The JAK2 46/1 haplotype influences PD-L1 expression

Abstract

Although described more than a decade ago, the mechanism by which the JAK2 46/1 haplotype increases the risk of developing JAK2-mutated myeloproliferative neoplasms (MPNs) remains unexplained. Inflammation and immunity are linked to MPN development and thus could be relevant to the mechanism by which 46/1 mediates its effect. Here, we show that programmed death-1 receptor ligand (PD-L1) expression is elevated in 46/1 haplotype, both in healthy carriers and in CD34+ cells from patients with MPN. Using circular chromosome conformation capture, we observed that PD-L1 and the neighboring PD-L2 loci physically interact with JAK2 in a manner that differs between 46/1 and nonrisk haplotypes. CRISPR/Cas9 genome editing identified a region within JAK2 intron 2 that influences both JAK2 and PD-L1 expression. We suggest that increased PD-L1 expression may be relevant to the mechanism by which 46/1 leads to an increased inherited risk of developing MPN.

Graphical abstract

Figure 1.

JAK2 46/1 haplotype influences PD-L1 expression levels. (A) The percentage of PD-L1–expressing cells, measured by flow cytometry, in different peripheral blood cell populations from healthy donors (46/1 nullizygous [blue] and 46/1 homozygous [red]; n = 37 each). ∗P < .05 (nonparametric Kruskal-Wallis test). (B) PD-L1 protein levels, determined by enzyme-linked immunosorbent assay and expressed as PD-L1/total protein (pg/μg) in CD34⁺ cells obtained from bone marrow of JAK2-positive patients with MPN (46/1 nullizygous [blue], n = 11; 46/1 homozygous [red], n = 9). Error bars indicate standard error of the mean. B, B lymphocytes; NK, natural killer cells; NKT, natural killer T cells; T, T lymphocytes.

Figure 2.

Chromatin contacts of JAK2 with the PD-L1 locus differ between the nonrisk and the 46/1 haplotype. Visualization of a 1.6 Mb region from human chromosome 9, spanning the JAK2 and PDL1/2 loci (hg19 chr9:4300000-5900000). Chromatin interactions established from the 7 distinct genomic regions or viewpoints (VP2, VP3, VP4, VP8, VP10, VP5, and VP6; genomic coordinates in supplemental Table 2) are shown as spider plots. Healthy individuals who were 46/1 nullizygous (green and light green, n = 2) or 46/1 homozygous (blue and light blue, n = 2) are shown for each VP, with differences between them around PD-L1/2 indicated. Regions corresponding to the 46/1 haplotype and the PDL-1/2 loci are boxed. Black horizontal arrows below the University of California Santa Cruz browser view indicate the position of all genes in the region, and the names of those referred to in this study are included. On the right of each spider plot, normalized read counts (or contacts) are shown for both PD-L1 (red) and PD-L2 (blue). As indicated, the left column for each locus represents 46/1 homozygous (Hom) and right column 46/1 nullizygous (Null). As an example, there is a marked reduction in normalized read counts between VP2 and PD-L1 on comparison of 46/1 homozygotes with 46/1 nullizygotes. There is also a reduction in the interaction between VP2 and PD-L2, but this is much more modest.

Figure 3.

JAK2 haplotype regulates PD-L1 expression as shown by CRISPR/Cas9 deletion. (A) Genomic locus of JAK2, indicating the region of intron 2 that was deleted to test its regulatory activity (hg19 chr9:5006961-5018796) based on the location of strongly linked 46/1 SNPs (see the supplemental Methods; supplemental Table 3). (B) Expression of JAK2 (left), PD-L1 (middle), and RIC1 (right) in control (green), heterozygous (red), and homozygous (blue) intron 2 element (E_i2) deleted K562 clones. (C) Expression of JAK2 (left), PD-L1 (middle), and RIC1 (right) in control (green), heterozygous (red), and homozygous (blue) intron 2 element (E_i2) deleted K562 clones treated overnight with 10 ng/mL of interferon gamma. In panels B-C, mRNA levels were normalized using GAPDH as an endogenous control, and data were analyzed by 1-way analysis of variance followed by Tukey multiple comparison test; ∗P < .05; ∗∗P < .01; ∗∗∗P < .005; ∗∗∗∗P < .001. Het, hetrorozygous; Hom, homozygous; IFN-γ, interferon gamma; mRNA, messenger RNA.

Figure 3.

JAK2 haplotype regulates PD-L1 expression as shown by CRISPR/Cas9 deletion. (A) Genomic locus of JAK2, indicating the region of intron 2 that was deleted to test its regulatory activity (hg19 chr9:5006961-5018796) based on the location of strongly linked 46/1 SNPs (see the supplemental Methods; supplemental Table 3). (B) Expression of JAK2 (left), PD-L1 (middle), and RIC1 (right) in control (green), heterozygous (red), and homozygous (blue) intron 2 element (E_i2) deleted K562 clones. (C) Expression of JAK2 (left), PD-L1 (middle), and RIC1 (right) in control (green), heterozygous (red), and homozygous (blue) intron 2 element (E_i2) deleted K562 clones treated overnight with 10 ng/mL of interferon gamma. In panels B-C, mRNA levels were normalized using GAPDH as an endogenous control, and data were analyzed by 1-way analysis of variance followed by Tukey multiple comparison test; ∗P < .05; ∗∗P < .01; ∗∗∗P < .005; ∗∗∗∗P < .001. Het, hetrorozygous; Hom, homozygous; IFN-γ, interferon gamma; mRNA, messenger RNA.