A hypermorphic SP1-binding CD24 variant associates with risk and progression of multiple sclerosis

Human samples

Demographic and clinical information of control and MS patients enrolled are listed in Table 1. The enrollment and study design have been approved by the Institutional Review Boards of the Ohio State University, University of Michigan, and the ACP. As controls, a total of 935 human subjects from the general Caucasian population were obtained from the American Red Cross between 1999 and 2006 using leftover peripheral blood samples, as described previously [20,21]. Sporadic Caucasian MS samples were collected during similar period. Samples used include those described in previous studies [18,19] and those provided by ACP. The clinical diagnosis of MS type and the EDSS score [29] were provided by either Dr. Kottil Rammoham or ACP.

Polymorphism identification

The CD24 locus was amplified by PCR. The PCR reactions were carried out in a 20μl volume containing ~20 ng of genomic DNA, 2.0μl 10X Pfx Amplification Buffer, 2.0μl 10X PCRx Enhancer solution, 0.6μl 10 mM mixture of deoxynucleotide triphosphates (dATP, dCTP, dGTP, and dTTP), 0.5μl 50 mM MgSO₄, 10 pmol of each primer, and 0.4 unit of Platinum Pfx DNA polymerase (Life Technologies, Grand Island, NY). After a 10-min initial denaturation step at 95°C, 35 cycles of PCR reaction consisting of 95°C for 30 sec, 55°C for 30 sec, and 68°C for 90 sec were carried out, followed by a 10-min final extension step at 68°C in a thermal cycler (Veriti 96 wells, Life Technologies, Grand Island, NY). Potential polymorphisms in the promoter, exon 1 and intron 1 were discovered by DNA sequencing of PCR products from 264 MS samples (21 additional cases excluded due to sequencing failure), 301 control samples (15 additional samples were excluded due to sequencing failure) and 325 MSGG samples (1 additional sample excluded due to sequencing failure). The variants were identified using Mutation Surveyor 3.01 software (SoftGenetics LLC, State College, PA).

Allelic discrimination

Genomic DNA isolated from peripheral blood leukocytes was used to identify CD24 polymorphisms. PCR amplification was performed with P-534C-F (GTG GCA ATG CAC TTG CTC CAG GAC C) and P-442C-R (GCG AGC CAC ACA CGC CGC GCT GGG). The PCR reactions were carried out in a 20μl volume containing 20 ng of genomic DNA, 2.0μl 10X High fidelity PCR Buffer, Amplification Buffer, 2.0μl 10X PCRx Enhancer solution, 0.6μl 10 mM mixture of deoxynucleotide triphosphates (dATP, dCTP, dGTP, and dTTP), 0.5μl 50 mM MgSO₄, 5 pmol of each primer, and 0.75 unit of Platinum Taq DNA polymerase High Fidelity (Life Technologies, Grand Island, NY). After a 5-min initial denaturation step at 95°C, 10 cycles of PCR reaction consisting of 95°C for 30 sec, 65°C for 30 sec, and 68°C for 30 sec and 25 cycles of PCR reaction consisting of 95°C for 30 sec, 60°C for 30 sec, and 68°C for 30 sec were carried out, followed by a 10-min final extension step at 68°C in a thermal cycler (Veriti 96 wells, Life Technologies, Grand Island, NY). The PCR products were electrophoresed on 2.5% agarose gels. Those with a band of 143 bp were scored positive for having at least one allele of P-534C-P-442C.

We tested the sensitivity the allele-specific PCR using 31 samples with known P-534C-P-442C haplotype and found the sensitivity to be 100%. However, among 630 subjects with no possible P-534C-P-246G haplotypes, 6 were found positive, yielding a false positive rate of 0.95%. These false positive samples, however, were all identified by subsequent sequencing (see below). Therefore the false positivity at this step does not affect the final results reported here.

The PCR products were then used in a Taqman Assay to determine the composition of P-492G/C in the PCR product. Samples were amplified in a model 7900-HT sequence system (Life Technologies, Grand Island, NY) using forward and reverse primers (P-492-Taqman-F, CGG AGG CGC GGA CTT T and P-492-Taqman-R, GGC CCA AGT TTC CTT TGT TTC C) along with FAM and VIC dye-labeled probes P-492G-probe-VIC (CTT ACC CCC CAA AAG A) P-492C-probe-FAM (TTA CCC CCG AAA AGA). If the genotype G or CG of P-492 was present, the haplotype CGC will be considered to be present on at least one allele. For those samples with P-492G genotypes, a new PCR reaction was carried out to amplified 1491 bp CD24 promoter, using primer F2 (GTT GGT CTG GAA CTC CTG ACC TCA GGT) and R2 (CCT CTG GGT GAA AGT GGG AA). Sanger sequencing was then used to determine homozygocity or heterozygocity of the CD24CGC allele. This step also identified the low rate (0.95%) of false positive samples in the allele-specific PCR reaction.

CD24 mRNA measurement

CD24 mRNA expression was analyzed using real-time PCR (Applied Biosystems ABI Prism 7500 Sequence Detection System, Life Technologies, Grand Island, NY). The average expression was determined by comparing the Ct (cycle number when a preset signal threshold is reached) of CD24 amplification with that of GAPDH. The primers are: CD24-realtime-F (TTC TCC AAG CAC CCA GCA) and CD24-realtime-R (TGG AAT AAA TCT GCG TGG GTA).

Promoter assay

The human CD24 promoter-luciferase gene vectors (pGL2-CD24) were constructed by inserting the DNA fragments from the proximal promoter region of CD24 into the pGL2 vector. The proximal promoter fragment (from 53 to 614 5’ of the translation starting site) covering six polymorphism sites (P-534, P-492, P-442, P-388, P-366 and P-246) were cloned into pGL2 to create human CD24 promoter luciferase plasmids. The luciferase activity was measured in a Veritas Microplate Luminometer (Turner BioSystems, Sunnyvale, CA) using a Dual Luciferase Assay System (Promega, Madison, WI).

Chromatin immunoprecipitation

ChIP of SP1 were carried out using breast cancer cell line MDA-MD453 according to published procedures [30]. The anti-SP1(Santa Cruz Biotechnology, sc-644) and anti-IgG antibodies (1 μg/mL) were used to pull down chromatin associated with SP1. The amounts of the specific DNA fragment were quantitated by real-time PCR and normalized against the genomic DNA preparation from the same cells. The sequences of ChIP real-time PCR primers are: P-534-F (AGAGATAACCCTGCCCGAG) and P-534-R (CCAAGTTTCCTTTGTTTCCC).

Statistical analysis

In the case-control population study, patients and normal controls were examined for any significant differences in their genotype (allele or haplotype) distributions in each of the CD24 polymorphisms at the population level. First, the Hardy-Weinberg equilibrium assumption was confirmed for each polymorphism the controls. We performed either Chi-square test or Fisher's exact test for each polymorphism by comparing the distribution of the genotypes of the cases to that of the normal controls. Then, using the counts of one of the genotypes as a reference, the ORs for the remaining genotype variants were computed. The associated 95% CIs for the ORs were obtained through a multivariate logistic regression model adjusted by age as a potential confounding factor using SPSS 16.0 software (SPSS Inc., Chicago, IL). The risk of MS associated with each haplotype having frequency >1% was compared with the most common haplotype using logistic regression models with the HAPSTAT software [31,32].

In survival analysis, we estimated the Kaplan-Meier survival curve for patients with each of the genotypes. We calculated the observed survival time of a patient as the time from the first day of symptoms to reaching EDSS 6.0 (in years) or to the day of the last follow-up visit if EDSS 6.0 had not been reached. In the latter situation, the recorded survival time was treated as a censored observation. Association between the estimated survival curves and the underlying genotypes or haplotypes were then assessed using a log-rank test. All analyses were performed by using SPSS 16.0 software (SPSS Inc., Chicago, IL).

A novel haplotype in CD24 promoter associate with risk and progression of MS

We first identified the full-length DNA sequence as it was not available (submitted to NCBI database, accession number FJ226006). The localization of CD24 to 6q21 was confirmed by fluorescence in situ hybridization using the intronic probe (Figure S1). By sequencing the entire CD24 locus of 50 general population controls, we identified 7 new SNPs that are tightly clustered within a 553 base-pair (bp) region in the CD24 promoter (Figure 1). Since no SNP exist the 2743 bp intron and only 4 previously reported SNPs were found in the 2017 bp exon 2 [20,21], the promoter is the most polymorphic region in the CD24 gene.

To determine whether the new SNPs are associated with MS risks, we carried out a case/control study by sequencing the CD24 promoter of 264 MS case and 301 race-, gender- and age-matched controls (Table 1) [20,21]. This analysis reveals that none of the SNPs in the promoter region is associated with MS risk (Table S1). However, using sequencing data of 301 control and 264 control Caucasians, we identified 2 haplotypes of CD24 promoters, CCGCCCG or CCGCCTG, that are significantly associated with MS risk (Table 2). To establish a high risk allele that can be easily genotyped, we searched for a minimal haplotype. As shown in Table 2, combination of P-534C, P-492G and P-442C constitutes a haplotype that associates with risks, which we refer as CD24^CGC. We therefore developed a PCR-based assay to identify heterozygous and homozygous samples. First, we amplified the samples that contained a P-534C-P-442C haplotype. The positive samples were further typed for the genotype of the P-442 position by Taqman-based approach. The promoter region of the samples that have at least one allele of CD24^CGC was completely sequenced to determine whether they are heterozygous or homozygous for the CD24^CGC allele. The cohort used consists of a total of 764 independent MS samples and 935 age-, race- and gender-matched Caucasian controls (Table 1). As shown in Table 3, 16.8% MS and 12.3% control samples are CD24^CGC/CGC, yielding an OR of 1.57 (P=0.003). The heterozygous CD24^CGC also showed a statistical trend (OR 1.21, P=0.08). The MS subjects are Caucasians in the United States and samples are obtained from two sources: the Ohio State University used in our previous study [20,21], and those from the Accelerated Cure Project (ACP, Waltham, MA 02451, USA). The distribution of the CD24^CGC genotypes of the MS samples obtained from OSU and ACP are not significantly different (Table S2).

To evaluate possible population stratification, we compared the frequency distribution of more than 1122 SNPs within a 30 megabase region in the X-chromosome using r = 0.8 as linkage threshold for marker selection and P = 0.001 as threshold for deviation from Hardy–Weinberg equilibrium in control group to delete questionable genotyping. The typing was done for a different project and the 216 cases and 236 controls were selected before the CD24 genotypes are known. As detailed in the Table S3, despite the high number of markers used, we failed to detect significant population stratification (λ < 1.0).

To determine whether the promoter variant is associated with more rapid progression of MS, we carried out survival analysis using Expanded Disability Status Scale (EDSS) 6.0 as the endpoint. As shown in Figure 2, while 50% of CD24^CGC/CGC patients lost ability for independent walking within 18 years after first MS symptom, those with other genotypes reached the same endpoint in approximately 27 years. The association between CD24 genotype and MS progression is statistically significant (P=0.016).

Hypermorphic promoter activity of the new risk variant

To determine whether the CD24^CGC variant affects CD24 mRNA levels in MS patients, we tested CD24 transcript levels in the gender and age-matched MS patients. As shown in Figure 3A, the levels of CD24 mRNA in the peripheral blood leukocytes associate with the gene dosages of the CD24^CGC allele (P=7x10^-4). To directly demonstrate that the polymorphism dictates promoter activity, we generated luciferase reporters using the CD24 sequence from 3 haplotypes. Compared with the major non-risk haplotypes (AGC and AGT), risk haplotype (CGC) exhibits significantly more activity (Figure 3B). Thus, the risk allele has a hypermorphic CD24 promoter.

CD24 promoter region contains multiple binding sites for GC- or GT-box-binding transcription factors [33], the prototype of which is SP1 (Figure 1). We hypothesized that differential activation of the CD24 gene by SP1 may contribute to differential MS risk. As the first test for the hypothesis, we silenced the SP1 gene using shRNA prior to transfection of the luciferase promoters. As shown in Figure 4A, SP1 silencing by two independent shRNA, but not scrambled shRNA, selectively reduced the hypermorphic CD24 promoter activity without significantly affecting the activity of the ACA promoter. Therefore, SP1 is required only for the enhanced promoter activity of the risk variant.

To determine whether different alleles of CD24 locus differentially bind to endogenous SP1, we searched a large panel of human cell lines and found a CD24^CGC/AGC cell line. This allowed us to directly compare endogenous interaction between SP1 and promoter of different alleles of CD24 by chromatin immunoprecipitation (ChIP). We carried out a chromatin immunoprecipitation (ChIP) with anti-SP1 mAb and compared the CD24 promoters DNA co-precipitated by the anti-SP1 mAb. As shown in Figure 4B, endogenous SP1 binds strongly to CD24 promoter. Remarkably, when the PCR products were genotyped by RFLP using P-534C-specific restriction enzyme (BsrFI), essentially all SP1-bound CD24 promoter products were found to be derived from the CD24^CGC risk allele. In toto, the data in Figure 4 demonstrate that the MS risk allele is hypermorphic because of its stronger interaction with SP1.