Article Text


Genetic variation analysis of MLP, TFAP2A, and CSK in patients with neural tube defects
  1. R Klootwijk1,
  2. F A Hol1,
  3. M Wu2,
  4. J J H T Willemen1,
  5. P Groenen3,
  6. B Hamel1,
  7. H Straatman3,
  8. R P M Steegers-Theunissen3,4,
  9. E C M Mariman5,
  10. B Franke1
  1. 1Department of Human Genetics, University Medical Centre Nijmegen, PO Box 9101, 6500 HB Nijmegen, The Netherlands
  2. 2Howard Hughes Medical Institute, Center for Cancer Research and Department of Biology, Massachusetts Institute of Technology, Cambridge, MA 02139, USA
  3. 3Department of Epidemiology and Biostatistics, University Medical Centre Nijmegen, PO Box 9101, 6500 HB Nijmegen, The Netherlands
  4. 4Department of Gynaecology and Obstetrics, University Medical Centre Nijmegen, PO Box 9101, 6500 HB Nijmegen, The Netherlands
  5. 5Department of Human Biology, Maastricht University, PO Box 616, 6200 MD Maastricht, The Netherlands
  1. Correspondence to:
 Dr B Franke, Department of Human Genetics, University Medical Centre Nijmegen, PO Box 9101, 6500 HB Nijmegen, The Netherlands; 

Statistics from

Neural tube defects (NTDs) are congenital malformations which arise from incomplete closure of the neural tube during early embryogenesis. The most common types found in humans are spina bifida and anencephaly. These can occur together in a single family and in some cases even in one person.1

About 70–80% of all human NTDs show complex or multifactorial inheritance patterns, indicating that both genetic and environmental factors play a part in the aetiology of this malformation.2 Folic acid supplementation reduces the incidence of human and mouse NTDs.3–7 Furthermore, supplementation of myo-inositol reduces the incidence of NTDs substantially in the NTD mouse model, curly tail.8

In curly tail embryos, supplementary myo-inositol increases the flux through the inositol/lipid cycle, stimulating protein kinase C (Pkc) activity. Protein kinase C phosphorylates and thus activates several proteins that are directly involved in the formation of the neural tube, such as Ap2-α and Mlp (fig 1).9,10Ap2-α null as well as heterozygous and chimeric knockout mice have exencephaly.11–13Mlp knockout mice have exencephaly and spina bifida.14,15

Figure 1

Signal transduction relations between the proteins TFAP2A, MLP, and CSK. PTK, protein tyrosine kinase receptor; DAG, diacylglycerol; PLC, phospholipase C; PIP2, phosphatidylinositol 4,5-biphosphate; IP3, inositol 1,4,5-triphosphate; PKC, protein kinase C; SRC, Src related protein tyrosine kinases.

Apart from the genes involved in inositol signalling, NTDs also occur in mice deficient for the Csk gene.16 This gene encodes a negative regulator of Src family tyrosine kinases and acts through phosphorylation of the Src family members at their C-terminal tyrosine residue (fig 1).17 Csk is involved in the organisation of the cytoskeleton.18

Key points

  • In this paper, we present the results of an extensive mutation analysis in the genes coding for TFAP2A, MLP, and CSK from a large panel of patients with neural tube defects.

  • With a TDT based log linear approach we found a relative risk (RR) of 5.3 for spina bifida aperta conferred by the 1257C allele in the TFAP2A gene (95% confidence interval (95% CI) 0.66 to 42.2). Sequence analysis of the entire coding region of MLP showed no genetic variations. Single strand conformation polymorphism (SSCP) analysis of the coding region and part of the introns of the CSK gene followed by sequence analyis identified three novel silent polymorphisms: 210C>T, 759C>T, and 792C>T. These three polymorphisms were equally distributed in patients and controls.

  • In conclusion, by a TDT based log linear approach we were not able to show a significant role of the TFAP2A gene in the aetiology of human neural tube defects. However, our results suggest the need for studies in larger samples of patients for a more powerful analysis. Furthermore, this study shows that CSK and MLP are not major risk factors for neural tube defects in humans.

As outlined above, Ap2-α, Mlp, and Csk are associated with NTDs in mice. In humans, Stegmann et al,19 using a case-control study found no involvement of the human orthologue of Ap2-α, TFAP2A, in the aetiology of human NTDs. Furthermore, with transmission disequilibrium test (TDT) analysis for a small set of simplex families, no association was found between the human MLP gene and spina bifida.20 In this paper, with DNA material from a large panel of patients with NTDs, we present the results of an extensive mutation analysis in the genes coding for TFAP2A, MLP, and CSK.


Ascertainment of patients

Patients with non-syndromic NTDs described by Hol et al21 were used for mutation analysis in this study (group I). The collection includes samples from 38 multiple case families, which were selected according to the criteria already described.22 In brief, each family had at least two affected members with a close degree of relationship (⩽3). The DNA of one affected member of each of these families was included in the present study. The types of NTDs in these familial cases were spina bifida aperta (SBA) (36), encephalocele (1), and craniorachischisis (1). Furthermore, material from 79 patients with sporadic NTDs was included in the analysis, consisting of patients with SBA (75), anencephaly (2), and with encephalocele (2). For the TFAP2A association study, a panel of 32 patients (group II) was used as well as the NTD collection of group I. All of the affected children of group II had SBA and were ascertained in collaboration with the Dutch Spina Bifida Teams (P Groenen, submitted).

RNA isolation and cDNA synthesis

For the analysis of MLP, total RNA was isolated from cultured lymphoblasts with the RNAzol B kit according to the protocol of the manufacturer (Tel-Test, Texas, USA). First strand cDNA synthesis was performed in a 20 μl volume containing 250 ng total RNA, 100 ng random hexamer primers (Amersham Pharmacia, Freiburg, Germany), 10 mmol/l Tris-HCl (pH 8.0), 50 mmol/l KCl, 5 mmol/l MgCl2, 0.01% (w/v) gelatin, 1 mmol/l dNTPs (Life Technologies, Breda, The Netherlands), 50 U RNAsin (Amersham Pharmacia), and 1 U MMLV reverse transcriptase (Life Technologies). Complementary DNA synthesis was completed by incubation at room temperature for 10 minutes followed by incubations at 37°C for 60 minutes and at 95°C for 6 minutes.

SSCP analysis

Polymerase chain reaction (PCR) amplification of TFAP2A and CSK fragments was carried out in a total volume of 25 μl PCR buffer (50 mmol/l KCl, 10 mmol/l Tris-HCl (pH 9.0), 0.01% (w/v) gelatin, 0.1% (w/v) Triton X-100, 1.5–6 mmol/l MgCl2) containing 50 ng of genomic DNA, 0.45 mmol/l of both forward and reverse primer, 0.1 mmol/l dCTP, 0.4 mmol/l dTTP, 0.4 mmol/l dATP, 0.4 mmol/l dGTP (Amersham Pharmacia), 2.5 U Taq DNA polymerase (Life Technologies) and 0.1 μl [α32P]dCTP (200 μmol/l; ICN Biomedicals B V, Zoetermeer, The Netherlands). Samples were denatured at 92°C for five minutes and then subjected to 35 cycles of amplification (92°C for one minute, 55°C for 50 seconds, 72°C for one minute) in a PTC-100 thermal cycler (MJ Research via Biozym, Landgraaf, The Netherlands). Primers used in these PCR reactions are summarised in table 1.

Table 1

Primers for amplification of the coding region and flanking sequences of the human genes TFAP2A, MLP, and CSK and part of the promoter region of TFAP2A. The genomic sequence of TFAP2A and CSK, and the protein coding sequence of MLP were retrieved from Genbank (accession No X77343, X74765 and NM_023009, respectively)

Aliquots of the amplified TFAP2A and CSK fragments were mixed with one volume of formamide dye buffer (95% (w/v) formamide, 0.05% (w/v) bromophenol blue, 0.05% (w/v) xylenecyanol, and 20 mmol/l EDTA), denatured at 95°C for five minutes, and placed on ice; 4 μl samples were loaded on a 5% non-denaturing polyacrylamide gel with and without 10% glycerol. Electrophoresis was performed for 3–7 hours at 30 W and 5°C. The gels were dried and exposed to Kodak X-omat S film overnight.

Sequence analysis

Genetic variations in TFAP2A and CSK causing altered SSCP band patterns were identified by standard sequence analysis of PCR products. Sequences were determined in two directions with the forward and reverse primers used for SSCP analysis. Positions of identified polymorphisms in TFAP2A, CSK, and MLP were numbered starting with 1 at the first nucleotide of the ATG translation initiation codon. Sequencing was carried out on an ABI Prism 377 automated sequencer using the DyeDeoxyterminator cycle sequencing kit of the manufacturer (Applied Biosystems, Nieuwerkerk a/d IJssel, The Netherlands). The complete coding region of MLP was PCR amplified from cDNA in a total volume of 50 μl containing PCR buffer (5.0 mmol/l Tris-HCl (pH 9.0), 0.15 mmol/l MgCl2 and 0.01% Triton X-100), 200–500 ng cDNA, 0.58 μmol/l primer MLP1 and 0.67 μmol/l primer MLP2 (table 1), 0.13 mmol/l of each type of dNTP (dATP, dCTP, dGTP, and dTTP; Life Technologies), and 1.0 U of Taq DNA polymerase (Promega, Leiden, The Netherlands). The PCR cycling conditions involved an initial step of denaturation at 94°C for four minutes followed by 30 cycles of one minute at 94°C, one minute at 55°C, and one minute at 72°C. The final step was a 10 minute extension step at 72°C. The PCR amplification of MLP fragments was performed with a Cetus thermal cycling machine (Perkin Elmer, Norwalk, USA). For sequence analysis primers MLP1–5 were used (table 1).

Pyrosequence analysis

Templates used for pyrosequencing were produced by PCR amplification from 200 ng genomic DNA in a total volume of 35 μl containing 10 pmol primers AP1pyro and AP2pyro (table 1), 0.1 mmol/l of each type of dNTP (dATP, dCTP, dGTP, and dTTP; Life Technologies), 2.9 mmol/l Tris-HCl (pH 8.0), 14.3 mmol/l KCl, 0.002% (w/v) gelatin, 0.72 mmol/l MgCl2, and 0.14 U AmpliTaq Gold (Perkin Elmer). PCR conditions for pyrosequencing were as follows: 50 cycles of 92°C for one minute, 60°C for one minute, and 72°C for one minute. Pyrosequencing was performed on a PSQ96 System with the AP3pyro primer (table 1) according to the protocol of the manufacturer (Pyrosequencing AB, Uppsala, Sweden).23

Statistical analysis

The genotype frequencies of the 1257C allele of the TFAP2A gene in group I, II, and the combined group were tested for Hardy Weinberg equilibrium using the standard goodness of fit test.

Association between variant alleles of TFAP2A and susceptibility to SBA was tested with the TDT based log linear approach of Weinberg et al24 with an extension allowing for missing parents in case-parent triads.25 Unlike the commonly used McNemar’s TDT, the log linear approach allows the detection of and discrimination between effects of an inherited genotype and effects of the maternal genotype, which presumably would be mediated by prenatal factors.24,25 The program modelled the expected count for each of the 15 triad types shown in table 2A by fitting a Poisson regression. A maternal effect was included in the model by two yes/no covariables for mothers with homozygosity or heterozygosity for the 1257C allele of the TFAP2A gene (genotypes CC or CT). Also an effect of the inherited genotype was included by two yes/no covariables for children with one or two 1257C alleles. Testing of maternal and inherited genotype was done with the likelihood ratio test (LRT). We fitted the log linear model with the GENMOD procedure of the general statistics package SAS, release 6.12. Relative risks (RRs) and 95% confidence intervals (95% CIs) were calculated directly from the model. Apart from the TDT based log linear approach, McNemar’s TDT test was also performed as described by Spielman et al.26

Table 2A

Genotype distribution of parent-affected child triads for the 1257C>T polymorphism in the TFAP2A gene


The TFAP2A gene was analysed for its involvement in human NTDs. The entire coding region, its flanking sequences and part of the promoter region of the gene were analysed by SSCP analysis in 117 patients. Two polymorphisms were identified. The first concerned 773–53delG, recently also described by Stegmann et al,19 the second was −803G>T, which had been found earlier by Kawanishi et al.27 The two polymorphisms, 773–53delG and −803G>T, had allelic frequencies of 1.6% and 4.2%, respectively, in our sample of patients. The frequencies of the aberrant alleles were considered too low to test for an association with NTDs with a TDT based strategy. A novel frequent single nucleotide polymorphism (SNP), a 1257C>T transition of the TFAP2A gene, that had not been detected by SSCP analysis, was identified by sequence analysis of samples from unrelated patients with spina bifida. The T allele had an allelic frequency of 15.5% in our panel of patients. This polymorphism does not change the primary structure of the TFAP2A protein (N419N). For association testing two groups of triads of parents and affected child were genotyped for this SNP by pyrosequence analysis (table 2A). In this analysis only cases with spina bifida aperta were included. The genotype frequencies in groups I, II, and in the combined group fitted the Hardy Weinberg proportions (I: χ2=1.15, p=0.28; II: χ2=1.14, p=0.29; I + II: χ2=2.0, p=0.15). The genotyping data were analysed by TDT for an association between the transmission of alleles and susceptibility to NTDs (data not shown). To be able to use the full information content of our data, we then carried out a log linear test as described by Weinberg et al.25 This test has several advantages for this type of analysis of association compared with the commonly used TDT. Firstly, it allows the inclusion of incomplete triads. Secondly, it is able to distinguish between effects caused by the maternal genotype and those caused by the genotype of the patient. Thirdly, codominant and recessive or dominant effects of a risk allele can be analysed separately. Lastly, the log linear approach provides an estimate for the RR conferred by the genotypes containing the risk allele. With this approach for the analysis of the 1257C>T SNP in the TFAP2A gene we found that none of the maternal genotypes contributed significantly to the risk of SBA in their children (p value of LRT model 2 versus 1=0.95; table 2B). Simplifying the log linear model by excluding a maternal contribution (model 2) showed the impact of the inherited CT and CC genotypes on the aetiology of human SBA (CTchild, RR=5.06, 95% CI 0.63 to 40.7; CCchild, RR=6.52, 95% CI 0.79 to 54.0, p value of LRT model 2 versus model with no covariables=0.09; table 2B). The fact that the RR of heterozygotes was already strongly increased indicated a dominant effect of the 1257C risk allele. Therefore, the separate contributions of the CT and CC genotypes to the log linear model were combined in one yes/no covariable for children with CT/CC versus TT in the final model (table 2B, model 3). This final model, which approached significance (p value of LRT model 3 versus model with no covariables=0.054; table 2B), showed that the RR of having SBA conferred by the 1257C allele of the TFAP2A gene is 5.3. The 95% CIs were still very large (0.66 to 42.2) owing to small numbers (only one TT child and seven TT parents in the study).

Table 2B

Relative risk (RR) versus reference group predicted from log linear analysis. Model 1 assumes covariables for mother and child. Model 2 includes two covariables for the child. Model 3 includes one covariable for the child

The second candidate gene we analysed for its involvement in human NTDs was MLP. This was done by sequence analysis of the entire coding region of MLP. However, no mutations or polymorphisms were found.

The third gene that we analysed was CSK. The SSCP analysis of the coding region and part of the introns of the CSK gene followed by sequence analysis of aberrant SSCP patterns, showed three novel silent polymorphisms: 210C>T (Gly70Gly) in exon 3, and 759C>T (Gly253Gly) and 792C>T (Ile264Ile) in exon 8. All three SNPs were detected in both patients with NTDs and controls with similar frequencies (table 3). Frequencies were too low to perform TDT based association studies in our triads of parents and a child with SBA.

Table 3

Distribution of CSK alleles in familial and sporadic spina bifida aperta patients (group I) and controls


Genetic variation analysis of MLP and CSK exclude these genes from being major risk factors for human NTDs, despite the fact that Mlp and Csk deficiency cause NTDs in mice. Our findings for MLP are in agreement with a study carried out by Stumpo et al20 in a white population. They found no evidence with TDT for linkage disequilibrium between a nearby microsatellite marker and spina bifida in a small sample of 43 patients.

Our results obtained by mutation analysis and association testing indicate that TFAP2A is not a major risk factor for human NTDs. However, the results of the study of association approached significance even though our sample of patients was relatively small. Our results are comparable with results obtained from a case-control study performed by Stegmann et al,19 who did not find significant associations between NTDs and the intronic genetic variants in TFAP2A (773–53delG and 892+284insCT) in 190 patients and 222 controls. For the 773–53delG the number of sequence variants in both their patient and control panels was very small, six and eight variants, respectively. Furthermore, the cohort with NTDs tested included about 90% of patients with SBA, whereas the rest of the cohort consisted of anencephalic and encephalic patients. As it is not yet clear if SBA and the other types of NTDs are caused by exactly the same genetic factors, studies of associations with a mix of types of NTDs might not show the importance of a specific genetic factor.

In conclusion, we showed by log linear analysis that the TFAP2A gene is not a major risk factor involved in the aetiology of human NTDs. However, our results show the need for more studies of associations with more patients to exclude a role of the gene in NTDs entirely. Furthermore, this study shows that MLP is not a major risk for human NTDs, which is in line with a previous report.20 This is the first study that provides information on the distribution of genetic variants in the CSK gene in patients with spina bifida and unrelated controls but it also shows that we can exclude the gene as a major risk factor for human NTDs.

Possible limitations of our study are a certain heterogeneity and lack of categorisation in our panel of patients, that is, SBA might further be subdivided into the phenotypes meningocele and myelomeningocele. Furthermore, we cannot rule out the possibility that some mutations or polymorphisms might have been overlooked in the mutation analysis, as the sensitivity of SSCP analysis is not 100%.28 Also variation attributing to the risk of NTDs might be present in the promoter region of the investigated genes, as was recently shown for the PDGFR gene.29


We thank H Brunner, P Joosten, and A Heister for helpful advice and valuable discussion. This work was supported by the Dutch Organisation for Scientific Research (NWO), grant 925–01–006, and the Dutch Prinses Beatrix Fonds, grant 97–0107.


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.