HLA Class Ib and MICA/MICB Expression in Human Tissues and Cell Types: Reshuffling Immune Players

Abstract

Abnormal expression of HLA class Ib, MICA and MICB molecules is associated with the evolution of pathological conditions and clinical settings. Here, we use RNA-sequencing data from two publicly-available projects, from different human organs and tissues and at single-cell level, to present their transcriptional expression throughout the human body, in comparison to that of HLA class Ia, HLA class II, their costimulatory molecules, and the main HLA transcription factors. Our analyses for 21 target genes reveal that median gene expression differs by orders of magnitude and that the classical/non-classical HLA distinction is not absolute for overall expression. Sixteen of the 21 target genes show correlated expressions, although careful analyses of individual expression patterns in tissues and organs highlight specificities. Tissue and organ expression patterns reveal that the lymphoid organs, lungs, and gastrointestinal tract organs display the highest expression of the HLA and HLA-related genes. At single-cell level, adipocytes, endothelial cells, and immune cells all have unexpectedly close expression patterns. The expression pattern of the 21 target genes in non-immune organs, such as the lung or colon, and in non-immune cells like adipocytes, questions the role of these organs and cell types in immune homeostasis and suggests additional, non-immune functions of these molecules. The lack of impact of the HLA transcription factors studied here on HLA regulation in non-immune tissues also supports a role for additional HLA transcription factors in these tissues. Finally, classical/non-classical HLA classification based on molecule structure and genetic polymorphism does not seem to extend to their expression.

Keywords: HLA; MICA/B; RNAseq; physiological expression; single‐cell.

FIGURE 1

Expression levels for the 21 HLA and HLA‐related genes investigated fall into five groups. mRNA expression for the 21 target genes in 95 individuals is quantified as reads per kilobase per million reads placed (RPKM) and presented as box and whiskers (min to max) with a line at the median. A Log₁₀ scale is used. Genes are sorted from left to right based on their magnitude of expression (median). Five statistical groups were defined through pairwise comparisons: for each group there is no statistical difference in expression between the gene with the highest expression and the gene with the lowest expression (p‐values are given in Table S4). Each group has a unique colour and the group number is given under the gene names. The level of confidence for the statistical differences between the highest values of each group is given with asterisks (Kruskal–Wallis test; one asterisk, a = 0.05, two asterisks, a = 0.01 and three asterisks a = 0.001). INFg was included in group V, even though its expression is significantly lower than the other genes of the group.

FIGURE 2

Expression patterns for the 21 HLA and HLA‐related genes investigated are significantly correlated for two blocks of genes representing a total of 16 genes. Correlation matrix heatmap between mRNA expression of 21 genes in 95 individuals. Genes are organised according to their correlation values. Asterisks correspond to adjusted p‐values < 0.01 (given in Table S5C). The first block includes HLA‐B to CD86 (11 genes) and the second HLA‐A, ‐B, ‐C, ‐E, ‐F, ‐H, DRB1, ‐DQB1 and ‐DPB1 (nine genes).

FIGURE 3

Principal component analysis (PCA) for the expression of the 21 target genes in 95 tissues separates lymphoid organs, lungs, and gastro‐intestinal tract organs from the other tissues. Principal components 1 and 2 represent, respectively, 49.15% and 10.68% of variance. Tissues are plotted by colour; for the gastro‐intestinal tract, letters identify the different tissues (A). PCA biplot illustrates the loadings of each variable (B).