Here, we studied the expression patterns of genes in Treg and Th cells from a large number of individuals from the BXD recombinant inbred strain collection. The ultimate goal of these studies was the identification of expression quantitative traits (eQTLs) and of co-regulated genes that will give new insights into gene-gene interactions and regulatory pathways. In addition, the large data set of expression profiles from 33 mouse lines allowed us to validate known and to newly identify so far unknown genes that are expressed specifically in Treg but not Th cells.
Until now, several studies have aimed to identify genes that are specific for Treg cells (additional file 3
, Table S3). Using our data set from 33 mouse stains allowed us to search for differences in expression levels in Treg versus Th cells with a so far unprecedented statistical power. Our analysis confirmed that many genes with a described Treg function are also differentially expressed. Of the 23 genes that were described in the literature as genes with an important function in Treg cells, 17 were also found to be Treg specific in our analysis. Of the remaining six genes, Galectin-10 does not have a homologue in humans. S1pr1a
were not represented on the array. Also a Treg function for LRRC32/GARP was so far only described in humans. The remaining three genes, Gzmb, Lag3
, were expressed at low levels in both Treg and Th but with no significant difference between the two cell types.
The genes Foxp3, Nrp1, Clta4, Tnfrsf18
have been reported to be hallmark genes for Treg cells [31
]. Indeed, all of them appear as DE-2fold genes specifically higher expressed in Treg cells. Since we discovered Nrp1
as surface marker of murine regulatory T cells [3
] comprehensive data have been obtained regarding its implications in Treg function [33
]. Whereas expression of other Treg-markers such as Cd25
are induced upon activation of Th cells, activated murine Th cells lack Nrp1
expression, suggesting fundamental differences in the transcriptional regulation of this molecule in Tregs when compared with other well described Treg markers.
In total, 608 probesets were found in our analysis as Treg-specific and with an expression difference of at least two-fold. Although a difference of two-fold in expression levels is often used in gene expression array studies in many studies, it may be considered as arbitrary. Our study allowed to find even more genes with statistical significant differences in expression levels and extended the known list to 14,117 probesets that are specific for differentiated Treg versus Th cells. Zheng et al. [38
] performed chromatin immunoprecipitation and genome tiling array profiling to identify direct targets of the Foxp3
transcription factor. About 270 genes were identified were also up-regulated in Foxp3+
T cells. Of these, 162 genes were also found as differentially expressed genes in our analysis.
Recently, a genome wide association study in a large cohort of patients suffering from alopecia areata, which is among the most highly prevalent human autoimmune diseases, leading to disfiguring hair loss due to the collapse of immune privilege of the hair follicle and subsequent autoimmune attack revealed an association with genomic regions containing several genes controlling the activation and proliferation of Tregs including Ctla4, CD25
, and Eos/Ikzf4
In conclusion, almost all genes described previously as Treg functional genes or in other screening assays were also found in our analysis. These findings confirm that our study is well suitable to identify genes that are important for Treg function, differentiation and/or maintenance of their differentiation status.
Cis-eQTL genes most likely carry a mutation in the promoter region or contain a structural variation in the transcribed region which results in low expression levels or unstable RNA of one but not the other allele. In our study, several cis-eQTL genes were found that were active both in Treg and Th cells and have a known function in the immune response. For example, Stx11
(syntaxin 11) has been associated with familial hemophagocytic lymphohistiocytosis 4 in humans [40
BXD strains that carry a very low expressing allele inherited from one of the parental strains are similar to genetic knock-down experiments or hypomorphic mutants [42
]. The function of such genes can be studied in a reverse genetics approach by comparing the mean phenotypes of BXD strains that have inherited either the low or high allele--a method referred to as reverse complex trait analysis. Prominent examples of genes that are nature "knock-downs" in the BXD strains, include Hc, Ahr, Gpnmb, Tyrp1, Sae1, Apoa2
, and several CLEC and KLRA gene family members [23
341 probesets represented cis-eQTLs shared by several tissues, in Treg and Th cells but also in brain, kidney and lung.
In Treg cells, 25 probeset were trans-regulated and represented genes that were differently expressed between Treg and Th cells and that showed a two-fold or higher level of expression (DE-2fold genes) in Treg cells. Several of them represented genes with a known immune-related function. For example, Abcb1a
(ATP-binding cassette, sub-family B (MDR/TAP), member 1A) is involved in the down-modulation of dendritic cell functions through the regulation of pro-inflammatory cytokine secretion [43
(GATA binding protein 1) has been shown to be an important regulator of mast cell differentiation [44
(mitogen-activated protein kinase binding protein 1) encodes a protein that enhances NF-kappaB activation induced by MAP kinase kinase kinase 7 and TNF Receptor-Associated Factor 2 [45
(macrophage receptor with collagenous structure) exhibits multiple functions in the innate immune response. MARCO, TLR2, and CD14 are required for macrophage cytokine responses to mycobacterial trehalose dimycolate and Mycobacterium tuberculosis [46
]. MARCO-deficient mice exhibit lower IL-12 production in responses to stimulation [47
]. A defect in Marco
results in an impaired clearance of apoptotic cells and a generalized defect in both endocytosis and phagocytosis [48
]. Expression of MARCO is required for TLR signaling [46
]. But Marco also exhibits suppressive functions by decreasing inflammation in lungs after exposure to ozone [49
(nuclear receptor subfamily 4, group A, member 2) represents a transcriptional mediator of inflammatory signals [50
]. Also, it plays an important role in modulating IL-8 expression [52
Two genes, Laptm4b
exhibited strong trans-eQTL signals in both cell types but at different chromosomal locations. Thus, the same genes are expressed in both cell types, but are likely regulated by different mechanisms. Laptm4b
(lysosomal-associated protein transmembrane 4B) is involved in cell proliferation and multidrug resistance [53
], whereas no biological function has yet been described for Lycat
(lysocardiolipin acyltransferase 1).
None of these genes represents a known Treg key gene. Our analysis thus allows expanding the list of genes with potentially important functions in Treg cells. We did not perform an extensive analysis for trans-eQTLs in Th cells but a similar result can be expected for Th cells.
We then analyzed several trans-eQTLs for potential regulator genes in order to propose possible novel gene-regulatory circuits in Treg and Th cells. These are by far not comprehensive but rather serve as examples to illustrate how our data set will allow searching for possible regulatory networks.
A QTL region was found on chromosome 4 that regulates the expression of the F2rl1
(coagulation factor II (thrombin) receptor-like 1) gene. F2rl1/Par2
gene expression has been associated with the activation and suppression of inflammatory responses. Overexpression of F2rl1
in allergic inflammation of the airway exacerbates eosinophil infiltration into the lumen and hyperreactivity of the airway, while F2rl1
deletion diminishes inflammatory cell infiltration and reduces hyperreactivity [54
]. Also, F2rl1
plays a protective role during influenza virus type A infection through IFN-gamma production and decreased excessive recruitment of inflammatory cells to lung alveoli [55
] and deletion of F2rl1
is associated with decreased clearance of P. aeruginosa
]. Several candidate genes in the chromosome 4 QTL interval, which regulates F2rl1
expression, exhibited a cis-eQTL and thus represent potential regulators of F2rl1
(Table S11). These include the Map3k7/Tak1
(mitogen-activated protein kinase kinase kinase 7) protein kinase gene which exhibits many immune modulator functions. It mediates activation of IKK (inhibitor of kappaB kinase) and silencing of Map3k7 suppressed T cell receptor-dependent IKK activation and interleukin-2 production in T cells [57
]. The Zfp292
(zinc finger protein 292) gene expression has been shown tocorrelate with growth hormone expression and may thus be involved in mediating proliferation signals [59
]. The analysis of candidate genes in the other QTL regions is discussed in the supplement data.
Our analysis of the chromosome 2 regions autoimmune trait [30
] identified several loci that exhibited a cis-eQTL in Treg cells and may thus be involved in regulating this trait: Slpi, Sys1/2610042O14Rik
, and Znf335
(secretory leukocyte peptidase inhibitor) exhibits many biological functions, including anti-bacterial, anti-fungal, anti-viral, anti-inflammatory, wound healing and immuno-modulatory activities [60
]. The Slpi protein represents a ligand for PLSCR1 and PLSCR4, which interact directly with the CD4 receptor at the cell surface of T lymphocytes [69
is a prominent innate immune protein of the respiratory tract with serine protease inhibitor activity [70
]. It attenuates excessive inflammatory responses resulting in a balanced innate immunity [71
]. Constitutive expression of Slpi
reduced the inflammatory response and improved lung function in an acute model of allergic asthma in Slpi
transgenic and knockout mice [73
]. The attenuation of the inflammatory response by Slpi
is mediated through macrophages that secrete an increased amount of Slpi when encountering apoptotic cells [74
(zinc finger protein 335, also NIF1
in human) is a cotransducer that regulates the activity of the nuclear hormone receptor coactivator NRC [76
] which may mediate the function of the CCR4 signal [77
]. No biological functions were identified so far for Sys1
(SYS1 Golgi-localized integral membrane protein homolog (S. cerevisiae
)). The discussion of candidate genes in the chromosome 4 auto-immune QTL interval can be found in the supplemental data.