| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
,1
Departments of
* Internal Medicine and
Obstetrics and Gynecology, University of Iowa and the
Veterans Affairs Medical Center, Iowa City, Iowa, USA
1Correspondence: Department of Internal Medicine and Graduate Program in Molecular and Cellular Biology, E300 GH, University of Iowa Carver College of Medicine, 200 Hawkins Dr., Iowa City, IA 52242-1081 USA. E-mail: christie-thomas{at}uiowa.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
4 Kb. Northern blot analysis and 3' rapid amplification of cDNA ends (RACE) in placenta confirmed the existence of distal intronic sFlt1 cleavage sites that give rise to a sFlt1 transcript of 7 Kb. The identity of the distal signal sequences were then confirmed by mutagenesis of putative signal elements in a polyadenylation reporter assay. We demonstrate the heterogeneity of human sFlt1 that arises from alternate splicing and from alternative polyadenylation directed by strong intronic poly(A) signal sequences leading to C-terminal variants and to an sFlt1 transcript with a large 3' UTR containing several AU rich elements and poly(U) regions that may regulate mRNA stability.—Thomas, C. P., Andrews, J. I., Liu, K. Z. Intronic polyadenylation signal sequences and alternate splicing generate human soluble Flt1 variants and regulate the abundance of soluble Flt1 in the placenta.
Key Words: angiogenesis • soluble VEGF receptor • preeclampsia
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2
3)
.
Flt1 is a transmembrane receptor, although a shorter variant produces a soluble form of this receptor (sVEGFR1 or sFlt1). The Flt1 and sFlt1 mRNAs arise from the gene FLT1 and have a common transcription start site but have distinct 3' ends with sFlt1 being generated by upstream intronic polyadenylation. The full-length Flt1 contains 30 spliced exons and is translated into a 200 kDa transmembrane protein with an extracellular N-terminal ligand-binding domain, a single membrane-spanning segment and a C-terminal intracellular segment that carries two tyrosine kinase domains. sFlt1 shares the first 13 exons with Flt1 and its translated protein has the ligand-binding domain but lacks the membrane-spanning and C-terminal domains (4
5
6)
. As a consequence, sFlt1 is secreted as a 100 kDa protein, which can bind VEGF with high affinity and function as a circulating VEGF antagonist.
Although the physiological functions of sFlt1 are not well established it is widely expressed by vascular endothelial cells and is thought to play a major role in regulating angiogenesis. Recent studies have begun to establish physiological functions of sFlt1 in diverse tissues such as the cornea and the uterus and demonstrate a pathophysiological role of sFlt1 in pregnancy. For example, vascular repair of the uterine endometrium after menstruation coincides with a rise in KDR phosphorylation and a fall in local sFlt1 levels (7)
. Second, selective expression of sFlt1 without corresponding expression of Flt1 appears to be necessary to maintain the avascularity of the cornea (8)
. Placental expression of sFlt1 increases in pregnancy, and there is emerging evidence that increased circulating sFlt1 levels may contribute to the development of hypertension and proteinuria in preeclampsia (9
10
11
12
13)
. Plasma levels of sFlt1 negatively correlate with the development of the ovarian hyperstimulation syndrome and appear to be a predictor of patient outcome in acute myeloid leukemia (14
, 15)
. It is thus important to understand the molecular underpinnings of sFlt1 expression in the placenta and in other tissues.
Although previous work in transfected HEK293 cells has identified cis-elements within the mouse flt1 gene that may regulate sFlt1 protein production, little is known about the regulation of sFlt1 in vascular endothelial cells or in cytotrophoblasts (6)
. In preeclampsia, the large increase in sFlt1 expression in placenta and in placental villus explants is not always matched by a corresponding increase in Flt1 expression and in some cases is associated with reduced Flt1 expression, suggesting that some part of the increase in sFlt1 expression seen in the placenta in preeclampsia may be post-transcriptional (16
17
18)
. Post-transcriptional regulation of sFlt1 may occur at the level of mRNA splicing, cleavage, polyadenylation, mRNA stability, or translational efficiency or a combination of these processes.
To begin to address the mechanisms that post-transcriptionally regulate sFlt1, we mapped human sFlt1 mRNA cleavage sites, identified polyadenylation signal sequences, and defined the 3' extent of these transcripts. We identified a previously unknown sFlt1 transcript that is cleaved within an alternatively spliced exon downstream of exon 14 and is predicted to encode a C-terminal variant of sFlt1 with an unusual polyserine tail. We also demonstrate that cleavage of the previously identified human sFlt1 transcripts is directed by three distinct intronic poly(A) signal sequences within intron 13 leading to a unusually large 3' untranslated region (UTR) containing several potential AU rich elements and poly(U) regions that may regulate mRNA stability.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Primary cytotrophoblasts were isolated from term placenta using a modification of the Kliman method as described by Nelson et al. (20
, 21)
. The HTR-8/SVNeo cell line was a gift from Charles Graham and was grown in DMEM/F12 (1:1) with 10% serum (22)
. Total RNA from HUVEC, primary cytotrophoblasts and HTR-8/SVNeo was prepared with the RNeasy Mini Kit (Qiagen, Valencia, CA, USA). RNA from human placenta was prepared using TRI reagent (Molecular Research Center, Cincinnati, OH, USA) according to manufacturer’s instructions.
Ribonuclease protection assay (RPA).
To measure Flt1 mRNA, an Flt1 cDNA was amplified from human placenta by PCR using the sFlt1_F primer and Flt1_R1 (see Table 1
). The 364 bp product was cloned into pcDNA3 (Invitrogen, Carslbad, CA, USA) and used as a template for synthesis of a cRNA probe. This cRNA probe is predicted to protect a 364 nt Flt1 product and a 269 nt sFlt1 product.
|
To measure sFlt1 mRNA levels in various tissues, an sFlt1 cDNA was amplified by PCR using primers: sFlt1_F and sFlt1_R. The 444 bp PCR product was cloned into pCRXL-topo (Invitrogen) digested at an internal BalI and used as a template for synthesis of an antisense cRNA probe. This cRNA probe is predicted to protect a 315 nt sFlt1 product and a 141 nt Flt1 product (probe 1,
Fig. 3
).
|
|
|
To map native sFlt1 mRNA cleavage sites, a genomic DNA fragment was amplified from human placenta using hFlt1_intron_F2 and R2, cloned into pCRXL-topo, digested at an internal EcoN1 site and then used as a template (probe 2, Fig. 3
) for cRNA synthesis. Another probe was made by subcloning a 400 bp HindIII fragment from cloned intron 13 (see below and probe 3, Fig. 3
) into pcDNA3, which was then linearized. To map mRNA cleavage from the polyadenylation signal reporter vector, pCβs-sFlt1distal, the constructs were linearized and used to synthesize cRNA probes as described (23)
. In all cases ribonuclease protection assays were performed as described previously (24)
.
To measure the abundance of a C-terminal sFlt1 variant (sFlt1-e15a), an sFlt1 cDNA was amplified by PCR using primers hFlt1_exon13F2 and hFlt1_exon15R1. The 398 bp PCR product was cloned into pCRXL-topo (Invitrogen) and used as a template for synthesis of an antisense cRNA probe. This cRNA probe is predicted to protect a 398 nt Flt1 product, a 291 bp sFlt1 variant (sFlt1-e15a) and a 144 nt sFlt1 (sFlt1-i13) product (
Fig. 5
A).
|
|
Northern blotting
A cDNA probe common to both Flt1 and sFlt1 was created by amplifying a 973 bp fragment from endothelial cell cDNA using primers hFlt1_F2 and R2 and an internal 618 bp Mfe1-NdeI fragment was isolated. To examine sFlt1 expression, a 6.2 Kb genomic fragment that included exon 13, intron 13, and exon 14 was amplified using primers FltintronF1 and FltintronR1 within flanking intron 12 and intron 14. From this larger DNA fragment, a 668 bp SmaI-PvuII piece (Probe 4, Fig. 3
) was digested and used as an sFlt1-specific probe. To determine the size of the C-terminal sFlt1 variant (sFlt1-e15a), a 327 nt PCR product was amplified from placenta with primers hFlt1_exon14F1 and hFlt1_exon15aR1 and used as a probe. Multiple tissue Northern blots obtained from Clontech (Mountain View, CA, USA) were hybridized to each radiolabeled probe and were hybridized as described previously (25)
3' Rapid amplification of cDNA ends (RACE)
Several primers were used to map the 3' end of sFlt1 mRNA including sFlt1_F, hFlt1intronF4, F5 and F7 (see Table 1
). The human placenta Marathon-ready cDNA (Clontech) has an adapter sequence ligated to both ends and was used as a template for 3' RACE. Gene-specific reverse primers were used with adapter-specific primers in PCR reactions using Advantage 2 Polymerase Mix (Clontech). Amplified products were cloned into pCRXL-topo and individually sequenced.
Polyadenylation reporter assay
A 562 bp EcoRV-NheI fragment that included the putative distal poly(A) signal sequences at the 3' end of intron 13 was isolated by restriction digestions from the previously amplified intron 13 genomic fragment and ligated into the multiple cloning site of the polyadenylation signal reporter vector, pCβS (gift from David Fritz) to create pCβS_sFlt1distal. pCβS clones were introduced into HTR-8/SVNeo by electroporation at 600 µF and 200 V with Gene Pulser II (Bio-Rad; Hercules, CA, USA) as described previously, then plated in standard culture media and RNA extracted 48 h later (26)
.
To determine the role of putative signal sequences, the distal sFlt1 sites were selectively mutated using the QuikchangeTM site-directed mutagenesis kit following manufacturer’s instructions. The 1st distal putative signal sequence element AATAAA (A1) was mutated to AAGCAA with the primers polyA1_mutF and R and the 2nd distal putative signal sequence AATAAA (A2) was mutated to AAGGGA with the primers polyA2_mutF and R (Table 1)
. Two other putative signal sequences adjacent to A1 were also individually mutated: ATTAAA (A3) to ATTCGA and TATAA (A4) to TACGTA (see Fig. 5
) using primers shown in Table 1
.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 5
, 27
, 28)
. The mRNAs have a common transcription start site with Flt1 containing 30 spliced exons. sFlt1 shares the first 13 exons but reads through intron 13 utilizing an alternative poly(A) signal sequence(s) within this intervening sequence to create a shorter transcript that lacks sequences from exon 14 to 30 (Fig. 1
A). The translation stop codon for sFlt1 is seen within a 100 bp of the skipped 5' splice site of exon 13 while the stop codon for Flt1 is within exon 30 (4
5
6)
. The sFlt1 mRNA cleavage sites and the precise extent of the sFlt1 3' UTR within the placenta and in vascular endothelium are not known.
Since sFlt1 and Flt1 are transcripts that arise from the same gene we designed RPA probes that would detect both transcripts simultaneously in expressing tissues. The first probe, which is transcribed from an Flt1 cDNA, includes sequences from exons 13 and 14 and protects two mRNAs, a larger fragment corresponding to Flt1 and a smaller fragment corresponding to sFlt1. We examined human placenta, primary human cytotrophoblasts, and HUVEC and demonstrated that in each of these sources, steady state levels of sFlt1 appeared to be more abundant than Flt1 (Fig. 1B
). To confirm the validity of this data a second probe was designed, which is transcribed from a sFlt1 cDNA and includes sequences from exon 13 and contiguous intron 13 (Fig. 1C
). This probe also distinguishes between sFlt1 and Flt1 with the smallest fragment corresponding to Flt1. Using this probe to detect FLT1 transcripts in these cells and tissues confirmed that sFlt1 mRNA is present at much greater levels than Flt1 even though they are transcribed from the same gene, have a common transcription start site and share transcriptional regulatory sequences. These data suggest that post-transcriptional regulation may be important in determining the relative abundance of sFlt1 mRNA in these sites.
We first attempted to define potential cleavage and polyadenylation sites for sFlt1 by analysis of the sequence of FLT1 for polyadenylation signal sequences. Since the 3' end of sFlt1 is derived from sequence within intron 13 we scanned this entire 5.5 Kb region for the signal sequence AAUAAA and the less used signal sequence AUUAAA. There were 5 copies of the classic signal sequence and another 5 copies of the 2nd signal sequence within intron 13 (Fig. 2
A). A review of sFlt1 cDNA sequences catalogued under FLT1 in Entrez Gene (http://www.ncbi.nlm.nih.gov/entrez) showed just one sequence with a poly(A) tail. This cDNA, BC039007 has a poly(A) tail that follows the signal sequence ATTAAA immediately after the sFlt1 stop codon. The stop codon TAA is itself a part of this poly(A) signal sequence and defines a 3' UTR that is just 30 bp long. To determine if any of other putative polyadenylation signal elements within intron 13 could function as true poly(A) signal sequences, we tested intron 13 with the poly(A) prediction tool, polyadq (http://rulai.cshl.org/tools/polyadq/polyadq_form.html) (29)
. The sequence analysis predicted that three signal sequences were strong poly(A) signal sequences. Of these sequence elements 2 of 3 were at the extreme 3' end of intron 13, which suggests that the sFlt1 3' UTR may be as much as 4 Kb in length. We then searched the human EST database using the 3' end prediction tool EST Parser (http://tagc.univ-mrs.fr/estparser/) and found several EST clones isolated/amplified from normal tissues or cells where an oligo-dT primer had been used for reverse transcription (Table 2
). These clones have 3' ends that correspond to mRNA cleavage sites at the 3' end of intron 13. These cleavage sites are downstream of a cluster of putative poly(A) signal sequences (1st distal) 4055 to 5005 nt into intron 13 or downstream of another classic signal sequence (2nd distal) 4224 nt into intron 13. This analysis suggests that in some cases the 3' end of sFlt1 may be far downstream of its translation stop codon (30)
.
|
Since our bioinformatic analysis suggested possible diversity in the length of the 3' UTR of sFlt1 arising from usage of alternative polyadenylation signal sequences within intron 13, we set out to systematically identify the 3' end of sFlt1 in placenta and in HUVEC. To begin to map mRNA cleavage sites by RPA we designed several cRNA probes from intron 13 sequence and adjacent portions of exon 13 or exon 14. Probe 1, complementary to exon 13 and contiguous portions of intron 13 simultaneously detects Flt1 (exon 13 protected fragment) and sFlt1 (exon 13 and intron 13 fragment), and we were able to detect both forms in HUVEC and in placenta (Fig. 2B
). The results demonstrate the heterogeneity in length of the sFlt1 3' UTR with two sFlt1 transcripts identified in HUVEC and placenta. The smaller transcript (sFlt1 small in Fig. 3B
) corresponds to the first polyadenylation site (proximal pA) noted in Entrez Gene (BC039007) and contributes to 30% of total sFlt1 transcripts in both HUVEC and placenta that terminate within intron 13. The larger transcript (sFlt1 large in Fig. 3B
) arises from one or more poly(A) signal sequences within intron 13 whose cleavage site is distal to the probe. Probe 2 was designed to overlap the region bearing the second classic poly(A) signal sequence and detects a single protected fragment corresponding to full length sFlt1 sequence contained within the probe indicating that the mRNA cleavage site is distal to this probe (Fig. 2C
). Probe 3 is from the 3' end of intron 13 that bears the distal putative poly(A) signal sequences predicted to be strong poly(A) signal sequences by polyadq. This probe protects two fragments, the smaller of which is stronger (sFlt large1), is downstream of a classic AAUAAA and accounts for the bulk (82–91%) of sFlt1 transcripts that end distally within intron 13 (Fig. 2D
). The larger, lighter fragment (sFlt large2) is just downstream of another classic AAUAAA and accounts for the remainder of sFlt1 transcripts. The principal sFlt1 transcript that ends within intron 13 thus has a predicted 3' UTR of almost 4 Kb, at least in placenta and in HUVEC.
We then performed Northern blot analysis in various tissues to determine if sFlt1 was indeed expressed as a large transcript. We first hybridized the tissue blot with a cDNA probe from the 5' end of Flt1 that would detect both Flt1 and sFlt1. Multiple bands were seen in human placenta corresponding to transcripts of 7.0 Kb, 3.4 Kb, and 2.6 Kb comprising 18, 47, and 35% of the total transcripts (Fig. 3A
). The 7.0 Kb and 2.6 Kb transcripts but not the 3.4 kb transcript were also faintly expressed in heart. We then hybridized the blot with an sFlt1 specific probe derived from the 5' end of intron 13, and we could demonstrate a signal at 7.0 Kb confirming that sFlt1 is indeed expressed as a 7.0 Kb transcript (Fig. 3B
). The 2.6 Kb transcript identified by Northern blotting using a 5' Flt1 cDNA probe corresponds to the smallest sFlt1 transcript identified in Fig. 2B
. This is not detected by the sFlt1 specific probe in Fig. 3B
as this cDNA probe corresponds to sequence downstream of the proximal polyadenylation site used by the shortest transcript.
Since we were unable to detect the 3.4 Kb sFlt1 transcript by Northern blot analysis using a cDNA probe derived from intron 13, we wondered whether other sFlt1 mRNAs could arise by alternate splicing or by polyadenylation further downstream. An analysis of human ESTs using the UCSC Genome Browser (Mar 2006 assembly) at www.genome.ucsc.edu identified an EST (AI188362) that included exon 13 and 14 and appeared to terminate within an alternatively spliced exon (exon 15a) downstream of exon 14. We were able to amplify this transcript by RT-PCR of placenta and cytotrophoblast RNA (Fig. 4
A). An analysis of nucleotide sequence within exon 15a using the Censor server at http://www.girinst.org/censor/index.php (31)
revealed the presence of a long trinucleotide repeat (TCAn) in the proximal portion of exon 15a and a full-length Alu sequence (Alu-Sq) in sense orientation in the distal portion of exon 15a (Fig. 4B
). Alu sequence elements are short interspersed repeats that contain a 3' poly(A) tail. Within the 3' end of this Alu sequence in exon 15a are tandem polyadenylation signal sequences that appear to direct mRNA cleavage and polyadenylation of this variant sFlt1 transcript. An open reading frame analysis of this alternatively spliced sequence indicates a translation stop codon within exon 15a giving rise to a 733 aa sFlt1 isoform. This isoform includes a 12 aa serine tail and diverges from 687 aa intronic sFlt1 isoform by 77 aa (Fig. 4C
).
To determine the relative abundance of this variant transcript, we performed an RPA using a cRNA probe that would simultaneously detect sFlt1 transcripts that were cleaved within intron 13 (sFlt1-i13), sFlt1 transcripts that were cleaved within exon 15a (sFlt1-e15a), and Flt1 transcripts. Using this strategy we determined that at least in placenta and in cytotrophoblasts, sFlt1-i13 and sFlt1-e15a are equally abundant (together comprising 93–97% of the total transcripts) and considerably greater than Flt1 transcripts (Fig. 5
A). To determine whether this sFlt1 variant could account for the abundant 3.4 Kb transcript seen in Fig. 3A, we
used the PCR fragment identified in Fig. 4A
as a probe for a multiple tissue blot by Northern blot analysis. The cDNA probe identified a specific 3.4 Kb band in placenta (Fig. 5B
). Many of the other tissues gave a broad smear, reflecting the poor specificity of the probe because of the repeat sequences found within it. Together, our data indicates that sFlt1 is expressed as 2.6, 3.4, and 7.0 Kb transcripts arising from intronic polyadenylation and from alternate splicing (Fig. 5C
).
Returning to the 7.0 Kb transcript, since the sFlt1 stop codon within intron 13 is 2310 nt from the transcription start site for sFlt1/Flt1, the Northern blot data suggested that the sFlt1 3' UTR may be as much as 4 Kb, consistent with the RPA data in Fig. 2D
. To confirm that the 3' UTR of sFlt1 extends to sequences at the 3' end of intron 13 and to identify mRNA cleavage sites, we then performed 3' RACE assays in placenta. PCR was performed first with a reverse primer corresponding to the adaptor that was ligated to the 3' end of the placental cDNA library, and a forward primer, sFlt1_F, upstream of the proximal polyadenylation signal sequence and the amplified products were cloned and sequenced to identify an mRNA cleavage site 21–29 nt downstream of this proximal signal sequence (Fig. 6
A, C). Another PCR was performed with a series of forward primers that were positioned at various distances upstream of the putative distal polyadenylation signal sequences. The far upstream primers A (hFlt1intronF4) and B (hFlt1intronF5) amplified a fragment of 3.2 and 2.6 Kb, respectively, that was compatible with the use of the distal polyadenylation signal sequences (Fig. 6A, B
). The most distal primer C (hFlt1intron F7) amplified fragments that were 450 and 550 bp in length that more clearly identified the 3' extent of these cDNAs and confirmed that they corresponded to the two distal cleavage sites identified earlier (see Fig. 2D
). Sequence analysis of several 3' RACE clones demonstrated that a cleavage site followed the 1st distal poly(A) signal sequence (A1) in all clones and is consistent with a 3' UTR of 4.1 Kb (Fig. 6B, C
).
|
We then performed a polyadenylation assay to confirm the identity of signal sequences that direct sFlt1 mRNA cleavage. We cloned a 600 bp genomic DNA fragment from FLT1 that included both distal mRNA cleavage sites into pCβS between the CMV promoter and the downstream BGH polyadenylation signal sequence (Fig. 7
A). This construct and the parent plasmid pCβS were transfected into HTR-8/SVNeo, a trophoblast cell line, and RNA preparations from transfected cells were analyzed for mRNA cleavage sites by RPA. Using an antisense cRNA probe derived from the sFlt1 poly(A) signal construct allows us to map mRNA cleavage that occurs downstream of the distal sFlt1 poly(A) signals and the cleavage that occurs downstream of the BGH poly(A) signal. Cells transfected with the parent plasmid showed a strong signal at 200 nt, 30 nt downstream of the BGH poly(A) signal (Fig. 7B
). RNA from cells transfected with distal sFlt1 poly(A) sequences showed strong signals at 240 nt and 330 nt, which are downstream of the putative distal poly(A) signals at 146 nt and 315 nt, respectively (Fig. 7B
).
|
Having confirmed that sFlt1 mRNA cleavage occurs in the primary transcript at two distal sites, we then set out to identify the poly(A) signal sequences that direct this cleavage. We individually mutated the 1st (A1) and 2nd (A2) classic poly(A) signal sequences as well as two other putative signal sequences adjacent to the 1st classic poly(A) signal, AUUAAA (A3) and UAUAAA (A4) (see Fig. 6C
) within the polyadenylation reporter vector, pcβS/sFlt1distal (Fig. 7C
). The constructs were transfected into HTR-8/SVNeo cells and the transcribed RNA hybridized to antisense cRNA probes derived from the transfected construct. When compared to wild-type constructs, neither mutation of A1 nor A3 had an effect on the 1st distal mRNA cleavage site while mutation of A4 and A2 abolished the 1st and 2nd distal mRNA cleavage sites respectively (Fig. 7D
). Interestingly, when A4 is mutated, two slower migrating bands are seen in the region of the 2nd distal mRNA cleavage and may represent aberrant cleavage directed by alternate signal sequences that are not ordinarily used. We thus demonstrate that the 1st distal cleavage is not directed by the classic poly(A) signal sequence (A1) present 80 nt upstream of the cleavage site but rather by a conserved variant signal sequence UAUAAA (A4) 35 nt upstream of the cleavage site. In contrast to the 1st distal cleavage however, the 2nd distal cleavage is directed by a classic signal sequence (A2) present upstream.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2
3)
. While both receptors are transmembrane proteins that bind VEGF with high affinity, KDR-1 appears to be the primary signaling receptor for VEGF, at least in endothelial cells, with Flt1 serving to regulate the amount of free VEGF available to bind to and activate KDR-1. The discovery of sFlt1 that can bind VEGF and PlGF indicates a mechanism to control VEGF-dependent signaling at its site of production and in distant sites (11
, 32)
.
In this manuscript we explore the expression of human sFlt1 in the placenta, in cytotrophoblasts and in vascular endothelial cells and define the 3' end of native human sFlt1. We use three complementary methods to demonstrate that the 3' UTR of sFlt1 is defined by multiple signal sequences within intron 13 that direct mRNA cleavage and a variant sFlt1 that arises by alternate splicing and termination within exon 15a. A series of RPA probes used to map sFlt1 mRNA cleavage sites demonstrated that there was one proximal cleavage site that accounted for 30% of total sFlt1 cleaved within intron 13 in placenta and HUVEC, while the remainder appeared to be cleaved at two distal sites near the 3' end of intron 13 (Fig. 2)
. These data are consistent with the existence of a cDNA species of 7.0 Kb. 3' RACE analysis in placental cDNA confirmed that the 3' UTR was in some cases as much as 4 Kb in length and by Northern blot analysis we identified a 7.0 Kb sFlt1 transcript in placenta (Fig. 3)
.
The 7.0 Kb sFlt1 transcript is considerably larger than the size of human sFlt1 transcripts previously noted. When first cloned, the 5' end of human Flt1 cDNA was used to identify 2.2, 3.0, and 8.0 Kb transcripts in various tissues, including the placenta (33)
. Others, using another 5' probe, identified 2.7, 3.4, 7.5, and 8 Kb transcripts in human placenta and in human vascular endothelial cells (34)
. Both groups presumed that the smaller transcripts corresponded to truncated forms of Flt1 mRNA although this was not pursued further. When sFlt1 was first identified as an alternate transcript of Flt1 in a vascular endothelial cell library, the 3' end of the mRNA was not identified although the cloned cDNA was 2.6 Kb and extended 400 nt beyond the sFlt1 stop codon (4)
. In a subsequent study, an sFlt1 cDNA containing a poly(A) tail that begins 20 nt downstream of a 2.3 Kb coding sequence was cloned from a human term placental library demonstrating that sFlt1 is expressed as 2.5–2.6 Kb transcript (27)
. Later studies exploring the expression and regulation of human Flt1 and/or sFlt1 have presumed that full-length Flt1 is the 7.5–8.0 Kb transcript present on Northern blot analysis with the smaller transcripts identified as sFlt1 (7
, 9
, 35)
. The 2.6 Kb transcript seen with the Flt1 common probe (Fig. 3)
is likely to correspond to the shortest sFlt1 transcript and is not detected with the sFlt1 probe because the probe is derived from sequence downstream to the cleavage site for this mRNA. The 3.4 Kb species corresponds to an alternatively spliced form of sFlt1 that terminates within exon 15a (Fig. 5)
. The termination of this transcript from exonic polyadenylation within exon 15a is a consequence of polyadenylation signal sequences introduced by an AluSq retrotransposition (36
, 37)
. The AluSq subfamily was inserted into the primate genome 35–48 myrs ago, and its restriction to the primate implies that this transcript is unlikely to be found in nonprimate species (38)
. This transcript is predicted to encode an sFlt1 C-terminal variant that is lacking a transmembrane domain and is likely to be secreted protein. Whether this form of sFlt1 with an unusual polyserine tail is translated and secreted and functional remains to be determined.
Our data demonstrate that human sFlt1 is also expressed as a large transcript that comigrates with Flt1 by Northern blot analysis. In a previous study multiple polyadenylation signal sequences for mouse sFlt1 mRNA within intron 13 was identified in native tissues and in HEK293 cells transfected with an sFlt1 minigene (6)
. In that study a distal cleavage and polyadenylation site was noted in native mouse lung cDNA 4000 bp into intron 13 by 3' RACE analysis and confirmed by 3' RACE of transcripts derived from expressed minigene constructs that included intronic elements. Interestingly, the region around the principal polyadenylation signal sequences share 85 to 89% homology between mouse and human Flt1 sequence although there is considerable divergence elsewhere within intron 13 between these two species (Fig. 2A
).
Heterogeneity in transcripts that arise from any given gene are principally from alternate promoter usage, alternate splicing, or from the use of alternate polyadenylation signal sequences. In the case of the FLT1 gene, heterogeneity in translated proteins comes from the utilization of polyadenylation signal sequences downstream of the translation stop codon in exon 30 to encode membrane bound Flt1 and upstream polyadenylation signal sequences within intron 13 and within an alternatively spliced exon 15a to encode two sFlt1 variants. Alternative polyadenylation is coupled to skipped splicing of exon 13 with a new translation stop codon downstream of the skipped splice site defining the 3' end of the open reading frame of one form of sFlt1. The heterogeneity in the length of sFlt1 transcripts encoding this form of the soluble receptor comes from the use of widely separated alternate polyadenylation signals within intron 13 that creates an sFlt1 with an exceedingly short 3' UTR and others with a unusually long 3' UTR.
The most widely used polyadenylation signal sequence in human mRNA 3' end formation is AAUAAA, accounting for 50 to 60% of all transcripts (30
, 39)
. The commonest variant, AUUAAA accounts for 15 to 20% of polyadenylation signal sequence while 9 other single nucleotide variants of AAUAAA are found less often and together account for the remainder. Interestingly, the hexamer UAUAAA that directs cleavage at the 1st distal sFlt1 site accounts for only 3.7% of polyadenylation signal sequences. In mRNAs that have alternate tandem polyadenylation sites, the more proximal poly(A) signal sequence is generally a "variant" sequence, while the 3' most distal site tend to use a canonical signal, a schema that appears to be true for sFlt1 (40)
.
Cleavage of mRNA typically occurs 11 to 24 nt downstream of the polyadenylation signal sequence which is directed by the binding of cleavage and polyadenylation factor (CPSF) to the polyadenylation signal sequence and the binding of cleavage stimulating factor (CstF) to a U/GU-rich element 10 to 30 nt downstream of the cleavage site (41
42
43
44)
. Unlike the upstream polyadenylation signal sequence these downstream elements are poorly defined and poorly conserved (43)
. We have identified putative conserved U/GU rich elements downstream of proximal and distal sFlt1 mRNA cleavage sites (Fig. 6C
). It has become clear through large scale genomic analysis that some transcripts have unexpectedly large 3' UTRs, although typically polyadenylation sites occur within 2 Kb of the stop codon (39
, 45)
. Conversely, there are examples of transcripts with exceedingly short or even an absent 3' UTR as in the case of the human 
-galactosidase transcript (46)
. Variations in length of the 3' UTR may result in distinct tissue-specific or temporal regulation of mRNA expression, effects on mRNA stability or effects on translational efficiency. An analysis of the long sFlt1 3' UTR shows a number of cis-acting elements that may regulate mRNA stability, including but not limited to AU-rich elements and poly(U) elements (47
48
49)
.
Flt1 and sFlt1 bind VEGF and PlGF with high affinity and regulate the biological actions of these growth factors. Cytotrophoblasts and vascular endothelial cells are an important source of placental sFlt1 and increased sFlt1 expression may underlie or contribute to some cases of preeclampsia. Post-transcriptional regulation of sFlt1 may account for the increased expression of sFlt1 compared to Flt1 in vascular endothelial cells and in cytotrophoblasts. In this regard, hypoxia, widely regarded as a pathogenic stimulus in preeclampsia, can differentially increase sFlt1 expression in cytotrophoblasts (50)
. The identification of post-transcriptional regulatory elements within sFlt1 defined here should provide the foundation to study the basis of regulated sFlt1 expression in various physiological and pathophysiological states.
|
ACKNOWLEDGMENTS |
|---|
Received for publication April 18, 2007. Accepted for publication May 31, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
B activation in human endothelial cells. Circulation 109,2221-2226 subunit of the epithelial sodium channel is an aldosterone-induced transcript in mammalian collecting ducts, and this transcriptional response is mediated via distinct cis-elements in the 5' flanking region of the gene. Mol. Endocrinol. 15,575-588This article has been cited by other articles:
|
|
 |
|
  S. Sela, A. Itin, S. Natanson-Yaron, C. Greenfield, D. Goldman-Wohl, S. Yagel, and E. Keshet A Novel Human-Specific Soluble Vascular Endothelial Growth Factor Receptor 1: Cell Type-Specific Splicing and Implications to Vascular Endothelial Growth Factor Homeostasis and Preeclampsia Circ. Res., June 20, 2008; 102(12): 1566 - 1574. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
