The zinc finger protein MAZR is part of the transcription factor network controlling CD4/CD8 cell fate decision of DP thymocytes

Abstract

CD4/CD8 lineage specification of thymocytes is linked with coreceptor expression. Previously, the transcription factor MAZR was identified as an important regulator of Cd8 gene expression. Here we show that variegated CD8 expression by loss of Cd8 enhancers is reverted in MAZR-deficient mice, confirming that MAZR negatively regulates the Cd8 loci during the transition to the double-positive (DP) stage. Moreover, loss of MAZR led to a partial redirection of MHC class I-restricted thymocytes into CD4⁺ helper-like T cells correlating with derepression of ThPOK, a central transcription factor for helper-lineage development. MAZR bound the ThPOK silencer, indicating direct regulation of the ThPOK locus by MAZR. Thus, MAZR is part of the transcription factor network regulating CD8 lineage differentiation of DP thymocytes.

INTRODUCTION

CD4 and CD8 coreceptor expression correlates with the functional phenotype of T cells. Understanding the transcriptional regulation of Cd4 and the adjacent genes Cd8a and Cd8b1 (Cd8) might therefore provide insight into the cell fate decision of DP thymocytes^1-3.

CD8 coreceptor expression is tightly regulated during thymocyte development by the activity of at least five different cis-regulatory elements^{4, 5}. Combined deletion of certain Cd8 enhancers leads to variegated expression of CD8 in DP thymocytes^6-8 and CD8 variegation correlates with an epigenetic "off" state⁹. These studies indicate a complex regulatory network of developmental stage- and subset-specific cis-regulatory elements and link Cd8 enhancer function with chromatin remodeling of the Cd8ab gene complex¹⁰. The transcriptional regulator MAZR appears to be one of the factors that are involved in keeping the local chromatin at the Cd8 gene loci in a transcriptional “off” state at the double-negative (DN) stage⁹. MAZR (also known as PATZ1 or Zfp278), encoded by the Patz1 gene (referred to here as Mazr), belongs to the family of BTB (bric-a-brac, tramtrack, broad complex; also known as POZ domain: poxvirus zinc finger) domain-containing zinc finger (ZF) factors^{11, 12}. Mazr is expressed in the brain, thymocytes, and B cells¹³. A recent study also reported that MAZR is essential for spermatogenesis¹⁴. In DN thymocytes MAZR interacts with several Cd8 cis-regulatory elements. MAZR expression is down-regulated during the transition from the DN to double-positive (DP) stage, and forced expression of MAZR results in variegated expression of CD8 in a proportion of DP thymocytes. Mechanistically, the transcriptional “off” state is linked to the nuclear corepressor NCoR, which interacts with the BTB domain of MAZR. Based on these observations MAZR has been proposed as an important negative regulator of Cd8ab gene complex activation during the DN to DP transition of thymocyte development⁹.

To further investigate the physiological function of MAZR during T cell development, Mazr^−/− mice were generated. While deletion of MAZR is not sufficient to activate CD8 expression in DN thymocytes, it partially reverted variegated expression of Cd8 genes in E8_I,E8_II enhancer doubly-deficient DP thymocytes. Furthermore, Mazr^−/− mice displayed elevated CD4 to CD8 ratios in mature thymocytes and in peripheral T cells. By using either MHC class I-restricted TCR transgenic mice or Mazr^−/− bone marrow (BM) chimeric MHC class II-deficient mice, we demonstrated that the increase in CD4 lineage cells in Mazr^−/− mice was in part due to redirected differentiation of MHC class I-restricted CD8 SP thymocytes into the CD4 helper lineage. In addition, we could show derepression of ThPOK (encoded by the Zbtb7 gene, referred to here as ThPOK), an essential molecule for helper lineage differentiation^{15, 16}, in MHC class I-signaled DP and CD4⁺CD8^lo Mazr^−/− thymocytes. MAZR bound the ThPOK locus, suggesting a direct control of ThPOK expression. Thus, our data provide genetic evidence that MAZR is part of the transcription factor network that controls CD4/CD8 cell fate decision of DP thymocytes by regulating ThPOK expression.

RESULTS

Altered CD4 to CD8 T cell ratios in Mazr^−/− mice

Mazr-deficient (Mazr^−/−) mice were generated by gene-targeting approaches (see Supplementary Fig. 1). Mazr^−/− mice were born at a reduced Mendelian frequency (Supplementary Fig. 2a) and were smaller in size compared to Mazr^+/+ littermates (Supplementary Fig. 2b,c). Since MAZR is expressed at high amounts in DN and DP thymocytes⁹, a comprehensive analysis of the T cell lineage was performed. The smaller size of Mazr^−/− mice correlated with an approximately 40% and 50% reduction of thymocyte and splenocyte numbers, respectively (Fig. 1a). CD4, CD8α and CD8β protein expression levels were normal in Mazr^−/− mice (Fig. 1b,c, and data not shown). However, the CD4SP to CD8SP thymocyte ratio was increased in Mazr^−/− compared to Mazr^+/+ mice, and was most notable in the CD3^hi population (Fig. 1b,d). In contrast, the CD4/CD8 profile of CD3^lo cells was similar between Mazr^+/+ and Mazr^−/− mice (Fig. 1b). The increased CD4SP to CD8SP ratio led also to an increased CD4⁺ to CD8⁺ T cell ratio in peripheral lymphoid organs of Mazr^−/− mice (Fig. 1c,d). However, the expression of several cell surface markers such as CD62L, CD44, CD25 and CD122 on peripheral CD4⁺ and CD8⁺ T cells was normal, with the exception of a small increase in the CD8⁺CD44^hi T cell subset in the absence of MAZR (Supplementary Fig. 3a and data not shown). Moreover, CD25⁺FoxP3⁺ regulatory T cells, CD1d-restricted invariant NKT cells, TCRγδ⁺ T cells and CD8αα positive TCRγδ⁺ intraepithelial lymphocytes (IEL) were present at similar frequencies in Mazr^+/+ and Mazr^−/− mice (Supplementary Fig. 3b-e). Mazr^−/− mice showed normal B cell development, and all major myeloid subsets were present under homeostatic conditions (data not shown).

Figure 1.

Altered CD4 to CD8 T cell ratio in the absence of MAZR. (a) Diagrams show thymocyte and splenocyte numbers of 6-8 weeks old Mazr^+/+, Mazr^+/− and Mazr^−/− female mice. Each symbol represents one mouse, and horizontal lines indicate average values. (b) Flow cytometric analysis of CD4 and CD8α expression on thymocytes isolated from Mazr^+/+ and Mazr^−/− littermates. Cells were gated on total thymocytes (upper panel), or on CD3^hi (middle panel) and on CD3^lo (bottom panel) subsets. Numbers indicate the percentage of cells in the respective quadrant. Data are representative of more than 10 independent experiments. (c) CD4 and CD8α expression pattern on lymph node cells isolated from Mazr^+/+ and Mazr^−/− littermates. Cells were gated on the CD3⁺ population. Numbers in the dot plots indicate the percentage of cells in the respective quadrant. Data are representative of more than 10 independent experiments. (d) Diagrams show CD4 to CD8 ratio of SP thymocytes (left panel) and of CD3⁺ lymph node T cells (right panel) from Mazr^+/+, Mazr^+/− and Mazr^−/− mice. Each symbol represents one mouse, and horizontal lines indicate average values.

MAZR affects CD8 expression

MAZR is proposed to function as a negative regulator of the activation of the Cd8ab gene complex during the DN to DP transition of thymocyte development⁹. DN thymocytes were still present in Mazr^−/− mice (Fig. 1b, lower panel), and the distribution of the DN1-4 subsets was similar between Mazr^+/+ and Mazr^−/− mice (Supplementary Fig. 3f,g). This indicates that early stages of thymocyte development are not affected by the loss of MAZR, and that the deletion of MAZR is not sufficient to activate Cd8a and Cd8b gene expression in DN thymocytes.

Combined deletion of the Cd8 enhancers E8_I and E8_II leads to CD8 variegation in DP thymocytes⁶, which correlates with epigenetic differences at the Cd8ab gene complex⁹. We hypothesized that in the absence of E8_I and E8_II, fewer positive factors are recruited to the Cd8 loci in DP thymocytes, thus changing the balance between a closed and an open state of chromatin. Since MAZR is one of the negative factors that controls the Cd8ab gene complex, one would predict reversion of CD8 variegation in E8_I,E8_II doubly-deficient mice in the absence of MAZR. To test our hypothesis, Mazr^−/− mice were intercrossed with E8_I,E8_II doubly-deficient mice. The numbers of viable Mazr^−/− offspring from heterozygous intercrosses dropped dramatically when mice were backcrossed to a C57BL/6 background (S.S. and W.E., unpublished observation). Therefore, E13.5 Mazr^+/+ and Mazr^−/− fetal liver (FL) cells (CD45.2⁺) on an E8_I,E8_II doubly-deficient background were transferred into irradiated CD45.1⁺ congenic mice (CD45.1⁺). As previously reported⁹, there was a strong CD8 variegation in Mazr^+/+,E8_I,E8 thymocytes. However, CD8 variegation was dramatically reduced in Mazr^−/− II ,E8_I,E8_II thymocytes (Fig. 2a, upper panel), which was even more evident within the CD3^lo subset that represents primarily DP thymocytes (Fig. 2a, lower panel and Fig. 2b). These findings provide genetic evidence that MAZR is an essential negative regulator of CD8 expression during DN to DP transition, as proposed in our previous studies.

Figure 2.

Reduced variegation of CD8 in E8_I,E8_II double-deficient mice in the absence of MAZR (a) CD4 and CD8α expression on CD45.2⁺ thymocytes isolated from Mazr^+/+,E8_I,E8_II and Mazr^−/−,E8 ,E8 + I II fetal liver (FL)-transplanted CD45.1 recipient mice (middle and right panel, respectively). Left panel shows thymic CD4 and CD8α expression profile of C57BL/6 mice. Cells were gated on total CD45.2⁺ thymocytes (upper panels), or on CD45.2⁺ CD3^lo thymocytes (lower panels). Numbers indicate the percentage of cells in the respective quadrant. Data shown are representative of five mice each. (b) Diagram showing the percentage of CD4⁺CD8⁻ (i.e. “CD8⁻”) DP thymocytes within the CD45.2⁺CD3^lo population in Mazr^+/+,E8_I,E8_II and Mazr^−/−,E8_I,E8_II FL-transplanted mice. Each dot represents one mouse, and horizontal lines indicate average values. The gating areas for CD4⁺CD8⁻ cells is shown in a (lower panel).

T cell-intrinsic defects in Mazr^−/− mice

The reduced thymocyte and T cell numbers could be the consequence of the smaller size of Mazr^−/− mice, or due to a T cell-intrinsic defect, or both. Furthermore, the thymic microenvironment could be altered in Mazr^−/− mice, and this could lead to altered CD4/CD8 ratios. To distinguish between these possibilities, competitive bone marrow (BM) reconstitution experiments were performed. For this, BM cells from wild-type CD45.1⁺ congenic mice were mixed in a 1:1 ratio with either Mazr^+/+ (CD45.2⁺) or Mazr^−/− (CD45.2⁺) BM cells and transplanted into irradiated CD45.1⁺ recipients. Mazr^−/− BM reconstituted the T cell lineage as well as the B cell lineage as efficiently as Mazr^+/+ BM in a competitive environment (Fig. 3a, upper panel and Fig. 3b). This indicates that reduced T cell numbers in Mazr^−/− mice is the consequence of the smaller size of Mazr^−/− mice. In contrast, while the CD4 and CD8 expression profile on thymocytes and peripheral T cells derived from Mazr^+/+ (CD45.2⁺) BM cells was normal, the CD4 to CD8 ratio was increased in the Mazr^−/− BM-derived (CD45.2⁺) T cell compartment (Fig. 3a, lower panel and Fig. 3c). This reveals that the altered CD4/CD8 ratio in Mazr^−/− mice is a T cell-intrinsic defect.

Figure 3.

T cell-intrinsic defects lead to an altered CD4 to CD8 ratio in Mazr^−/− mice (a) Mazr^+/+ (left panel) or Mazr^−/− (right panel) BM (CD45.2⁺) was mixed in a 1:1 ratio with wild-type BM (CD45.1⁺) and injected into irradiated CD45.1⁺ recipients. Histograms show CD45.2 expression on CD3⁺ lymph node cells in BM chimeric mice. Numbers indicate the percentage of CD45.2⁻ and CD45.2⁺ cells. Dot plots indicate CD4 and CD8α expression on CD45.2⁻ and CD45.2⁺ lymph node T cells. Data are representative of 9 mice each. (b) Panels show the percentage of CD45.2⁺ thymocytes (left panel), T cells (CD3⁺ splenocytes; middle panel), and B cells (B220⁺ splenocytes; right panel) in BM chimeric mice. (c) Diagrams indicate CD4 to CD8 ratio in CD3^hi thymocytes (left panel) and CD3⁺ splenocytes (right panel) in the CD45.1⁺ wild-type and CD45.2⁺ Mazr^+/+ or CD45.2⁺ Mazr^−/− subsets of BM chimeric mice. In b and c, each symbol represents one mouse, and horizontal lines indicate average value.

Partial redirection of MHC class I-restricted thymocytes

To investigate the reason for an increased CD4 to CD8 T cell ratio, Mazr^−/− mice were either crossed with OT-I or OT-II TCR transgenic mice, which express TCRs (V_β5, V_α2) restricted to MHC class I or class II, respectively^{17, 18}. The thymic CD4/CD8 profile in Mazr^−/−,OT-II mice was similar to the one in Mazr^+/+,OT-II cells (Fig. 4a and Supplementary Fig. 4a). In the periphery, there was a mild increase in the percentage of CD4⁺ T cells and a corresponding decrease in CD8⁺ T cells in Mazr^−/−,OT-II mice, however, the percentage of transgenic V_α2-expressing T cells was similar (Fig. 4a,b and Supplementary Fig. 4b). This is in contrast to Mazr^−/−,OT-I TCR transgenic mice, which showed dramatic changes in the CD4/CD8 expression profile. While Mazr^+/+,OT-I thymocytes were predominantly selected into the cytotoxic T cell lineage, Mazr^−/−,OT-I thymocytes displayed a marked increase in the percentage of CD4SP cells and a corresponding reduction of the CD8SP population (Fig. 4c, upper and middle panels, and Supplementary Fig. 4c). In the periphery of Mazr^+/+,OT-I mice, only few CD4⁺ T cells were present (Fig. 4c, lower left panel). In contrast, a substantial number of CD4⁺ T cells appeared in Mazr^−/−,OT-I mice (Fig. 4c, lower right panel and Supplementary Fig. 4d). Moreover, these CD4⁺ T cells predominantly expressed the V_α2 TCR chain, suggesting that these T cells were selected on the basis of MHC class I-restricted OT-I TCR (Fig. 4d,e).

Figure 4.

T cell development in MHC class I- and class II-restricted TCR transgenic mice in the absence of MAZR (a) CD4 and CD8 expression pattern on thymocytes and lymph node T cells isolated from Mazr^+/+,OT-II and Mazr^−/−,OT-II littermates. Numbers indicate the percentage of cells in the respective quadrants. Data are representative of 3 independent experiments. (b) Histograms show V_α2 expression on CD4⁺ and CD8⁺ lymph node T cells isolated from Mazr^+/+,OT-II and Mazr^−/−,OT-II littermates. Numbers show the percentage of V_α2^hi cells within the indicated region. Gating areas for CD4⁺ and CD8⁺ T cell populations are shown in the lower panel of a. Data are representative of 3 independent experiments. (c) CD4 and CD8α expression on total (upper panel) and CD3^hi (middle panel) thymocytes, and on lymph node T cells (lower panel) isolated from Mazr^+/+,OT-I and Mazr^−/−,OT-I littermates. Numbers indicate the percentage of cells in the respective quadrant. Data are representative of 7 independent experiments. (d) Histograms show V_α2 expression on CD4⁺ and CD8⁺ lymph node T cells isolated from Mazr^+/+,OT-I and Mazr^−/−,OT-I littermates. Numbers show the percentage of V_α2^hi cells within the indicated region. Gating areas for CD4⁺ and CD8⁺ T cell populations are shown in the lower panel of c. Data shown are representative of 7 independent mice. (e) Panels showing the percentage of V_α2^hi cells in CD4⁺ lymph node T cells isolated from Mazr^+/+,OT-I and Mazr^−/−,OT-I mice. Each symbol represents one mouse, and horizontal lines indicate average values.

To exclude the possibility that Mazr^−/−,OT-I cells were selected on MHC class II-restricted TCRs consisting of an endogenous V_α2 chain, Mazr^−/−,OT-I mice were crossed onto a Rag2^−/− background. E13.5 FL cells (CD45.2⁺) from either Mazr^+/+Rag2^−/−,OT-I or Mazr^−/− Rag2^−/−,OT-I embryos were transferred into irradiated wild-type CD45.1⁺ congenic mice. Similar to Mazr^−/−,OT-I mice, a significant population of peripheral V_α2⁺CD4⁺ T cells developed in Mazr^−/−Rag2^−/−,OT-I chimeric mice (Fig. 5a). This resulted in a marked increase in the absolute number of CD4⁺ T cells in Mazr^−/−Rag2^−/−,OT-I chimeric mice compared Mazr^+/+Rag2^−/−,OT-I chimeric mice (Fig. 5b). Moreover, the MHC class I-restricted CD4⁺ T cells in Mazr^−/− Rag2^−/−,OT-I chimeric mice displayed characteristic features of helper T cells, such as expression of ThPOK and upregulation of CD154 (CD40L) upon activation, but produced low amounts of IFN-γ and did not express Prf1 (encoding perforin) (Supplementary Fig.5a,b,c). Of note, most of the Mazr^−/−Rag2^−/−,OT-I CD4⁺ T cells were CD44^loCD62L^hi, suggesting that they are not innate-like T cells (such as iNKT cells) (Supplementary Fig. 5d).

Figure 5.

Redirected differentiation of MHC class I-restricted thymocytes into helper lineage cells in the absence of MAZR. (a) CD4 and CD8α expression on lymph node T cells isolated from Mazr^+/+Rag2^−/−,OT-I and Mazr^−/−Rag2^−/−,OT-I fetal liver (FL)-transplanted mice. Cells were gated on the CD45.2⁺V_α2⁺ population. Numbers indicate the percentage of cells in the respective quadrants. Data are representative of 7 independent experiments. (b) Panels show CD45.2⁺V_α2⁺CD4⁺ and CD45.2⁺V_α2⁺CD8⁺ T cell numbers isolated from the spleens of Mazr^+/+Rag2^−/−,OT-I and Mazr^−/−Rag2^−/−,OT-I FL-transplanted mice. Each symbol represents one mouse, and horizontal lines indicate average values. (c) T cell-depleted BM cells from Mazr^+/+ and Mazr^−/− mice (CD45.2⁺) were transferred into MHC class II-deficient (H2-Ab1^−/−, CD45.1⁺) recipient mice. Dot plots shows CD4 and CD8α expression pattern of CD45.2⁺ CD3⁺ T cells in the spleen of Mazr^+/+ (left) and Mazr^−/− (right) BM chimeric MHC class II-deficient mice. To exclude invariant NKT cells, CD1d:PBS57 tetramer-positive cells were excluded from the analysis. Data are representative of 10 independent experiments. (d) Diagrams show CD45.2⁺CD3⁺CD4⁺ and CD45.2⁺CD3⁺CD8⁺ T cell numbers isolated from the spleens of Mazr^+/+ and Mazr^−/− BM chimeric MHC class II-deficient (H2-Ab1^−/−) mice. Each symbol represents one mouse, and horizontal lines indicate average values.

Next, we investigated whether MHC class I-restricted MAZR-deficient CD4⁺ T cells also develop in a non-TCR transgenic setting. Thus, Mazr^+/+ and Mazr^−/− BM cells (CD45.2⁺) were transplanted into MHC class II-deficient recipients (CD45.1⁺) and the number of peripheral T cells and thymocytes upon reconstitution was determined. There was a significant increase in the percentage and absolute number of CD3⁺CD4⁺ T cells (CD45.2⁺) and a corresponding decrease of CD3⁺CD8⁺ T cells in the spleen and thymus of Mazr^−/− chimeric mice compared to Mazr^+/+ chimeras (Fig. 5c,d, and Supplementary Fig.6a,b). To exclude innate-like T cells such as invariant NKT cells, only CD1d tetramer-negative cells were analyzed. A similar increase in CD4 lineage cells was also detected on TCRαβ⁺-gated T cells (to exclude γδT cells, data not shown). We noticed that the majority of Mazr^+/+ and Mazr^−/− CD4⁺ T cells were CD44^hi, most likely due to altered homeostasis in an MHC class II-deficient environment. However, there was no difference in the percentage of IFN-γ-producing CD44^hi T cells upon PMA/ionomycin stimulation between Mazr^+/+ and Mazr^−/− MHC class II-deficient chimeras, indicating that the increase in Mazr^−/− CD4⁺ T cells is not due to an increase of innate-like T cells (Supplementary Fig. 6c).

Together, the data observed from either MHC class I-restricted TCR transgenic mice or BM chimeric MHC class II-deficient mice indicate that a fraction of Mazr^−/− MHC class I-restricted thymocytes redirects into the helper T cell lineage.

Derepression of ThPOK in MAZR-deficient thymocytes

Molecules such as ThPOK, Runx1, Runx3, GATA-3 and Tox are implicated in CD4/CD8 lineage commitment of DP thymocytes^{15, 16, 19-24}. To investigate whether loss of MAZR induces alterations in the expression of these factors, qRT-PCR experiments on sorted CD69⁻V_α2^loDP (DP CD69⁻; i.e. pre-selected DP cells), CD69⁺V_α2^int-hi DP (DP CD69⁺; i.e. signaled DP cells) and CD69⁺V_α2^int-hiCD4⁺CD8^lo (CD4⁺CD8^lo CD69⁺) thymocyte subsets from Mazr^+/+Rag2^−/−,OT-I and Mazr^−/−Rag2^−/−,OT-I FL-transplanted mice were performed. An increased expression of ThPOK gene in DP CD69⁻ and DP CD69⁺ subsets of Mazr^−/−Rag2^−/−,OT-I chimeric mice compared to Mazr^+/+Rag2^−/−,OT-I chimeric mice was observed. In contrast, Runx1, Runx3, Gata3, and Tox expression was similar in Mazr^−/−Rag2^−/−,OT-I and Mazr^+/+Rag2^−/−,OT-I DP (both CD69⁻ and CD69⁺) subsets, while Runx3 expression was reduced in CD4⁺CD8^loCD69⁺ subsets (Fig. 6a and Supplementary Fig. 7).

Figure 6.

Derepression of ThPOK in Mazr^−/−Rag2^−/−,OT-I thymocytes and CD8⁺ T cells (a) Quantitative real-time PCR showing ThPOK expression in DP CD69⁻ (DP 69⁻), DP CD69⁺ (DP 69⁺), and CD4⁺CD8^loCD69⁺ (4⁺8^lo) thymocytes isolated from Mazr^+/+Rag2^−/−,OT-I (black bars) and Mazr^−/−Rag2^−/−,OT-I (white bars) FL-chimeric mice. As a reference, the expression of ThPOK in wild-type CD4SP (4S) and CD8SP (8S) thymocytes is shown at the right (gray bars). Values are relative to Hprt1 expression. ThPOK expression in Mazr^+/+Rag2^−/−,OT-I DP CD69⁻ thymocytes is set as 1. Data are representative of 3 independent experiments (b) Histograms show GFP expression in DP CD69⁻ (DP 69⁻), DP CD69⁺ (DP 69⁺) and CD4⁺CD8^loCD69⁺ (4⁺8^lo69⁺) subsets, in CD4SP and CD8SP thymocytes, and in splenic CD4⁺ (4T) and CD8⁺ (8T) T cells. Gating areas are shown in Supplementary Fig. 8a. Upper and lower panels show thymocyte and splenocyte subsets analyzed from Mazr^+/+ThPOK^+/GFP,OT-I and from Mazr^−/−ThPOK^+/GFP,OT-I mice, respectively. Numbers show percentage of cells in the indicated region. Data are representative of 3 independent experiments.

To investigate in detail the expression pattern of ThPOK in MAZR-deficient thymocytes, Mazr^−/−,OT-I mice were crossed with a ThPOK GFP reporter mouse, in which a cDNA encoding GFP is inserted into the ThPOK gene locus²⁰. Thus, GFP expression is driven by the endogenous ThPOK regulatory elements. ThPOK expression is primarily induced by MHC class II signaling, and after positive selection a large fraction of CD4⁺CD8^loCD69⁺ thymocytes expresses ThPOK15, 16, 20, 21, 25. However, some ThPOK-expressing CD4⁺CD8^loCD69⁺ cells retain the potential to develop into the cytotoxic cell lineage^{21, 25}. During CD4 lineage differentiation, ThPOK mRNA expression is upregulated and all CD4SP thymocytes and CD4⁺ T cells express high levels of ThPOK, while only a small population of CD8SP thymocytes and peripheral CD8⁺ T cells expressed very low levels of ThPOK^{21, 25}. In Mazr^+/+ThPOK^+/GFP,OT-I mice, GFP⁺ cells were detected in DP CD69⁺ and in CD4⁺CD8^lo CD69⁺ populations (Fig. 6b and Supplementary Fig. 8a). The percentage of GFP⁺ cells increased almost to 100% within the CD4SP subset, whereas CD8SP thymocytes terminated GFP expression. Consistent with the qRT-PCR data, a substantial number of GFP⁺ cells appeared in DP CD69⁻ thymocytes of Mazr^−/−ThPOK^+/GFP,OT-I mice, indicating derepression of ThPOK (Fig. 6b and Supplementary Fig. 8a). Moreover, the percentage of GFP⁺ cells within the DP CD69⁺ and CD4⁺CD8^loCD69⁺ populations of Mazr^−/−ThPOK^+/GFP,OT-I thymocytes was increased compared to Mazr^+/+ThPOK^+/GFP,OT-I thymocytes (Fig. 6b). In CD8SP cells, endogenous ThPOK expression was already decreased compared to CD4SP cells (Supplementary Fig. 9a,b), even though a certain fraction of CD8SP remained GFP-high (Fig. 6b), most likely due to slow GFP protein degradation, Notably, peripheral CD8⁺ T cells in Mazr^−/−ThPOK^+/GFP,OT-I mice expressed GFP⁺, although at lower levels compared to CD4⁺ T cells (Fig. 6b), consistent with the observation that ThPOK expression was detected in CD8⁺ T cells of Mazr^−/−Rag2^−/− ,OT-I mice (Supplementary Fig.5c).

To determine whether derepression of ThPOK is also observed in a non-TCR transgenic setting, we analyzed GFP expression in Mazr^−/−ThPOK^+/GFP mice. GFP expression (or ThPOK mRNA expression) was not detected in Mazr^−/− DP CD69⁻ thymocytes (Fig.7a, and data not shown), suggesting that MAZR is not essential for ThPOK repression in pre-signaled DP cells. However, an increased percentage of GFP⁺ DP CD69⁺ and CD4⁺CD8^loCD69⁺ thymocyte subsets (i.e. CD69⁺TCRβ^int-hiDP and CD69⁺TCRβ^int-hiCD4⁺CD8^lo subsets) was observed in Mazr^−/−ThPOK^+/GFP mice compared to Mazr^+/+ThPOK^+/GFP mice, while CD8SP cells terminated GFP expression (Fig.7a and Supplementary Fig. 8b). Similar to Mazr^−/− ThPOK^+/GFP,OT-I mice, a fraction of peripheral CD8⁺ T cells in Mazr^−/−ThPOK^+/GFP mice re-expressed GFP (Fig. 7a), indicating that MAZR is required to maintain ThPOK repression in CD8⁺ T cells. Next, we determined whether TCR interactions with either MHC class I or MHC class II induced increased percentages of GFP⁺ cells. Therefore, BM cells from Mazr^+/+ThPOK^+/GFP and Mazr^−/−ThPOK^+/GFP mice were transferred either into irradiated MHC class I- or MHC class II-deficient mice. Similar amounts of GFP were expressed in DP CD69⁺ and CD4⁺CD8^loCD69⁺ thymocyte subsets derived from Mazr^+/+ThPOK^+/GFP and Mazr^−/− ThPOK^+/GFP BM in MHC class I-deficient recipients. In contrast, the percentage of GFP⁺ cells in those subsets in MHC class II-deficient recipients was much higher in the absence of MAZR (Fig. 7b and Supplementary Fig. 8c). This indicates derepression of ThPOK in Mazr^−/− thymocytes in MHC class I-signaled thymocytes.

Figure 7.

Derepressed ThPOK expression in MHC class I-signaled Mazr^−/− thymocytes (a) Histograms show GFP expression in DP CD69⁻ (DP 69⁻), DP CD69⁺ (DP 69⁺) and CD4⁺CD8^loCD69⁺ (4⁺8^lo69⁺) subsets, in CD4SP and CD8SP thymocytes, and in splenic CD4⁺ (4T) and CD8⁺ (8T) T cells. Gating areas are shown in Supplementary Fig. 8b. Upper and lower panels show thymocyte and splenocyte subsets analyzed from Mazr^+/+ThPOK^+/GFP and from Mazr^−/−ThPOK^+/GFP, respectively. Numbers show percentage of cells in the indicated regions. Data are representative of 5 independent experiments. (b) T cell-depleted BM cells from Mazr^+/+ThPOK^+/GFP and Mazr^−/−ThPOK^+/GFP mice were transferred into MHC class II-deficient (H2-Ab1^−/−) or MHC class I-deficient (B2m^−/−) recipient mice. Histograms show GFP expression in DP CD69⁺ (DP 69⁺) and CD4⁺CD8^loCD69⁺ (4⁺8^lo69⁺) thymocyte subsets of Mazr^+/+ThPOK^+/GFP (upper panels) and Mazr^−/−ThPOK^+/GFP (lower panels) BM chimeric MHC class II- (left) and MHC class I-deficient (right) mice. Gating areas are shown in Supplementary Fig. 8c. Numbers show percentage of cells in the indicated regions. Data are representative of 4 mice from 2 independent experiments. (c) Chromatin immunoprecipitation analysis of the ThPOK silencer. Mazr^+/+ and Mazr^−/− thymocytes chromatin was immunoprecipitated with anti-MAZR serum (Ab) or pre-serum (Pre), followed by PCR with primers specific for the ThPOK silencer region (Silencer) or for a non-MAZR binding region from the Cd8ab gene complex (Ctrl). Input DNA was PCR amplified undiluted or at a dilution of 1:5 or 1:25 (wedges) to ensure PCR quantification in a non-saturated amplification range. Data are representative of three independent experiments.

To provide mechanistic insight into how MAZR is involved in ThPOK repression, we tested whether MAZR could bind a newly identified ThPOK silencer, which is indispensable for the helper lineage-specific expression of ThPOK^{20, 26}. Chromatin immunoprecipitation assays (ChIP) revealed that MAZR binds to the ThPOK silencer region (Fig 7c). This result strongly suggests a direct regulation of the ThPOK gene by MAZR.

DISCUSSION

In this study we provide evidence that the transcriptional regulator MAZR has important functions at two distinct stages of T cell development. MAZR is a negative regulator of the Cd8ab gene complex during the DN to DP transition, and it represses ThPOK expression in MHC class I-signaled thymocytes during the CD4/CD8 cell fate decision.

First, we observed that the deletion of MAZR almost completely reversed CD8 variegation in Cd8 enhancer E8_I,E8_II doubly-deficient mice. Thus, our data provide strong genetic evidence that MAZR negatively regulates the activation of the Cd8 gene loci during the DN to DP transition. MAZR binds to Cd8 enhancers and interacts with a the nuclear corepressor NCoR complex in thymocytes⁹. The co-repressor NCoR is a component of multiple protein complexes that also contain histone deacetylases that repress target genes²⁷. Therefore, MAZR might repress Cd8 expression via recruitment of repressor complexes containing NCoR. Based on this model, one would predict reversion of CD8 variegation in E8_I,E8_II doubly-deficient mice in the absence of MAZR. This was indeed observed. Thus, our results indicate that the balance between negative and positive factors at the Cd8ab gene complex can determine the initiation of Cd8a and Cd8b gene expression. Even though the Cd8 gene loci are released from MAZR-mediated repression in the absence of MAZR, positive factors required for CD8 expression might not be sufficiently available in DN thymocytes. In contrast, in the absence of E8_I and E8_II, less positive factors are recruited to the Cd8 loci in DP thymocytes, leading to CD8 variegation. However, deletion of the negative regulator MAZR in E8_I,E8_II doubly-deficient DP cells shifts the balance at the Cd8 gene loci towards an “on” state and thus facilitates CD8 expression.

A second important result of our study is the demonstration that MAZR is part of the transcription factor network that controls CD4/CD8 cell fate specification of DP thymocytes. Several transcription factors (such as ThPOK, Runx1, Runx3, GATA-3 and Tox) act in a complex regulatory network that orchestrates CD4/CD8 cell fate decision^{15, 16, 19-24, 28-30}. Among those, ThPOK is considered as a “master” CD4 commitment factor, in part by shutting off lineage inappropriate genes^{21, 22, 31, 32}. Therefore, the regulation of ThPOK expression is crucial for correct lineage choice. We now show that in the absence of MAZR, a substantial proportion of MHC class I-restricted thymocytes redirects into CD4⁺ T cells with helper lineage characteristics due to derepression of ThPOK. This was demonstrated either by using transgenic mice expressing MHC class I-restricted TCR or by generating Mazr^−/− BM chimeric MHC class II-deficient mice. Although only a minor subset of Mazr^+/+ DP 69⁺ and CD4⁺CD8^loCD69⁺ cells expresses low levels of ThPOK mRNA in response to MHC class I stimulation^{15, 16, 20, 21, 25}, a higher percentage of these cells derepressed ThPOK in the absence of MAZR, providing a potential molecular explanation for the redirected differentiation of Mazr^−/− thymocytes. This suggests a role for MAZR in repressing ThPOK expression in MHC class I-signaled thymocytes during positive selection.

The ThPOK silencer is required for ThPOK repression and cytotoxic lineage development^{20, 26}. Runx complexes are essential for ThPOK repression at the pre-selection DP CD69⁻ stage via binding to the ThPOK silencer^{20, 21, 29}. We cannot formally exclude the possibility that MAZR regulates ThPOK expression indirectly, e.g by regulating the expression of Runx3 in CD4⁺CD8^lo cells, which would then lead to an altered expression of ThPOK. Alternatively, MAZR might have pleiotropic effects on gene expression, e.g. it might regulate the expression of a factor involved in TCR signaling. This could lead to qualitative and/or quantitative changes in TCR signaling and as a consequence changes in ThPOK expression. However, our ChIP result showing that MAZR binds to the ThPOK silencer clearly suggests direct control of the ThPOK locus by MAZR. It is tempting to speculate that MAZR represses ThPOK expression in cooperation with Runx complexes in MHC class I-signaled cells. Runx1 and Runx3 expression was normal in DP thymocytes, and Runx complexes were still bound to the ThPOK silencer in the absence of MAZR (data not shown), indicating that the recruitment of Runx complexes is independent of MAZR. In preliminary experiments we detected an association of MAZR with Runx1 but not Runx3 upon overexpression in HEK293T cells (see Supplementary Fig. 10). Thus, MAZR might be recruited to the ThPOK silencer via interaction with Runx complexes that are bound already to the ThPOK silencer. Alternatively, MAZR recruitment to the ThPOK silencer could occur independently of Runx complexes, however the interaction between MAZR and Runx complexes might be necessary for the recruitment of additional factors to repress ThPOK. Further studies are required to investigate potential mechanism(s) of ThPOK repression mediated by MAZR and Runx interactions. Moreover, since Runx1 promotes the expression of CD8 in unsignaled DP thymocytes¹⁹, and since MAZR regulates the activation of the Cd8ab gene complex, further studies should also address whether MAZR-Runx1 interactions might regulate CD8 expression during the DN to DP transition.

We also observed that a fraction of peripheral MAZR-deficient CD8⁺ T cells expressed ThPOK. This suggests that MAZR might also be part of the molecular machinery that maintains ThPOK silencing in CD8⁺ T cells. A recent study demonstrated that certain DNase I hypersensitive sites within the ThPOK locus preferentially appear in CD4SP thymocytes, supporting the notion that ThPOK expression is accompanied by structural changes of chromatin²⁶. Based on the variegated expression of ThPOK in Mazr^−/−ThPOK^+/GFP CD8⁺ T cells, MAZR might regulate ThPOK expression at the epigenetic level, in a similar manner to which it regulates Cd8 gene expression⁹. In contrast to CD8⁺ T cells, CD8SP thymocytes in Mazr^−/−ThPOK^+/GFP mice did not express ThPOK. This suggests different mechanisms of ThPOK silencing in CD8SP and in CD8⁺ T cells, a situation that is reminiscent of the transcriptional control of CD4 silencing, where the Cd4 silencer is required in different ways in CD8SP versus CD8⁺ T cells^{19, 33, 34}. Further analysis is required to elucidate the molecular mechanism(s) by which MAZR shuts-off ThPOK expression and/or maintains ThPOK silencing in CD8 lineage cells. Thus, in MHC class I-signaled CD4⁺CD8^lo thymocytes, MAZR is involved in keeping the ThPOK locus repressed during CD8 lineage differentiation, and/or in peripheral CD8⁺ T cells MAZR is likely to function in establishment and/or maintenance of ThPOK silencing (see Supplementary Fig. 11a,b for a working model of MAZR function). Of note, MHC class I- and MHC class II-signaled CD4⁺CD8^lo thymocytes expressed similar levels of Mazr mRNA (data not shown), and MAZR bound the ThPOK silencer in DN, DP, CD4SP, and CD8SP cells (data not shown). These data suggest that ThPOK repression by MAZR in MHC class I-signaled cells is not due to differences in MAZR expression or differential recruitment to the ThPOK locus.

We observed that ThPOK was derepressed in a proportion of MHC class I-signaled Mazr^−/−ThPOK^+/GFP thymocytes, and that ThPOK mRNA expression levels were broadly distributed leading to cells expressing higher and lower levels of ThPOK. It might be possible that only those thymocytes in which derepressed ThPOK expression levels remain above a certain threshold redirect into the helper lineage, thus leading to a partial redirection phenotype. In contrast, if derepression of ThPOK is below a certain threshold level, their differentiation into the cytotoxic lineage is impaired. ThPOK expression might even be regulated by TCR strength. Indeed, in contrast to the OT-I TCR transgenic system, thymocytes expressing the MHC class I-restricted P14 transgenic TCR are not redirected by loss of MAZR, although CD8 lineage development is severely impaired (data not shown). A similar dependency on the transgenic TCR system was observed in GATA-3-deficient mice, where redirected differentiation of MHC class II-restricted thymocytes occurs on the 5CC7 TCR transgenic background, but not on AND or DO11.10 TCR transgenic background²². Further studies are required to clarify the potential relation between TCR signaling strength and ThPOK derepression in the absence of MAZR, and the subsequent impact for CD4/8 cell fate decision.

Taken together, our data provide genetic evidence that MAZR controls the Cd8 gene loci. Moreover, our study reveals that loss of MAZR leads to derepression of ThPOK causing redirected differentiation of MHC class I-restricted thymocytes into helper T cells. Thus, MAZR is part of the transcription factor network that controls cell fate decision of DP thymocytes.