CD33 Alzheimer's risk-altering polymorphism, CD33 expression, and exon 2 splicing

Abstract

Genome-wide association studies are identifying novel Alzheimer's disease (AD) risk factors. Elucidating the mechanism underlying these polymorphisms is critical to the validation process and, by identifying rate-limiting steps in AD risk, may yield novel therapeutic targets. Here, we elucidate the mechanism of action of the AD-associated polymorphism rs3865444 in the promoter of CD33, a member of the sialic acid-binding Ig-superfamily of lectins (SIGLECs). Immunostaining established that CD33 is expressed in microglia in human brain. Consistent with this finding, CD33 mRNA expression correlated well with expression of the microglial genes CD11b and AIF-1 and was modestly increased with AD status and the rs3865444C AD-risk allele. Analysis of CD33 isoforms identified a common isoform lacking exon 2 (D2-CD33). The proportion of CD33 expressed as D2-CD33 correlated robustly with rs3865444 genotype. Because rs3865444 is in the CD33 promoter region, we sought the functional polymorphism by sequencing CD33 from the promoter through exon 4. We identified a single polymorphism that is coinherited with rs3865444, i.e., rs12459419 in exon 2. Minigene RNA splicing studies in BV2 microglial cells established that rs12459419 is a functional single nucleotide polymorphism (SNP) that modulates exon 2 splicing efficiency. Thus, our primary findings are that CD33 is a microglial mRNA and that rs3865444 is a proxy SNP for rs12459419 that modulates CD33 exon 2 splicing. Exon 2 encodes the CD33 IgV domain that typically mediates sialic acid binding in SIGLEC family members. In summary, these results suggest a novel model wherein SNP-modulated RNA splicing modulates CD33 function and, thereby, AD risk.

Figure 1.

CD33 immunohistochemistry in human brain. CD33-immunopositive cell profiles (arrows) show morphology consistent with microglia in both AD and non-AD samples (a, b). Immunofluorescence was used to help distinguish the CD33-immunopositive cell types. c/d, e/f, and g/h show the same microscope fields. Sections were immunostained for CD33 and counterstained for IBA-1 (a microglial/macrophage lineage marker) or GFAP (an astrocyte lineage marker), as indicated. Microglia with rounded morphologies tended to colocalize with both CD33 and IBA-1 (blue arrows in a–d), with more ramified IBA-1-immunopositive microglia staining less positively for CD33 (yellow arrows); these results suggest that CD33 expression is increased in ameboid microglia. In addition to apparent double labeling in microglial cells, there are areas in which GFAP-positive label colocalizes with CD33 immunopositivity, indicating either astrocytic expression of CD33 or the engulfment of CD33-positive cells by astrocytes (note the two DAPI-labeled nuclei present in the CD33-positive/GFAP-positive area indicated by arrows in g, h). Sections are from superior/middle temporal gyri of individuals with AD pathology (a, c–h) or without (b) AD pathology. Ctrl, Control. Scale bar: a, b, 50 μm; c–h, 20 μm.

Figure 2.

CD33 isoform expression relative to AD status. CD33 expression correlated well with microglial gene expression (a, presented as geometric mean of CD11b and AIF-1, r2=0.64), as well as each microglial mRNA individually (data not shown). An association between CD33 and AD status was visualized by considering the ratio of CD33 to the geometric mean of the microglial reference genes (b). D2-CD33 correlated well with CD33 expression (c; r² = 0.88, 0.67, and 0.51 for the AA, CA, and CC genotypes, respectively). The percentage of CD33 expressed as D2-CD33 was strongly associated with rs3865444 genotype (d, p = 1.2 × 10⁻¹³).

Figure 3.

CD33 and TREM2 represent opposing forces in microglial activation. Sialic acid binding to CD33 results in activation of SHP1 phosphatase, which inhibits immune cell activation. The D2-CD33 isoform lacks the exon that encodes the apparent sialic acid binding domain and hence is proposed to be nonresponsive to sialic acid and inactive. In contrast, an unknown ligand binds TREM2 that signals through DAP12 to activate the tyrosine kinase Syk, resulting in microglial activation. Genetic evidence suggests that loss of CD33 function decreases AD risk, whereas loss of TREM2 function increases AD risk.