Reduced Numbers of Mature Medullary Thymic Epithelial Cells in SKG Mice

Attenuated T cell receptor (TCR) signalling contributes to the susceptibility for autoimmunity as shown via mutants of PTPN22 and Zap70 genes. We here set out to investigate the effect of an attenuated TCR signal on the composition of the thymic epithelial cell (TEC) compartment. To that extent, we combined flow cytometry and histology and compared the TEC subpopulations of Zap70 wild type with SKG mutant mice. We found an increased cortical TEC compartment in SKG thymi at the expense of reduced numbers of mature medullary TECs and a 4.8‐fold reduced medulla area. We also found reduced proportions of CD69+‐activated thymocytes among double‐negative, double‐positive and CD4−CD8+ single‐positive stages, reduced absolute numbers of single‐positive thymocytes, diminished expression of Lta and Ltb by CD4−CD8+ single‐positive thymocytes and a diminished expression of Ccl19, a target gene of the lymphotoxin‐b‐receptor. While the reduced thymocyte numbers together with the attenuated TCR signal explain the diminished expression of lymphotoxins, the latter is required for an AIRE‐independent expression of tissue‐restricted antigens as well as attracting positively selected thymocytes to the medulla. Our results describe altered TEC compartments in SKG mice that are likely to support the development of autoimmunity.


Introduction
The aetiology of autoimmunity is based on the interaction between genetic and environmental factors. For human autoimmune diseases, the major histocompatibility complex (MHC) confers the main genetic risk to individuals [1]. However, over the last years, evidence has accumulated that genes involved in T cell receptor (TCR) signalling can also contribute to the susceptibility. For instance, variations in the PTPN22 gene that encodes the protein tyrosine phosphatase Lyp have been associated with rheumatoid arthritis (RA), systemic lupus erythematodus, type I diabetes and Graves disease [2]. Another molecule involved in TCR signalling and susceptibility for autoimmunity is the zeta-chain-associated protein kinase 70 (ZAP70).
Murine mutant strains of Zap70 provided important insights into the role of the TCR signal strength in shaping the TCR repertoire and leading to T cell-driven autoimmunity. As of yet, three distinct mutations of Zap70 have been described and analysed for their effects on thymic T cell selection and subsequent immunological consequences in the periphery. Siggs and colleagues analysed two variants of the Zap70 gene. On the one hand, the murdock variant mildly decreased TCR signalling which was tolerated on the systemic level. On the other, the mrtless variant led to a strongly diminished TCR signalling and an abolished positive selection with subsequent immunodeficiency [3]. In addition, Tanaka and colleagues investigated the skg variant and made use of Zap70 +/+ , Zap70 skg/+ , Zap70 skg/skg , Zap70 skg/À and Zap70 À/À mutant strains that are also characterized by a graded attenuation of TCR signalling. They proposed a 'selection shift' model whereby the extent of the TCR signal attenuation controls the selection of autoimmune TCR clones and subsequent autoimmunity. According to this model, T cell clones selected under mildly diminished signalling conditions led to predominantly organ-restricted autoimmunity while a strongly attenuated TCR signalling led to the selection of T cell clones that mediated severe systemic autoimmunity [2,4].
Interestingly, in all these models, the counterparts of the thymocytes during thymic selectionthe thymic epithelial cells (TEC)have been neglected. However, the first selection process during T cell development is mediated by cortical thymic epithelial cells that present a specific set of antigens in an MHC-restricted fashion [5]. Thymocytes recognizing self-MHC and binding ubiquitous self-antigens with low affinity are positively selected. These thymocytes then migrate into the thymic medulla and are subjected to negative selection mediated by medullary TECs and dendritic cells [6,7]. Importantly, medullary TECs express genes that outside the thymus are tissue restricted, a process called promiscuous gene expression [8]. Within the thymus, dendritic cells and medullary TECs cooperate to eliminate autoreactive thymocytes with high affinity to self-proteins [9]. In contrast, thymocytes with lower affinities for self-antigens are allowed to escape into the periphery for immune surveillance. And finally, thymocytes with medium affinity to self-antigens are driven to differentiate into regulatory T cells that will mediate suppression [10].
Up to now, knowledge about the impact of high and low TCR signal strength on the TEC compartment is scarce. However, a cross-talk between thymocytes and TECs has been described to be essential for mTEC maturation [11][12][13]. We therefore postulated that the reduced TCR signal strength in SKG mice will be reflected in the medullary TEC compartment.

Materials and methods
Ethics statement. All animal experiments were performed in accordance with the guidelines of the local animal use and care committee. All organ removals were reported to the responsible authority 'Landesamt f€ ur Landwirtschaft, Lebensmittelsicherheit und Fischerei Mecklenburg-Vorpommern' in line with §7(2) German animal protection law. Animal housing was carried out by professional animal keepers, and all efforts were made to minimize suffering.
Mice. Balb/c (Charles River, Sulzfeld, Germany), C57BL/6J (Charles River), NMRI (Charles River), DBA/ 1J (Charles River), MRL/MpJ (Jackson Laboratory, Bar Habor, ME, USA) and SKG (kind gift from Ulf Hamann, DRFZ in Berlin, Germany; the skg genotype was verified by sequencing in our laboratory) mice were maintained in a specific pathogen free unit on a 12 h light / 12 h dark cycle with 30 min twilight period. The ambient temperature was 21 AE 2°C, the humidity was 60 AE 10%, and the room air change is 20-fold. Mice were housed using a stocking density of 3-5 mice per cage. Mice were given water and ssniff R/M-H diet (ssniff Spezialdi€ aten GmbH, Soest, Germany) ad libitum.
Histology. Thymi from SKG and Balb/c mice were excised and fixed in 4% paraformaldehyde for 5 days. Paraformaldehyde was removed under floating tap water for 30 min, tissues were paraffin-embedded, and 5 lm thin sections were prepared. Sections were deparaffinized and rehydrated prior to staining with haematoxylin/eosin (H&E). Stained sections were digitalized using an AxioCam MRc5 on an Axio Imager.M2 microscope with an EC Plan-Neofluar 20 9 /0.50 M27 objective (all from Carl Zeiss Jena GmbH, Jena, Germany). Regions were quantified using ImageJ (v 1.51n).
Immunofluorescence. Thymi from SKG and Balb/c mice were excised, embedded in TissueTek (SakuraFinetek Europe B.V., Alphen aan den Rijn, Netherlands), mounted on steel blocks and immediately frozen in liquid nitrogen. Cryosections of 6 lm were performed with a Carl Zeiss Cryotome (Carl Zeiss Jena GmbH, Jena, Germany) and transferred to glass slides. Sections were fixes in À20°C precooled acetone (Merck Millipore, Darmstadt, Germany) for 10 min and were then allowed to dry for 20 min at RT. Slides were washed in 1 9 TBS for 5 min each, and thymi were encircled with a wax pen (Thermo Fisher Scientific, Waltham, MA, USA). Blocking was performed for 1 h in a humidified chamber at RT with 1 9 blocking buffer (1 9 TBS containing 2,5% skim milk (USBiologicals, Salem, MA, USA) and 2,5% FCS (Biochrom, Berlin, Germany). Given dilutions of primary reagents were administered after removing blocking buffer and slides were incubated overnight in a humified chamber at 4°C. For the detection of mTEC areas biotinylated UEA-1 (Vector Laboratories, Burlingame, CA, USA; dilution 1:100) was used, detection of terminally differentiated mTECs was performed using rabbit-anti-mouse-involucrin (BioLegend, San Diego, CA, USA; dilution 1:1000). Afterwards, the slides were washed in 19 TBS for 5 min each and incubated for 1 h in a humidified chamber at RT with the given dilutions of secondary reagents: NeutrAvidin Alexa488 (Molecular Probes Inc., Eugene, OR, USA; dilution 1:500) and goat-anti-rabbit Alexa 546 (Molecular Probes Inc., OR, USA; dilution 1:500). The slides were washed two times in 1 9 TBS, and the nuclei were counter-stained for 15 min at RT with DAPI (Molecular Probes Inc., Eugene, OR, USA; 1:1000). Finally, the slides were washed two times with 1 9 TBS, four times each under floating tap water and 5 min with distilled water. After carefully removing the excess of water, the sections were embedded in mounting medium (DiaSorin Inc., Stillwater, MN, USA) and the coverslips were encircled with nail polish to protect them from dehydration. Fluorescence imaging was performed with Confocal Microscope LSM780 (Carl Zeiss Jena GmbH, Jena, Germany) and Zen2011 Software.
Enrichment of TEC. Thymic cells were isolated from 4-to 5-week-old male mice [14]. Thymi from five mice were carefully excised and adjacent connective tissue was removed. Thymi were cut into small pieces and digested in 500 ll RPMI containing 0.02% collagenase B (Roche, Basel, Switzerland), 2 U/ml dispase (BD Bioscience, Heidelberg, Germany) and 100 U/ml DNase (Roche) per thymus. Thymic pieces were incubated at 37°C, and cells were dissociated by careful pipetting using glass pipettes with decreasing opening diameter. Single cell suspension was centrifuged, resuspended in 4 ml dense Percoll (density: 1.115 g/ml, Sigma Aldrich, Germany) and transferred into a 15 ml tube. Cell suspension was overlain with 2 ml less dense Percoll (density: 1.065 g/ml) and 2 ml D-PBS (Gibco, Life Technologies, Germany). Density gradient was established by centrifugation at 1450 g at 4°C for 30 min. TEC enriched in the second interphase was used for subsequent flow cytometry analysis.
Flow cytometry analysis and cell sorting. TEC enriched and TEC depleted cell populations were stained for surface markers in ice-cold PBS pH 7.4, 0.5% bovine serum albumin, 0.1% sodium azide. The following antibodies were used for TEC staining and sorting: Ly-51:FITC (clone: 6C3, BD Bioscience, Germany), CD80:PE (clone: 16-10A1, Biolegend, Germany), CD45:PerCP (clone: 30-F11, Biolegend), EpCAM:Alexa647 (clone: G8.8, Biolegend). The cell numbers for the analysed TEC populations were calculated by adding up the respective cell numbers contained in the TEC enriched and TEC depleted cell fractions after the Percoll density gradient divided by the number of thymi used to isolate the cells. TEC data were obtained in two independent experimental procedures with different settings during acquisition of the data. Therefore, data were normalized for these different experimental settings.

SKG harbour reduced numbers of mature medullary TECs
We here investigated the cellular composition and size of the murine TEC compartment comparing SKG mice harbouring a mutant Zap70 gene and Balb/c as the corresponding background strain. We also analysed additional mouse strains with a wild type Zap70 gene including models for induced and spontaneous autoimmunity, respectively. To that extent, we performed flow cytometry and calculated both, absolute numbers and percentages of cortical as well as immature and mature medullary TECs in the thymus (Fig. 1A). TECs were identified as EpCAM + CD45 À cells. Among them, cortical TECs are identified by a high expression of Ly51 whereas medullary TECs express none or low levels of Ly51. Medullary mTECs are further differentiated into a mature (CD80 high ) and an immature stage (CD80 low ).
For the cortical TECs, we found a significant increase in numbers as well as percentages in SKG mice compared to Balb/c (Fig. 1B). Consequently, there were significantly decreased percentages alongside a trend towards lower numbers of medullary TECs in SKG mice. The reduction in the numbers of medullary TECs was limited to the mature medullary TEC compartment as shown by a 75% decrease in mature medullary TECS and constant numbers of immature medullary TECs. These data resulted in a significant increase in the percentages of immature medullary TECs and a marked decrease in the percentages of mature medullary TECs among medullary TECs in SKG mice. Of note, the percentages ( Figure S1) and numbers ( Figure S2) of mature medullary TECs in SKG mice are also reduced compared to the mouse strains C57BL/6J, DBA/1J, MRL and NRMI.
We further investigated the thymic morphology by histology and performed H&E stainings of thymic thin sections (Fig. 1C). In line with the reduced numbers of mature medullary TECs, the proportion of the medulla in SKG mice was significantly reduced in comparison with Balb/c and appears as sparsely scattered patches. Thus, the reduction seen in histology is even more prominent than the reduction in cell numbers of mature medullary TECs.
We here considered CD80 low medullary TECs as immature. However, terminally differentiated medullary TECs also downregulate CD80. To exclude the possibility that the increased percentages of CD80 low medullary TECs result from an increase in terminally differentiated medullary TECs, we performed immunofluorescence on thymic thin sections and stained with involucrin, a specific marker for terminally differentiated medullary TECs. As Scandinavian Journal of Immunology, 2018, 87, 28-35 expected, we did not see an increase in involucrin + cells in SKG compared to Balb/c mice ( Figure S3).
In summary, our data show a relative and absolute increase in cortical TECs alongside a decrease in mature medullary TECs in autoimmune-prone SKG mice with a mutant Zap70 gene.

SKG thymocytes show a reduced capacity to drive maturation of medullary TECs
Next, we investigated the mechanism that may lead to a delay in the differentiation of medullary TECs from the immature to the mature stage. It has been shown previously that positively selected thymocytes provide signals which drive medullary TEC maturation [15]. Therefore, we analysed the thymocyte compartment of SKG and Balb/c mice in more detail.
Second, we sorted DN, DP, SP4 and SP8 thymocytes and analysed their expression of molecules known to be crucial for medullary TEC maturation (Fig. 3A). We found a significant reduction in the expression of Lta, Ltb and Tnfsf11 in DN and a significantly lower expression of Lta and Ltb in SP8 thymocytes in SKG compared to Balb/c mice.
Third, we sorted medullary TECs from SKG and Balb/c thymi and analysed the expression of selected Ltb receptor target genes (Fig. 3B). We found a significantly lower expression of Ccl19 in mature mTECs from SKG mice. However, no difference could be detected for Ccl21 and Fezf2. Furthermore, we determined the expression of Fezf2dependent (Resp18 and Krt10) and Aire-dependent (Ins2) TRAs and found no difference in either expression. The same holds true for the expression of Aire. However, all genes were significantly differentially expressed between mature medullary TECs and immature medullary TECs independent of the mouse strain.
In summary, the activation of thymocytes as well as their expression of signalling molecules for medullary TEC maturation is reduced in SKG mice.

Discussion
We here demonstrate a marked decrease in the mature medullary TEC compartment in SKG mice in comparison with BALB/c and four additional mouse strains carrying the Zap70 wt gene. This decrease in SKG mice resembles an intermediate phenotype between Zap70 wt and knockout mice, the latter showing a complete lack of medullary structures as well as severe systemic autoimmunity. In contrast, SKG mice show a more restricted autoimmunity resulting in arthritis and some systemic manifestations such as reduced bone density, rheumatoid nodules and inflammation of the lungs [4,[16][17][18].
We aimed to investigate what caused the reduction in mature medullary TECs in SKG mice. Previous studies already suggested an intricate cooperation between thymocytes and TEC compartments by showing that the development of mature medullary TECs is crucially dependent on the expression of signalling molecules expressed by positively selected thymocytes [11,12].
The present manuscript describes reduced numbers of both, SP and CD69 + -activated DP thymocytes. These reduced numbers together with the attenuated TCR signal strength in SKG mutant mice are predicted to result in a net reduction in signalling molecules. Indeed, we here found a reduced expression of the signalling molecules Lta and Ltb in SP8 thymocytes. Of note, a reduction in lymphotoxin b is expected to lead to two different effects, the first one being a reduction in the AIRE-independent expression of several TRAs in murine medullary TECs, including collagen type II [19][20][21][22]. The second effect implies a reduction in lymphotoxin b-regulated chemokines that have been described to attract positively selected thymocytes to the medulla [20]. Our results thus combine quantitative and qualitative findings that provide an explanation for both, the lack of positively selected thymocytes in the medulla of SKG mice and autoimmunity resulting from inaccurate selection [21,23].
It remains enigmatic, whether autoimmunity in SKG mice is caused or driven by the attenuated TCR signal strength, the diminished expression of lymphotoxins, the reduction in mature medullary TECs or by a combination of all above mechanisms. In favour of a role of the reduced numbers of mature medullary TECs, it was shown that a single medullary TEC expresses only a small group of 200-300 tissue-restricted antigens (TRA). In order for the whole spectrum of TRAs to be represented within the thymus, medullary TECs need to establish a mosaic expression pattern [24]. This mosaic expression pattern in turn requires a high mobility of the thymocytes so that they get contact to all possible TRAs and become selected. A reduced number of mature medullary TECs will simply increase the likelihood for autoreactive thymocytes to evade negative selection in the medulla. Indeed, a recent study could demonstrate that the specific depletion of medullary TECs led to an overt autoimmune hepatitis paralleled by autoantibodies against liver, lung, kidney and intestine as well as inflammatory infiltrations within lungs and kidneys [25]. Thus, a marked reduction in medullary TECs by itself seems to suffice to induce a restricted autoimmunity with systemic manifestations.
We here not only found a decrease in mature medullary TECs, but we also demonstrated an absolute and relative increase in cTECs in the SKG mice. Even though some knowledge exists about the factors driving medullary TEC maturation, the information on cortical TEC differentiation is rather poor. It has been shown that the presence of DN1-3 thymocytes drives cortical TEC maturation by as of Scandinavian Journal of Immunology, 2018, 87, 28-35 yet unknown signals [26]. Another study suggested that thymocytes at the DP stage can induce cortical TEC maturation [27]. Indeed, we found an increase in the proportion of early CD69 À DP thymocytes in SKG mice without observing any quantitative changes in the DN compartment. Both, the increase in early DP thymocytes and an altered expression of signalling molecules in DN thymocytes may therefore be responsible for the increase in cortical TECs in SKG mice.
As of yet, it appears that autoimmunity in SKG mice may be induced by a combination of different parameters, among them the attenuated TCR signalling, a reduced medullary TEC compartment and a diminished Lta and Ltb expression. All of these should be considered when studying the effect of altered TCR signalling on autoimmunity.

Supporting Information
Additional Supporting Information may be found online in the supporting information tab for this article: Figure S1 Percentages of TEC populations in SKG, Balb/c, C57BL/6J, DBA/1J, MRL/MpJ and NMRI mice. Figure S2 Number of TEC populations in SKG, Balb/c, C57BL/6J, DBA/1J, MRL/MpJ and NMRI mice. Figure S3 Comparable numbers of involucrin-positive terminally differentiated mTECs in SKG and Balb/c mice.