The structural and functional analysis of the human HSPA2 gene

HSPA2 is a human counterpart of the testis-specific rodent Hst70/Hsp70.2 gene. In contrast to the latter, the expression of the human HSPA2 gene is not limited to the testis, and recent data show that human tumor cells can express this gene at significant levels. The characteristics of HSPA2 expression suggests that it can influence the phenotype and survival of cancer cells similarly as overexpression of major members of the HSP70 gene family. Until now, neither the structure of the transcription unit of the human HSPA2 gene has been established nor a functional analysis of its promoter performed. In this study we established that the human HSPA2 gene, in contrast to its rodent counterparts, is intronless and has a single transcription start site. We also show that the same type of HSPA2 transcripts are synthesized in the testes and in cancer cell lines. In order to perform a functional study of the HSPA2 promoter, we used a transient transfection assay and found that the 392 bp fragment upstream of the ATG codon was a minimal region required for efficient transcription, while a 150 bp deletion from the 5' end of this region dramatically reduced the promoter activity. Delineation of the minimal promoter is a basic step toward identifying the cis and trans elements involved in the regulation of the HSPA2 gene expression in cancer cells.


INTRODUCTION
HSPA2, a member of the HSP70 family of heat shock genes is a human counterpart of mouse and rat HspA2 genes (called previously Hsp70.2 in mouse and Hst70 in rat), which are specifically and highly expressed in spermatocytes (Krawczyk et al., 1988a;1988b;Zakeri et al., 1988;Bonnycastle et al., 1994).Rodent HspA2 genes encode a molecular chaperone essential for dissociation of synaptonemal complexes, progression of meiosis and male fertility (Dix et al., 1996;1997;Zhu et al., 1997).
Human HSPA2 gene is highly expressed in spermatids and tails of mature spermatozoa, suggesting its involvement in late stages of spermatid maturation (Huszar et al., 2000).The HSPA2 protein was identified as a testis-specific creatinine kinase type M (CK-M), the decreased activity of which has been connected with male infertility (Yesilli et al., 2005;Huszar et al., 2006;Cedenho et al., 2006).Relatively high levels of HSPA2 transcripts were found in various human non-testicular tissues (Bonnycastle et al., 1994), but the corresponding protein was essentially undetectable (Son et al., 1999).In contrast to normal tissues, a sig-

2007
W. Pigłowski and others nificant expression of the HSPA2 protein was found in various human cancer cell lines and primary nonsmall-cell lung cancer tissues (Ścieglińska et al., unpublished).Since another group (Rohde et al., 2005) reported that silencing of the HSPA2 gene in cancer cells led to growth arrest, HSPA2 has been considered important for cancer cell survival.
In order to study the mechanism regulating expression of the HSPA2 gene in cancer cells it is essential to establish the structure of the gene transcription unit and the minimal promoter sequence.In our previous papers we reported the structure of rodent HspA2 genes and their promoter regions.We established that their transcription is initiated at two major start sites, T1 and T2, (the latter being localized within the intron) that are found approx.350 bp (T1 site) and 115 bp (T2 site) upstream of the ATG codon.Transcription initiated at the T1 site generates transcripts containing intronic sequences that are subsequently spliced out, whereas transcripts initiated at the T2 site are not spliced and probably are not translated (Ścieglińska et al., 2001; 2004).We also determined that major regulatory sequences responsible for efficient expression of the rat HspA2 (Hst70) gene are present within a 306 bp DNA fragment (-368/-62) upstream of the ATG translation start codon (Widłak et al., 1994;1995;Ścieglińska et al., 2001).The regulatory elements indispensable and sufficient for spermatocyte-specific activity of the rat HspA2 (Hst70) gene are localized within a 165 bp fragment between the T1 and T2 transcription start sites, encompassing exon 1 and the 5' part of the intron (Ścieglińska et al., 2004).It seems that an octamer sequence, localized directly downstream of the T1 transcription start site, is responsible for regulating the developmental activation of the gene (Ścieglińska et al., 2004).
So far, neither the structure of human HSPA2 gene promoter region nor its functionality have been established.In this study, using 5' RACE and RT-PCR, we found that human HSPA2 gene is intronless and that its transcription is initiated from a single start site.We also found that transcription of the HSPA2 gene is initiated at the same site in cancer cells and in testes.Using transient transfection of cancer cell lines with diffent HSPA2 transcription levels we also performed functional studies of the HSPA2 promoter and determined that a 392 bp fragment upstream of the ATG codon functions as a minimal promoter of HSPA2.

RNA isolation.
Total RNA was prepared using the guanidine isothiocyanate method (Chomczyński & Sacchi, 1987) from human normal testes and cancer cell lines indicated below.RNA samples were purified from contaminating DNA by on-column treatment with RNase-free DNase I (Qiagen).After inactivation of the enzyme, an aliquot of the RNA sample was digested with RNase A and control PCR for DNA contamination was performed using GAPDH, as described earlier (Ścieglińska et al., 1997).The integrity of purified DNase I-treated RNA was analyzed using Bioanalyzer (Ambion).In RT-PCR only RNA with the RNA Integration Number (RIN) higher than 8 was used.RNA concentration was measured using NanoDrop ND1000 (NanoDrop Technology).
5' RACE analysis.The 5' RACE analysis of HSPA2 transcripts was done using GeneRacer TM Kit (Invitrogen) and total testicular RNA as a template.This kit enables RNA ligase-mediated rapid amplification of 5' cDNA end (RLM-RACE) and ensures amplification of only full-length mRNA via elimination of truncated molecules from the amplification process.To accomplish this, total RNA was treated with calf intestinal phosphatase (CIP) to remove the 5' phosphates.As CIP has no effect on full-length capped mRNA, the digestion eliminates truncated mRNA and non-mRNA from subsequent ligation with an RNA oligonucleotide linker.The procedure was performed according to the manufacturer's protocol.An outline of the RLM-RACE procedure is shown in Fig. 1.To increase specificity of RACE amplification for reverse transcription, an HSPA2specific oligonucleotide (Nested) complementary to the HSPA2 coding region was used.The first round of RACE-PCR was performed with the following primers: Gene Racer 5' (supplied with the kit) and HSPA2-specific primer (Nested).The second round of PCR was performed with the following primers: Gene Racer 5' Nested (supplied with the kit) and HSPA2-specific primer (2-as).The sequences of the HSPA2-specific oligonucleotides and their positions in the HSPA2 gene are given in Table 1.The conditions of the RACE-PCR reactions were as in the manufacturer's protocol; primer annealing temperatures optimal for HSPA2-specific primers were used and the reaction mixture contained 1 M betaine and the proofreading DNA polymerase Pfu (2 U, Fermentas).To confirm the specificity of RLM-RACE, products of the second round of RACE-PCR were electrophoretically resolved in 2% agarose and transferred onto Hybond-N membrane (Amersham) by capillary blotting.Hybridization was performed overnight at 58°C with 20 pM oligonucleotide probe (T2, Table 1) 5' labeled with [γ-32 P]dATP and T4 polynucleotide kinase (Roche), as described previously (Ścieglińska et al., 2001).The amplification products were also cloned into pCR-Blunt vector using PCR-Blunt-Topo Cloning Kit (Invitrogen), according to the manufacturer's protocol, and the insert was sequenced with BigDye Terminator v3.1 Cycle Sequencing Kit (PE Applied Biosystem).
Lower-case letters indicate nucleotides not complementary to the HSPA2 gene sequence.

Oligonucleotides sequence
Position in the HSPA2 gene
For transient transfection, A549 and HepG2 cells were seeded at 1-2.5 × 10 4 cells per 35 mm culture dish 24 h before transfection.The cells were incubated with 2.5 μg of pHSPA2-CAT plasmids mixed with 50 ng of pGL3-Control plasmid (Promega) and 10 μl of lipofectin (Invitrogen) in serum-free growth medium according to the manufacturer's protocol.The transfection mixtures were removed after 8 h of incubation and replaced by growth medium.Cells were harvested 48 h after transfection and lysed in reporter lysis buffer (Promega) according to the manufacturer's protocol.Crude lysates were clarified by centrifugation (14 000 rpm, 5 min at 4°C) and total protein content was determined using a Protein Assay Kit (BioRad).
For 5-azacytidine (5-aza) treatment, HepG2 cells were seeded at 30% confluence in 60 mm culture dishes and after 24 h normal growth medium was replaced by medium containing 2 μM 5-azacytidine (Sigma).The medium supplemented with 5-aza was replenished at 2-day intervals.Cells were harvested after 10 days of culture and RNA or proteins were isolated.
Luciferase and CAT reporter gene assay.To perform CAT assays supernatants were heated at 60°C for 10 min and centrifuged again (14 000 rpm, 10 min at 4°C).An aliquot of the clarified cell extract containing 10-20 μg of protein was added to the same volume of reaction mixture containing 0.25 M Tris/HCl (pH 7.8), 1 mM EDTA, 4 mM acetyl-CoA (Sigma) and 2.5 μl of [ 14 C]chloramphenicol (2.5 μCi/ml, ICN).The samples were incubated for 2 h at 37°C.The acetylated forms of chloramphenicol were by thin-layer chromatography as described earlier (Widłak et al., 1995).CAT activity was expressed as the percentage of acetylated prod-ucts formed per one hour per milligram of protein, as described previously (Widłak et al., 1995), normalized according to the activity of an internal control (luciferase acivity) and expressed in relative activity units.
To perform the luciferase assay, 1 μl of nonheated protein extract was diluted in reporter lysis buffer (Promega) and 20 μl was added to 100 μl of luciferase assay substrate (Promega) prepared according to the manufacturer's instruction.Luciferase activity was measured using a Berthold Lumat LB95d instrument and calculated per 1 mg of protein.For estimating the transfection efficiency the luciferase activity was taken as internal control.

The structure of HSPA2 transcripts
To determine the structure of the 5' region of the HSPA2 gene we initially assumed that the general transcription unit structure is similar in the rodent and human homologues, i.e. the mouse HspA2 (Hsp70.2), the rat HspA2 (Hst70) and the human HSPA2 genes.We found this assumption justified by the fact that all of these genes share high nucleotide sequence similarity, not only within the coding region but also upstream of the ATG codon, including the splicing donor and acceptor sites (Bonycastle et al., 1994;Widłak et al., 1995;Ścieglińska et al., 2001).The basic issue to resolve was to find out if the HSPA2 gene transcription was initiated at two sites and whether the gene contained an intron upstream of the ATG codon.
In order to localize the transcription start site of the HSPA2 gene, we performed 5' RACE analysis using mRNA isolated from normal human testes.For selective detection of full length 5' UTR ends of the HSPA2 transcripts we choose the RLM-RACE method (described in Materials and Methods).To enhance the detection specificity for cDNA synthesis and subsequent RACE-PCR reactions, the HSPA2specific Nested and the 2-as antisense oligonucleotide were used (Fig. 1A; for nucleotide sequences and positions of the oligonucleotides see Table 1).The nucleotide sequences of the RLM-RACE products indicate that the majority (eleven out of twelve) of amplified 5' ends of the HSPA2 transcripts start from nucleotide 109 upstream of the ATG codon (Fig. 1D).
To characterize the 5' UTR structure of the HSPA2 transcripts synthesized in testes we performed RT-PCR reactions according to the schedule shown in Fig 2A .As expected, the RT-PCR reactions gave products of the same length (145 bp) as the PCR products obtained from the total DNA template using the sense primer (Int-2) complementary to sequences directly downstream of the detected HSPA2 transcription start site T(-109) and the antisense primer matching a region downstream of the ATG codon (2-as) (Fig. 2B).By RLM-RACE we did not detect transcripts initiated at the transcription start site corresponding to the T1 site of rodent HspA2 genes.No specific amplified product was obtained by RT-PCR using the sense primer (1-s) complementary to sequences directly downstream of the putative T1 transcription start site and the antisense primer (2-as) complementary to a region downstream of the ATG codon (Fig. 2C).Only unspecific products were present, the nucleotide sequences of which were complementary to those of the gene coding for 28S rRNA (Fig. 2C).These data indicate that transcription of human HSPA2 gene is not initiated in the region corresponding to the T1 transcription start site of rodent HspA2 (Hst70/Hsp70.2) genes.Thus, unlike rodent HspA2 genes, the 5' region of human HSPA2 gene is devoid of intronic sequences.The HSPA2 transcripts detected by us are initiated at a transcription start site which corresponds to the T2 transcription start site of rodent HspA2 genes (compare Figs. 1D and Fig. 2A).
We have also identified by 5' RACE some cDNA clones which possibly represent less abundant HSPA2 transcript synthesized in testes from some additional transcription start site localized about 200 bp upstream of the main T(-109) initiation site.Its 5' UTR region corresponds exactly to the nucleotide sequence of DNA, indicating that this less prominent transcript is not subjected to splicing.This tran-script was also detected by RT-PCR with the sense primer (2-s) corresponding to the HSPA2 sequence directly upstream of the putative donor splicing site and the antisense primer (2-as) located downstream of the ATG codon (see Fig. 1C).These data, together with our data presented earlier (Widłak et al., 1995;Ścieglińska et al., 2001) show that, in contrast to its rodent counterparts, human HSPA2 gene does not contain an intron at its 5' end and is transcribed from the main transcription initiation site preceded by a canonical TATA-box.
A remarkable feature of mammalian spermatogenesis is that many genes in spermatogenic cells exhibit conspicuously different expression patterns than in somatic cells.A large number of mRNAs in spermatogenic cells differ in size and structure from transcripts of the same genes in somatic cells by usage of spermatogenic cell-specific transcription start sites, as well as alternative splicing and polyadenylation sites.There are multiple reports that the altered testicular transcripts encode proteins of different functions to those performed in somatic cells (Eddy, 2002).To find out whether the structure of the HSPA2 mRNA synthesized in cancer cell lines differs from that produced in testicular cells we performed an RT-PCR analysis according to the schedule shown in Fig. 2A.
Total RNA was isolated from a panel of cancer cell lines originating from human skin (Me45), colon (HCT116), liver (HepG2), breast (MCF-7), lung (A549, NCI-H1299, NCI-H358, NCI-H292), and nontumorigenic epithelial cell lines from breast (HBL-100, MCF-10A) and lung (BEAC-2B).Examples of this analysis are shown for total mRNA isolated from HCT116 and MCF-7 cell lines (Fig. 2B, 2C, 2D).Abundant RT-PCR products were obtained only in the reaction with the sense primer (Int2-s) and total RNA template isolated from the majority of the cell lines tested.By RT-PCR reaction with the sense primer (2-s) we detected also very low amounts of HSPA2 transcripts beginning probably at minor additional transcription start site.The specificity of the RT-PCR products was confirmed by Southern hybridization with an internal specific probe (not shown).The results obtained by RT-PCR analysis indicate that in the analyzed cancer cell lines transcription of HSPA2 starts at the same initiation site as in testicular cells.
The primers used in the described RT-PCR analysis enabled amplification of only the 5' region of the HSPA2 transcripts.Using primers complementary to sequences within the 5' UTR and to the 3' UTR regions (primers Int2-s and 3-as; for sequences of primers see Table 1) we show by RT-PCR that the HSPA2 transcript of the expected length (i.e.testicular one), albeit at different levels, could be detected in the majority of the cell lines studied (Fig. 3), with the exception of HepG2 hepatoma and H292 lung mucoepidermoid carcinoma.These results indicate that in testes and in cancer cell lines the same HSPA2 protein is produced.

Functional study of the HSPA2 gene promoter
To obtain a preliminary characterization of the HSPA2 gene promoter region we analyzed the regulatory elements required for expression of the HSPA2 gene localized directly upstream of the transcription start site.Fig. 4A shows the localization of putative regulatory elements in a 860-bp-long 5'flanking region upstream of the HSPA2 transcription start site.To confirm its promoter functionality, the pHSPA2(-868/-4)CAT6 reporter plasmid was constructed that contains the HSPA2 fragment from nucleotides -868 to -4 fused to the chloramphenicol acetyltransferase (CAT) reporter gene.To test this promoter's activity we used A549 cells in which we had previously found high expression of the endogenous HSPA2 gene (Ścieglińska et al., not shown).As a control we used the HepG2 cell line in which we had found endogenous HSPA2 gene activity to be fully repressed.The pHSPA2(-868/ -4)CAT6 plasmid was transiently cotransfected with the luciferase-encoding pGL3-Control vector (Promega).Surprisingly, in both cell lines we found similar, high levels of the promoter activity.Subsequently, we analyzed expression of two groups of the pHSPA2-CAT plasmids.The first group included pHSPA2(-530/-4)CAT6, pHSPA2(-392/-4)CAT6 and pHSPA2(-243/-4)CAT6 constructs containing promoter fragments shortened from the 5' end while the second included pHSPA2(-868/-247)CAT6 and pHSPA2(-868/-351)CAT6 constructs with promoter fragments shortened from the 3' end.The absolute level of CAT activity was normalized to luciferase activity (internal control) and expressed in relative units.The normalized activity of the pHSPA2(-868/-4)CAT6 construct (the longest promoter fragment analyzed) was set as 100% and the activity of the other pHSPA2-CAT constructs was normalized against that value.
Truncation of the HSPA2 promoter from the 5' end to nucleotide -43 (pHSPA2(-243/-4)CAT6 construct), which additionally eliminated three consensus elements recognized by the SP1-like transcription factor, dramatically reduced the reporter gene activity to approx.10% of that observed with the full-length HSPA2 promoter.Reduction of the promoter activity to nearly background levels was also observed when sequences surrounding the T(-109) transcription start were removed from the HSPA2 promoter fragments (pHSPA2(-868/-247)CAT6 and pHSPA2(-868/-351)CAT6 constructs).
Functional analysis of the deletion mutants revealed that the HSPA2 gene fragment between nucleotides -392 and -4 contains the regulatory elements indispensable for efficient transcriptional activity of the HSPA2 promoter.It can be thus speculated that distal Sp1 sequences are necessary for transcription of the HSPA2 gene.
Despite the lack of endogenous expression of the HSPA2 gene in HepG2 cells (Fig. 3) a significant activity of pHSPA2-CAT expression construct could be observed after transient transfection (Fig. 4).Thus, in HepG2 hepatoma cells the pHSPA2-CAT transcription units can be efficiently transcribed when present in the episomal form which suggests repression of the chromosomal HSPA2 gene dependent on the chromatin structure.However, it seems that the mechanism of the HSPA2 repression is independent of DNA methylation, as treatment of HepG2 cells with the DNA methylation inhibitor 5-azacytidine did not induce expression of the endogenous HSPA2 gene (not shown).
Our earlier study on the transcription regulation of the rat HspA2 gene also suggests an involvement of chromatin structure in the repression/derepression of genes highly expressed in the testes (Widłak et al., 2003).We demonstrated that a transgene driven by certain fragments of the promoter region of the rat HspA2 gene was not transcribed in the testes of transgenic mice, whereas it was highly expressed after transfection into rat FTO-2B hepatoma cells.And, conversely, some other rat HspA2 promoter fragments were highly active in transgenic mice but significantly less active in transfected FTO-2B cells.
It is well established that certain genes which are highly expressed in spermatocytes and repressed in somatic cells can be activated in tumor cells.Such genes are known as so-called cancer-testis antigen genes (Scanlan et al., 2004).These genes attract significant interest especially if the corresponding protein is expressed on the surface of cancer cells, being a potential target for immunotherapy.Considering the expression pattern of the HSPA2 gene, i.e. its highest activity in the testis, repression or insignificant activity in somatic cells and enhanced activity in cancer cells it can be proposed that HSPA2 may encode a novel cancer-testis antigen.It seems significant in this context that a recent global profiling of the cancer cells proteome led to identification of the HSPA2 protein on the surface of several cancer cell lines (Shin et al., 2003).Among the latter were also A549 cells derived from non-small-cell lung carcinoma, in which we found a significant expression of the HSPA2 protein (Ścieglińska et al., unpublished).However, whether the HSPA2 gene does indeed exhibit all the features attributed to genes belonging to the cancer-testis an- CAT reaction mixture contained 20 µg of protein and the reaction time was 2 h.The absolute level of CAT activity was normalized to the activity of the control luciferase activity.CAT activity for individual constructs is expressed as the percentage of the pHSPA2(-868/-4)CAT6 construct (the longest promoter fragment analyzed) and represents the mean of three experiments ± S.D. using independent plasmid preparations.Details in the Materials and Methods.
W. Pigłowski and others tigen group requires further study.The mapping of the transcription initiation site and preliminary characterization of the potential promoter region of the HSPA2 gene performed in this study will enable us to further characterize the mechanisms involved in the activation of the HSPA2 gene in tumor cells.

Figure 1 .
Figure 1.Detection of the HSPA2 gene transcription start site by RLM-RACE method.(A) Scheme of procedure using GeneRacerTM Kit (Invitrogen) described in more detail in Materials and Methods.(B) RLM-RACE products obtained after second round of PCR amplification and separated in 2% agarose gel.Legend: T, RLM-RACE reaction with testicular RNA as template; C, control RLM-RACE reaction without template.Numbers on the left-hand side of photograph indicate the size (in bp) of DNA marker.(C) Southern hybridization of RLM-RACE products (from panel B) with HSPA2-specific T2 oligonucleotide probe (for its nucleotide sequence and position in HSPA2 gene see Table 1).(D) Scheme showing structure of the HSPA2 gene promoter region and localization of the main transcription start site marked T(-109).The position of the T(-109) start site was identified by determining the nucleotide sequence of RLM-RACE product cloned into pCR-Blunt vector (Invitrogen).

Figure 2 .
Figure 2. Detection of HSPA2 transcripts synthesized in testis and in selected cancer cell lines.(A) Upper part shows structure of rodent HspA2 genes.Vertical arrows indicate the position of two main transcription start sites marked T1 and T2.Lower part shows position of oligonucleotides (marked as horizontal arrows) used in the analysis of HSPA2 transcripts by PCR and RT-PCR (for their nucleotide sequence and position in the HSPA2 gene see Table 1).Vertical arrow indicates position of HSPA2 main transcription start site marked T(-109).(B) Examples of RT-PCR results showing transcripts initiated at the T(-109) main transcription start site of the HSPA2 gene.(C) RT-PCR analysis showing that HSPA2 transcription is not initiated in region corresponding to T1 initiation site of rodent HspA2 gene.Only unspecific bands are visible; their nucleotide sequence was verified by sequencing and is complementary to gene coding for 28S rRNA.(D) RT-PCR analysis showing transcript initiated at an additional minor transcription start site from which intronic sequences are not spliced out.DNA, products of PCR reaction obtained on DNA matrices using indicated primers."−" denotes control RT-PCR reactions with RNA digested with DNase I and RNase A (for details see Materials and Methods); "+" denotes RT-PCR reactions with total RNA as template.Numbers on the left-hand side of each photograph indicate the sizes (in bp) of DNA markers.Under each photograph are indicated primers used in RT-PCR analysis.

Figure 3 .
Figure 3. Analysis of HSPA2 transcription in various human cell lines.(A) A scheme showing structure of the HSPA2 gene (upper part), positions of oligonucleotide primers (horizontal arrows) used in RT-PCR and PCR analysis of HSPA2 transcripts and expected size (in bp) of RT-PCR product (lower part).Vertical arrow indicates position of the main transcription start site.For nucleotide sequences and positions of the primers see Table 1.(B) Detection of HSPA2 transcripts in cancer cell lines and (C) in immortalized epithelial cell lines by RT-PCR analysis.Control RT-PCR reactions were made with primers corresponding to GAPDH gene (bottom picture)."-", control RT-PCR reactions with RNA digested with DNase I and RNase A (for details see Materials and Methods); "+", RT-PCR reactions with total RNA as template.Numbers on the right side indicate sizes (in bp) of DNA markers.

Figure 4 .
Figure 4. Functional analysis of the HSPA2 promoter by transient transfection experiments.(A)Possible regulatory elements localized upstream of the ATG codon of the HSPA2 gene; ERE(nf) -estrogen response element found to be non-functional (nf)(Krawczyk et al., 1993), T(-109) -the HSPA2 gene transcription start site.(B) Structure of the pHSPA2-CAT constructs.The open box represents the CAT reporter gene.The thick line represents the promoter of the HSPA2 gene.Numbers are coordinates of the nucleotide sequence with respect to nucleotide +1 (A in the ATG codon).Details on the construction of expression vectors are available on request.(C) CAT activity in transiently transfected A549 (dark gray box) and HepG2 (light gray box) cells.Cells were transfected with a mixture containing 2.5 µg of pHSPA2-CAT plasmid and 50 ng of pGL3-Control plasmid (Promega).CAT reaction mixture contained 20 µg of protein and the reaction time was 2 h.The absolute level of CAT activity was normalized to the activity of the control luciferase activity.CAT activity for individual constructs is expressed as the percentage of the pHSPA2(-868/-4)CAT6 construct (the longest promoter fragment analyzed) and represents the mean of three experiments ± S.D. using independent plasmid preparations.Details in the Materials and Methods.