Alternative initiation of transcription of the human presenilin 1 gene in SH-SY5Y and SK-N-SH cells The role of Ets factors in the regulation of presenilin 1 Martine Pastorcic 1 and Hriday K. Das 1,2,3 1 Department of Pharmacology & Neuroscience, 2 Department of Molecular Biology & Immunology and 3 Institute of Cancer Research, University of North Texas Health Science Center at Fort Worth, TX, USA We have identified DNA seq uences required fo r the expres- sion of the presenilin 1 (PS1) gene. A promoter region has been mapped in SK-N-SH cells and includes sequences be- tween )118 and +178 flanking the major initiation site (+1). The PS1 ge ne is also efficiently transcribed in the SH-SY5Y subclone of SK-N-SH cells. However the promoter appears to be utilized in alternative ways in both cell types. Sequences both upstream as well as downstream from the initiation site mapped in SK-N-SH cells were shown by 5¢-and3¢- deletion analysis to play a crucial role in both cell lines. However, in SH-SY5Y cells either upstream or downstream sequences are sufficient to direct transcription, whereas in SK-N-SH cells 5¢-deletions past the +1 site eliminate over0 95% of transcription. Several Ets motifs (GGAA) 11 as well as Sp1 motifs [(G/T)GGCGGRRY] 22 are juxtaposed both up- stream and downstream from +1. To understand how the promoter may be utilized alternatively in different cell types we have examined the effect of point mutations in these elements. Altering an Ets motif at )10 eliminates 80% of transcription in SK-N-SH cells whereas the same mutation has only a minor effect in SH-SY5Y cells. Conversely, mutation of the Ets element at +90, which eliminates 70% of transcription in SH-SY5Y cells, has a lesser effect in SK-N- SH cells. In both cell types a promoter including mutations at both )10 and +90 sites loses over 90% transcription activity indicating the crucial importance of these two Ets motifs. The effect of Sp1 mutations appears to be similar in both cell types. Hence the differential e xpression in each cell type may be at least partially determined by Ets factors and the )10/ +90 sites. We have identified several Ets factors that recognize specifically the )10 Ets motif by the yeast one- hybrid selection including avian erythroblastosis virus E 26 oncogene homologue 2, Ets-like gene 1, Ets translocation variant 1 and Ets related molecule (ERM) 3 . We show here that ERM specifically recognizes Ets motifs on the PS1 promoter located at )10 as well as downstream a t +90, +129 and +165 and activates PS1 transcription with pro- moter fragments containing or not the )10 Ets site. Presenilins (PS1 and PS2) are highly homologous multipass transmembrane proteins [ 1–3]. T hey are required for the protease activity of a multiprotein complex termed c-secr- etase, which includes presenilin, nicastrin, Aph-1 and Pen-2 that are all necessary for proteolytic activity [4–7]. It appears that presenilin acts as a catalyst or a required cofactor. c- Secretase cleaves the amyloid precursor protein (APP), resulting in the production of the Ab peptide which appears to be central in the pathogenesis of Alzheimer’s disease [8– 10]. Indeed PS1 mutations 4 have been linked to many cases of early onset familial Alzheimer’s disease [11–13]. Similarity in the processing of APP a nd the Notch receptor protein, which controls signaling and cell–cell communication in development, have largely contributed to the understanding of the role of presenilins. The Notch receptor, a type 1 membrane protein, is also cleaved by c-secretase. The stimulation of Notch by its ligand leads to the intramem- brane proteolysis of the receptor, freeing the Notch intracellular domain, which translocates to the nucleus and regulates gene expression [14–16]. Increasing evidence indicates that p resenilin and c-secretase cleave a variety of Type 1 transmembrane proteins which all release intracel- lular fragments with the ability to interact with transcription coactivators [17]. CD44, a ubiquitous cell adhesion protein, is cleaved by c-secretase and releases an intracellular domain that activates CBP/p300 [18]. Neuronal cadherin (N-cadh- erin), mediates Ca 2+ -dependent cell–cell adhesion and recognition, and plays a crucial role in neurogenesis, tissue development a nd homeostasis. The proteolytic cleavage of N-cadherin releases an intracellular C -termin al fragment that suppresses CRE-dependant activation of transcription by promoting CBP degradation by the proteasome [19]. Hence it appears that presenilins may affect the expression of many genes through intramembrane proteolysis. This mechanism appears to be modulated by neuronal activity [19] and may represent an important aspect in the pathology of Alzheimer’s disease. Furthermore, PS1 appears to play a crucial role in the normal metabolism of APP as well as in the pathological increase of the Ab42 [20]. APP is a Correspondence to H. K Das, University of North Texas Health Sci- ence Center at Fort Worth, 3500 Camp Bowie Boulevard, Fort Worth, TX 76107, USA. Fax: +1 817 735 2091, Tel.: +1 817 735 5448, E-mail: hdas@hsc.unt .edu Abbreviations: APP, amyloid precursor protein; CAT, chloramphen- icol acetyltransferase; Elk1, Ets-like gene 1; ERM, Ets related mole- cule; ER81, alias for ETV1 (Ets translocation variant 1); Ets2, avian erythroblastosis vi rus E 26 oncogene homologue 2; PS1, presenilin 1. (Received 30 July 2004, accepted 30 September 2004) Eur. J. Biochem. 271, 4485–4494 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04453.x ubiquitously expressed cell surface protein and its process- ing by c-secretase cleavage generates not only Ab but also an intracellular domain of APP [21] that may function in nuclear signaling both in normal as well as pathological signal transduction [22–24]. We have studied the transcriptional regulation of PS1 in different cell types and in particular the role of Ets and Sp1 sequence motifs flanking the major initiation site mapped in SK-N-SH cells [25]. A crucial Ets element located at )10 has been used as bait in yeast one-hybrid screening of a human brain cDNA library to isolate Ets factors that may control PS1 expression in neu ronal cells [26,27]. The specificity of Ets elements is specified by four factors: their site of synthesis and chemical modification, the presence of specific Ets promoter motifs and the set of nuclear factors with which they interact in the tissue being considered [28,29]. Hence they should be a powerful determinant in the degree of activation, the initiation site, as well as the regulation of the PS1 gene. Materials and methods Transfection assays SK-N-SH and SH-SY5Y cells were transfected with PS1CAT fusion genes containing various fragments of PS1 sequences flankin g the transcription initiation site [25]. Cells were seeded at a density of 10 4 Æcm )2 2 days before transfec- tion. On the day of transfection, medium was replaced with serum-free Dulbecco’s modified Eagle’s medium 2 h p rior the addition of DNA. Usually 6 lg of PS1CAT and 4 lgof pSV-b-galactosidase control vector (Promega, Madison, WI) 5 were mixed in 112 lLH 2 O. CaCl 2 (12.5 lL) was then added to the DNA mixture which was slowly added to 125 lLof2· Hank’s buffered salt solution 6 [25]. Precipitate wasallowedtoformat37°C for 40 min and added to the cells (on 5 cm plates). After 4 –5 h i ncubation at 37 °Ca glycerol shock was performed. The medium was removed and 1.5 mL of 12.5% (v/v) glycerol was added for 90 s on SK-N-SH cells and 60 s on SH-SY5Y cells. Glycerol was removed after dilution with 2 m L NaCl/P i and the plates were washed three times with 3 mL NaCl/P i . Dulbecco’s modified Eagle’s medium containing 10% (v/v) fetal bovine serum was finally added and the cells were returned to the incubator for 16–18 h before harvesting. 7 Again, the medium was removed, cells were washed three times with 3 mL NaCl/P i and harvested in 500 lLNaCl/P i and pelleted at 1000 g for 4 min. The pellets were resuspended into 100 lL of 250 m M Tris/HCl pH 7.5 and lysed by three cycle s of freeze/thaw and cellular debris were pelleted at 12 000 g for 4 min. Supernatants were stored at )70 °C. For chloram- phenicol acetyltransferase 8 (CAT) assays, 50 lgproteinwas heat-treated at 60 °C for 10 min, centrifuged for 4 min at 12 000 g, incubated in 100 lL reaction containing 0.25 mgÆmL )1 n-butyrylCoA and 0.1 lCi [ 14 C]chloram- phenicol at 37 °C [25]. CAT assays were incubated for 3– 4 h and extracted with 250 lL mixed xylenes. The xylene phase was t hen back extracted twice with 2 50 m M Tris pH 7.5 and the 14 C in the xylene phase was then counted. For b-galactosidase assays 12–20 lg o f p rotein extract (non heat-treated) was incubated in 0.1 M sodium phos- phate pH 7.5, 1 m M MgCl 2 ,50m M 2-mercaptoethanol, and 3 m M 2-nitrophenyl-b- D -galactopyranoside 9 . Reaction s were stopped with 800 lLof1 M Na 2 CO 3 and A 420 was measured in the visible spectrum. Promoter activity in different samples was usually compared using b-galactosidase activity as an internal control, except in the experiments testing the activity of Ets related molecule (ERM) 10 cDNA to limit possible competi- tion between promoters. Each experiment was r epeated three times, with a minimum o f triplicate te sts of each construct and treatment. Point mutagenesis of the PS1 promoter Promoter mutants were generated in the context of the ()118/+178)PS1CAT constr uct by PCR-based site-direc- ted mutagenesis, using the QuickChange k it from Strata- gene (La Jolla, CA) 11 and t he primers listed b elow. The following point mutations were generated with the primers listed and each primer was used together with its re verse complementary strand. The point mutation at the )70 Sp1 motif, m()70), was obtained with p)70F (5¢-GGCCGGA GGCCTCGAAGCCTTCCTCCTGG-3¢) and its reverse complement p)70R. Mutation m()50) was obtained with p)50F (5¢-CTCCTGGCTCCTCAAGTCCTCCGTGG-3¢) and the corresponding p)50R; m()30) was obtained with p)30F (5¢-CCCTCCTCCGTGATGAGGCCGCC AACGACG-3¢)andp)30R; m(+20) with p+20F (5¢-GTGAGGGTTCTCGGGCTCATCCTGGGACAG GCAGCTC-3¢) and p+20R; m(+65) with p+65F (5¢-GCGGTTTCACATCCTAGACAAAACAGCG-3¢)and p+65R; m(+90) with p+90F (5¢-GGCTGGTCTGTGA CTAACCTGAGCTACG-3¢) and p +90R; m(+129) with p+129F (5¢-CGGCGGCAGCGGGGCGGCGACTAA GCGTATGTGCGTGATG-3¢); and p+129R using p()118, +178)CAT as the template. Rapid amplification of cDNA ends (RACE) 12 in SH-SY5Y and SK-N-SH cells transfected with the PS1 promoter Total RNA was prepared from SK-N-SH cells and SH-SY5Y cells transfected with ()119, +178)PS1CAT by the guanidinium/cesium chloride method [25]. Rapid ampli- fication of cDNA ends (RACE) was performed using the BD Bioscie nces (Clontech, P alo Alto, CA) 13 Smart R ace cDNA amplification kit with the primer (5¢-CCGGAT GAGCATTCATCAGGCGG-3¢) specific for the CAT gene in order to detect RNA encoded from the transfected plasmid. Amplification products were gel purified, inserted into KS bluescript vector (Stratagene) and sequenced. Cloning of ERM Two independent clones containing the entire ERM cDNA were obtained by yeast one-hybrid screening of a human brain cDNA library from Clontech [26]. The cDNA was then inserted into pC1 (Promega) by generating a subclone using Pfu DNA polymerase and the following forward (F) and reverse (R) primers: ERM-F: 5¢-gatcacgcgtCTCAGG AGGATCCCTTTTC-3¢ and E RM-R: 5¢-gatcgtcgacGCG GGTACTAACCTGAACAAGA-3¢. PCR conditions were according to the manufacturer’s recommendations at 94 °C for denaturation, 62 °Cfor 4486 M. Pastorcic and H. K. Das (Eur. J. Biochem. 271) Ó FEBS 2004 annealing and 72 °C for extension, 14 for 30, 30 and 60 s, respectively. Primers were designed to incorporate sites for the restriction enzymes MluI (forward primer) and SalI ( reverse primer), to direct integration into the cloning vector (restriction sites 5¢ flanking sequence shown in lowercase) 15 . PCR products were purified, digested with MluIandSalI and in serted into the corresponding sites of pCI. Nuclear extracts Nuclear extracts from SH-SY5Y cells were prepared from cells growing exponentially as described previously for SK- N-SH [25]. Compacted nuclei were extracted with buffer C [20 m M Hepes pH 7.9, 600 m M NaCl, 1.5 m M MgCl 2 , 0.2 m M EDTA, 0.5 m M dithiothreitol, and 25% (v/v) glycerol]. The volume of buffer C was adjusted to obtain an ionic strength equivalent to 300 m M NaCl in the homogenate. Protein concentration of the nuclear extracts was 5–6 mgÆmL )1 . Analysis of specific DNA binding by electrophoretic gel mobility shift assays (EMSAs) To assay the binding of Ets factors to the PS1 promoter, the proteins were synthesized from the corresponding pCI- based vectors by in vitro transcription–translation as recommended by the manufacturer (Promega). For e lec- trophoretic gel mobility shift assays (EMSAs) 16 , aliquots (2 lL) from in vitro translation reactions were added to 20 lL of DNA binding mixtures including 12 m M Hepes pH 7 .5, 50 m M NaCl, 1 m M dithiothreitol, 0.1 m M EDTA, 1% (w/v) Igepal CA-630 (Sigma, S t Louis, MO) 17 , 12% (v/v) glycerol, 10–50 pg of end-labeled DNA probe, and 1 lgof poly(dI-dC)Æpoly(d I-dC). Reactions were incubated at 24 °C for 30 min and analyzed by electrophoresis on native 4.5% (w/v) polyacrylamide gels containing 0.1% (w/ v) Igepal at 4 °C a s described previously [25]. Nuclear extracts from SK-N-SH or SH-SY5Y cells were prepared as described above an d binding reactions were carried out by incubating 0.1–0.2 ng of probe with 5 lg of nuclear extracts in the presence of 2 lg of poly(dI-dC)Æpoly(dI-dC) 18 in 10 m M Hepes pH 7.9, 50 m M NaCl, 0 .7 5 m M MgCl 2 , 0.1 m M EDTA, 1 m M dithiothreitol, 1% (w/v) Igepal CA- 630 (Sigma) and 10% (v/v) glycerol for 30 min at 4 °C. The goat polyclonal antibody sc-1955X (Santa Cruz Biotechnology, Santa Cruz, CA) raised against a 20 amino acid C-terminal peptide of the human ETV5/ERM was used in supershift assays, and was preincubated with either the protein extracts for 60 min at 4 °C. EMSAs included 32 P-labeled p robes c ontaining either wild type or mutant promoter sequences generated by PCR as described previously [25]. The +50/+107 probe was synthesized using primers p19 and labeled p 26 [25], digesting the )22/ +107 fragment wit h SacII and purifying t he +50/+107 fragment by e lectrophoresis on 12% (w/v) polyacrylamide gels. The +107/+178 probe was generated with the p19 and labeled p27 primer p air, digestion with SacII followed by gel purification of the + 107/+178 fra gment. Mutant probes were obtained by substituting mutant templates instead o f w ild typ e. The probe containing a mutation at +165 was generated by substituting p27m (5¢-gat ctctagaCGGTGCCTGACTGGCTTGC-3¢) instead of p27. Results Differential effect of 5¢-and 3¢ -deletions in SH-SY5Y cells as compared to SK-N-SH cells The )118/+178 promoter fragment (Fig. 1) produced the maximum level of expression in SK-N-SH a s well as SH-SY5Y cells (Fig. 2). In both cell types deletions of sequences upstream from )22 had little effect (Fig. 2; [25]). The minor effects observed f rom )687 to )22 in SK-N-SH c ells have been discussed previously [25]. Deletions of the )22/)6 region reduced transcription drastically in SK-N-SH, however, in SH-SY5Y it a ffected transcription by less than 50% and l arger deletions to +2 did not result in further decrease in PS1 expres sion. Hence in SH-SY5Y cells the )10 Ets site does not appear to be crucial for transcription. It is probable t hat in SH-SY5Y c ells elements downstre am from +1 play a major role in directing transcription and that initiation is also shifted further downstream. The pattern of the effects of 3¢-dele tions presents similar- ities in both cell types. Independently from the 5¢-end point of the fragment tested, 3¢ deletions from +178 to +107 markedly decreased transcription by about fivefold in Fig. 1. Th e PS1 promoter. The sequences from )119 to +178 flanking the major transcription initiation site (+1) in SK-N-SH cells included in the PS1CAT reporter fusion vector are shown. The binding sites for known transcription factors Sp1 and Ets are indicated with brackets defined by footprinting with SK-N-SH cells (25) as well as site C which corresponds to an unknown binding protein. Arrowheads indicate the position of DNase I hypersensitive sites obser ved in footpr inting experiments. The Ets and Sp1 consensus motifs are underlined. Arrows indicate the end points of 5¢ or 3¢-deletions. SacII restriction enzyme sites use d in the pre paration of the probes used i n EMSAs are boxed. Ó FEBS 2004 Regulation of the presenilin 1 gene (Eur. J. Biochem. 271) 4487 SH-SY5Y cells and similarly by about tenfold in SK-N-SH and cells. F urther deletion to +42 increased transcription in both cells lines: about fivefold in S H-SY5Y cells and twofold in SK-N-SH cells. Further deletions to +6 did not result in any change in expression in any cell type. Hence element(s) required for transc ription in both cell types are located between +178 and +107. Element(s) repressing transcription, at least in the context of a promoter truncated to +107, are contained between +107 and +42. In SH-SY5Y cells alternative initiation mechanisms are clearly indicated by the significant level of promoter activity conferred by either of the two fragments ()118 to +6) or (+ 2 to +178). Effects of point mutations at Ets and Sp1 motifs on PS1 transcription The effects of point mutations in Ets motifs at positions )10, +65, +90 and +129 as well as Sp1 motifs at )70, )50 and +20 have been examined in both SK-N-SH and SH-SY5Y cells (Fig. 3A). Altering an Ets motif at )10 eliminated 80% of transcription i n S K-N-SH ce lls whereas the mutation had only a minor effect (30%) in SH-SY5Y cells. Conversely, the Ets element at +90 which eliminated 70% of transcription i n SH-SY5Y cells had a lesser e ffect (40%) in SK-N-SH cells. In both cell types a prom oter including mutations at both )10 and +90 Ets sit es lost o ver 9 0% transcription a ctivity indicating the crucial importance of these two Ets motifs. Mutations at +65 and +129 resulted in a mild 25–30% decrease in SH-SY5Y cells and 40–50% decrease in SK-N-SH cells. T he double m utation at +65 an d +129 actually increased transcription markedly in SH-SY5Y cells to 180% of the wild type promoter. Similarly it is possible that the double mutant 19 at +90 and +129 resulted in activity comparable to wild type promoter, hence the double mutation appears to reverse the effect of either of the single mutations alone. These results considered toge ther with the effect of 3¢-deletions to +107 and +42 suggest that deletion of sequences downstream from +107 or mutations at the +129 site in particular does not just result in loss of function, but rather produces a cis-acting n egative effect, such as an abortive protein complex. In S K-N-SH cells the effect of the double mutants was not as striking and showed the same activity as each of the single mutations independ- ently. However, the absence of an additive effect between each mutation may suggest some interaction also between each pair of sites in SK-N-SH cells. The effect of Sp1 mutations appeared similar in both cell types. The mutation at )70 had the most deleterious effect decreasing transcription by 65–70%. The m()50) mutation affected transcription by 45% and 60% in SK-N-SH and SH-SY5Y c ells, respectively. It is interesting t o note that )70 had more effect than )10 in SH-SY5Y cells, suggesting that )70 is also required for )10 independent initiation. The same case may apply t o )50. Furthermore the double mutant )70,)10 appears to result i n a more than additive effect, in contrast with the double mutant )50,)10 in SH-SY5Y cells. This may reflect a cis -negative effect in addition to simply the loss of function, su ch as squelching 20 of a factor in limiting amount or steric hindrance of an abortive complex. Mutation at the Sp1 site at +20 did not alter transcription and the )10/+20 double mutant appeared to reflect the effect of the )10 mutation alone (Fig. 3A). The double mutant )10/+65 appears to reflect also a simple additive effect. The )30 site appears to bind a nuclear factor that has not been identified [25]. Mutation at this site alone or together with the )10 mutation affected transcription only mildly as compared to the wild type or the )10 mutant alone. The higher requirement for the +90 Ets site would be consistent with the ability of SH-SY5Y cells to direct transcription from the +2/+178 promoter fragment. Hence the differential expression in each cell type may be at least partially determined by Ets factors and the )10 and +90 sites. Efficient initiation requires a t least one of the two sites. It appears that the )70 Sp1 motif is required whether initiation is directed by either of th e )10 or +90 Ets site. Fig. 2. D eletion mapping of the PS1 promoter. The positions of the 5¢-and3¢-ends of each deletion fragment are indicated on the left (5¢D) and on the right (3¢D). PS1CAT plasmids (6 lg) were cotransfected with 3 lg b-galac- tosidase expression vector. Promoter activity was expressed as the ratio of CAT to b-ga- lactosidase activity for each transfected plate. The mean values for each construct (n ¼ 3or 4) are indicated. SD values were 10–20% in all cases. All constructs were tested in at least three different e xperiments. The )118/+178 construct was taken as 100%. 4488 M. Pastorcic and H. K. Das (Eur. J. Biochem. 271) Ó FEBS 2004 The effect of the point mutation at )70 appears in contrast with the effect of the 5¢-deletion to )35 or )22, which do not decrease transcription in either cell type (Fig. 2). We have not examined which sequences in the )118/)35 fragment aside from the )70 motif account for this difference. In addition, in both cell types the requirement for the )10 Etssiteishigherinshorterpromoterfragments 21 ( 21 Table 1). In the )22/+178 promoter fragment a mutation altering the )10 Ets site eliminated over 90% transcription in SK-N-SH cells [25]. In the context of the +118/+178 fragment the same mutation was slightly less deleterious, reducing transcription by a bout 80% (Table 1). In SH-SY5Y cells the )10 mutation showed a 66% reduction in the )22/+178 context whereas in the )118/+178 context the same mutation only had a minor effect, reducing transcription by 30%. This suggests complex interactions between upstream and downstream sequences and the )10 and +90 Ets sites. In SH-SY5Y cells the requirement for the )10 Ets element appears most strongly in very short promoter fragment )22/+42. This is consistent with the greater importance of +90 in SH-SY5Y cells. A B Fig. 3. Effect of PS1 promoter point mutations on the efficiency of transcription initiation and the position of the start site(s) 23 . (A) Effect of point mutations in Ets and Sp1 motifs on PS1 promoter activity. The positions of Ets (d) and Sp1 (h)motifsinthe)119/+178 region of the PS1 promoter are indic ated. Th e p osi- tions relative to the +1 site of point mutations (X) eliminating Ets or Sp1 consensus in each motif are indicated on the right together with the corresponding promoter activity in SK- N-SH or SH-SY5Y cells. (The mutant DNA sequences are reported in Experimental pro- cedures.) b-Galactosidase activity was used as an internal standard. The mean values for each construct (n ¼ 3 or 4) are indicated and SD values were 10–20% in all cases. All con- structs were tested in at least three different experiments. The )119/+178 construct was taken as 100%. (B) RACE in SH-SY5Y and SK-N-SH cells transfected with the ()199/ +178) PS1 promoter. RACE was performed as described above using the CAT4 primer (5¢- CCGGATGAGCATTCATCAGGCGG-3¢) specific for the CAT gene (solid box). RNA was prepared from SH-SY5Y cells transfected with wild type PS1()119/+178) CAT (lane 1) (open box) o r the same CAT reporter con- taining the )10Etsmutation(lanes2)orSK- N-SHcellstransfectedwithwildtypePS1 ()119/+178) CAT (lane 3). PCR products were analyzed by electrophoresis on 2% agarose gels. The size of molecular mass markers (lane 4) is indicated in bp. Table 1. Effects of )10 Ets GGAA fi TTAA mutation. The activity of PS1CAT constructs including the promoter fragments shown on the left containing (m) or not (wt) the GGAA to TTAA mutation at the )10 Ets site was assayed by transfection in SK-N-SH and SH-SY5Y cells. Each construct was tested in three transfections, each with n ¼ 3. Promoter fragment SK-N-SH cells SH-SY5Y cells wt m wt m )119/+178 100 ± 12 21 ± 5 100 ± 5 70 ± 20 )119/+143 73 ± 10 10 ± 4 133 ± 30 70 ± 6 )22/+178 91 ± 10 3.4 ± 2.6 100 ± 20 28 ± 7 )22/+4 23 ± 5 150 ± 20 9 ± 2 Ó FEBS 2004 Regulation of the presenilin 1 gene (Eur. J. Biochem. 271) 4489 RACE in SH-SY5Y and SK-N-SH cells transfected with the PS1 promoter RACE PCR w ith RNA from SK-N-SH cells transfected with the wild type [)119, +178] PS1 promoter yielded a single band of about 450 bp (Fig. 3B). RACE with the RNA from SH-SY5Y cells transfected with the wild type PS1 promoter produced two bands. The larger product appeared to a have a size similar to that obtained in SK-N- SH cells. The smaller product appeared to be about 50– 80 bp shorter. RNA from SH-SY5Y cells transfected with the PS1 promoter containing a mutated Ets site at )10 produced only the shorter RACE product. Sequencing of RACE PCR products from the lower band showed the 5¢-end points at +63 (Fig. 1). Hence sizing and sequencing of the RACE P CR products from SH-SY5Y cells is consistent with the g reater importance of s equences down- stream from the +1 start site originally mapped in SK-N- SH cells [25] and with a higher frequency of downstream (+63) initiation events in the )10 mutant. Nuclear factors in SH-SY5Y and SK-N-SH cells specifically recognize Ets motifs on the PS1 promoter upstream and downstream from the +1 site The binding of nuclear factors present in SK-N-SH and SH-SY5Y cells to various fra gments of the PS1 promoter was examined by EMSAs. We have compared binding A B Fig. 4. Nuclear factors from SK-N-SH and SH-SY5Y cells recognize specifically Ets motifs flanking the +1 site. (A) The binding of nuclear factors is eliminated by mutations in Ets motifs. The binding of nuclear factors present in SK-N-SH (K) or SH-SY5Y (H) nuclear extracts over the PS1 promoter was assayed by EMSAs. 32 P-labeled probes included sequences )22/+6 (lanes 1–6), +50/+107 (lanes 7–15) and +107/+178 (lanes 16–24). Either the wild type probes or the same fragments containing mutant Ets sites at positions )10 (lanes 4–6), +90 (lanes 10–12), +65 (lanes 13–15), +129 (lanes 19–21) and +165 (lanes 22–24) were incubated with 5 lg nuclear extracts or in absence of extract (O). The position of complexes t hat a re signific antly affected by mutatio ns is indicated by d ots. (B) The binding of nuclear factors is specific ally co mpete d b y a heterologous Elk1 binding site. The specific binding in the p romoter regions )22/+6 (lanes 1–6), +50/+107 (lanes 7–18), +107/+178 (lanes 19– 30) was tested by competition with the cold oligonucleotide E74 containing the Drosophila Elk1 binding motif (25) (+) or E74 containing a mutated Ets consensus (m). The position of c omplexe s define d in (A) is indicated by arrows or lines. 4490 M. Pastorcic and H. K. Das (Eur. J. Biochem. 271) Ó FEBS 2004 between wild type fragments a nd fragments containing the point mutations at the )10, +65, +129 and +165 Ets motifs that have been tested in transfection experiments (Fig. 4A). The binding of several nuclear factors to the probe including sequences )22/+6 appears to be eliminated by the )10 Ets mutation: complexes at levels A, B, C, D and E are decreased in lanes 5 and 6 as compared to lanes 2 and 3. With the +50/+107 probe, complex F d ecreases with m(+90) (lanes 11 and 12) whereas G decreases w ith m(+65) (lanes 14 and 15) as compared to binding to the wild type probe (lanes 8 and 9), suggesting that A and B may specifically recognize the +90 and +65 E ts sites, respectively. With the +107/+178 probe, c omplex I i s abolished by m(+129) (lanes 20 and 21) as compared to the wild type probe (lanes 17 and 18) and may represent specific recognition of this site. The specificity of the binding was assessed further by competition assays using E74 (a heterologous DNA sequence containing a known Ets-like gene 1 [Elk1] binding site [25]) as competitor. Among complexes formed with )22/+6 (Fig. 4B), B and E were eliminated by competition with wild type E74 competitor (lanes 2 a nd 5) but were unaffected by E74 competitor containing a mutated Ets motif (lanes 3 and 6). With the +50/+107 m(+65) probe, the complex(s) in region F are decreased selectively with the wild type E74 (lanes 8 and 11) as compared with the mutant E74 (lanes 9 and 12). This confirms the specificity of complex F and indicates that the +90 site is recognized by a nuclear factor related to Ets. With the +50/+107 m(+90) probe none of the complexes appear selectively affected by competition with the wild type E74 (lanes 14 and 17) as compared with the mutant E74 (lanes 15 and 18). Hence complex G is not competed although it is eliminated by the +65 mutation. We do not know whether G represents Ets specific binding to the +65 site that does not bind stably to the Ets com petitor used here, or if it represents specific recognition by nuclear factor(s) different from Ets. With the +107/+178 m(129) probe no competition is observed (Fig. 4B lanes 19–24) and this is consistent with the absence of an effect of m(+165) on the binding pattern (Fig. 4A, lanes 23 and 24 compared to lanes 17 and 18). This suggests that the +165 site does not bind any Ets related protein. With the +107/+178 m(+165) probe the complex in region H is efficiently competed by both wild type (lanes 26 and 29) as well as mutant E74 (lanes 27 and 30). Hence the competition data is inconclusive for complex H because it is possible that the factor may have a strong affinity for DNA ends present in large excess with both competitors. ERM transactvates PS1 and does not require the )10 Ets element We have begun to identify the Ets factors controlling the expression of PS1 using the yeast one-hybrid selection and the )10 Ets site as bait. We have found that avian erythroblastosis virus E 26 oncogene homologue 2 (Ets2), Elk1 and Ets translocation variant 1 (ER81) recognize PS1 and act as trans-effectors [26–28]. ERM was also identified by one-hybrid selection and we have examined its effects on PS1 transcription. Cotransfection of SK-N-SH cells with a )119/+178 promoter fragment fused to a CAT r eporter revealed that ERM acts as an activator of PS1 (Fig. 5A). Fig. 5. Ac tivation of PS1 transcription by ERM and the effect of promoter sequences on PS1 transcription 24 . (A) ERM activates the PS1 promoter in SK-N-SH cells. The ()119/+178) PS1CAT reporter (6 lg) was cotransfecte d with various amounts of ERM expression vector integrated in pC1 (ERM) or the same amount of empty vector (pC1).Theamountofproteinineachassaywasusedasaninternal standard. (B) PS1 sequences upstream or downstream from the +1 site confer transactivation by ERM. PS1CAT constructs (6 lg) containing various fragment o f the PS1 promoter w ere cotransfected with 3 lg pC1 expression vector including or not ERM. Promoter activities were compared using the c oncen tration of protein in the extract as an internal standard. The activity of ()119/+178) PS1CAT in the pres- ence of the empty vector was taken as 100%. The values represented in the graph are indicated below. In both A and B each data po int cor- responds to the average from threeplatesineachexperiment.Each data point was retested in three independent experiments. Ó FEBS 2004 Regulation of the presenilin 1 gene (Eur. J. Biochem. 271) 4491 We also asked which PS1 sequences are required for activation by ERM. Various promoter fragments were cotransfected with ERM (Fig. 5B). All the PS1 fragments assayed were transactivated by ERM, including )22/+178 containing a mutated )10 Ets, the +2/+178 3¢-deletion eliminating the major initiation site, and the )22 /+42 where all downstream Ets sites are absent. H ence it appears that either sequences upstream or downstream from the +1 site are sufficient to confer transactivation by ERM. ERM specifically recognizes Ets motifs upstream and downstream from the +1 site We first detected specific interactions between various portions of the PS1 promoter and ERM by supershifting the DNA–protein complexes appearing i n EMSAs using an anti-ERM Ig (Fig. 6A). In vitro translated ERM was preincubated with anti-ERM Ig. PS1 promoter probe was then combined with the protein mix and the protein–DNA interactions were analyzed by EMSAs. All portions of the PS1 promoter tested ()22/+6) ( lanes 1–4), (+50/+107) (lanes 5–7), (+107/+178) (lanes 8–10) formed complexes with in vitro translated ERM which could be supershifted by the antibody, indicating the presence of ERM binding site(s) in each of the three probes. We also examined the pattern of complexes formed o n the same promoter fragments inclu- ding individual point mutations at +65, +90, +129 and +165, with in vitro translated ERM as compared to t he control t ranslation mix containing the empty pC1 vector (Fig. 6B). All ERM-containing lanes (M) showed com- plexes that were absent in the control reactions (O) (dots) with the +50/+107 probe (lanes 1–6), suggesting the binding of ERM to the Ets sites present on this fragment. The binding profile on the +107/+178 probe (lanes 7–12) was not clearly modifed b y the addition of in vitro translated ERM (lanes M) as compared to control binding reactions (lanes O). The amount of complex formed may be small and/or not form a sharp band due to a lack of stability or a change in conformation during electrophoresis. Specific ERM–DNA complexes can, however, be captured a s supershifted complexes with this probe (Fig. 6A, lane 9) indicating that ERM may indeed recognize Ets sites +129 and/or +165. Discussion The PS1 promoter appears to be used alternatively in SK- N-SH cells and its SH-SY5Y subclone. Two Ets elements at )10 and +90 app ear to play an essential role although they are required to different extents in each cell type. Hence it is likely that Ets elements play a major role in determining both the level of expression and the location of the start of transcription of PS1 . In the Ets family of transcription AB Fig. 6. ERM and nuclear factors recognize specifically several Ets elements on the PS1 promoter 25 . (A) Supershift of specific DNA–protein complexes by anti-ERM Ig. In vitro translated ERM was preincubated with anti-ERM Ig. PS1 promoter probe fragments were combined with the protein mix and the protein–DNA complexes formed were analyzed by EMSA. Various portions of the PS1 promoter were tested for the presence of ERM binding site(s): )22/+6 (lanes 1–4), +50/+107 (lanes 5–7), +110/+178 (lanes 8–10). Lanes 1, 4 and 7 contained no antibody or antibody buffer, lanes 2, 5 and 8 (aEtv5) contained 2 lL ERM antibody (Santa Cruz Biotech sc-1955X), lanes 3, 6 and 9 (C) contained control antibody buffer. An arrow marks the position of supershifted complexes. (B) The binding of in vitro translated ERM is eliminated by mutations in Ets motifs. The binding of in vitro translated ER M o ver t he PS1 promoter was a ssayed by E MSAs inc luding the 32 P-labeled probes +50/+107 (lanes 1–6) and +107/+178 (lanes 7–12). Either the wild type probes or the same fragments containing mutant Ets sites at positions +65 (lanes 3, 4), +90 (lanes 5, 6), +129 (lanes 9, 10), +165 (lanes 11, 12) were incubated with 2 lL ERM translation mixture (M) or in absence of ERM protein (O). The position of complexes that are present only w ith t ranslation reactions containing ERM expression vector is indicated by dots. 4492 M. Pastorcic and H. K. Das (Eur. J. Biochem. 271) Ó FEBS 2004 factors a highly conserved 85 amino acid DNA binding domain called the ETS domain determines the specific recognition of the sequence (GGAA/T). However, se- quences immediately flanking this core element determine the recognition of target sites in different genes. In addition promoter sequences flanking the Ets motif determine activity t hrough protein–protein interactions between Ets factors and factors binding at adjacent sites, as w ell as cofactors that do not bind DNA themselves [29,30]. Hence the specific set of factors present in a given cellular context probably determines th e recognition a nd activation of promoter sequences by Ets f actors. Binding sites for S p1 factors are interdigitated with Ets m otifs on the PS1 promoter. The sites at )70 a nd )50 appear to have some degree of activity in both cell types, whereas the point mutation at )20 did not indicate any importance for the downstream motif. Sp1 factors have been known to activate transcription and to form synergistic i nteraction with Ets factors b inding at adjacent sites [31]. The specific factors that transactivate PS1 through binding at the Sp1 motifs remain to be identified. We do not know what major d ifferences between SH-SY5Y and SK-N-SH affect PS1. The results mostly demonstrate the potential for flexibility in the utilization of the promote r. Ana lysis of the start sites of the major species of PS1 mRNAs present in brain and human placenta revealed a set of 5¢ ends, all present between positions +1 and +90 defined in SK-N-SH ce lls [32]. H ence initiation within the first exon may indeed occur in vivo.Inthehuman presenilin 1 gene the 5¢-UTR is encoded by exons 1–4. Most full length cDNAs characterized and discussed above were initiated in exon 1 and directly spliced to exon 3 [32]. The absence of a 3¢ splice sequence in exon 2 suggests that its presence on a minority of mRNAs results from alternative initiation site(s) at e xon 2 rather than alternative splicing [32]. Such flexibility in the mechanism of initiation presents several advantages. First it is consistent with the ubiquitous expression o f the gene and increases its potential for expression in a variety of cellular systems. Secon d it increases the potential for regulation by various signal transduction cascades. Third alternative initiation of shorter transcripts may have implications for additional regulation at the level of translation. The 5¢-UTR of PS1 mRNA appears to be naturally unfavorable for translation, due to a number of AUG codons 22 present upstream from the actual initiation site and the potential for secondary structure of the RNA [32,33]. The relatively low translatability of the message may provide potential mechanisms for the rapid induction of PS1. A shift in the transcription initiation site and a shorter 5¢-UTR may increase translation by simply reducing the path of ribosome scanning. Alternatively it may alter secondar y structure and/or regulatory pr otein binding sites on the 5¢-UTR, which are likely to result in changes in the affinity of the cap binding the translation initiation complex. Transiently produced shorter RNAs may have a crucial r egulatory role during differentiation. Indeed promoter activity from cryptic promoters is found in long 5¢-UTRs of genes with a crucial role in the regulation of other genes [34]. Ets factors are likely to be important determinants in the choice of initiation site in a particular cellular context. The specificity of Ets factors is determined by their site of expression, their selective target sequence recognition, their selective abilities to interact with other transcription factor coactivators, and their selec tive modification by s ignal transduction kinases [29,30]. We have s hown here that ERM transactivates the PS1 promoter. In our assay system Ets elements either upstream o r downstream from + 1 appear to confer response t o ERM and downstream sequences retain a significant fourfold activation. Hence it is possible that in a particular cellular context ERM may also direct a significant level of initiation of shorter transcripts in vivo . ERM belongs to the PEA3 subfamily of Ets transcription factors that includes only three mem- bers (ERM, ER81 and PEA3), ER81 and ERM being the most related [35]. PEA3 members all contain the highly conserved ETS domain. ERM contains two DNA binding inhibitory domains flanking the Ets domain: the central 87 amino acid inhibitory domain is poorly conserved a nd virtually specific to ERM, and the C-terminal domain that is present in all three PEA3 members. Toge ther with a 32 residue N-terminal acidic stretch which is also conserved within the PEA3 group the C-terminal domain contains transactivation properties of the P EA3 factors [35]. There- fore, among the Ets factors that transactivate PS1 [27,28] ERM should confer certain specific regulatory properties. Although ERM mRNA is found in many tissues, its expression is remarkably high in the brain [35] and t his also suggests its importance for the regulation of PS1 i n vivo in this tissue. Acknowledgements This research was supp orted by a grant f rom the National I nstitutes of Health (AG 18452). References 1. Rogaev, E.I., Sherrington, R., Rogaeva, E.A., Levesque, G., Ikeda, M., Liang, Y., Chi, H., Lin, C., Holman, K., Tsuda, T., Mar., L., Sorbi, S., Nacmias, B., Piacentini, S., Chumakov, I., Cohen, D., Lannfelt, L., Fraser, P.E., Rommens, J.M. & St George-Hyslop, P.H. (1995) Cloning of a gene bearing missense mutations in early-onset familial Alzheimer’s disease. Nature 376, 775–778. 2. Li, X. & Greenwald, I. (1998) Additional evidence for an eight- transmembrane-domain topology for Caenorhabditis elegans and human presenilins. Proc.NatlAcad.Sci.USA95, 7109–7114. 3. Dewji, N.N., Valdez, D. & Singer, S.J. (2004) The presenilins turned inside out: implications for their structures and functions. Proc.NatlAcad.Sci.USA101, 1057–1062. 4. De Strooper, B. (2003) Aph-1, Pen-2, and nicastrin with presenilin generate an active c-secretase complex. Neuron 38, 9–12. 5. Kimberly, W.T., LaVoie, M.J., Ostaszewski, B.L.YeW., Wolfe, M.S. & Selkoe, D.J. (2003) Gamm a-secre tase is a me mbra ne protein complex comprised of pre senilin, nicast rin, Aph-1, and Pen-2. Proc. Natl Acad. Sci. USA 100, 6382–6387. 6. Edbauer, D., Winkler, E., Regula, J.T., Pesold, B., Steiner, H. & Haass, C. (2003) Reconstitution of gamma-secretase activity. Na t. Cell Biol. 5, 486–488. 7. Takasugi, N., Tomita, T., Hayashi, I., Tsuruoka, M., Niimura, M., Takahashi, Y ., Thinakaran,G.&Iwatsubo,T.(2003)The role of presenilin cofactors in the c-secretase complex. Nature 422, 438–441. 8. De Strooper, B., Saftig, P., Craessaerts, K., Vanderstichele, H., Guhde, G., Annaert, W., Von Figura, K. & Van Leuven, F. (1998) Ó FEBS 2004 Regulation of the presenilin 1 gene (Eur. J. Biochem. 271) 4493 Deficiency of presenilin-1 inhibits the normal cleavage of amyloid precursor protein. Nature 391, 387–390. 9. Price, D.L. & Sisodia, S.S. (1998) Mutant genes in familial Alz- heimer’s disease and transgenic models. Annu. Rev. Neurosci. 21, 479–505. 10. Levy-Lahad, E., Wasco, W., Poo rkaj, P., Ro mano, D.M., Oshima , J. & Pettingell, W.H., YuC.E., Jondro, P.D., Schmidt, S.D. & Wang, K. (1995) Candidate gene for the chromosome 1 familial Alzheimer’s disease locus. Science 269, 973–977. 11. Tanzi, R.E., Kowacs, D.M., Kim, T W., Moir, R.D., Guenette, S.Y. & Wasco, W. (1996) The gene defects responsible for familial Alzheimer’s disease. Neurobiol. Dis. 3, 159–168. 12. Sherrington, R., Rogaev, E.I., Liang, Y., Rogaeva, E.A., Levesque,G.,Ikeda,M.,Chi,H.,Lin,C.,Li,G.,Holman,K., Tsuda, T., Mar., L ., Foncin, J F., Bruni, A.C., Mo ntesi, M.P., Sorbi,S.,Rainero,I.,Pinessi,L.,Nee,L.,Chumakov,I.,Pollen, D., Brookes, A., Sanseau, P., Polinsky, R.J., Wasco, W., Da Silva, H.A.R.,Haines,J.L.,Pericak-Vance,M.A.,Tanzi,R.E.,Roses, A.D., Fraser, P.E., Rommens, J.M. & St George-Hyslop, P.H. (1995) Cloning of a gene bearing missense mutations in early-onset familial Alzheimer’s disease. Nature 375, 754–760. 13. Goate, A., Chartier-Harlin, M.C., Mullan, M., Brown, J., Craw- ford,F.,Fidani,L.,Giuffra,L.,Haynes,A.,Irving,N.,James,L., Mant, R., Newton, P., Rooke, K., Roques, P., Talbot, C., Pericak- Vance, M., Roses, A., Williamson, R., Rossor, M., Owen, M. & Hardy, J. (1991) Segregation of a missense mutation in the amy- loid precursor protein gene with familial Alzheimer’s disease. Nature 349, 704–706. 14. Fortini, M.E. (2001) Notch and Presenilin: a proteolytic mech- anism emerges. Curr. Opin. Cell Biol. 13, 627–634. 15. Kopan, R. & Goate, A. (2000) A common enzyme connects Notch signaling and Alzheimer’s disease. Genes Dev. 14, 2799–2806. 16. Weinmaster, G. (2000) Notch signal transduction: a real rip and more. Curr. Opin. Genet. Dev. 10, 363–369. 17. Fortini, M.E. (2002) G amma-secretase-mediated prote olysis in cell-surface-receptor signalling. Nat. Rev. Mol. Cell Biol. 9, 673– 684. 18. Okamoto, I., Kawano, Y., Murakami, D., Sasayama, T., Araki, N., Miki, T., Wong, A.J. & Saya, H. (2001) Proteolytic release of CD44 intracellular domain and its role in the CD44 signaling pathway. J. Cell Biol. 155, 755–761. 19. Marambaud,P.,Wen,P.H.,Dutt,A.,Shioi,J.,Takashima,A., Siman, A. & Robakis, N.K. (2003) A CBP binding transcrip- tional repressor produced by the PS1/-cleavage of N-cadherin is inhibited by P S1 F AD mu tations. Cell 114, 635–645. 20. De Strooper, B., Saftig, P., Craessaerts, K., Vanderstichele, H., Guhde, G., Annaert, W., Von Figura, K. & Van Leuven, F. (1998) Deficiency of presenilin-1 inhibits the normal cleavage of amyloid precursor protein. Nature 391, 387–390. 21. Passer, B., Pellegrini, L., Russo, C., Siegel, R.M., Lenardo, M.J., Schettini, G., B achmann, M ., T abaton, M . & D’Adamio, L. (2000) Generation of an apo ptotic i ntracellular peptide by c- secretase cleavage of Alzheimer’s amyloid b protein precursor. J. Alzheimers Dis. 2, 289–301. 22. Cao, X. & Sudhof, T.C. (2001) A transcriptionally active complex of APP with Fe65 and histone acetyltransferase Tip60. Science 293, 115–120. 23. Baek, S.H., Ohgi, K .A., Rose, D.W., Koo, E.H., Christopher, K., Glass, C.K. & Rosenfeld, M.G. (2002) Exchange of N-CoR cor- epressor and Tip60 coactivator complexes links gene expression by NF-jBandb-amyloid precursor protein. Cell 110, 55–67. 24. Scheinfeld, M.H., Ghersi, E., Laky, K ., Fowlkes, B.J. & D’Adamio, L. (2002) P rocessing o f b-amyloid precur sor-like protein-1 and -2 by c-secretase regulates transcription. J. Biol. Chem. 277, 44195–44201. 25. Pastorcic, M. & Das, H.K. (1999) An upstream element contain- ing an ETS binding site is crucial for transcription of the human presenilin-1 gene. J. Biol. Chem. 274, 24297–24307. 26. Pastorcic, M. & Das, H.K. (2003) An abbreviated procedure for the cloning and identification of Ets t ranscription factors r eg- ulating the expression of the human presenilin 1 gene. Brain Res. Brain Res. Protoc. 12, 35–40. 27. Pastorcic, M. & Das, H.K. (2003) Ets transcription factors ER81 and Elk1 regulate the transcription of the human presenilin 1 gene promoter. Brain Res. Mol. Brain Re s. 113, 57–66. 28. Pastorcic, M. & Das, H.K. (2000) Regulation of transcription of the human presenilin-1 gene by ets transcription factors and the p53 protooncogene . J. Biol. Chem. 27 5, 34938–34945. 29. Sharrocks, A.D. (2001) The ETS-domain transcription factor family. Nat. Rev. Mol. Cell Biol. 11, 827–837. 30. Verger, A. & Duterque-Coquillaud, M. (2002) When Ets tran- scription factors meet their partners. Bioessays 24, 362–370. 31. Gegonne, A., Bosselut, R., Bailly, R.A. & Ghysdael, J. (1993) Synergistic activation of the HTLV1 LTR Ets-responsive region by transcription factors Ets1 and Sp1. EMBO J. 12 , 1169– 1178. 32. Rogaev, E.I., Sherrington, R., Wu, C., Levesque, G., Liang, Y., Rogaeva,E.A.,Ikeda,M.,Holman, K., Lin, C., Lukiw, W.J., de Jong, P.J., Fraser, P .E., Rommens, J.M. & St George-Hyslop, P. (1997) Analysis of the 5¢ sequence, genomic structure, and alter- native splicing of the presenilin-1 gene (PSEN1) associated with early onset A lzheimer disease. Genomics 40, 415–424. 33. Kozak, M. (2000) Do the 5¢untranslate d domains of human cDNAs challenge the rules for initiation of translation (or is it vice versa). Genomics 70, 396–406. 34. Han, H. & Zhang, J T. (2002) Regulation of gene expression by internal ribosom e entry sites or cryptic p romoters: the eIF4G story. Mol. Cell. Biol. 22, 7372–7384. 35. DeLaunoit, Y., Baert, J.L., Chotteau, A., Mont e, D., Defossez, P.A., Coutte, L., Pelczar, H. & Leenders, F. (1997) Structure- function relationships of the PEA3 group of Ets-related tran- scription factors. Biochem. Mol. Med. 61, 127–135. 4494 M. Pastorcic and H. K. Das (Eur. J. Biochem. 271) Ó FEBS 2004 . Alternative initiation of transcription of the human presenilin 1 gene in SH-SY5Y and SK-N-SH cells The role of Ets factors in the regulation of presenilin. fragment SK-N-SH cells SH-SY5Y cells wt m wt m )11 9/ +17 8 10 0 ± 12 21 ± 5 10 0 ± 5 70 ± 20 )11 9/ +14 3 73 ± 10 10 ± 4 13 3 ± 30 70 ± 6 )22/ +17 8 91 ± 10 3.4 ± 2.6 10 0