-λR1 sequences are very similar, only two of three tyrosine residues of the human receptor intracellular domain are conserved in the mouse orthologue. In addition, the mouse receptor contains three additional tyrosine residues. There is also a stretch of negatively charged residues close to the end of the human receptor intracellular domain. This region in the mouse receptor is significantly altered by a short insertion and substitutions of several amino acid residues, resulting in a longer and more negatively charged region in the mouse receptor (18 of 20 amino acids are negatively charged; Fig. 1D). Two tyrosines, Tyr343 and Tyr517, of human IFN-λR1 (hIFN-λR1) can independently mediate STAT2 activation by IFN-λs ( 23). Interestingly, the Tyr341-based motif of mIFN-λR1 (YLERP) shows similarities with that surrounding Tyr343 of hIFN-λR1 (YIEPP). In addition, the COOH-terminal amino acid sequence of mIFN-λR1 containing Tyr533 (YLVRstop) is very similar to the COOH-terminal amino acid sequence of hIFN-λR1 containing Tyr517 (YMARstop). Therefore, both the mouse and human IFN-λR1 chains contain similar docking sites for STAT2 recruitment and activation, YΦEXP and YΦXRstop (where Φ is hydrophobic). Thus, Tyr341- and Tyr533-based motifs on mIFN-λR1 are also likely to mediate STAT2 recruitment and, therefore, mediate ISGF3 activation, which is responsible for most of the IFN-λ-induced biological activities ( 23).
Cloning and sequence analysis of murine IFN-λs and IFN-λR1 revealed that mIFN-λ2 and mIFN-λ3 are very similar to their human orthologues and that mouse genome does not encode functional IFN-λ1 protein. Because there were variations in the intracellular domain of the mIFN-λR1 in comparison with hIFN-λR1, which could lead to differences in signaling and biological activities, we studied the mouse IFN-λ ligand-receptor system.
Signaling by mouse IFN-λs. We first used a previously described hamster cell line that expresses a modified human IFN-λ receptor complex ( 15) to show that mIFN-λs can signal through human IFN-λ receptor complex. Hamster cells transfected with chimeric human IFN-λR1/γR1 and IL-10R2 were responsive to both human and mouse IFN-λs as measured by STAT1 activation in EMSA and up-regulation of MHC class I antigen expression ( Fig. 2A, middle , and data not shown). Parental hamster cells were unresponsive to either human or mouse IFN-λs (ref. 15; Fig. 2A, left). Interestingly, expression of murine IFN-λR1/γR1 alone rendered hamster cells responsive to both human and mouse IFN-λs as determined by EMSA with a GAS DNA probe ( Fig. 2A, right), indicating that hamster IL-10R2 can dimerize with murine IFN-λR1 to mediate signaling in response to either human or mouse IFN-λs. These experiments show that mouse and human IFN-λs are not species specific.
In a search for mouse cell lines responsive to IFN-λs, we screened several mouse cell lines by measuring up-regulation of MHC class I antigen expression and STAT activation in response to mIFN-λs. Several mouse fibroblast-like cell lines, such as L929 and NIH 3T3, did not respond to mIFN-λs but responded well to mIFN-α ( Figs. 2C and 3A , and data not shown). This correlated with earlier findings that several human cell types, such as primary fibroblasts and human umbilical vein endothelial cells, do not express IFN-λR1 and therefore are not responsive to IFN-λs. 5 In contrast, mouse B16 melanoma cells strongly responded to both mouse IFN-λ and IFN-α ( Figs. 2B and 3A). Using EMSA with either GAS or ISRE DNA probes, we showed that both type I and type III mIFNs induced formation of similar STAT-DNA-binding complexes in intact B16 cells ( Fig. 2B). Treatment of B16 cells with either mouse IFN-λ or IFN-α stimulated formation of the ISGF3 complex, which binds to ISRE probe and is specifically induced only by type I and type III IFNs ( Fig. 2B, right). EMSA with the GAS probe showed IFN-induced activation of STAT1 and STAT3 DNA binding complexes ( Fig. 2B, left). STAT identity was defined by supershifting the complexes with STAT1 or STAT3 antibodies ( Fig. 2B, left). In addition to B16 cells, mouse keratinocyte-like PAM212 cells and mouse primary keratinocytes were also responsive to IFN-λs as shown by EMSA ( Fig. 2C; ref. 17 and data not shown). In contrast, IFN-α, but not IFN-λ, induced STAT activation in mouse bone marrow–derived macrophages and mouse primary splenocytes ( Fig. 2C). Therefore, we concluded that similar to their human orthologues, mouse type I and type III IFNs use common signaling pathways to activate gene transcription. However, the target cells for IFN-λ and IFN-α only partially overlap.
Immunomodulatory and antiviral activities of mouse IFN-λs. We next determined whether the mouse IFN-λ system is capable of mediating biological activities similar to those induced by human IFN-λs, including antiviral protection and up-regulation of MHC class I antigen expression. We first showed that MHC class I antigen expression was strongly up-regulated in mouse B16 cells and PAM212 cells, and not in mouse fibroblast-like cell lines such as L929 and NIH 3T3 cells, in response to treatment with either mouse or human IFN-λs ( Fig. 3A and data not shown). In addition, both mouse IFN-λ2 and IFN-λ3 were capable of up-regulating MHC class I antigen expression in several human cell lines such as epitheloid carcinoma HeLa S3 cells, lung carcinoma A549 cells, keratinocyte HaCaT cells, hepatoma HuH7 cells, and colorectal carcinoma HT-29 cells, which are responsive to hIFN-λs (ref. 15; Fig. 3A and data not shown). These experiments again showed that mouse and human IFN-λs are active between these species. In contrast, murine and human type I IFNs do not have substantial activity on each other's cells ( 30).
Consistent with our initial findings ( 15), we also found that mIFN-λs induced antiviral protection in mouse B16 cells or human HT29 cells infected with VSV, an RNA rhabdovirus ( Fig. 3B and data not shown). Antiviral potency of mIFN-λ2 on B16 cells against VSV was comparable with that of mIFN-λ (∼107 IFN-α-relevant units/mg). We also showed that, similar to their human orthologues, expression of mIFN-λ genes was induced in response to viral infections. Consistent with their role in antiviral protection, the expression of mIFN-λ2/IFN-λ3 mRNAs was detected in B16 cells infected with VSV ( Fig. 3C).
Antitumor activities of mIFN-λs. Type I IFNs are approved for the treatment of various malignancies including melanomas. Therefore, we used a gene therapy approach to investigate whether type III IFNs may also possess antitumor activities. Cytokine gene therapy has shown many advantages in comparison with systemic administration, which requires the injection of high doses of a cytokine, often leading to adverse side effects ( 31– 33). Modified tumor cells constitutively producing a cytokine at the tumor site have been shown to be highly effective in treating solid cancers ( 34, 35).
B16 cells were transfected with plasmid pEF-mIFN-λ2, and G418-resistant cells were selected, pooled together, and designated B16.IFN-λ2 cells. The plasmid contains a strong elongation factor 1α promoter that provides constitutive production and secretion of mIFN-λ2 from B16.IFN-λ2 cells. To assess IFN-λ production and activity in B16.IFN-λ2 cells, we tested expression of MHC class I antigen, which is induced by IFN-λs in B16 cells ( Fig. 3A). As shown in Fig. 4A , a significant increase of MHC class I antigen expression was observed in B16 cells transfected with pEF-IFN-λ2 plasmid (B16.IFN-λ2 cells) in comparison with B16 cells transfected with control pEF plasmid (B16.vector), suggesting a secretion of IFN-λ from B16.IFN-λ2 cells. Production of IFN-λ2 in the culture medium was confirmed by the treatment of parental B16 cells with conditioned medium obtained from B16.IFN-λ2 cells after 3 days of cell culture. The expression level of MHC class I antigen was markedly up-regulated in B16 cells after treatment with the B16.IFN-λ2 cell–conditioned medium in a dose-dependent manner ( Fig. 4B). These results show that B16.IFN-λ2 stable transfectants constitutively secrete and respond to autocrine mIFN-λ2. However, we did not observe any difference in the in vitro proliferation rate by counting the cells over a 5-day period between B16.IFN-λ2 cells and parental B16 cells or B16.vector cells transfected with an empty vector (data not shown).
Because IFN-λs signal in a similar manner as type I IFNs and possess overlapping biological activities ( 15), we hypothesized that the constitutive expression of IFN-λ at the tumor site may affect the in vivo tumorigenicity of B16 cells, as described for IFN-α ( 34). To examine whether the constitutive secretion of IFN-λ inhibited tumor growth in vivo, 0.5 × 106 B16.IFN-λ2 cells, parental B16 cells, or B16.vector cells were injected s.c. into immunocompetent syngeneic C57BL/6 mice. All mice injected with parental B16 cells or B16 cells transfected with an empty vector (B16.vector) developed tumors in <20 days whereas the tumorigenicity of IFN-λ-producing B16.IFN-λ2 cells was highly impaired or completely abolished ( Fig. 4C). Although some mice developed B16.IFN-λ2 tumors, the tumors appeared later and grew about thrice slower than parental B16 tumors.
To examine if the presence of B16 cells producing IFN-λ2 in the same anatomic site used for the tumor implantation could affect the growth of parental B16 cells, we mixed both types of cells at equal amounts and injected them into the mice. The tumor onset was delayed in 100% of the mice injected with a mixture of 0.5 × 106 B16 cells and 0.5 × 106 B16.IFN-λ2. Furthermore, 30% of the mice remained tumor-free over a 3-month observation period ( Fig. 4D).
The level of IFN-λ production correlates with the level of MHC class I antigen expression in B16.IFN-λ2. To study the correlation between the level of IFN-λ secretion from B16.IFN-λ2 cells and the tumorigenicity, we selected three different clones of B16.IFN-λ2 cells based on the level of MHC class I antigen expression in these clones. The clones were respectively designated B16.IFN-λ2L (low; Fig. 5A ), B16.IFN-λ2M (intermediate; Fig. 5B), and B16.IFN-λ2H cells (high; Fig. 5C). The level of MHC class I antigen expression in these clones correlated with the level of IFN-λ production, which was evaluated by comparing the ability of conditioned medium from the clonal cells to induce MHC class I antigen expression in B16 cells relative to the standard recombinant IFN-λ2. The level of secreted IFN-λ2 was estimated to be 1 to 5 ng for B16.IFN-λ2L cells, 20 to 50 ng for B16.IFN-λ2M cells, and 100 to 150 ng for B16.IFN-λ2H cells produced by 106 cells every 24 hours.
We next assessed the tumorigenicity of the clones in vivo. As shown on the Fig. 5 (right), a close correlation existed between the level of IFN-λ production and the inhibition of tumor growth in vivo. B16.IFN-λ2H cells producing the highest amount of mIFN-λ2 were rejected at a higher rate than B16.IFN-λ2M and B16.IFN-λ2L cells secreting lower amount of mIFN-λ2. These experiments showed that tumorigenicity of B16.IFN-λ2 cells negatively correlated with the increase of mIFN-λ2 production. We again observed that in animals that developed B16.IFN-λ2 tumors, the tumors appeared later and grew slower than parental B16 tumors. Importantly, B16.IFN-λ2 tumor cells extracted from animals that succumbed to the tumors still preserved their in vitro preimplantation characteristics: the cells maintained mIFN-λ2 production and retained up-regulation of MHC class I antigen expression (data not shown).
Our results showed that IFN-λ played an important role against the establishment of B16 melanoma tumors. IFN-λ might inhibit tumor formation by acting directly on B16 melanoma cells or through indirect mechanisms such as stimulation of antitumor immune responses or inhibition of angiogenesis. To separate direct and indirect effects of IFN-λ on tumor growth, we generated B16.IFN-λ2 cells which were unresponsive to IFN-λ. Clones of B16.IFN-λ2 cells were selected for their low level of MHC class I antigen expression comparable to that of parental B16 cells. The clones were further screened for sustained IFN-λ2 production. A clonal cell population, designated B16.IFN-λ2Res, was selected which maintained IFN-λ2 production but the cells lost responsiveness to IFN-λ. As shown in Fig. 6A , treatment of parental B16 cells with the supernatant from B16.IFN-λ2Res cells resulted in the induction of MHC class I antigen expression, indicating that B16.IFN-λ2Res cells secreted IFN-λ2. However, B16.IFN-λ2Res cells failed to up-regulate the expression of MHC class I antigen, establish antiviral state, and activate STATs in the presence of either their endogenous secreted mIFN-λ2 or exogenous mIFN-λ2 ( Fig. 6A and data not shown). In contrast, treatment with mIFN-α induced STAT activation, antiviral protection, and up-regulation of MHC class I antigen expression in B16.IFN-λ2Res cells to the same level as in parental B16 cells, indicating that the MHC class I antigen expression and the Jak-STAT signaling pathway were still functional in B16.IFN-λ2Res cells ( Fig. 6A and data not shown). Transfection of B16.IFN-λ2Res cells with plasmid encoding mIFN-λR1 restored responsiveness of the cells to IFN-λ (data not shown). Therefore, B16.IFN-λ2Res cells produced IFN-λ2 but were resistant to IFN-λs due to mIFN-λR1 deficiency.
B16.IFN-λ2Res cells (0.5 × 106) were injected s.c. into mice and tumor development was monitored. A significant delay (∼2-fold) in tumor appearance or a complete inability of B16.IFN-λ2Res tumors to develop was observed ( Fig. 6B), suggesting that IFN-λ can act through indirect mechanisms mediated by the host to suppress tumor growth. In the mixed transplantation assay, when the mixture of 0.5 × 106 B16 cells and 0.5 × 106 B16.IFN-λ2Res cells was injected into mice, tumors were delayed or completely rejected ( Fig. 6C), similar to results obtained with the mixture of B16 and B16.IFN-λ2 cells ( Fig. 4D). These experiments indicated that IFN-λ produced by tumor cells induced indirect antitumor effects mediated by the host.
To study whether B16.IFN-λ2 cells, sensitive or resistant to IFN-λ treatment, generate a long-lasting memory antitumor response, mice that had rejected B16.IFN-λ2 or B16.IFN-λ2Res tumors were challenged by injection of 0.5 × 106 parental B16 cells in the opposite flank. Whereas 100% of naive mice developed tumors, ∼90% of mice that had previously rejected transfected modified B16 melanomas succumbed to parental B16 tumors ( Fig. 6D), implying that mice in which B16.IFN-λ2 tumors were eliminated did not develop a significant sustained antitumor immunity.
When B16.IFN-λ2 cells, sensitive or resistant to IFN-λ, were injected into mice, some animals developed tumors albeit with significant delay compared with tumors from parental B16 cells. In addition, B16.IFN-λ2 cell–derived tumors seemed to grow about thrice slower than parental B16 tumors. To elucidate a mechanism(s) explaining the inhibition of tumor growth in vivo, tumors removed from several mice in each treatment group were examined by light microscopy. Photomicrographs of B16 and B16.IFN-λ2 tumors are shown in Fig. 7 . Histologically, tumors from mice injected with B16.IFN-λ2 cells, which displayed delayed growth, differed in several aspects from the parental B16-derived lesions. B16 tumors were large and composed of solid nests of neoplastic cells supplied by numerous small blood vessels. Focal areas of the lesion showed prominent cellular degeneration and areas of necrosis. In contrast, tumors derived from B16.IFN-λ2 cells were less vascular than B16 tumors, with nests of viable tumor cells clustered around a thin-walled, centrally located vessel. Large confluent areas of necrosis were evident throughout the mass. The mitotic rate of the B16 tumors was more than double those from B16.IFN-λ2 cells (14 versus 6 per 40× field, respectively). No tumor-infiltrating immune cells were detected in tumors from any group of mice, including ones that received the tumor vaccines and were challenged with parental B16 cells.
Following the discovery of human IFN-λ proteins ( 18), we searched the mouse genome to identify murine IFN-λ orthologues. Although there are three genes encoding highly homologous but distinct human IFN-λ proteins (IFN-λ1, IFN-λ2, and IFN-λ3), the search of the mouse genome revealed the existence of only two genes, representing mouse IFN-λ2 and IFN-λ3 gene orthologues, encoding intact proteins ( Fig. 1). The mouse IFN-λ1 gene orthologue contains a stop codon in the first exon and, therefore, it does not encode an intact protein. Mouse IFN-λs (mIFN-λ2 and mIFN-λ3) and IFN-λ receptor (mIFN-λR1) orthologues were cloned and found to be quite similar to their human counterparts ( Fig. 1). Experiments showed that similar to their human counterparts, mIFN-λ2 and mIFN-λ3 signal through the IFN-λ receptor complex, activate ISGF3, and are capable of inducing antiviral protection and MHC class I antigen expression in several cell types ( Figs. 2 and 3). The results confirmed our previous observation that type III IFNs (IFN-λs) engage a unique receptor complex, composed of IFN-λR1 and IL-10R2 subunits, to induce signaling and biological activities similar to those of type I IFNs.
Type I IFNs are recognized not only for their antiviral role but also for their antitumor activity. IFN-α is used clinically as a treatment for several malignancies, including melanoma ( 1, 34). Recently, the target cells for the antitumor action of both type I and type II IFNs were highlighted in animal experiments with transplanted and carcinogen-induced tumors ( 36, 37). Using mice deficient in IFN response, it was shown that whereas endogenously produced type I and type II IFNs target immune cells to enhance antitumor responses, only the direct action of type II IFN on cancer cells seemed to be important to mobilize an effective antitumor response ( 37). Because cell signaling by IFN-λs is similar to type I IFNs, which have shown antitumor effects, we investigated whether type III IFNs also possess antitumor activities using a gene therapy approach in the mouse B16 melanoma model.
B16 cells respond to IFN-λs, as shown by up-regulation of MHC class I antigen expression and antiviral activity ( Figs. 2 and 3). By gene transfer, we engineered B16 cells, which constitutively produced mIFN-λ2 (B16.IFN-λ2 cells). In response to their secretion of IFN-λ, B16.IFN-λ2 cells exhibited constitutively high level of MHC class I antigen expression (autocrine mIFN-λ2 effect; Fig. 4A).
B16 melanoma is an aggressive and highly malignant murine tumor. One hundred percent of immunocompetent C57BL/6 syngeneic mice, injected with parental B16 cells, developed tumors within 3 weeks ( Figs. 4– 6). However, the constitutive production of mIFN-λ2 by B16.IFN-λ2 cells markedly affected tumorigenicity of the cells ( Figs. 4 and 5). B16.IFN-λ2 cells were either rejected by the host or grew at a slower rate than control parental B16 cells. The antitumor effect of IFN-λ was dose dependent ( Fig. 5). B16.IFN-λ2 cells also inhibited the growth of parental B16 cells when both cell types were injected together ( Fig. 4). B16 melanoma is a poorly immunogenic tumor, characterized by inefficient MHC-restricted antigen presentation ( 38). The high level of constitutive MHC class I antigen expression in B16.IFN-λ2 cells may render the cells more immunogenic and promote adaptive antitumor immune responses. However, B16.IFN-λ2 tumors did not display increased lymphocytic infiltrates ( Fig. 7) and failed to induce a strong memory response ( Fig. 6D), suggesting that up-regulation of MHC class I antigen expression on the surface of B16 cells was not sufficient to significantly increase the immunogenicity of B16.IFN-λ2 cells.
To examine whether the antitumor effect of mIFN-λ2 was due to a direct action on B16 cells or mediated by a host response, we developed B16.IFN-λ2Res cells, which produced mIFN-λ2 but were resistant to IFN-λ treatment ( Fig. 6A). B16.IFN-λ2Res cells displayed reduced tumorigenicity and repressed the growth of parental B16 cells in vivo to a level comparable to IFN-λ-sensitive B16.IFN-λ2 cells ( Fig. 6B and C). The impaired tumor growth of IFN-λ-resistant B16.IFN-λ2Res cells, which still secreted mIFN-λ2, strongly suggested that host-defense mechanisms played a major role in mediating IFN-λ-induced antitumor activity. However, direct action of IFN-λs on the host is rather limited. Several cell types, including primary lymphocytes and macrophages, the major players in specific antitumor immunity, were found to be unresponsive to IFN-λ ( Fig. 2), suggesting that immune cells are not the primary targets of IFN-λ. Therefore, the function of immune cells is unlikely to be directly altered by IFN-λ. In contrast, virtually all cell types respond to type I IFNs, indicating that mechanisms of antitumor activities of type I and type III IFNs are not identical. Our experiments also showed that only ∼10% of mice that rejected the B16.IFN-λ2 cells, either sensitive or resistant to IFN-λ, survived the parental tumor challenge ( Fig. 6D), suggesting that the development of long-lasting immunity in this mouse tumor model was relatively weak. Therefore, the involvement of antitumor mechanisms in response to IFN-λ, which are likely to play an important role independent of adaptive immunity, can be proposed.
The immunosurveillance function of CTL, which can kill tumor cells, is clearly important ( 36). Although neither lymphocyte recruitment in established B16.IFN-λ tumors ( Fig. 7) nor strong protective immunity in animals which rejected B16.IFN-λ cells was detected ( Fig. 6D), these results cannot be extrapolated to other tumors. For example, B16 cells expressing IFN-α also showed decreased tumorigenicity and failed to elicit protective immunity to subsequent challenge with B16 cells ( 39). In contrast, IFN-α-expressing TS/A and MC38 tumors were effective in the induction of protective immunity ( 40, 41). Therefore, the ability of IFN-λ to induce adaptive immune mechanisms in other tumor models remains to be investigated.
Tumor rejection is a complex process which integrates both immune and nonimmune mechanisms toward tumor destruction. In addition to adaptive immune mechanisms, tumor development and growth is also affected by tumor stromal cells, a mixture of various cell types that reside at the tumor site and those recruited to the site, which are required to sustain tumor growth by providing growth factors, extracellular matrix support, blood supply, and waste removal ( 42). For example, under physiologic conditions, keratinocytes control melanocyte growth and behavior through cell interaction and paracrine mechanisms, whereas under melanoma conditions, normal communications are altered ( 43). We found that normal keratinocytes are highly responsive to IFN-λ ( Fig. 2C; ref. 15). Therefore, mIFN-λ2 produced by either IFN-λ-resistant or IFN-λ-sensitive B16.IFN-λ2 melanomas could affect neighboring keratinocytes and, perhaps, other tumor stromal cells, and could inhibit their tumor-supportive function. Histologic analysis ( Fig. 7) showed large areas of tumor necrosis associated with reduced vascularity in B16.IFN-λ2 tumors. IFN-λs may contribute to antiangiogenic mechanisms by inhibiting the abilities of tumor or stroma to stimulate angiogenesis, which is required to maintain tumor survival and proliferation in vivo ( 44). However, the precise mechanisms and mediators of IFN-λ antitumor activities should be further investigated and compared with mechanisms induced by other types of IFNs for which the antiangiogenic effects have been reported in several studies ( 1, 45).
This study represents the first report of the use of IFN-λs in a model of cancer therapy. The results clearly showed that IFN-λs are effective in the mouse B16 melanoma model and engaged host mechanisms to exert their antitumor functions. Although the precise mechanism(s) responsible for the antitumor activities of type III IFNs remains to be elucidated, this study reveals the potential therapeutic antitumor properties of IFN-λs and suggests their potential use in cancer therapy.
Grant support: National Institute of Allergy and Infectious Diseases, USPHS grants RO1 AI051139 and AI057468 and American Heart Association grant AHA #0245131N (S.V. Kotenko) and ES05022 grant support for the Molecular Pathology Core.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Jerry Langer and Martin Schwarz for the critical review of the text and helpful suggestions.
↵5 Unpublished observations.
- Received October 10, 2005.
- Revision received January 3, 2006.
- Accepted February 14, 2006.
- ©2006 American Association for Cancer Research.