A cryptic promoter in the first exon of the SPG4 gene directs the synthesis of the 60-kDa spastin isoform

Abstract

Background: Mutations in SPG4 cause the most common form of autosomal dominant hereditary spastic paraplegia, a neurodegenerative disease characterized by weakness and spasticity of the lower limbs due to degeneration of the corticospinal tract. SPG4 encodes spastin, a microtubule-severing ATPase belonging to the AAA family. Two isoforms of spastin, 68 and 60 kDa, respectively, are variably abundant in tissues, show different subcellular localizations and interact with distinct molecules. The isoforms arise through alternative initiation of translation from two AUG codons in exon 1; however, it is unclear how regulation of their expression may be achieved.

Results: We present data that rule out the hypothesis that a cap-independent mechanism may be involved in the translation of the 60-kDa spastin isoform. Instead, we provide evidence for a complex transcriptional regulation of SPG4 that involves both a TATA-less ubiquitous promoter and a cryptic promoter in exon 1. The cryptic promoter covers the 5'-UTR and overlaps with the coding region of the gene. By using promoter-less constructs in various experimental settings, we found that the cryptic promoter is active in HeLa, HEK293 and motoneuronal NSC34 cells but not in SH-SY-5Y neuroblastoma cells. We showed that the cryptic promoter directs the synthesis of a SPG4 transcript that contains a shorter 5'-UTR and translates the 60-kDa spastin isoform selectively. Two polymorphisms (S44L and P45Q), leading to an early onset severe form of hereditary spastic paraplegia when present in heterozygosity with a mutant allele, fall a few nucleotides downstream of the novel transcriptional start site, opening up the possibility that they may exert their modifier effect at the transcriptional level. We provide evidence that at least one of them decreases the activity of the cryptic promoter in luciferase assays.

Conclusion: We identified a cryptic promoter in exon 1 of the SPG4 gene that selectively drives the expression of the 60-kDa spastin isoform in a tissue-regulated manner. These data may have implications for the understanding of the biology of spastin and the pathogenic basis of hereditary spastic paraplegia.

Figure 1

Schematic representation of the SPG4 first exon. Translation of spastin initiates from two in-frame start codons (+1 and +259). A uORF overlaps with the first in-frame AUG and may serve to divert some ribosomes to the downstream start site.

Figure 2

Experiments with dicistronic vectors reveal a cryptic promoter. (a) The sequence under analysis was cloned in a dicistronic vector between two different luciferases from Renilla and firefly. HeLa and SH-SY-5Y cells were transfected with the indicated constructs. Cell lysates were prepared 24 hours post-transfection and the activity of the firefly luciferase was normalized to that of the Renilla luciferase. For each construct at least three independent experiments were performed. (b) The same sequence was cloned into a vector lacking the SV40 promoter (pRFΔP). Cell lysates were prepared 24 hours post-transfection and the activity of the firefly luciferase was normalized to that of the Renilla luciferase. For each construct at least three independent experiments were performed. (c) In vitro transcribed dicistronic mRNAs were synthesized from the indicated linearized constructs. HeLa cells were transfected with the capped dicistronic mRNAs and Renilla and firefly activities were measured 8 hours after transfection. Error bars represent standard error of the mean.

Figure 3

Analysis of the SPG4 minimal promoter. (a) Sequence comparison using mVISTA between human and mouse genomic regions upstream of the first ATG of SPG4 demonstrates extensive sequence conservation. (b) Schematic representation of the constructs used. Different genomic sequences were cloned upstream of the firefly luciferase reporter gene. The arrow indicates the transcriptional start sites of the reference SPG4 sequence. The position of a putative CAAT box is shown. (c) HeLa, SH-SY-5Y and HEK293 cells were cotransfected with the indicated constructs and with a CMV-Renilla luciferase plasmid. Cell lysates were prepared 24 hours post-transfection and the activity of the firefly luciferase was normalized to that of Renilla luciferase. For each construct at least three independent experiments were performed using different DNA preparations. Error bars represent standard error of the mean.

Figure 4

Identification of a cryptic promoter in SPG4 exon 1. (a) Schematic representations of the firefly luciferase reporter constructs used. The position of the predicted Sp1 sites is indicated. (b) HeLa, SH-SY-5Y and HEK293 cells were cotransfected with the indicated constructs and with a CMV-Renilla luciferase plasmid. Cell lysates were prepared 24 hours post-transfection and the activity of the firefly luciferase was normalized to that of Renilla luciferase. For each construct at least three independent experiments were performed using different DNA preparations. (c) Mutation of each and both predicted Sp1 sites were generated in the S -207/+259 construct and tested in HeLa cells as described above (n = 3). Error bars represent standard error of the mean. The P-value of Student's t test is shown.

Figure 5

The cryptic promoter mediates expression of the short spastin isoform in vivo. (a) HeLa cells were transfected with a CMV-spastin-GFP, a CMV-spastin-ΔM1 or a spastin-GFP- ΔCMV construct. Cell lysates were prepared 48 hours post-transfection and subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Immunoblotting for transfected spastin was performed with the S51 polyclonal antibody. The CMV-spastin-GFP plasmid drives expression of two spastin isoforms starting from the first and second methionine, as previously described [11]. Consistently, the CMV-spastin- ΔM1 construct produces only the shorter isoform. Notably, the promoter-less spastin construct synthesizes the short isoform, albeit at lower level, indicating that the cryptic promoter of spastin is active in vivo. Below each lane, the amount of transfected cell lysate loaded is indicated. (b) Similar results were obtained in the murine immortalized motoneuronal cell line NSC34. (c) The empty CMV-EGFP vector and CMV-EGFP-STOP-spastin construct were transfected in HeLa cells. Immunofluorescence was performed 48 hours after transfection. Transfected cells were detected by enhanced green fluorescent protein epifluorescence, while spastin was revealed using the S51 polyclonal antibody. Cells expressing high levels of GFP also synthesize low levels of spastin. Note the different pattern of GFP (diffuse) and spastin staining (discrete, as described previously [11]).

Figure 6

The role of c.131C>T and c.134C>A polymorphisms on cryptic promoter activity. (a) Schematic representations of the firefly luciferase reporter constructs used. The position of both polymorphisms is indicated. (b) HeLa cells were transfected with the indicated constructs together with a plasmid containing CMV-Renilla luciferase. Cell lysates were prepared 24 hours post-transfection and the activity of the firefly luciferase was normalized to that of Renilla luciferase. For each construct at least three independent experiments were performed using different DNA preparations. Error bars represent standard error of the mean. The P-value of Student's t test is shown.

Figure 7

An endogenous SPG4 transcript specific for the 60-kDa spastin isoform. (a) 5'-end RACE performed on total RNA extracted from HeLa cells detects a specific product of about 250 base pairs that is lacking in the minus tobacco acid pyrophosphatase control sample. (b) Sequence of the SPG4 first exon starting from the transcriptional start site as defined in public databases and ending with the second ATG in position 259–261. Sp1 sites are underlined, the position of the c.131C>T and c.134C>A polymorphisms and the two in-frame ATGs are shown in bold, while arrows indicate the traditional and novel transcriptional start site found in this study.