Human surfactant protein A (SP-A) is encoded by two functional genes (SFTPA1, SFTPA2) with a high degree of sequence identity. Sequence differences among these genes and their genetic variants have been observed at the 5′ and 3′ untranslated regions (UTRs). In this work, we studied the impact on translation of the SFTPA1 (hSP-A1) and SFTPA2 (hSP-A2) gene 5′ UTR splice variants and 3′ UTR sequence variants, in the presence or absence of poly(A) tail. We generated constructs containing the luciferase reporter gene flanked upstream by one of the hSP-A 5′ UTR splice variants and/or downstream by one hSP-A 3′ UTR sequence variant. mRNA transcripts were prepared by in vitro transcription and used for either in vitro translation with a rabbit reticulocyte lysate or transient transfection of the lung adenocarcinoma cell line NCI-H441. The luciferase activity results indicate that hSP-A 5′ UTR and 3′ UTR together have an additive effect on translation. In this context, the hSP-A1 6A3 and 6A4 3′ UTR variants exhibited higher translation efficiency than the 6A2 variant (P <0.05), whereas no significant difference was observed between the two hSP-A2 3′ UTRs studied (1A0, 1A3). Further sequence analysis revealed that a deletion of an 11-nucleotide (nt) element in both the 6A3 and 6A4 3′ UTR variants changes the predicted secondary structure stability and the number of putative miRNA binding sites. Removal of this 11-nt element in the 6A2 3′ UTR resulted in increased translation, and the opposite effect was observed when the 11-nt element was cloned in a guest 3′ UTR (6A3, 6A4). These results indicate that sequence differences among hSP-A gene variants may account for differential regulation at the translational level.
- surfactant protein A
- untranslated regions
- NCI-H441 cells
pulmonary surfactant is a complex mixture of proteins and lipids that is essential for respiratory function, as it prevents alveolar collapse at low lung volumes. Surfactant protein A (SP-A) is the most abundant protein of this complex, and its functions include participation in innate lung host defense, surfactant structure and physiology, and parturition (7, 54, 73). In humans, SP-A is encoded by the two functional genes SFTPA1 (hSP-A1) and SFTPA2 (hSP-A2) located in the long arm of chromosome 10 (5, 25), with a highly conserved structure (Fig. 1). Based on sequence differences within the coding region, several variants have been identified for hSP-A1 and hSP-A2 with variable frequency in the population (8, 17). At the mRNA level, both splice variants of the 5′ untranslated region (UTR) and sequence variants of the 3′ UTR have been identified for hSP-A1 and hSP-A2 (32, 34). Several 5′ UTR exons have been identified that are alternatively spliced to give rise to specific SP-A1 and hSP-A2 patterns [SP-A1: AD′ (major), ACD′, AB′D′; SP-A2: ABD, ABD′] (Fig. 1). Despite the high degree of sequence identity and structure similarity between hSP-A1 and hSP-A2, functional, structural, biophysical, and biochemical differences among their products have been observed in several in vitro and in vivo studies (17, 19, 49, 50, 64, 65, 69, 70). Regulatory differences among hSP-A 5′ UTR and 3′ UTR variants have been studied, and a differential impact on translation efficiency and mRNA stability has been observed (66, 67). In addition, SP-A gene-specific and variant-specific expression has been shown to be differentially regulated by a number of factors (e.g., hormones), although the mechanisms underlying this regulation remain unclear (4, 25, 31, 67).
Specific hSP-A variants and polymorphisms within the hSP-A gene sequences have been correlated with the development of, or protection against, various lung diseases. Work from our laboratory and others demonstrated that genetic variants of surfactant proteins associate with respiratory distress syndrome and bronchopulmonary dysplasia in the prematurely born infant (12–14, 24, 30, 41, 53, 57). Moreover, aberrant expression of SP-A is a characteristic of many pulmonary diseases and disorders, including asthma, tuberculosis, idiopathic pulmonary fibrosis, and lung cancer, among others (15, 58, 59, 71). SP-A protein levels, as well as the SP-A1/SP-A ratio, have been shown to differ as a function of lung disease and/or age (61). Together, the available literature indicates that the activities and properties of the two human SP-A genes and their variants may not be equivalent, and the overall SP-A function may depend on the relevant levels of hSP-A1 and hSP-A2. It is possible that variant-specific regulatory elements differentially regulate SP-A expression under diverse conditions (e.g., lung disease).
Gene expression is a complex and highly regulated process. Among the numerous mRNA regulatory elements that participate in posttranscriptional regulation, the UTRs have been extensively studied as modulators of protein translation and mRNA stability (28, 35, 39). In recent years, many regulatory elements such as cis-acting factors, AU-rich elements (AREs), microRNA target binding sites, among others, have been identified and characterized in the UTRs, and their role in mRNA translation and stability has been reported in several organisms (2, 20). Furthermore, mutations in these regions have been associated with a number of human diseases, providing further evidence for a role of the 3′ UTR in the control of a variety of biological processes (6, 47, 55). In addition, the poly(A) tail located at the 3′ end of most eukaryotic mRNAs has been extensively studied as a regulator of translation efficiency, mRNA stability, and other functions, including mRNA translocation and turnover (72).
Human SP-A expression is controlled developmentally and by tissue specificity at both transcriptional and translational levels (3, 46). Work from our laboratory has demonstrated a differential impact of the hSP-A1 and hSP-A2 5′ UTR variants on translation efficiency and mRNA stability (66). In this study, we investigated the regulation of translation by the 5′ UTR and 3′ UTR of several hSP-A1 and SP-A2 variants, in combination with other structural mRNA elements known to control translation [i.e., the poly(A) tail]. Our goal was to determine whether these UTRs singly or in combination impart differential regulation among hSP-A1 and hSP-A2 variants, and identify specific regions that may contribute to these differences. Our findings indicate that variant-specific sequences located at the UTRs may differentially regulate SP-A expression at the translational level. We have demonstrated that an 11-nt element, present in all the hSP-A2 variants and the 6A2 hSP-A1 variant, was able to modulate the translation efficiency when placed in a guest 3′ UTR sequence.
MATERIALS AND METHODS
The human lung adenocarcinoma cell line NCI-H441 was obtained from American Type Culture Collection (Manassas, VA) and grown in RPMI 1640 medium (Invitrogen Life Technologies, Carlsbad, CA) with 10% FBS (Summit Biotechnology, Ft. Collins, CO), 1× antimycotic-antibiotic solution (Sigma, St. Louis, MO), and 1% l-glutamine (Sigma). Cells were maintained at 37°C in 5% CO2 atmosphere and subcultured weekly in 10-cm dishes to 80–90% confluence.
The previously generated rpcDNA3/5′-UTR/Luc vector (66) was modified as indicated in Fig. 2A. These modifications included: 1) digestion with HindIII and NruI to replace the SV40 promoter by the T7 promoter; 2) cloning of the 3′ UTR sequence of the human hSP-A1 (6A2, 6A3, 6A4) or hSP-A2 (1A0, 1A3) gene variants downstream of the luciferase cassette; and 3) removal of the bovine growth hormone factor 3′ UTR sequence from the original vector by overlap PCR and site-directed mutagenesis. The final constructs contain two restriction enzyme sites downstream of the reporter gene (SpeI and XhoI). SpeI digestion results in template vectors containing the hSP-A 3′ UTR downstream of the luciferase cassette, whereas XhoI digestion generates plasmids linearized downstream of the luciferase stop codon, without 3′ UTR (Fig. 2A). A control vector lacking hSP-A 5′ UTR (con-Luc) was generated by digestion with HindIII and BamHI followed by T7 polymerase treatment and blunt-end ligation. Overall, a total of 12 constructs containing all possible combinations of hSP-A1, hSP-A2, and control UTRs were generated and used as templates for the in vitro transcription of 16 different experimental mRNAs containing 1) one hSP-A (AD′, ABD, ABD′) or control 5′ UTR upstream of the luciferase gene (4 transcripts); 2) one hSP-A 3′ UTR (1A0, 1A3, 6A2, 6A3, 6A4) downstream of the luciferase (5 transcripts), or 3) both hSP-A1 or hSP-A2 5′ and 3′ UTR combinations (7 transcripts) (Fig. 2B). The modified vectors were confirmed by DNA sequencing (Molecular Genetics Core Facility at Pennsylvania State University College of Medicine).
In Vitro Transcription
After linearization of the constructs with either XhoI or SpeI, in vitro transcription was performed using the mMESSAGE mMACHINE T7 Ultra kit (Ambion). To generate capped mRNA, 1 μg of linearized and purified DNA was incubated with 15 mM NTPs in the presence of ARCA (anti-reverse cap analog) and GTP in a ratio 4:1 (ARCA to GTP) and T7 RNA polymerase enzyme mix for 2 h at 37°C. After treatment with TURBO DNase for 15 min, transcripts were purified by phenol-chloroform extraction and isopropanol precipitation. RNA quality was assessed by 1.5% agarose denaturing gel electrophoresis. A second reporter mRNA, Renilla luciferase, was prepared by in vitro transcription and used as an internal control in the translation reactions.
Poly(A)+ RNA was obtained by mixing 2 μg of each experimental transcript with E-PAP buffer (Ambion), 25 μM MnCl2, and 10 μM ATP in a final volume of 20 μl. The reaction mixture was fractioned into two aliquots, and one unit of E-PAP enzyme was added [poly(A)+ RNA] or omitted [poly(A)− RNA]. These were incubated at 37°C for 30 min, and a dilution was used for either in vitro translation or mRNA transfection. To confirm the addition of poly(A), and the quality of the mRNA after the reaction, we analyzed all the transcripts before translation/transfection by electrophoresis on a 0.8% denaturing agarose gel. A representative gel is shown in Fig. 2C.
In Vitro Translation
Experimental mRNAs (0.36 pmol) were mixed with 20 μM of a methionine-free amino acid mixture, 20 μCi of [35S]methionine (Perkin Elmer), 40 units of RNase inhibitor (Invitrogen), 12.5 fmol (5 ng) of Renilla luciferase mRNA (control), and 20 μl of rabbit reticulocyte lysate (RRL) mix (Promega) in a final volume of 40 μl. The mixture was incubated at 30°C for 90 min.
Luciferase Activity Assay
Luciferase activity was measured in an aliquot using the Dual-Luciferase reporter assay system kit (Promega). One microliter of reaction was mixed with 100 μl of luciferase assay reagent II, and Firefly luciferase activity was recorded with an FB12 luminometer (Zylux, Maryville, TN). Stop & Glo reagent (100 μl) was added to measure Renilla luciferase activity. To evaluate the newly synthesized luciferase protein, a 5-μl aliquot of the reaction was subjected to SDS-PAGE (10%), the gel was dried and exposed to Kodak film, and the intensity of the bands was measured by densitometry (data not shown). Results are shown as the Firefly luciferase activity to Renilla luciferase activity ratio.
Transient mRNA Transfection
Approximately 24 h before transfection, NCI-H441 cells were subcultured into six-well culture plates (1 × 106 cells/well) and incubated overnight. Transfection was performed using the TransIT-mRNA transfection kit (MirusBio). Briefly, 2.7 pmol of experimental mRNA were preincubated with the mRNA boost and TransIT-mRNA reagents in serum-free RPMI 1640 medium, in the presence of 0.5 pmol (200 ng) of Renilla luciferase control mRNA. The complexes were incubated for 3 min at room temperature and delivered to the cells in complete growth medium. Cells were harvested after 90 min, washed with 1× PBS, and rocked for 15 min at room temperature in 500 μl of Dual-Luciferase reporter assay system kit (Promega) passive lysis buffer (1×). Lysates were transferred to 1.5-ml tubes and centrifuged for 1 min at 4°C. Luciferase activity was measured as previously described, in a 50-μl aliquot. The harvesting time point and the amount of mRNA used were chosen from time-course and titration experiments we performed. We optimized the conditions for transfection so that we could use the minimum volume of the poly(A) reaction that would still provide us with adequate levels of luciferase activity. An mRNA concentration of 1.5 μg/μl (2.7 mM) and a harvesting time of 90 min were optimal for transfection.
RNA Secondary Structure Analysis
The RNAfold online program (http://rna.tbi.univie.ac.at/) (42) was used to predict the secondary structures of the 3′ UTR of the hSP-A1 and hSP-A2 variants. This software follows the algorithm published by Zuker and Stiegler (75) to estimate the minimum free energy and the minimum total base pair distance of the RNA secondary structure at 37°C. The centroid secondary structure, with the best possible combination of paired bases, and the minimum free energy (dG) were obtained for the experimental 3′ UTR sequences. We considered unpaired bases and GU wobbles allowance as parameters for the analysis because they resulted in more stable secondary structures.
3′ UTR MicroRNA Binding Sites Prediction
To identify putative human microRNA (miRNA) target binding sites in the hSP-A 3′ UTRs (and potential differences among hSP-A1 and hSP-A2 variants), we analyzed them with the RegRNA (27) and the PITA (Probability Interaction by Target Accessibility) (33) online prediction tools. We followed standard parameter settings to maximize the specificity of the interactions: eight nucleotides minimum seed size (100% complementarity at the 5′ end portion of the interaction), with single mismatches and single GU wobbles allowed (36). Results were considered when the minimal free energy for the predicted binding was below −10 Kcal/mol, since these are likely to be functional in endogenous miRNA expression levels (1).
miRNA Expression Analysis
NCI-H441 cells were cultured and maintained as described. Approximately 1 × 105 cells were recovered by trypsinization, pelleted, and washed with 1× PBS before miRNA preparation with the mirVANA miRNA Isolation Kit (Ambion). The same procedure was followed in a 200-μl aliquot of the commercial RRL. The small RNA fraction obtained was treated with the DNA-free kit (Ambion) to remove possible contaminating DNA. The quality of the small RNA fractions was confirmed by visualization on a 15% denaturing polyacrylamide gel stained with ethidium bromide. Thirty-five nanograms of purified small RNA were poly(A) tailed and reverse transcribed, and a 1:10 dilution of the cDNA was amplified by endpoint PCR with the All-in-One miRNA qRT-PCR detection kit (GeneCopoeia, Rockville, MD) with the 3′ universal reverse primer (GeneCopoeia) and specific forward primers designed based on the miRNA sequences available in the miRBase online database (http://www.mirbase.org/) (23), as follows: hsa-mir-183: TATGGCACTGGTAGAATTCACT, hsa-mir-194-1: TGTAACAGCAACTCCATGTGGA, hsa-mir-194-2: CCAGTGGGGCTGCTGTTATCTG, hsa-mir-449b: AGGCAGTGTATTGTTAGCTGGC, hsa-mir-612: GCTGGGCAGGGCTTCTGAGCTCCTT, hsa-mir-654-5p: TGGTGGGCCGCAGAACATGTGC, hsa-mir-767-3p: TCTGCTCATACCCCATGGTTTCT, hsa-mir-885-3p: AGGCAGCGGGGTGTAGTGGATA, and the small RNA U6 (positive control) CGCAAGGATGACACGCAAATTC. The PCR products were separated on a 3.5% agarose gel stained with ethidium bromide. Negative control reactions in which the reverse transcriptase was omitted were carried out for all the samples and primers.
Deletion-Insertion of the 3′ UTR 11-nt Element
The 11-nt element was removed from the 3′ UTR of the 6A2 variant by site-directed mutagenesis, with specific primers and the QuikChange II XL Site-Directed Mutagenesis Kit (Agilent Technologies, Santa Clara, CA). The same procedure was followed to insert the 11-nt element (CTGCCTGCCCA) at position 405 of the 3′ UTR sequences of the 6A3 and 6A4 variants. In all cases, the resulting constructs were confirmed by sequencing. New mRNA transcripts were prepared as described, and NCI-H441 cells were transfected for 90 min with the wild-type and mutant variants, to compare the luciferase activity among the transcripts.
At least three independent experiments were carried out for poly(A) tailing and translation. Triplicate assays were performed for luciferase activity. Data were analyzed using the standard software program SigmaStat version 3.5. Differences among groups were assessed by the ANOVA test or multiple comparison ANOVA (Tukey test). Results are expressed as means ± SE. Statistically significant differences were considered when P < 0.05.
We used two previously described strategies, in vitro translation using a RRL (18), and mRNA cell transfection (62), to compare the translation efficiency of the Luc reporter gene flanked by different UTR variants of hSP-A1 and hSP-A2. For each transcript, the efficiency of translation was calculated as the ratio of mRNA Firefly to Renilla luciferase activity.
Human SP-A 5′ UTR, 3′ UTR, and Poly(A) Tail Affect Translation in a RRL In Vitro System
hSP-A 5′ UTR splice variants and poly(A) tail regulate in vitro translation of a reporter gene.
The translation efficiency of poly(A)+ and poly(A)− mRNA transcripts containing different hSP-A 5′ UTR variants upstream of the luciferase reporter gene was examined in a RRL system. In this system, both poly(A)+ and poly(A)− mRNAs containing 5′ UTRs showed increased translation compared with the control mRNA lacking 5′ UTR, and the poly(A)+ mRNAs showed a higher level of translation compared with that of their respective poly(A)− mRNAs (Fig. 3). Although both hSP-A1 (AD′) and hSP-A2 (ABD, ABD′) variants displayed higher translation than the control without 5′ UTR (con-Luc) in the presence and absence of poly(A), the hSP-A2 5′ UTRs stimulated translation with significantly higher efficiency than the hSP-A1 5′ UTR (ABD = ABD′ > AD′ > con-Luc; P < 0.05). The Renilla luciferase absolute values did not differ greatly among samples [mean = 93.7 ± 2.5 × 105 relative luciferase units (RLUs)]. Similar results were observed for 35S total in vitro protein synthesis detected by fluorography (data not shown).
hSP-A 3′ UTR increased in vitro translation in the absence of poly(A).
Translation of poly(A)− transcripts in a RRL system was improved (compared with con-Luc) when different hSP-A 3′ UTR variants were placed downstream of the luciferase gene, in the absence of 5′ UTR. However, this effect was not observed for the equivalent poly(A)+ mRNAs (Fig. 3). The newly translated 35S-labeled luciferase showed a similar effect (data not shown).
hSP-A 5′ UTR, 3′ UTR, and the poly(A) tail together contribute to the translational control of hSP-A gene variants in vitro.
An interesting effect was observed when the luciferase reporter gene was flanked by both hSP-A 5′ UTR and 3′ UTR. In the absence of poly(A), in vitro translation over control (con-Luc) was significantly higher for all the 5′ UTR and 3′ UTR hSP-A1 and hSP-A2 combinations compared with mRNAs containing only 5′ UTR or only 3′ UTR, although this effect was not additive (Fig. 3). These mRNAs did not show significantly different activities over control when they contained poly(A) tail. Interestingly, translation of hSP-A2 transcripts containing both 5′ UTR and 3′ UTR displayed a reduced translation compared with hSP-A2 poly(A)+ mRNAs containing only 5′ UTRs (Fig. 3). In vitro 35S protein synthesis also followed a similar pattern for these transcripts (data not shown).
The UTRs of the hSP-A1 and hSP-A2 Gene Variants Positively Affect Luciferase Expression in Cell Culture
The lung adenocarcinoma cell line NCI-H441 model was chosen for this study because it expresses SP-A protein; therefore, it is likely to contain any potential regulatory factors specific to SP-A. Cells were cotransfected with the Renilla luciferase control RNA and poly(A)+ or poly(A)− experimental transcripts containing hSP-A UTRs, as described in materials and methods. Time-course experiments were run to determine the optimal harvesting time (Fig. 4), as well as concentration-course experiments (not shown). A total of 2.7 pmol of mRNA and a harvest time of 90 min after transfection were selected for subsequent experimentation, as described in materials and methods.
Figure 5 shows the results of translational efficiency measured as the Firefly/Renilla luciferase activity ratio. As observed for the RRL method, in this system the Renilla luciferase values did not show variation among samples (mean = 79,995 ± 2,214 RLUs). In agreement with the results obtained with the RRL system (Fig. 3), NCI-H441 cells transfected with poly(A)+ transcripts containing hSP-A1 and hSP-A2 5′ UTR variants displayed 1) an increase in translation compared with con-Luc (i.e., lacking 5′ UTR) and 2) a higher efficiency of translation for transcripts containing the hSP-A2 5′ UTR variants (ABD, ABD′) compared with those containing the hSP-A1 variant (AD′) upstream of the luciferase gene (Fig. 5A). In the absence of poly(A) tail, translation of transcripts containing hSP-A 5′ UTRs was higher than control (without 5′ UTR), although the luciferase activity levels for these transcripts were very low and we did not consider them in the statistical analysis. In all cases, the addition of a poly(A) tail had a remarkably positive effect on translation, when the 3′ UTR was omitted (Fig. 5A).
When the effect of positioning different hSP-A 3′ UTRs downstream of the luciferase gene was investigated, a positive effect on translation was observed, although no differences among variants were detected (Fig. 5B). The addition of a poly(A) tail did not modify this effect in most of the 3′ UTRs studied, except for the hSP-A1 6A3 variant, which displayed a significantly higher luciferase activity (Fig. 5B). When both hSP-A 5′ UTR and hSP-A 3′ UTR variants were included in the same luciferase transcript, a differential impact on luciferase translation was observed among the variants tested (Fig. 5C). The poly(A) tail did not alter translation in transcripts of hSP-A2 variants, a finding similar to that observed in the RRL system (Figs. 3 and 5C). However, for some of the hSP-A1 variants, the inclusion of the poly(A) tail showed a significant enhancement in translation (Fig. 5C). The translational efficiency was uneven among the hSP-A1 variants, with the AD′-Luc-6A3 and AD′-Luc-6A4 showing higher activities. Moreover, the addition of a poly(A) tail displayed an additive effect on the translation of these two transcripts, and with a significantly higher effect being observed for the hSP-A1 6A3 variant.
An 11-nt Deletion-Insertion Element Alters mRNA Secondary Structure Properties
To identify potential regulatory elements in the experimental transcripts that may explain the experimental data above, we compared the sequences of the SP-A 3′ UTR variants with the online alignment software ClustalW2 (http://www.ebi.ac.uk/Tools/clustalw2). In agreement with our previously published data (67), sequence similarity among variants ranged from 98.4% to 99.4% for hSP-A1, and the similarity was 99.3% between SP-A2 variants. Specific single nucleotide polymorphisms were also identified throughout the 3′ UTR. As previously described (26, 56, 67), at position 405 in the 3′ UTR, an 11-nt sequence element is present in the hSP-A2 1A0 and 1A3 variants, and the hSP-A1 6A2 variant, but is absent in both hSP-A1 6A3 and 6A4 variants. To determine whether the 11-nt element modified the mRNA configuration, we analyzed the hSP-A 3′ UTR sequences with the RNAfold online software (see materials and methods). We compared the optimal minimum free energy (dG) secondary structures of the complete 3′ UTR variants, and the specific region containing the 11-nt element, because this region displayed a differential secondary structure pattern among variants (Fig. 6). The 3′ UTR variants showed similar minimum dG values for their secondary structures, as follows: 1A0 (1,308 nt): −434.93 kcal/mol, 1A3 (1,308 nt): −438.17 kcal/mol; 6A2 (1,304 nt):−438.64 kcal/mol, 6A3 (1,292 nt): −438.24 kcal/mol and 6A4 (1,293 nt): −435.04 kcal/mol. However, when the sequence surrounding the 11-nt deletion-insertion element was analyzed among hSP-A1 variants (nt 383–500/383–489), a higher dG was observed in the variants that contain the 11-nt element [1A0 (117 nt):−32.68 kcal/mol, 1A3 (117 nt):−33.00 kcal/mol, 6A2 (117 nt):−32.5 kcal/mol, 6A3 (106 nt): −46.7 kcal/mol and 6A4 (106 nt): −44.9 kcal/mol]. In addition to the lower stability displayed by 6A2, a secondary structure with lower base pair probabilities and higher positional entropy was observed for this variant in the region studied (Fig. 6).
Identification of Variant-Specific miRNA Predicted Binding Sites at the hSP-A 3′ UTR
Since associations between enrichment and depletion of miRNA binding sites at the 3′ UTR and mRNA expression have been demonstrated in various systems (60), we performed in silico predictions to determine whether the 11-nt element induced differential miRNA binding sites among variants. We screened the complete 3′ UTR sequence by two different online miRNA binding site prediction tools (RegRNA and PITA) to exclude possible algorithm-specific artifacts. Among the nearly 60 putative binding sites identified for all the variants, 19 were common for hSP-A1 and hSP-A2, 12 were specific for hSP-A1, and 20 were specific for hSP-A2 (Table 1). In addition, both programs predicted seven specific miRNA binding sites only in the 1A0, 1A3, and 6A2 variants: hsa-mir-183, hsa-mir-194, hsa-mir-449b, hsa-mir-612, hsa-mir-654–5p, hsa-mir-767–3p, and hsa-mir-885–3p. Interestingly, these sites were all located in positions adjacent to the 11-nt element (Fig. 7).
miRNA Expression in NCI-H441 Cells
Next, we investigated the expression of the predicted miRNAs whose binding sites were specific for the variants that contain the 11-nt element (1A0, 1A3, and 6A2) in NCI-H441 cell extracts and in the RRL system, since the RRL is supplemented by tRNAs and other small RNAs (Promega, personal communication). The presence of small RNAs was confirmed in both NCI-H441 cell lysates and RRL by PAGE as described in materials and methods (Fig. 8A). The expression of the miRNAs tested was negative in the RRL extract, as well as in the controls where the reverse transcriptase was omitted (Fig. 8B). The expression of hsa-mir-183, hsa-mir-449b, hsa-mir-654–5p, and hsa-mir-885–3p was confirmed in NCI-H441 cell extract. A lower but positive expression of hsa-mir-612 was also detected in the lung cell extract (Fig. 8B).
The 11-nt Insertion Element Negatively Affects Translation of the Reporter Gene
To investigate whether the 11-nt element played a role in translation, we cloned the 11-nt sequence into the 3′ UTR of the hSP-A1 variants that lacked this element (6A3, 6A4) and obtained new transcripts for NCI-H441 cell transfection. In addition, we removed the 11-nt element from the hSP-A1 6A2 variant 3′ UTR. Transient transfection experiments were carried out with these transcripts, and Luc activity was determined. Comparison of translation efficiency revealed that the 11-nt element negatively affects translation in the presence and absence of poly(A) tail (Fig. 9).
The human SP-A locus consists of two functional genes (hSP-A1 and hSP-A2). These have been identified with extensive heterogeneity at several regions, including splice variants at the 5′ UTR and sequence variants at the 3′ UTR. The goal of this study was to investigate the ability of the hSP-A1 and hSP-A2 UTRs, and the poly(A) tail, to regulate translation. By using a RRL and a lung cell culture system, we found a positive impact of the hSP-A UTR sequences on the translation of the luciferase reporter gene. However, differences in the response were observed between the UTR sequences of the hSP-A1 and hSP-A2 genes, and among hSP-A1 gene variants. Specifically, with regards to 3′ UTR, the presence or absence of an 11-nt element differentially affected the luciferase translation efficiency between variants that had and variants that lacked this 11-nt element. Secondary structure and miRNA binding prediction analysis indicated significantly different stabilities and variant-specific binding sites for several miRNAs, respectively, pointing to the importance of this 11-nt element in the differential translation between certain groups of hSP-A variants. Moreover, the removal of this element from the 3′ UTR of the SP-A1 6A2 variant, or its positioning at the 3′ UTR of the SP-A1 variants where it is naturally absent, had opposite effects on translation. Interestingly, this effect was more evident when the transcripts (6A2 and 6A3) contained a poly(A) tail. Recent studies have shown that both poly(A)-dependent and poly(A)-independent mechanisms mediate the inhibition of protein translation by miRNAs, but a stronger repression is observed for polyadenylated mRNAs (11). In addition to the miRNA-mediated mRNA decay, a direct interaction between specific proteins from the RNA-induced silencing complex (RISC) (e.g., GW182) and the poly(A) binding protein (PABP) has been demonstrated and proposed as another mechanism used by miRNAs to decrease translation, involving deadenylation (10). Thus, if the presence of the 11-nt element in our experiments is associated with a specific miRNA inhibition mechanism, we cannot discard a role of the poly(A) tail in this repression. Future experiments are needed to explore the mechanisms underlying the effects observed in our translation system.
Interactions between cis-acting mRNA elements (usually located in the UTRs) and trans-acting factors (e.g., proteins, miRNA) are known to influence translation (40), and functional and structural communications between the 5′ and 3′ RNA termini, mediated by cis- or trans-acting factors, have been proposed in a number of gene expression studies (28, 44). Trans-acting factors have been shown to interact with several RNA structures such as the 5′ CAP, the poly(A) tail, and the UTRs (40), and it is well known that the poly(A) tail functions as a positive regulator of translation and mRNA stability in the majority of the eukaryotic transcripts (21).
In the presence of hSP-A 5′ UTRs, the Luc activity increase was additive with the poly(A) tail, indicating the possibility of interactions between these two RNA regulatory structures, or among translation factors that may interact with these structures (e.g., via mRNA circularization). However, in the presence of both hSP-A 5′ and 3′ UTRs, the poly(A) tail significantly increased the translation efficiency in only two of the hSP-A1 variants tested (AD′-Luc-6A3 and AD′-Luc-6A4), and this effect was only observed in the cell lung system. It is possible that sequence differences among the hSP-A 3′ UTR variants determine regulatory binding elements that, via interactions with specific factors, bring about the differential translation observed. In the absence of 3′ UTR, but in the presence of poly(A), transcripts containing hSP-A splice variants with exon B at the 5′ UTR (ABD, ABD′) showed a higher translation efficiency compared with those containing the AD′ variant, indicating that exon B may act as a translation enhancer, in agreement with our previous findings (66).
In the absence of 5′ UTR, the results indicated that the impact of the hSP-A 3′ UTR sequences on translation differs between the two experimental systems. In the RRL system, transcripts containing 3′ UTRs variants of the hSP-A1 or hSP-A2 genes displayed similar translation levels. The luciferase activity of transcripts containing both 3′ UTR and poly(A) was not significantly different from that of transcripts containing only hSP-A 3′ UTR or poly(A). However, in the lung cell system, an additive effect of 3′ UTR and poly(A) was observed in transcripts with the hSP-A1 6A3 3′ UTR variants, in the presence or absence of 5′ UTR. Variant-specific interactions among these regulatory regions, as well as the presence of specific regulatory factors present in the lung cell line, could explain the results observed.
It is known that the biological activities of the mRNA regulatory elements depend on both their primary and secondary structures (48). We have previously demonstrated that the hSP-A 5′ UTR splice variants exhibit differential stability and secondary structures (66) that may account for differences in translation. Here, we extended our analysis to the 3′ UTR to investigate whether the natural sequence differences observed among variants could influence the mRNA secondary structure and translation. The 3′ UTR functions include the regulation of translation and mRNA stability, as well as targeting transcripts to specific compartments in the cell (2). Moreover, because of its ability to interact with numerous protein factors and miRNAs that may be altered in human diseases, it has been postulated that the 3′ UTR is a pathological “hot spot” (45). Furthermore, a cross talk between the 3′ UTR and the poly(A) tail has been previously proposed and related to posttranscriptional events (9, 21, 28). It is possible that the ability of the variants to bind regulatory factors at the UTRs differs, and this may account for differential protein expression of hSP-A1 and hSP-A2 variants under normal or compromised conditions, such as lung disease. In a previous work, variations in the hSP-A1 levels have been correlated with pulmonary health status and age (61). Although comparisons among genotypes and SP-A levels had not been determined, this is a worthwhile future endeavor.
Sequence alignments and comparisons among the hSP-A variants confirmed the previously described elements within the 3′ UTRs (67). The minimum AU-rich element motif UUAUUUAUU described to effectively destabilize mRNA (74) is present at position 926–935 in all the hSP-A2 variants, but not in the hSP-A1 variants. In this study, we focused the analysis on an 11-nt sequence element that is present in all the hSP-A2 and the hSP-A1 variant 6A2, but it is absent in the hSP-A1 6A3 and 6A4 variants (which exhibited higher levels of translation). The secondary structure analyses indicate that the presence or absence of the 11-nt element promoted the generation of secondary structures with differential stabilities. Interestingly, the variants in which this 11-nt element was absent displayed lower free energies and higher base pair probabilities, indicating more stable RNA secondary structures. It is possible that this mRNA element could bind specific trans-acting factors that regulate translation and/or increase or decrease the overall mRNA stability. In addition, the 11-nt element generated seven specific predicted miRNA interacting sites for the 1A0, 1A3 (hSP-A2), and the 6A2 (hSP-A1) variants. These miRNA binding sites were not predicted for the hSP-A1 6A3 and 6A4 variants, indicating that these sites may provide variant-specific targets for trans-acting factors. It is possible that some, if not all of them, participate in the observed differential regulation of luciferase translation among variants.
MicroRNAs are small non-protein coding RNAs that interact with the 3′ UTR of a target gene and regulate (repress) translation and/or target the mRNA for degradation (11, 52). The interaction of the RISC complex and mRNA binding proteins such as PABP, and the initiation factor complex eIF4F has been described for multiple miRNAs, and both CAP and poly(A) structures were reported to be essential for miRNA-mediated repression of translation (11, 29, 43, 63). There are more than 800 human miRNAs described (22), and their function has been studied by 1) the use of software prediction tools, 2) protein translation analysis by in vitro systems that overexpress the miRNA of interest, and 3) cloning/mutation of the miRNA target site in the 3′ UTR of a reporter gene (51). By the use of bioinformatic prediction tools, we found hSP-A1- and hSP-A2-specific miRNA target sites as well as variant-specific target sites. As mentioned above, seven putative miRNA binding sites were specific for the hSP-A1 6A2 and all the SP-A2 variants, and their binding sites are located within the 3′ UTR region that contains the 11-nt sequence element. Expression of most of these predicted miRNAs has been confirmed in human and animal lung and cell models (38), and some of these miRNAs have been shown to regulate protein expression in lung cancer tissues (37, 68). In this work, the seven miRNA target sites of interest are within the 3′ UTR cloned downstream of a reporter gene (Luc), and when they are naturally deleted, as is the case with the 6A3 and 6A4 variants, or experimentally deleted, as in AD′-Luc-6A2 del (Fig. 9), the luciferase translation is improved, indicating that it is likely that these miRNAs are responsible for the differential translation observed among variants in the NCI-H441 cell system. In addition, we demonstrated the presence of four miRNAs that potentially interact with the 11-nt element in the human lung cell system, and the absence of these potential inhibitors of translation in the RRL. This difference, together with the presence/absence of other modulators of translation (e.g., lung cell-specific factors), could explain the differences in the translation levels obtained with the two experimental designs for the SP-A variants.
In summary, we investigated whether mRNA regulatory elements [i.e., the UTRs and the poly(A) tail] and/or their possible interactions play a role in the translation of hSP-A1 and hSP-A2 genes. By using two different in vitro systems, we have demonstrated that the UTRs of the hSP-A1 and hSP-A2 gene variants, and the poly(A) tail, differentially regulate the translation of a reporter gene. A specific 11-nt element, absent in the 3′ UTR of hSP-A1 6A3 and 6A4 variants, but present in the rest of the 3′ UTR variants tested, 1) correlated with decreased translation efficiency (if present), 2) differentially affected the predicted mRNA secondary structure, and 3) contained seven specific sequence binding sites for miRNAs. We have demonstrated that four of these seven miRNAs were present in the NCI-H441 cells and absent from RRL. We speculate that variant-specific sequence elements located at the UTRs of hSP-A, together with cell-specific regulatory factors (i.e., miRNA), are responsible for the differential regulation observed between hSP-A1 and hSP-A2 and/or between groups of hSP-A variants with and without the 11-nt element.
This work was supported by National Heart, Lung, and Blood Institute Grant HL-34788.
No conflicts of interest, financial or otherwise, are declared by the author(s).
We thank Susan DiAngelo and Joseph Bednarczyk for expert technical support, and the Pennsylvania State University College of Medicine core facility for DNA sequencing and oligonucleotide synthesis.
- Copyright © 2010 the American Physiological Society