The absence of the autoimmune regulator gene (AIRE) impairs the three-dimensional structure of medullary thymic epithelial cell spheroids

Background Besides controlling the expression of peripheral tissue antigens, the autoimmune regulator (AIRE) gene also regulates the expression of adhesion genes in medullary thymic epithelial cells (mTECs), an essential process for mTEC-thymocyte interaction for triggering the negative selection in the thymus. For these processes to occur, it is necessary that the medulla compartment forms an adequate three-dimensional (3D) architecture, preserving the thymic medulla. Previous studies have shown that AIRE knockout (KO) mice have a small and disorganized thymic medulla; however, whether AIRE influences the mTEC-mTEC interaction in the maintenance of the 3D structure has been little explored. Considering that AIRE controls cell adhesion genes, we hypothesized that this gene affects 3D mTEC-mTEC interaction. To test this, we constructed an in vitro model system for mTEC spheroid formation, in which cells adhere to each other, establishing a 3D structure. Results The comparisons between AIRE wild type (AIREWT) and AIRE KO (AIRE−/−) 3D mTEC spheroid formation showed that the absence of AIRE: i) disorganizes the 3D structure of mTEC spheroids, ii) increases the proportion of cells at the G0/G1 phase of the cell cycle, iii) increases the rate of mTEC apoptosis, iv) decreases the strength of mTEC-mTEC adhesion, v) promotes a differential regulation of mTEC classical surface markers, and vi) modulates genes encoding adhesion and other molecules. Conclusions Overall, the results show that AIRE influences the 3D structuring of mTECs when these cells begin the spheroid formation through controlling cell adhesion genes.

The thymic 3D architecture offers a particular microenvironment for the positive and negative selection of developing thymocytes [4][5][6]. The thymic cortex is implicated in the positive selection of thymocytes that express functional T-cell receptors (α/β TCRs). The medulla is involved in the negative selection of autoreactive thymocytes, which avidly recognize self-peripheral tissue antigens (PTAs) presented on the surface of mTECs. This unique pattern of gene expression allows mTECs to express more than 19,000 protein-encoding genes, including the "ectopic" genes that encode PTAs, a phenomenon known as promiscuous gene expression (PGE) [4,[7][8][9][10]. At present, no other cell type is known that expresses such a large set of genes [11][12][13]. The positive and negative selection processes are crucial for the induction of central immune tolerance that prevents aggressive autoimmunity [9,14].
The autoimmune regulator gene (AIRE) is the primary transcriptional regulator of PTAs in mTECs, whose encoded Aire protein unleashes stalled RNA Pol II in the chromatin to proceed with the elongation phase of the transcription [15,9,10]. A second controller of PGE in mTECs is the forebrain embryonic zinc finger-like protein 2 (Fezf2) that plays a role as a classical transcription factor and thus binds directly to DNA in specific promoter regions [4,16].
Given the non-specificity of RNA Pol II in transcribing genes and given that the Aire protein in mTECs unleashes this enzyme on chromatin, AIRE ends up controlling the expression of a diverse range of mRNAs. In addition to PTAs, AIRE also regulates the expression of genes involved in cell adhesion, such as the extracellular matrix (ECM) constituent Lama1, the CAM family adhesion molecules Vcam1 and Icam4, which control mTEC-thymocyte adhesion [17,18]. Cell adhesion corresponds to an essential biological process in the structure and function of the thymus. The adequate adhesion of TECs increases the efficiency for T-cell development [19,20]. Besides adhesion molecules, the formation of a dense cellular network is necessary for the maturation of thymocytes, which is composed of ECM proteins, such as laminins, integrins, collagens, and fibronectins, as well as soluble molecules such as hormones, cytokines, chemokines, and growth factors that are also mediated by AIRE [21]. AIRE −/− mice have small and disorganized medulla of the thymus [22], and little is known about whether Aire influences the mTEC-mTEC adhesion in a 3D thymus structure. Two experimental strategies enable the study of the in vitro formation of the thymus's 3D structure that reproduces its microenvironment with the extracellular matrix and thymic epithelial cells. One of these strategies is the thymus re-aggregation [19,20], and the other is the 3D organotypic culture [23]. In the re-aggregation model, the thymus tissue is devoid of cells (decellularized), but retains most of the microenvironment of the extracellular matrix that could support the re-aggregation between TECs. When transplanted into athymic nude mice, the re-aggregate thymus organoids can receive lymphocyte progenitors derived from bone marrow and develop a diverse and functional T-cell repertoire [19,20].
In the search for an experimental model that could mimic the thymic microenvironment closest to that found in vivo, Pinto et al. [23] adapted a 3D organotypic co-culture, preserving the main characteristics of mTECs, such as proliferation and differentiation. This strategy helped to identify molecular components and pathways involved in the mTEC differentiation and promiscuous gene expression. Other studies used stem cells to form "thymospheres, " permitting the study of the factors necessary for thymocyte development [24,25]. Recently, a 3D culture substrate has been reported that allowed TECs to survive and proliferate, using electrospun fibrous meshes (eFMs) functionalized with fibronectin. The mTECs presented increased proliferation, viability, and protein synthesis when cultured on fibronectin-functionalized eFMs (FN-eFMs) [26].
In the present study, we asked whether AIRE regulates the adhesion between mTECs during the in vitro formation of a 3D structure. For this, we developed a model system in which mTECs are grown in non-adherent agarose micro-wells and adhere to form spheroids. The comparison between AIRE wild type (AIRE WT ) versus AIRE −/− mTECs allowed us to evaluate the influence of AIRE in the adhesion between these cells during the spheroid formation.

Results
The absence of AIRE disorganizes the initial phases of spheroid formation Figure 1A shows the beginning of spheroid growth (0 h), comparing AIRE WT mTEC 3.10 vs. AIRE −/− mTEC 3.10E6. The cultures started from the seeding of mTEC cells (0 h) until the complete 3D spheroid formation (24 h). A compact real-time video of the dynamics of the  showing the spheroid formation in agarose micromolds starting from the deposition of 1 × 10 5 mTECs, B Spheroid growth curves from 0 to 108 h, time-point values correspond to the mean and ± SD from three independent replicates. C Spheroid cell viability from 0 to 108 h of growth. One representative experiment of n > 3 is shown had a cell count higher than AIRE WT spheroids. Differently, the development of the AIRE WT spheroids drew a classical growth curve, permitting the identification of the exponential, stationary, and decay phases. The slope of the AIRE −/− spheroid growth curve indicates that in the absence of AIRE, the growth is increased at the beginning, and then the growth abruptly decreases. This suggests that in addition to controlling adhesion between cells, AIRE might influence the cell cycle of mTECs (Fig. 1B).
Besides the growth curve, we evaluated the cell viability during spheroid growth. The AIRE −/− spheroids increased viability during the first 12 h and decreased thereafter, whereas the AIRE WT progressively reached viability starting from 12 h and maintaining thereafter (Fig. 1C).

AIRE −/− spheroids influence the cell cycle kinetics and the cell death
After 12 and 24 h of adhesion, AIRE WT and AIRE −/− spheroids were dissociated and analyzed for the distribution of cells at the different cell cycle phases.
The cell counting performed at 12 h of spheroid formation indicated a significant increase in the percentage of cells at the G0/G1 phase, 46% in AIRE WT and 57% in AIRE −/cells, showing an accumulation of AIRE −/− cells at G0/G1 ( Fig. 2A). Interestingly, a significant reduction was observed in the sub-G1 fraction (7%) for AIRE −/− cells compared to AIRE WT (22%). Within 24 h of adhesion, a similar pattern was seen, with a significant increase in the percentage of cells at G0/G1 phase (65%) observed for AIRE −/− cells, as well as a lower percentage (5%) of the subG1 fraction compared to AIRE WT (46% G0/G1; 28% SubG1). In addition, a significant increase (18%) in AIRE −/− cells was observed at the G2/M transition compared to AIRE WT (13%). Taken together, these results show that the absence of AIRE causes a significant alteration in the cell cycle progression, as demonstrated by the high number of cells undergoing the G0/G1 and G2 phases.
Since AIRE −/− spheroids showed lower percentage of sub-G1 fraction than AIRE WT , and since these fractions indicate a certain percentage of cell death, we analyzed the percentage of apoptosis and necrosis. After 12 h of spheroid formation, AIRE −/cells showed a significant increase in the population of apoptotic cells (15%) compared to the AIRE WT (1.96%) (Fig. 2B). On the other hand, a significant reduction of necrotic cells was observed in AIRE −/− (8.3%) spheroids analyzed after 12 h compared to AIRE WT (15.26%). At 0 h (prior to spheroid formation) and 24 h, we did not find significant differences between AIRE −/and AIRE WT spheroids regarding the percentage and the profile of cell death.

Spheroid dead-cell center and spheroid morphology are impaired in the absence of AIRE
In the center of the spheroids, dead cells were observed for both AIRE WT or AIRE −/− cells along the growth; however, AIRE −/− spheroids exhibited an increased rate of dead cells along the growth, maximizing at 48 h, as shown in Fig. 3A. Quantitative analysis allowed us to determine the relative fluorescence intensity for both live (greencolored) and dead (red-colored) cells (Fig. 3B).
The high-resolution optical microscopy of the 1 μm histologic cuts allowed the observation of the internal structure of spheroids. The AIRE WT or AIRE −/− spheroids showed significant differences in their areas, especially at the first 24 h of growth. Although AIRE −/− spheroids present a larger area than AIRE WT spheroids, AIRE −/− cells are dispersed and compact similarly to AIRE WT spheroids only after 24 h of adhesion (Fig. 4A, B).
Scanning electron microscopy (SEM) accessed the detailed external surface of the spheroids. Twenty-fourhour AIRE WT spheroids were more compact, exhibiting a well-defined contour, while AIRE −/− spheroids showed an irregular surface (Fig. 4C).

The loss of AIRE impairs the strength of the cell-cell adhesion
To observe the strength of cell-cell adhesion, we induced an enzymatic spheroid disaggregation at 12 h of culture.  indicating that the lack of AIRE gene decreased the strength and spheroid adhesion.

Spheroid RNA-Seq analysis
Comparative transcriptome analyses were performed for AIRE WT and AIRE −/− spheroids at 12 h. We identified a set of 1210 differentially expressed (DE) mRNAs, of which approximately 83% (1104) corresponded to protein-coding mRNAs, and about 17% (106) corresponded to non-protein-coding RNAs (Fig. 7A). Among the 1104 DE mRNAs, 606 were upregulated and 498 downregulated (Fig. 7B). Figure 7C shows the hierarchical clustering of the DE mRNAs. Figure 8 shows the differences observed for the topten biological functions of downregulated ( Fig. 8A) and upregulated mRNAs (Fig. 8B). Figure 9A shows the differentially expressed genes involved in the positive regulation of epithelial cell proliferation. Figure 9B confirms by RT-PCR the differential expression of some top-ten DE mRNAs, including the downregulated PARVB that encodes a protein associated with the cell adhesion pathway and the PLCB2 and P2RX7 that are involved in the calcium signaling pathway in AIRE −/− when compared to AIRE WT spheroids. In contrast, the VEGFC and ID1 transcripts, associated with the regulation of epithelial cell proliferation pathway, were upregulated in AIRE −/− spheroids. We took advantage of the use of the STRING algorithm, which establishes interactions among the encoded proteins based on the validated data retrieved from the literature, to understand the interaction of these genes. Figure 9C shows that the Aire protein may interact with other proteins associated with the immunological response (Foxn1 and CD80), autoimmune disorders (Ptpn22), adhesion receptor (Itgb7), collagen (Col6a1 and Col6a2), and apoptosis (Tfeb).

Discussion
Considering that: i) the AIRE gene, besides controlling the expression of genes that encode tissue-specific antigens (TSAs) [7,9,14], also controls the expression of adhesion molecule genes [17,18], ii) studies regarding the role of AIRE in the negative selection have primarily focused on the interaction mTEC-thymocyte, and iii) little attention has been devoted to the role of AIRE on the mTEC-mTEC adhesion, in the present study we used 3D spheroids to evaluate the role of the AIRE gene on the mTEC-mTEC adhesion. We showed that the absence of the AIRE gene: i) disorganizes the 3D structure of mTEC spheroids, ii) increases the proportion of cells at the G0/ G1 phase of the cell cycle, iii) increases the rate of mTEC apoptosis, iv) decreases the strength of mTEC-mTEC adhesion, v) promotes a differential regulation of mTEC classical surface markers, and vi) modulates genes encoding adhesion and other molecules.
Since the consequences of AIRE deletion in the intercellular adhesion remain to be elucidated, we studied the interaction between mTECs when these cells were seeded in a non-adherent substrate such as agarose, where 3D spheroids may be observed at the various stages of development. Once in an agarose substrate, mTECs aggregate and adhere to each other to form spheroids, which may be helpful for the study of mTEC-mTEC adhesion. Then, we evaluated the consequences arising from the absence of AIRE, evaluating the: i) mTEC-mTEC physical interaction at regular periods, ii) growth curve of mTEC spheroids, iii) dynamics of 3D mTEC spheroid formation and We reported that AIRE interferes with the early process of spheroid formation, since its absence yielded a delay on the spheroid growth and on cell viability ( Fig. 1A-C). Although AIRE is a pro-apoptotic factor in mature mTECs [32], in this study we observed that during the 3D spheroid formation AIRE −/− spheroids presented increased apoptosis rate. Indeed, the absence of AIRE produced a significant impact on the progression of cells through the cell cycle ( Fig. 2A), mainly at 12 h. The higher percentage of cells undergoing the G0/ G1 phase in AIRE −/− spheroids are possibly related to the higher percentage of apoptotic cells when compared with AIRE WT spheroids (Fig. 2B). These results suggest that the lack of AIRE function may affect the cascade of signaling pathways, leading to alterations in the cell cycle progression (checkpoint mechanism) and apoptosis induction in proliferating cells. Corroborating this idea, we observed that the several genes related to apoptotic processes were upregulated in the AIRE −/− spheroids, including the PERP , PDCD4, TNFRSF21,  SLC5A8, S100A9, PPID, CYCS, CARD14, ALDOC,  CKAP2, STK17B, POLB, CDCA7, GADD45B, APLP1,  CASP1, CASP4, BNIPL, TNS4, AIM2, IL24, STEAP3,  FAM3B, SRGN, EPB41L3, EPHA7, FGFR2 and S100A8 genes. Notably, the CDCA7 and GADD45B genes are involved in cell division and mediate Fas-induced apoptosis, respectively. Regarding the physical structure of the 3D spheroids, dead cells were observed in the center areas of both spheroid types; however, the number of these cells progressively increased along time, predominating at 48 h in AIRE −/− spheroids (Fig. 3A-B). Besides, in the model system of this study, it was possible to observe a distinct morphology of the mTEC interactions; i. e., the AIRE WT forms a typical spheroid at 12 h, whereas the AIRE −/− initially forms an ellipsoid form instead of a spheroid shape (Fig. 4A/B). Noteworthy, the scanning electron microscopy showed that AIRE WT spheroids are well-compacted with a well-defined contour. In contrast, AIRE −/− spheroids exhibited an irregular shape and ill-defined surfaces (Fig. 4C), indicating that AIRE influences the internal and external spheroid structure.
Since the 3D AIRE −/− spheroids exhibited a delayed formation at 12 h, as shown in Fig. 1A, we also evaluate the strength of spheroid dissociation, using a dispase assay. Indeed, AIRE −/− spheroids dissociated faster than AIRE WT spheroids and exhibited increased number of disaggregated mTECs (Fig. 5 A-B). Taken in concert, these in vitro results corroborate the in vivo finding of a disorganized thymic medulla in AIRE −/− mouse [22] and add information regarding the dynamics of mTEC growth and mTEC-mTEC interactions. In addition to the spheroid structure, we also evaluate the role of AIRE on the mTEC surface markers (Fig. 6A/B). Both AIRE WT or AIRE −/− spheroids maintain their characteristic CD45 − Ly51 − medullary profile, i.e., confirming that they are not hematopoietic cells (CD45 + ) nor cTECs (Ly51 + ). The absence of AIRE decreased the expression of the EpCAM and UEA-1 cells, which are classical markers for epithelial cells and medullary thymic cells, respectively. Noteworthy, the expression of double positive (CD80 + MHCII + ) was increased in AIRE −/− spheroids (0.38% in AIRE WT to 9.77% in AIRE −/− ). Three major stages can be observed along mTEC maturation: i) the immature mTEC presents MHC-II low , CD80 low and Aire low , ii) during TRAs, mTECs exhibit their mature stage, presenting MHC-II high , CD80 high and Aire high , and iii) at the post-Aire stage, mTECs are MHC intermediary , CD80 intermediary , accompanied by an increased rate of apoptotic cells [2]. As observed in Fig. 6A/B, the AIRE WT spheroids present features of an immature mTEC constitutively exhibiting low expression of AIRE, as previously reported by our group [17]. In contrast, AIRE −/− spheroids present an intermediate expression of CD80 together with increased rate of dead cells, features that are characteristic of the post-Aire stage. These findings indicate that in the absence of Aire, mTECs present a terminally differentiated mTEC during the first 12 h of 3D spheroid culture, which may affect the interaction with thymocytes.
Considering that AIRE controls more than 3300 genes in mTECs [33], we initially evaluated the mRNA transcript profiles of AIRE WT or AIRE −/− spheroids at 12 h of mTEC-mTEC adhesion, showing a distinct pattern of gene expression (Fig. 7A-C). Among the DE mRNAs, we identified genes associated with cell adhesion, positive regulation of cell proliferation, apoptotic process, and response to hypoxia (Fig. 8A/B). Among the DE mRNAs related to cell adhesion, we observed a set of transcripts that encode proteins belonging to the cadherins (PCDHGB7), collagen (COL15A1), integrin (CIB3), fibronectin (FLRT3), or extracellular matrix protein families (MMP19), which were downregulated in AIRE −/− spheroids and are putatively involved on the dysregulation of the spheroid formation, as observed in this study. In contrast, VEGFC and ID1 genes, involved in the positive regulation of epithelial cell proliferation, were upregulated (Fig. 9A/B), a finding that may be associated with the increased number of cells at 12 h of growth (Fig. 1B). The network generated by the STRING tool (Fig. 9C) revealed that Aire interacts with important proteins involved in the immunological response, autoimmune disorders, adhesion receptor, formation of collagen, and apoptosis, indicating that the lack of Aire impairs the 3D spheroid conformation.
Noteworthy, we observed a set of repressed mRNAs associated with the calcium signaling pathway, and among these, the ADCY3, ADCY1, NOS1, P2RX7, ADRB2, PLCB4, ATP2B4, ATP2A3, P2RX3, CACNA1G, ADRA1B, PLCB1, PLCB2, and F2R transcripts are associated with the intercellular calcium waves. Since intercellular communication used by mTECs is mediated by intercellular calcium waves, requiring functional gap junctions and P2 receptors [34,35], these genes are potential candidates to be further studied. Even though operating through different ways, AIRE and mTEC-mTEC adhesion culminate into related activities that regulate a cascade of mRNAs that encodes cell adhesion molecules and other relevant molecules that maintain the 3D spheroid formation, opening new perspectives for studies of molecular mechanisms that control the 3D thymic medulla organization and mTEC adhesion and communication.

Conclusion
While most studies have focused on the mTEC-thymocyte interaction to evaluate the negative selection, this study focused on the role of AIRE on the mTEC-mTEC interaction as the primary step to support the adequate thymic medullary structure. Considering the 3D spheroid model, this study reported that the absence of AIRE disorganizes the 3D structure of mTEC spheroids, promotes a differentially regulation of mTEC classical surface markers, and modulates genes encoding adhesion and other molecules.

Spheroid formation
We use a precast agarose mold with non-adherent 600 μm diameter microwells, making the mTEC cells adhere once seeded in these compartments. The AIRE WT mTEC 3.10 and the AIRE −/− mTEC 3.10E6 cell lines were initially cultured as monolayers in RPMI 1640 medium (Gibco, Darmstadt, Germany) supplemented with 10% inactivated fetal bovine serum in 75 cm 2 polystyrene plastic bottles (Corning, New York, NY) in an incubator at 37 °C with 5% CO 2 atmosphere. After acquiring confluence, mTECS were trypsinized and seeded in agarose molds, using 2% low electroendosmosis agarose (Sigma-Aldrich, Saint Louis, MO), sterilized in 70% ethanol, washed twice in sterile PBS, and irradiated under germicidal UV light for 15 min. Spheroid growth was observed through a Cytosmart ® (Lonza Group AG, Basel, Switzerland) inverted microscope to produce a real-time movie of the culture.

Growth curve
To better characterize the 3D culture model, we quantified the cells that form the spheroids at different time points (determinations at every 12 h). At each time point, the spheroids were removed from the agarose microwells and dissociated using trypsin. Isolated cells were counted using a Cellometer ® Auto T4 Bright Field Cell Counter (Nexcelom Bioscience, Lawrence, MA). Triplicates were performed for each time point. We have thus drawn a spheroid growth curve to identify the exponential, stationary growth, and decline phases.

Analysis of cell cycle
Analysis of cell cycle kinetics was performed by flow cytometry using the Guava cell cycle analysis. AIRE WT and AIRE −/− spheroids were stained with propidium iodide (Sigma-Aldrich), a fluorescent nuclear marker.

Live-dead assay
We used the LIVE/DEAD ® Viability/Cytotoxicity Kit (Thermo-Fisher, Waltham, MA) to assess the proportion of live and dead cells in the spheroids, following the manufacturer's instructions.

Histological analysis of spheroids
The spheroids were fixed in 10% formaldehyde buffered in PBS, then dehydrated, passing through a battery of aqueous ethanol solution (75 to 100% ethanol) and included in historesin (HistoResin standard kit, Biosystems Switzerland AG, Muttenz, Switzerland). The resin blocks were cut to 1 μm thickness and deposited on microscope slides. After deparaffinization and rehydration, the slides were stained with hematoxylin-eosin (H&E) for further microscopic examination of the spheroid morphology.

Transcriptome analysis through RNA-Seq
We followed a protocol previously described by St-Pierre et al. [11]. Briefly, paired-end (

Functional enrichment of differentially expressed mRNAs
The list of the DE mRNAs was analyzed in terms of functional enrichment through the Database for Annotation, Visualization, and Integrated Discovery (DAVID) annotation tool (https:// david. ncifc rf. gov/) to the identification of the main biological processes and pathways represented by DE mRNAs. A functional category was considered significant if it comprised at least five mRNAs and a score of p < 0.005 with Benjamini-Hochberg correction. The STRING algorithm (Search Tool for the Retrieval of Interacting Proteins database: https:// string-b. org/ cgi/ input? sessi onId= brAz1 0J8qu fA& input_ page_ show_ search= on), which establishes interactions among the gene encoded proteins based on the validated data retrieved from the literature, was used to construct networks.

Reverse transcription quantitative real-time PCR (RT-qPCR)
The validation of the transcriptional expression of selected DE mRNAs was assayed by reverse transcription-quantitative real-time PCR (RT-qPCR) of the respective cDNAs. The expression level of each target mRNA was normalized to the housekeeping mRNA Hprt, which is commonly used as a reference. The Primer Blast (https:// www. ncbi. nlm. nih. gov/ tools/ primerblast/) web tool was used to select pairs of oligonucleotide primers spanning an intron/exon junction with an optimal melting temperature of 60 °C. The respective sequences were retrieved from the NCBI GenBank database (https:// www. ncbi. nlm. nih. gov/). The forward (F) and reverse (R) primer sequences (presented in the 5′-3′ orientation) were as follows: Hprt Gene expression was quantified using a StepOne Real-Time PCR System apparatus (Applied Biosystems). The analyses were performed using the cycle threshold (Ct) method, which allows for quantitative analysis of the expression of a factor using the formula 2 − ΔΔCt, in which ΔCt = Ct target gene − Ct of the housekeeping gene Hprt, and ΔΔCt = ΔCt sample − ΔCt. Experiments were performed in three independent replicates.

Statistical analyses
Depending on the distribution of the variables in each comparison, statistical analyses were performed using the unpaired T or the Mann-Whitney U-tests, considering significant p-values < 0.05.