Overexpression of cathepsin L, a cysteine protease, and consequently procathepsin L secretion switch the phenotype of human melanoma cells to highly tumorigenic and strongly metastatic. This led us to identify the DNA regulatory sequences involved in the regulation of cathepsin L expression in highly metastatic human melanoma cells. The results of the present study demonstrated the presence of regulatory sequences in the 3′ region downstream of the cathepsin L gene and in the 3′- and 5′-flanking regions of GC/CCAAT sites of its promoter. In addition, we established that the 5′-UTR (untranslated region) was the most important region for cathepsin L expression. This 5′-UTR integrated an alternative promoter and sequences involved in post-transcriptional regulation. Transfection experiments of bicistronic reporter vectors and RNAs demonstrated that the cathepsin L 5′-UTR contained a functional IRES (internal ribosome entry site). This complete IRES was present only in one of the three splice variants, which differed in their 5′-UTR. Then, we analysed cathepsin L expression in this human melanoma cell line grown under hypoxia. We demonstrated that under moderate hypoxic conditions (1% O2) intracellular expression of cathepsin L was up-regulated. Hypoxia significantly increased only the expression of the transcript which contains the complete IRES, but inhibited promoter activity. These results suggest that the presence of an IRES allowed cathepsin L mRNA translation to be efficient under hypoxic conditions. Altogether, our results indicated that in vivo a tumour hypoxic environment up-regulates cathepsin L expression which promotes tumour progression.
- cathepsin L
- internal ribosome entry site (IRES)
- tumour progression
Cathepsin L, a lysosomal cysteine protease, plays an important role in the catabolism of proteins and consequently in the behaviour of human cells. Evidence supports that the increase of cathepsin L expression is involved in tumour progression. Indeed, we previously demonstrated that cathepsin L overexpression in human melanoma cells transfected with cathepsin L cDNA switches their phenotype to highly tumorigenic and strongly metastatic [1,2]. In addition, malignant melanoma cells of patient tumours expressed higher concentrations of cathepsin L than normal cells . Specific inhibitors of cathepsin L activity inhibited distant bone metastasis  and invasive capacity of melanoma cells . Down-regulation of cathepsin L by antisense cDNA in melanoma cells significantly reduced invasiveness .
As a consequence of cathepsin L overexpression, procathepsin L secretion increases and contributes to the switch to a highly tumorigenic and metastatic phenotype of human melanoma cells. Indeed, pretreatment of tumorigenic and highly metastatic human melanoma cells with a polyclonal anti-cathepsin L antibody, which acts at the extracellular level, strongly inhibited tumour formation and significantly decreased lung metastasis in nude mice . Moreover, focusing on procathepsin L secretion, we generated a ScFv (single chain variable fragment) from a monoclonal anti-cathepsin L antibody prepared in our laboratory . We demonstrated that intracellular expression of the anti-cathepsin L ScFv in human melanoma cells strongly inhibited: (i) procathepsin L secretion, without modifying the intracellular amount or processing pattern of cathepsin L; and (ii) human melanoma tumour growth and metastasis development in nude mice .
Secreted procathepsin L, by acting on a wide range of extracellular components, contributes to human melanoma progression, facilitates invasion, angiogenesis and inhibits the immune system. For instance, we demonstrated that secreted procathepsin L cleaves human C3, the third component of complement, consequently inhibiting complement-mediated cell lysis and thus contributing to human melanoma resistance to the immune system [7,10,11]. Furthermore, procathepsin L generated C3 fragments which could trigger regulatory functions on normal and tumour cells by interacting with specific surface receptors. C3 fragments are involved in a large spectrum of biological functions and regulations .
Thus these results emphasize the need to identify the molecular events involved in the regulation of the human cathepsin L gene. We previously characterized the human cathepsin L promoter and identified a region, essential for basal cathepsin L transcription, which contains one CCAAT motif and two GC boxes (GC/CCAAT sites), localized 60 bp upstream of the major transcription initiation site. Transcription factors such as NF-Y (nuclear factor Y) and members of the Sp family (Sp1, Sp2 and Sp3) bind to these sites . More recently, we demonstrated that distinct molecular mechanisms regulate expression of cathepsin L in tumour cells. Indeed, cathepsin L gene silencing in lymphoma cells was due to genomic DNA methylation. Cathepsin L overexpression in melanoma cells involved either gene amplification or unidentified transcriptional mechanism at the promoter level . Expression of several cathepsin L mRNA splice variants, which differed in their 5′-untranslated ends, could also constitute another level of regulation in melanoma cells .
Taken together these results led us to analyse the regulation of the human cathepsin L gene, under normoxic and hypoxic conditions, in A375SM, a highly tumorigenic and metastatic melanoma cell line. In the present study we demonstrate that: (i) the 5′-UTR (untranslated region) of cathepsin L mRNA is essential for its expression; (ii) the cathepsin L 5′-UTR contains an alternative promoter and an IRES (internal ribosome entry site); (iii) hypoxia up-regulates expression of cathepsin L; and (iv) this increase correlates with higher expression of the splice variant which contains the IRES.
MATERIALS AND METHODS
Cell lines and culture conditions
The A375SM and DM-4 human melanoma cell lines were kindly provided by Dr M. Bar-Eli (M. D. Anderson Cancer Center, Houston, TX, U.S.A.) and the H1299 lung carcinoma cell line was obtained from the ATCC. The highly tumorigenic and metastatic human melanoma A375SM cells were established from nude mice lung metastases produced by the A375 human cell line isolated from a lymph node . Cells were maintained in culture as adherent monolayers in DMEM (Dulbecco's modified Eagle's medium) supplemented with 10% (v/v) fetal calf serum, Glutamax-1 (2 mM), non-essential amino acids (0.1 mM), and penicillin (100 units/ml)/streptomycin (100 μg/ml) (Invitrogen) at 37 °C in a CO2 incubator.
Cells were maintained under hypoxic conditions at 37 °C within a modular incubator chamber (Billups-Rothenberg) filled with 1% O2 and 5% CO2 (balanced with N2) for 26 h. A standard tissue culture incubator was used for a normoxic environment (21% O2, 5% CO2). For transient transfected cells, hypoxic conditions were applied 24 h after transfection.
The mammalian expression vectors encoding a fusion protein of procathepsin L with a C-terminal V5 epitope (pcDNA3.1-V5-His-ProCTSL and pcDNA3.1-V5-His-5′-UTR-ProCTSL) were generated from preprocathepsin L cDNA, amplified by PCR using a 5′ primer complementary either to the coding region or the 5′-UTR of cathepsin L cDNA (Table 1) and cloned into pcDNA3.1-V5-HisA (Invitrogen), using BamHI and XhoI. The cathepsin L promoter–firefly luciferase constructs pGL3-P4/U/1-(−1/−3263), pGL3-P1/U/I-(−1/−1646), pGL3-U1/I-(−1/−1388) and pGL3-P1-(−1465/−1646) and GAPDH (glyceraldehyde-3-phosphate dehydrogenase) promoter–firefly luciferase construct pGL3-GAPDH have been previously described . Other monocistronic reporter constructs (pGL3-U/I-(−1/−1489), pGL3-P1/U-(−1211/−1646), pGL3-P2-(−1465/−1929), pGL3-P3-(−1465/−2085) and pGL3-P4-(−1465/−3263) were generated by PCR using specific primers (Table 1) and by subcloning in the promoterless pGL3-Basic vector (Promega) digested with BglII and MluI. The 1500 bp genomic fragment at the 3′ of the cathepsin L gene was generated by PCR from the genomic DNA of melanoma cells using a primer downstream of the stop codon and a primer complementary to the 3′ region (Table 1). PCR products were subcloned into the promoterless pGL3-Basic, pGL3-P1/U/I-(−1/−1646) and pGL3-P4/U/1-(−1/−3263) digested with BamHI and SalI, to generate pGL3-Basic+1500, pGL3-P1/U/I-(−1/−1646)+1500 and pGL3-P4/U/1-(−1/−3263)+1500. Mutagenesis of the CCAAT motif and of the two GC boxes to generate pGL3-P1/U/I-(−1/−1646)Mut and pGL3-P1-(−1465/−1646)Mut have been described previously . To create pRL-FL and pSV-RL-FL bicistronic vector constructs, the coding region of RL (Renilla luciferase) without or with the SV40 early promoter was amplified by PCR from the pRL-SV40 vector (Promega) and subcloned into the promoterless pGL3-Basic vector digested with SacI and MluI. DNA fragments of the cathepsin L 5′-UTR and the EMCV (encephalomyocarditis virus) IRES were generated by PCR using specific primers (Table 1) from pcDNA3.1-V5-His-5′-UTR-ProCTSL or pIRES2-AcGFP1 vectors (Clontech) respectively. PCR products were then inserted into pRL-FL and pSV-RL-FL vectors at MluI and BglII sites to generate bicistronic vector constructs used in the present study. Amplification reactions were performed using the GeneAmp high-fidelity PCR system (Applied Biosystems) and constructs were confirmed by sequence analysis (Genome Express). Nomenclature used for functional elements in reporter constructs is: P1–P5 for cathepsin L promoter fragments; I for intron 1 of the cathepsin L gene; U and U1–U5 for the 5′-UTR of cathepsin L cDNA.
DNA transfection and reporter assays
Expression and monocistronic reporter vectors were transfected into melanoma cells using Lipofectin® reagent (Invitrogen) as previously described . Transfections were performed with 600 ng of reporter constructs and 200 ng of phRL-null (Promega) or with 600 ng of cathepsin L expression vectors and 80 ng of pRL-CMV (Promega; where CMV is cytomegalovirus). Reporter vectors which encode RL were used for normalization of transfection efficiency. Bicistronic reporter vectors (800 ng) were transfected using the Gene Porter 2 reagent (Gene Therapy System) in accordance with the manufacturer's protocol. Cell extracts were prepared 48 h after transfection. Quantification of firefly luciferase and RL activities was performed with the Dual-Luciferase Reporter Assay System (Promega). The relative luciferase activity was calculated by normalizing transfection efficiency with the RL activity. The transfections were performed in triplicate, and similar results were obtained from at least three independent experiments.
RT (reverse transcription)-PCR analysis
To estimate mRNA expression, RT-PCR was performed as previously described . Briefly, total RNA was isolated from melanoma cells 48 h after transfection using TRIzol® reagent (Invitrogen) and then reverse transcribed. The PCR reaction was carried out with 24 cycles of amplification for GAPDH, 25 cycles for firefly luciferase and 30 cycles for RB18A. Sequences of specific primers have been previously described for GAPDH . Primers for firefly luciferase were 5′-TTGCTTTTACAGATGCAC-3′ (forward) and 5′-AGGATCTCTGGCATGCGA-3′ (reverse), and for RB18A 5′-ATGGAGCATCACAGTGGTAGT-3′ (forward) and 5′-ACTGCTTTTCATCTTCCCTG-3′ (reverse). To analyse cathepsin L mRNA splicing, PCR was performed with 30 cycles of amplification using primers complementary to sequences downstream and upstream of the first intron: 5′-GACAGGGACTGGAAGAGAGGAC-3′ and 5′-AAAGGCAGCAAGGATGAGTGTAGGATTCAT-3′. To analyse splicing of bicistronic mRNA, cDNAs were prepared from cells transfected with bicistronic constructs and PCR was carried out with 30–50 cycles of amplification using a primer 5′-GAGCTATTCCAGAAGTAGTGAGGAG-3′ complementary to the RL 5′-UTR and a primer 5′-GCCTTATGCAGTTGCTCTCCAGCGGTTCC-3′ specific for the firefly luciferase coding region. To quantify mRNA expression, real-time PCR was performed by Cogenics (Meylan, France) with an Applied Biosystems 7500 Fast Real-Time PCR System using SYBR Green. Luciferase amplification primers were 5′-GCCTGAAGTCTCTGATTAAGT-3′ (forward) and 5′-ACACCTGCGTCGAAGATGT-3′ (reverse). The amount of luciferase transcript in each RNA sample was quantified and normalized to GAPDH content. Relative amounts of luciferase cDNA were calculated by the relative comparative 2−ΔΔCt method and are expressed as a percentage of the luciferase cDNA measured in cells transfected with pGL3-P1/U/I-(−1/−1646).
Western blot analysis
Protein expression and secretion were analysed in conditioned media and cell extracts by Western blot analysis as previously described , using an anti-cathepsin L mAb (monoclonal antibody; clone 3D8) prepared in our laboratory, an anti-GAPDH mAb (mAb374; Chemicon International), an anti-RB18A mAb (anti-TRAP220, C19; Santa Cruz Biotechnology), an anti-lamin A/C mAb (MCA1429; Serotec) or an anti-V5-Tag mAb (MCA1360; Serotec). Bound antibodies were detected with a peroxidase-linked secondary antibody (Dako) and the ECL® (enhanced chemiluminescence) system (Amersham Biosciences).
In vitro RNA synthesis and RNA transfection
In vitro synthesis of capped RNAs was carried out using the mMESSAGE mMACHINE kit (Ambion). Briefly, pSV-RL-FL and pSV-RL-U1-FL containing bicistronic transcription units under the control of the T7 promoter were linearized downstream from the 3′ end of the firefly luciferase coding sequence at the XbaI restriction site. The transcript was synthesized from 1 μg of the linearized plasmid DNA with T7 RNA polymerase in the presence of a cap analogue. Poly(A) tails were added to the newly synthesized mRNAs using the poly(A) tailing kit using the manufacturer's protocol (Ambion). Capped and polyadenylated mRNAs were examined for integrity by electrophoresis in a denaturing agarose gel and the concentration was estimated by measuring the absorbance at 260 nm. RNA transfections were carried out using Lipofectamine™ 2000 (Invitrogen) reagent according to the manufacturer's protocol. A375SM melanoma cells were seeded in 24-well tissue culture plates, cultured for 24 h and transfected with 0.8 μg of in vitro synthesized mRNA and 2 μl of the Lipofectamine™ 2000 reagent. After incubation for 6 h, the cells were washed and the transfection medium was replaced by culture medium. Cells extracts were prepared 16 h later and luciferase activities were measured.
DNA regulatory sequences of the human cathepsin L gene
We previously demonstrated that in A375SM human melanoma cells cathepsin L overexpression correlated with high promoter activity and that the CCAAT motif and GC boxes were not directly involved in this regulation . This led us to identify others DNA regulatory sequences involved in cathepsin L gene expression.
Thus we subcloned the 3′- and 5′-flanking region of the cathepsin L gene in the luciferase reporter vector pGL3-Basic. These constructs were co-transfected into A375SM cells with the promoterless phRLnull plasmid, used as an internal control and luciferase activities were quantified (Figure 1). As shown in Figure 1(A), deletion mutants in the 3′ of GC/CCAAT sites were compared with the pGL3-P1/U/I-(−1/−1646) reporter vector, which contained the 1646 bp sequence upstream of the translation start site of the human cathepsin L gene . Deletion of the first intron (−1210/−1) of the cathepsin L gene induced a 1.7-fold decrease of luciferase activity (Figure 1A). Deletion of the −1211/−1464 sequence corresponding to the 5′-UTR of cathepsin L mRNA induced an additional 5.2-fold decrease of luciferase activity (Figure 1A). Then, reporter vectors containing fragments of the 5′-flanking region of GC/CCAAT sites were compared with the pGL3-P1-(−1465/−1646) construct which contained the 181 bp sequence upstream of the major transcription start site (−1489) of the cathepsin L gene . The 1.4-fold decrease in luciferase activity measured with the construct pGL3-P4-(−1465/−3263) supported the presence of negative regulator sequences between −2086 and −3263 (Figure 1B).
Then, the presence of a transcriptional regulatory region in the 3′ of the cathepsin L gene was investigated. The 1500 bp sequence in the 3′ region of the stop codon was subcloned downstream of the luciferase reporter gene into the promoterless pGL3-Basic vector and into pGL3-P1/U/I-(−1/−1646) and pGL3-P4/U/I-(−1/−3263) vectors to generate pGL3-Basic+1500, pGL3-P1/U/I-(−1/−1646)+1500 and pGL3-P4/U/I-(−1/−3263)+1500. As shown in Figure 1(C), luciferase activity was not detected with the pGL3-Basic+1500 construct, supporting that the 3′ region of the cathepsin L gene was not able to initiate transcription. A 1.4-fold decrease in luciferase activity was observed in cells transfected with the pGL3-P1/U/I-(−1/−1646)+1500 and pGL3-P4/U/I-(−1/−3263)+1500 vectors, in comparison with the pGL3-P1/U/I-(−1/−1646) and pGL3-P4/U/I-(−1/−3263) vectors respectively (Figure 1C). Taken together, these results supported the presence of regulatory sequences in the 3′- and 5′-flanking regions of GC/CCAAT sites and in the 3′ region of the cathepsin L gene.
The 5′-UTR is involved in regulation of cathepsin L gene expression
As the deletion of the −1211/−1464 sequence of the 5′-UTR induced the highest inhibition of luciferase activity, we focused on the analysis of its contribution in cathepsin L gene expression. Thus A375SM cells were transfected with reporter vector constructs in which the cathepsin L promoter was deleted, i.e. pGL3-U/I-(−1/−1489) and pGL3-U1/I-(−1/−1388) constructs, containing only sequences downstream of the major transcription initiation site (Figure 2A). No significant luciferase activity was detected with the pGL3-U1/I-(−1/−1388) construct, as compared with the promoterless pGL3-Basic vector. However, a 58% luciferase activity was detected with the pGL3-U/I-(−1/−1489) vector, as compared with the activity of the pGL3-P1/U/I-(−1/−1646) construct. These results indicated that the region between −1389 and −1489 contained a promoter. Then, A375SM cells were transfected either with wild-type pGL3-P1-(−1465/−1646) and pGL3-P1/U/I-(−1/−1646) vectors or with mutated pGL3-P1-(−1465/−1646)Mut and pGL3-P1/U/I-(−1/−1646)Mut vectors in which the CCAAT motif and GC boxes were disrupted . Luciferase activities showed (Figure 2A) that these mutations totally inhibited promoter activity of the P1 region (−1465/−1646) and induced a 76% decrease of the activity of the P1/U/I region (−1/−1646). Differences in relative luciferase activities of pGL3-U/I-(−1/−1489) and pGL3-P1/U/I-(−1/−1646)Mut, i.e. 58% and 24% respectively, suggested that the promoter in the 5′ region of the major transcription initiation site inhibits the activity of the promoter localized between −1389 and −1489. Then, we analysed expression of firefly luciferase mRNA by real-time quantitative RT-PCR in cells transfected with reporter vector constructs (Figure 2B). Transfection efficiencies, determined by measuring RL activity, were similar. The amount of firefly luciferase transcript in each RNA sample was normalized to the GAPDH content and was expressed as a percentage of luciferase cDNA measured in cells transfected with pGL3-P1/U/I-(−1/−1646). The results demonstrated that luciferase mRNA expression levels correlated with luciferase activities in cells transfected with pGL3-P1/U/I-(−1/−1646) and pGL3-U/I-(−1/−1489), in agreement with a transcriptional regulation (Figure 2B, columns 1 and 2). In contrast, despite a strong 87% difference (shown in Figure 2A) between luciferase activities of cells transfected with pGL3-P1/U/I-(−1/−1646) or pGL3-P1-(−1465/−1646), luciferase mRNA expression levels were similar (Figure 2B, columns 1 and 3). Identical results were observed when firefly luciferase mRNA levels were analysed by semi-quantitative PCR analysis (Figure 2C). Indeed, intensities of amplified luciferase PCR products are very similar between cells transfected with pGL3-P1/U/I-(−1/−1646) or pGL3-P1-(−1465/−1646) (Figure 2C, lanes 1 and 2). Thus both quantitative and semi-quantitative PCR analyses demonstrated that inhibition of luciferase activity induced by the deletion of the 5′-UTR was not supported by a transcriptional mechanism.
In addition, cathepsin L cDNA was subcloned with or without the 5′-UTR in an expression vector under the control of the CMV promoter. The pcDNA3.1-V5-His-ProCTSL and pcDNA3.1-V5-His-5′-UTR-ProCTSL vectors encode a fusion preprocathepsin L with a C-terminal V5 epitope. Expression of ectopic procathepsin L, characterized by a 44 kDa molecular mass, was analysed in cell extracts and in conditioned media prepared from transfected A375SM cells by Western blotting using an anti-V5 mAb (Figure 2D). To take into account tranfection efficiency variation, RL reporter vector was co-transfected with cathepsin L expression vectors and samples were loaded according to their RL activities. The results demonstrated a strong increase of the intracellular expression and secretion of procathepsin L when the cathepsin L coding region was associated to its 5′-UTR. Taken together, these results supported that the 5′-UTR was involved in the regulation of cathepsin L expression independently of the transcriptional mechanism.
Presence of an IRES in the 5′-UTR of cathepsin L
The 5′-UTR of cathepsin L mRNA is 293 bp long and 68% GC-rich. Folding of the 5′-UTR RNA using the Mfold program  allowed prediction of a very stable structure [ΔG=−126 kcal/mol (1 kcal≈4.184 kJ)], suggesting that cap-dependent translation will be inefficient. This analysis and the ability of the cathepsin L 5′-UTR to stimulate expression suggested that translation might be proceeded by internal ribosome recruitment. IRES, first identified in the 5′-UTR of poliovirus and EMCV, has also been described in several cellular mRNAs . To investigate the putative presence of an IRES in the cathepsin L 5′-UTR, we used pSV-RL-FL, a bicistronic vector containing two reporter genes: the RL (upstream cistron driven by the SV40 promoter) and firefly luciferase (downstream cistron). DNA fragments of the cathepsin L 5′-UTR were subcloned within the intercistronic spacer between these two coding regions (Figure 3A). As a control, we used the pRL-FL, a SV40 promoter-less bicistronic vector which allowed us to detect the presence of a cryptic promoter in the subcloned region. The bicistronic vectors pSV-RL-U1-FL and pRL-U1-FL (U1 corresponds to −1/−1388 region, Figure 3A) were transfected into A375SM cells. As shown in Figure 3(B), RL activity was detected only with vectors which contained the SV40 promoter (pSV-RL-FL and pSV-RL-U1-FL) supporting transcription of the bicistronic mRNA. Firefly luciferase activity was 5.3-fold higher in cells transfected with pSV-RL-U1-FL than with control pSV-RL-FL (Figure 3C). RL (Figure 3D) and firefly luciferase (Figure 3E) activities were also determined in A375SM cells transfected with pSV-RL-U2-FL and pRL-U2-FL (U2 corresponds to −1371/−1489 region, Figure 3A). High firefly luciferase activity detected with both constructs (Figure 3E) also supported the presence of a promoter in the cathepsin L 5′-UTR between −1371 and −1489, as described in Figure 2. These results suggested the presence of an IRES in fragment −1/−1388 of the cathepsin L 5′-UTR.
However, the presence of cryptic splice sites in the cathepsin L 5′-UTR could also lead to firefly luciferase activity, without the presence of an IRES. To clarify this point, RT-PCRs were performed on total RNA isolated from A375SM cells transfected with the pSV-RL-U1-FL vector or pRL-U1-FL control vector, using primers specific to the 3′ end of the SV40 promoter sequence (downstream of the transcription start site) and to the firefly luciferase coding region. A single amplified product was obtained from transfected cells with pSV-RL-U1-FL, the size of which corresponded to the full-length biscistronic RNA, approx. 1600 bp (Figure 4A, lane 4). Absence of smaller PCR products using 30, 40 and 50 cycles of amplification supported the absence of splicing. Then, as direct transfection of dicistronic RNAs were previously used to confirm IRES activity in cells [17,18], we generated in vitro capped and polyadenylated mRNAs from control pSV-RL-FL and pSV-RL-U1-FL (Figure 4B) for transfection experiments into A375SM cells. As shown in Figure 4(C), relative luciferase activity obtained from A375SM cells transfected with RL-U1-FL mRNA was 2.6-fold greater than that obtained from cells transfected with control RL-FL mRNA. RNA transfections indicated that the observed firefly luciferase activity was RNA-dependent and not due to the presence of a cryptic promoter or a splicing. Thus taken together these results strongly support the presence of an IRES in the cathepsin L 5′-UTR.
To localize more precisely the region necessary for IRES activity, deletions were introduced in the U1 fragment (−1/−1388) and were subcloned into the pSV-RL-FL bicistronic vector (Figure 5A). These constructs were then transfected into A375SM cells and their relative luciferase activities were measured (Figure 5B). A decrease in IRES activity of 34% was observed when the −1302/−1388 DNA fragment was deleted from the 5′ end of the U1 fragment (−1/−1388), to generate the U4 fragment (−1/−1301). The U3 fragment (−1301/−1388) showed a very slight activity. In addition, deletion of the −1/−11 fragment corresponding to exon 2 at the 3′ end of U1 fragment, to generate the U5 fragment (−1211/−1388), did not modify IRES activity. These results support that the IRES was mainly localized in the first exon between −1211 and −1388 and suggested that progressive, rather than abrupt, loss of activity occurs when 5′-UTR fragments are deleted, consistent with the modular property of cellular IRES elements, as described previously .
We next compared activities of the cathepsin L IRES and EMCV IRES, one of the most powerful elements known for internal initiation of translation in a broad range of host cells . Thus the cathepsin L IRES and the EMCV IRES, subcloned into pSV-RL-FL bicistronic vector, were transfected into A375SM melanoma cells and H1299 lung carcinoma cells (Figure 5C). Results demonstrated only a 2-fold difference between cathepsin L IRES activity and EMCV IRES activity in both cell lines with the IRES activity lower in H1299 cells. Thus cathepsin L IRES activity was not restricted to melanoma cells and was efficient in initiating translation.
Regulation of human cathepsin L gene expression in hypoxic conditions
An IRES has been identified in genes induced under hypoxic conditions such as VEGF (vascular endothelial growth factor) [21,22] and HIF-1α (hypoxia-inducible factor 1α) . The present demonstration of the presence of an IRES in cathepsin L mRNA and the involvement of cathepsin L in angiogenesis  led us to analyse its regulation by hypoxia.
Thus cathepsin L expression was analysed by Western blotting in A375SM cells grown under hypoxic conditions for 26 h (Figure 6A). Results demonstrated that both cathepsin L forms (i.e. the 29 kDa cathepsin L mature and the 41 kDa procathepsin L forms) were clearly increased. GAPDH expression, usually used as control, was slightly increased under hypoxic conditions (Figure 6A), as described in other cell types . Thus we used two others controls, i.e. RB18A (TRAP220/DRIP205) and lamin A/C, the expression of which was not modified between normoxic and hypoxic conditions (Figure 6A). Similar results were obtained with DM-4, another human melanoma cell line. Indeed, in DM-4 cells, hypoxic conditions also induced a significant increase of cathepsin L forms, especially for the procathepsin L form (Figure 6B). Patterns of intracellular 41 kDa procathepsin L, 34 and 29 kDa cathepsin L mature forms were different between these two melanoma cell lines, as described previously .
Furthermore, human melanoma cells express three splice variants differing in their 5′-UTR . These transcripts, named hCATL-A, hCATL-A II and hCATL-A III, are the result of splicing of intron 1 using the same 3′ splice site located 11 nucleotides upstream of the ATG and different 5′ splice sites located 1210, 1300 and 1355 nucleotides upstream of the ATG respectively. Analysis by semi-quantitative RT-PCR showed that hypoxia induced a significant increase of only hCATL-A expression (Figure 6C). Interestingly, the complete IRES (−1211/−1388) is enclosed only within the hCATL-A transcript variant. In control, expression of RB18A mRNA was not modified under hypoxia (Figure 6C). These results demonstrated that hypoxia induced modification in the ratio of splice variants by increasing expression of the transcript, which contains the IRES.
In addition, we compared the cathepsin L promoter activity under both normoxic and hypoxic conditions. The pGL3-P1/U/I-(−1/−1646) construct, which contained the firefly luciferase gene driven by the cathepsin L promoter, was transfected into A375SM cells, with the reporter vector pRL-SV40, used as an internal control. We also verified that SV40 promoter activity was not modified in A375SM cells grown under hypoxic conditions (results not shown). Hypoxia induced a 37% decrease of relative luciferase activity in transiently transfected cells (Figure 6D, part 1). In comparison, GAPDH promoter activity was increased under hypoxic conditions (Figure 6D, part 2). In contrast with cathepsin L, up-regulation of GAPDH mRNA levels (Figure 6C) correlated with the increase of its promoter activity, which supported that GAPDH was regulated by transcriptional mechanisms in melanoma cells, as described previously in endothelial cells [24,25]. Results demonstrated that up-regulation of cathepsin L expression under hypoxic conditions was not due to transcriptional mechanisms. Using the pSV-RL-U1-FL biscistronic construct (−1/−1388 fragment of the cathepsin L 5′-UTR), cathepsin L IRES activity was analysed in A375SM cells grown under normoxic or hypoxic conditions. As shown in Figure 6(D), part 3, relative luciferase activities were identical, supporting the fact that the cathepsin L IRES is functional under both conditions.
First, we focused on the identification of DNA regulatory sequences involved in cathepsin L gene overexpression in A375SM, a highly tumorigenic and metastatic human melanoma cell line.
We established that cathepsin L gene regulation is certainly complex, as the 5′-flanking sequence, the 3′-flanking sequence and untranslated exons/intron regions were involved. Indeed, our results demonstrated the presence of: (i) inhibitory sequences of transcription in the 5′ region of the cathepsin L promoter, upstream of the GC/CCAAT sites; (ii) silencers in the 3′ region that regulated cathepsin L promoter activity. This was supported by cloning the 1500 bp sequence from the end of the stop codon of cathepsin L in the 3′ region of the luciferase gene downstream of the SV40 late poly(A) signal. In these conditions, the luciferase mRNA sequence was not modified, which allowed us to measure only transcriptional regulation and to exclude the effect on mRNA stability or translation by the 3′-UTR of the cathepsin L gene; and (iii) negative regulatory sequences localized in the intron and the two exons upstream of the translation initiation site. Inside this region, the 5′-UTR of cathepsin L mRNA is the most important for cathepsin L expression. It was described that long 5′-UTRs often confer an inhibitory effect on gene expression due to the inhibition of cap-dependent translation by a stable secondary structure, upstream open reading frames or the presence of specific sequences for inhibitory proteins such as the iron-responsive element . In contrast, the cathepsin L 5′-UTR increased expression, as shown using luciferase reporter constructs and expression vectors encoding an ectopic cathepsin L cDNA with or without the 5′-UTR.
In addition, we demonstrated that the 5′-UTR contained an alternative promoter localized between −1389 and −1489 from the translational start site. Transfection of reporter constructs containing mutations in GC boxes and the CCAAT motif demonstrated that the main promoter in the 5′ region of the major transcription initiation site inhibited the activity of the alternative promoter. However, we previously identified minor transcription initiation sites between −1371 and −1382 using RLM-RACE (RNA-ligase-mediated rapid amplification of cDNA ends) in melanoma cells . Indeed, this alternative promoter seems to be active in melanoma cells and allows synthesis of shorter cathepsin L mRNAs. This −1389/−1489 promoter could mediate approx. 20% of cathepsin L transcription in melanoma cells, based on the percentage of the sequenced RLM-RACE products (22%) , the 24% luciferase activity to the pGL3-P1/U/I-(−1/−1646)Mut construct and the absence of activity of the pGL3-P1-(−1465/−1646)Mut construct. The −1389/−1489 promoter is also functional in other cell types, as shorter cathepsin L mRNAs lacking nucleotides at the 5′-end were identified in a human malignant kidney tumour .
In the present study, we also demonstrated, for the first time, the presence of an IRES in the 5′-UTR of cathepsin L or a member of the lysosomal cysteine protease family, using the bicistronic reporter strategy. Indeed, transfection experiments of bicistronic reporter vectors or RNAs confirmed that the cathepsin L 5′-UTR contains a functional IRES, which is involved in translational regulation. Several human cathepsin L mRNA species differing in the length of the 5′-UTR have been reported previously [27,28]. However, we demonstrated that human melanoma cells synthesize three different splice variants, hCATL-A, hCATL-A II and hCATL-A III . The complete IRES (−1210/−1301) is enclosed only in the hCATL-A transcript variant. Interestingly, cellular distribution analysis in some normal and tumour tissues and cell lines showed that hCATL-A is widely expressed .
Secondly, we analysed cathepsin L regulation under hypoxic conditions, as IRES involvement was described in the regulation of genes induced under hypoxic conditions . In the present study, we demonstrated, for the first time, that under moderate hypoxic conditions (1% O2) cathepsin L intracellular expression was up-regulated in human melanoma cells. Interestingly, in rodent cells grown under severe hypoxic conditions (<0.1% O2), an increase of cathepsin L expression was mentioned [30,31].
In addition, we demonstrated that hypoxia increased cathepsin L mRNA expression, especially of the hCATL-A transcript variant which contains the complete IRES. However, promoter activity was inhibited under hypoxic conditions supporting that cathepsin L was not up-regulated at the transciptional level. Specific up-regulation of the hCATL-A transcript suggested that hypoxia induced either selective modification of mRNA stability of individual splice variants or modification in cathepsin L splicing. Indeed, both mechanisms have been shown to be involved in gene regulation under hypoxia: (i) increase of the stability of VEGF mRNA contributes to its induction , and this stabilization depended on the cooperation of elements in the 5′-UTR, the coding region and the 3′-UTR ; and (ii) differential splicing was also described for others genes, such as telomerase  and X-box binding protein 1 . It should be noted that in both cases, splicing was linked to up-regulation of transcription. Indeed, from our results, down-regulation of cathepsin L transcription suggested that the hCATL-A variant contains elements required for efficient mRNA stabilization by hypoxia.
Our results supported the fact that hCATL-A, the only cathepsin L transcript in which the complete IRES is present, is involved in up-regulation of cathepsin L expression by hypoxia. However, cathepsin L IRES activity was not modified by hypoxia. A similar discrepancy was also observed under hypoxic conditions with other genes which contained an IRES. Indeed, IRES activity of VEGF [21,22] and HIF  were not increased under hypoxic conditions and IRES activity of Tie2 was only slightly stimulated  in bicistronic reporter assays. However, these authors showed using polysome profile analysis that the presence of the IRES allowed translation of these genes during hypoxic conditions [18,22,23]. It is well known that hypoxia down-regulates protein synthesis by inhibiting cap-dependent translation, which accounts for the translation of most mRNAs. Then, translation of factors that are critical for the hypoxic response is maintained by initiation via an internal ribosomal entry mechanism . Taken together, these results suggest that the IRES in the hCATL-A transcript mediates cathepsin L mRNA translation under hypoxic conditions and contributes to the up-regulation of cathepsin L expression in melanoma cells. Further analyses are needed to explore whether other molecular mechanisms are involved in cathepsin L regulation by hypoxia.
In conclusion, our results clearly demonstrate that the 5′-UTR region of cathepsin L mRNA plays a major role in its regulation. For the first time, the presence of an IRES was identified within this 5′-UTR. In addition, hypoxia up-regulates intracellular expression of cathepsin L with a selective increase of the variant which contains this IRES. The increase of cathepsin L expression induced under hypoxic conditions, should be related to the properties of cathepsin L to regulate angiogenic factors [36–38] and to contribute to neovascularization in ischaemia  and in tumour progression . Taken together, these results emphasize that the hypoxic environment of tumours up-regulates cathepsin L expression which promotes in vivo tumour angiogenesis, tumour growth and metastasis.
This work was supported by INSERM (Institut National de la Santé et de la Recherche Médicale), INCa (Institut National Contre le Cancer), ARC (Association de Recherche Contre le Cancer), Ligue National contre le Cancer (Comité de Paris) and MENRS (Ministère de l'Education Nationale et de la Recherche Scientifique).
Abbreviations: CMV, cytomegalovirus; EMCV, encephalomyocarditis virus; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; HIF, hypoxia-inducible factor; IRES, internal ribosome entry site; mAb, monoclonal antibody; RL, Renilla luciferase; RLM-RACE, RNA-ligase-mediated rapid amplification of cDNA ends; RT, reverse transcription; ScFv, single chain variable fragment; UTR, untranslated region; VEGF, vascular endothelial growth factor
- © The Authors Journal compilation © 2008 Biochemical Society