Levels of human platelet-derived soluble CD40 ligand depend on haplotypes of CD40LG-CD40-ITGA2

Increased circulating soluble CD40 ligand (sCD40L) is commonly associated with inflammatory disorders. We aimed to investigate whether gene polymorphisms in CD40LG, CD40 and ITGA2 are associated with a propensity to secrete sCD40L; thus, we examined this issue at the level of human platelets, the principal source of sCD40L. We performed single polymorphism and haplotype analyses to test for the effect of twelve polymorphisms across the CD40LG, CD40 and ITGA2 genes in blood donors. ITGA2 presented a positive association with rs1126643, with a significant modification in sCD40L secretion (carriers of C allele, P = 0.02), unlike the investigated CD40LG and CD40 polymorphisms. One CD40LG haplotype (TGGC) showing rs975379 (C/T), rs3092952 (A/G), rs3092933 (A/G) and rs3092929 (A/C) was associated with increased sCD40L levels (1.906 μg/L (95% CI: 1.060 to 2.751); P = 0.000009). The sCD40L level was associated with the inter-chromosomal CD40LG/CD40/ITGA2 haplotype (ATC), displaying rs3092952 (A/G), rs1883832 (C/T) and rs1126643 (C/T), with increased sCD40L levels (P = 0.0135). Our results help to decipher the genetic role of CD40LG, CD40 and ITGA2 with regard to sCD40L levels found in platelet components. Given the crucial role of sCD40L, this haplotype study in a transfusion model may be helpful to further determine the role of haplotypes in inflammatory clinical settings.

The latter was previously associated with some individual variation in platelet expression levels of GPIa/IIa (integrin α 2β 1), the preferential platelet receptor for collagen, which plays a crucial role in platelet adhesion and activation. This marker has further been defined as an independent predictor for the release of sCD40L 20,21 .
Therefore, the present study sought to highlight a possible genetic association (single markers and haplotypes) between 10 single nucleotide polymorphisms (SNPs) of CD40LG, rs1883832 of CD40 and rs1126643 of ITGA2, which display enhanced in vitro sCD40L release in PCs during storage.

Results
The genotype distribution for all investigated polymorphisms was found to be in Hardy-Weinberg equilibrium.
Correlation between sCD40L levels and single polymorphisms. Relevant CD40LG, CD40 and ITGA2 polymorphisms were assessed.
No significant correlation was detected between the investigated CD40LG polymorphisms and sCD40L levels in the PCs, neither at D 0 nor at D del , for all ten investigated CD40LG polymorphisms (Table 1).
There was also no association found between sCD40L levels and rs1883832 of the CD40 gene, although this polymorphism has been shown elsewhere to regulate CD40L expression 19,22 .
However, there was a significant association with rs1126643 of ITGA2; the platelets of C-allele carriers (CC and CT) secreted elevated levels of sCD40L upon storage in shelf-life conditions-that is, with no deliberate stimulation-compared with the non-C carriers (TT homozygous), P = 0.08 at D 0 and P = 0.02 at D del (Table 1).
Haplotype association with sCD40L levels. Five CD40LG haplotypes accounted for 97.6% of all potential combinations, including rs975379 (C/T), rs3092952 (A/G), rs3092933 (A/G) and rs3092929 (A/C). The association between CD40LG haplotypes and CD40L secretion leading to sCD40L is reported in Table 2. One haplotype (H 4 : TGGC; frequency: 2.6%) was associated with the largest increase in sCD40L levels at the day of PC delivery, i.e., 1.906 μg/L (95% CI: 1.060 to 2.751; P = 0.000009), compared with the reference haplotype H 1 (CAGA). None of the other four CD40LG haplotypes was associated with any difference in sCD40L secretion and plasma levels.
Eight haplotypes accounted for 100% of potential multigene CD40LG (rs3092952)/CD40 (rs1883832)/ITGA2 (rs1126643) combinations. These haplotypes were tested for association with PC sCD40L levels. A significant association was shown for haplotype H C at the time of preparation [H C : ATC; frequency: 10.8% with increased sCD40L level, i.e., 0.189 μg/L (95% CI: − 0.182 to 0.560; P = 0.0135)], compared with the most common haplotype H A (ACC; Table 3). It is interesting to note that haplotype H G (GTC), which differs from H C (ATC) by the first allele (A instead of G of the rs3092952 polymorphism), was also associated with a small non-significant increase in sCD40L levels at D del , i.e., 0.553 μg/L (95% CI: 0.235-0.87; P = 0.071). No other haplotype was consistently associated with sCD40L levels.
These results are reported after full adjustments for covariates, i.e., gender, age, number of platelets (10 9 /L) and the number of days of storage (inventory condition) prior to delivery. The major contributing factor exerting a positive effect on sCD40L levels was the platelet count (P = 0.000406 and P < 0.000001 at D 0 and D del , respectively) for the CD40LG haplotype analysis. The same covariate was identified in the inter-chromosomal haplotype analysis (P = 0.001457 and P < 0.000001 at D 0 and D del , respectively).

Discussion
SNP association studies did not reveal any association with sCD40L levels measured in PCs at either D 0 or D del. These results were consistent with our previous findings 16 showing that no particular pattern of CD40LG in individuals who donated platelets by single apheresis and those in whom PCs induced an adverse transfusion reaction. It does seem, in light of the results and limitations of this study, that there are no so-called "regulator polymorphisms of CD40LG", at least in healthy individuals.
Our findings did disagree with the observations of Malarstig et al., who showed in a large cohort of patients with cardiovascular disease that carriers of the G allele of rs3092952 1 had a 10% higher sCD40L level. However, to the best of our knowledge, this finding has yet to be consistently reproduced and despite this apparent correlation, rs3092952 did not confer an increased risk of cardiovascular adverse events. Regarding CA repeats in the 3′ -UTR of CD40LG, our results corroborate those of Bugert et al., indicating that neither the sizes of the alleles nor the genotypes of the CA repeat polymorphism were associated with plasma sCD40L levels 23 (Table 1). Dai et al. found no association between CA repeats and mRNA expression in CD4 + T cells 24 . Only studies by Perez-Aciego et al. found that CD40L expression (membrane and mRNA) decreases in CD4 + T cells, but only in those with 24 CA alleles 25,26 . Taken together, these results indicate that the regulation of membrane CD40L expression and sCD40L levels is complex and implies not only genetic variations in CD40LG. Furthermore, one cannot exclude the possibility of environmental influences, cell-related post-translational regulation, catabolic regulation and/ or polygenic control.
Extensive literature reported the association of the CD40 -1C/T polymorphism (rs1883832) with CD40L expression. However, published results are difficult to compare because either the T or the C allele is implicated, with relatively equal frequencies between them 19,22 . The present study found no significant association with this polymorphism. Some of our unpublished data, however, confirm a positive correlation between surface protein CD40L expression and the rs1883832 genotypes present in the CD40 gene in T lymphocytes, in line with the findings of Zhang B et al. 22 .
The only genotype associated with sCD40L levels in PCs was found with C-allele carriers (CC and CT) of ITGA2, with a positive association (P = 0.02). This finding is discordant with the work of Antoniades et al., who defined the T allele as an independent predictor for the release of sCD40L in healthy subjects but only in the subgroup with von Willebrand factor greater than or equal to the median 21 . Thus, again, an association with a specific allele (T or C) seems to be an ambiguous result, similar to other polymorphisms, e.g., CD40 19,22 . This polymorphism may affect either mRNA stability or a regulatory genetic region, with a subsequent change in the density of the expressed molecule on the platelet surface and, consequently, an alteration of platelet adhesion and activation, although this polymorphism does not alter the functional status of the protein 20 . This possibility may explain why the association between the ITGA2 genotype and sCD40L levels was only found on the day of delivery, suggesting the requirement for sustained platelet activation over time, due to the preparation and storage process 27,28 .
We identified a highly significant association with the CD40LG haplotype at the day of PC delivery (H 4 : TGGC; P = 0.000009; Table 2). This result may be explained by progressive sCD40L release during platelet storage 13,27 . Moreover, we also identified one inter-chromosomal CD40LG/CD40/ ITGA2 haplotype, H C (ATC) from the rs3092952, rs18828232 and rs1126643 genotypes, associated with sCD40L levels (P = 0.0135; Table 3). This result highlights the importance of the association of several polymorphisms in different genes that are involved in the complex regulation of this immuno-modulatory molecule that is released after platelet activation. Notably, the frequency of H C was nearly 11% of all the investigated CD40LG/CD40/ ITGA2 haplotypes.
H G was also associated with a relative increase in sCD40L levels at D del , although the difference did not reach statistical significance. This similarity could be linked to the similarity of the two haplotypes, H G (GTC) and H C (ATC), given that they show a difference in only the allele of the first rs3092952 polymorphism, which presents a regulatory function; however, that function was not identified in our study, which considered the polymorphism alone. The apparent controversies between individual polymorphisms and haplotype analysis are explained by their different biological values and a higher informative analysis was attributed to haplotype investigation 29 . Sequential nucleotide variants may catch subtle changes in protein function, regardless of the presence of nucleic acid changes in the coding region 29,30 .
We identified two haplotypes associated with high levels of plasma sCD40L. However, a large fraction of CD40L is known to be carried by extracellular vesicles, including microvesicles and exosomes, after platelet activation 31 . As most methods used to assay sCD40L (i.e., ELISA and Luminex technologies) don't distinguish between free sCD40L and microvesicle-carried-CD40L 32,33 , we believe that the levels of sCD40L measured in our study are not an underestimate.
To the best of our knowledge, this is the first study to suggest the haplotypic and polygenic control of sCD40L release from platelets stored for transfusion. This highly informative in vitro model contributes to a better understanding of the genetic role of CD40LG, CD40 and ITGA2 with regard to sCD40L platelet secretion and the subsequent levels in PCs and, consequently, the role of sCD40L in AEs. However, we cannot exclude that the biological response may also be affected by pathophysiologic aspects and co-morbidities in the transfusion recipients. The replication of the associations we found, as well as additional functional studies in pathological situations with inflammatory disorders, is required to confirm our findings in a larger number of individuals and must be extended to other clinical settings (e.g., in vivo models).

Methods
Subjects. Ethics statement. The study was carried out in accordance with the Helsinki Declaration and approved by the ethical committee of the F. Hached University Hospital, Sousse, Tunisia. Informed and written consent was obtained from all the healthy donors who participated in this study.
Study population. The studied cohort comprised 142 volunteer blood donors, including 52 males and 90 females, 26 ± 10 years of age (mean ± SD), who donated whole blood at the Transfusion Centre of Sousse. Individual PCs were derived from each donation as described 34 . None of the blood donors were family-related; donors entered the study randomly, on the sole basis of the timing of their donations; no selection criteria specific to this study were applied.  (rs975379, rs3092952, rs3092933, rs3092929) Haplotype interaction with sCD40L  Table 2. Frequency distribution of CD40LG haplotypes in PCs and their interaction with sCD40L level, considering sCD40L level, on the day of preparation (D 0 ) and on the day of PC delivery (D del ). Data represent the mean and relative 95% CI of the difference in platelet supernatant sCD40L levels (μg/L) observed in one copy of each haplotype configuration compared with the reference haplotype. Haplotypes (Hn) are indicated in ACGT format with rs975379 (C/T), rs3092952 (A/G), rs3092933 (A/G) and rs3092929 (A/C). The data are boldfaced if the corresponding haplotypes show a significant P-value (< 0.05). All data were adjusted for gender, age and platelet count. Data on D del were further adjusted for the number of days before delivery.  Table 3. Frequency distribution of CD40LG/CD40/ ITGA2 haplotypes in PCs and their interaction with sCD40L levels. Data represent the mean and relative 95% CI of the difference in platelet supernatant sCD40L levels (μg/L) observed in one copy of each haplotype configuration compared with the reference haplotype. Haplotypes (Hn) are indicated in ACGT format with rs3092952 (A/G), rs1883832 (C/T) and rs1126643 (C/T). The data are boldfaced if the corresponding haplotypes show significant P-values (< 0.05). All data were adjusted for gender, age and platelet count. Data on D del were further adjusted for number of days before delivery.
Scientific RepoRts | 6:24715 | DOI: 10.1038/srep24715 Genotyping. Genomic DNA was obtained from peripheral venous blood using the FlexiGene DNA Kit (Qiagen, Paris, France). For CD40LG, rs3092952 A/G and CA repeats were genotyped as previously described using denaturing High Performance Liquid Chromatography (dHPLC) and capillary electrophoresis, respectively 16 . The remaining 8 CD40LG polymorphisms were genotyped via multiplex PCR amplification in two groups of quadruplex Tetra primer Amplification Refractory Mutation System-PCR (T-ARMS-PCR) followed by fragment analysis (the first group included rs3092945 C/T, rs975379 C/T, rs3092929 A/C and rs3092920 G/T; the second group included rs3092948 C/G, rs3092927 A/G, rs715762 C/T and rs3092933 A/G) [35][36][37][38][39] . Each of the two groups of primers used for the amplification of twelve fragments of different sizes was pooled in a single 25-μL reaction volume, which contained 12.  40 . Genotype data were integrated with two additional SNPs, mapping at CD40 (rs1883832 C/T) and at ITGA2 (rs1126643 C/T), genotyped by single T-ARMS-PCR under the same technical conditions, with annealing temperatures of 67 °C and 67.5 °C for CD40 and ITGA2, respectively (Supplementary Tables S1 and S2).
Soluble CD40L assay. Non-leukodepleted, individual PCs were prepared according to the standard protocol of the blood bank of Sousse 34 . Under sterile conditions, 4 mL was derived from PCs, first on the day of blood donation and platelet bag preparation (D 0 ) and again on the day of delivery (D del ). The samples were centrifuged at 180 g for 10 min and supernatants were collected and frozen at − 80 °C and then thawed immediately before being assayed at room temperature. Soluble CD40L in PC supernatants was measured using a commercially available method by Luminex Technology according to the manufacturer's instructions (Milliplex Map Kit Millipore, Darmstadt, Germany). Statistical analysis. All statistical analyses were performed with XLSTAT TM software (Addinsoft, Paris, France) using non-parametric methods. The Hardy-Weinberg equilibrium test was used to control the genotyping results. We used the Kruskal-Wallis test to evaluate the association between the genetic polymorphisms and sCD40L levels.
To test for the prevalence of genotypes of the CA repeat polymorphism, we defined three types as referred to Table 1 as follows: (i) genotype 'only 26′ with the 26 CA allele exclusively (this group comprised females with two 26 alleles and males with one 26 allele, because the gene is X-linked); (ii) genotype '26+ ' with any allele other than 26 in addition to one 26 allele (heterozygous females); and (iii) genotype 'no 26' with no 26 allele (any genotype without the 26 allele).
For haplotype association, we used Haploview 4.2 software and employed the algorithm proposed by Gabriel 41 in order to choose tagged SNPs based on our previous study 16 . The haplotype tagger function was used to identify redundant SNPs, which were considered redundant if the pairwise LD (r 2 ) was ≥ 0.8. Haplotypes with a frequency ≤ 1% were excluded from the analysis (Fig. 1).
The consequences of displaying haplotypes on the propensity for platelet secretion of sCD40L were evaluated using THESIAS v3.1 software 42 . The association of each haplotype with sCD40L levels was measured using a regression parameter and the 95% confidence interval (CI), where the effect of each haplotype is compared with the most frequent haplotype, termed the 'reference' in the regression model. Adjustments were performed for different covariates, such as gender, age, platelet count and storage length prior to delivery (considered when testing for sCD40L at D del ).
All significance thresholds were set at P < 0.05.