The role of host genetic markers in HIV-1-specific CrNA remains largely unknown. In our present study we performed a GWAS in HIV-1 infected individuals of the ACS on the neutralization titers of CrNA in serum at ~35 months post seroconversion. The prevalence of CrNA in this cohort (27%) was similar to previously described cohorts 
Although none of the signals reached genome-wide significance, our GWAS revealed a considerable number of SNPs within the MHC region to be potentially associated with CrNA in HIV-1 infected individuals. Six of the SNPs that ranked in the top of associations with CrNA are located in the MICA gene region on chromosome 6, in two separate LD blocks. MICA is distant, but a structural homolog of MHC class I molecules, and a ligand for NKG2D, which is expressed on natural killer (NK) cells, most NKT cells, all CD8 T cells and subsets of γδ T cells and CD4 T cells 
. NKG2D is a potent activating NK cell receptor and it has been shown that the antiviral factor APOBEC3G and HIV-1 Vpr both upregulate NKG2D-ligands on HIV-1 infected cells, enhancing NK-cell mediated killing 
. Furthermore, elevated levels of soluble MICA have been shown to impair NK cell function in chronic HIV-1 infection 
. MICA is also a ligand for NKG2D on γδ T cells, which have been shown to increase neutralizing antibodies in the absence of CD4 αβ T cells against vesicular stomatitis virus (VSV), foot-and-mouth disease virus, as well as Borrelia burgdorferi,
the causative agent for lyme disease 
SNP rs2284178 which was found to be associated with CrNA in this study is located in the coding region of HCP5. However, it is not in high LD with the minor allele of SNP rs2395029 identified to be associated with HIV-1 viral load control via linkage with HLA-B57 
. Furthermore, the association of SNP rs2284178 with CrNA was even stronger after correction for viral load, suggesting that this polymorphism does not exert its role in the emergence of CrNA solely through controlling viral load, even though the geometric mean IC50
titer of sera across the viral panel did associate with viral load at set-point.
The outcome of our GWAS prompted us to compare the prevalence of all HLA-types with the CrNA neutralization titers in these individuals. Conversely to the association between HLA-B*57 and lower CrNA titers, the HLA-B*07 allele was more prevalent among individuals with higher titers of CrNA, although not significant after correction for multiple testing. Interestingly, HLA-B*07, which is part of the HLA-B7 supertype, is more prevalent among HIV-1 infected individuals who experience a more rapid disease progression and a higher viral load at set-point 
. The higher prevalence of a non-protective HLA-B type and the lower prevalence of protective HLA-B types in individuals with higher CrNA titers fit the observations that CrNA does not protect from HIV-1 disease progression 
. This observation is further strengthened by the fact that none of the top SNPs for CrNA was associated with disease progression in the ACS.
SNP rs10079250 in the coding region of CSF-1R, seems to be associated with CrNA. This gene was identified in a recent genome-wide analysis to play a role in HLA class II restricted antigen presentation 
. In our study no HLA class II alleles were significantly associated with CrNA. However, other genes involved in the HLA class II antigen presentation process may still be involved. CSF-1R, the natural ligand of this receptor, controls the production, differentiation and function of macrophages which may indirectly influence B-cell functioning. However, the exact mechanism by which this SNP in CSF1R may influence CrNA remains to be established.
The results from our present study are based on HIV-1 infected MSM from the Netherlands, infected with HIV-1 subtype B. The identified signals should be replicated in other cohorts to confirm their association with HIV-1 specific CrNA. In addition, it may be interesting to see if these SNPs are associated with the quality of humoral immunity in general.
In conclusion, we have identified several host genetic factors that may potentially be associated with the development of HIV-specific CrNA in serum collected at approximately 35 months post-seroconversion from HIV-1 infected individuals. Although genome-wide significance was not reached, the highest associations were seen for SNPs in or near genes in the HLA-region that may influence the eliciting of CrNA, which might be partially antigen load dependent.
These results indicate that a comprehensive analysis at the system-level can greatly facilitate the screening for host factors that associate with neutralizing responses in natural infection, This may provide clues for the optimalization of antigens and antigen formulations (i.e. adjuvanted antigens) to induce the immune pathways that are required for the desired adaptive immune response.