Integrated role of human thymic stromal cells in hematopoietic stem cell extravasation

Abstract The human thymus is the site of T‐cell maturation and induction of central tolerance. Hematopoietic stem cell (HSC)‐derived progenitors are recruited to the thymus from the fetal liver during early prenatal development and from bone marrow at later stages and postnatal life. The mechanism by which HSCs are recruited to the thymus is poorly understood in humans, though mouse models have indicated the critical role of thymic stromal cells (TSC). Here, we developed a 3D microfluidic assay based on human cells to model HSC extravasation across the endothelium into the extracellular matrix. We found that the presence of human TSC consisting of cultured thymic epithelial cells (TEC) and interstitial cells (TIC) increases the HSC extravasation rates by 3‐fold. Strikingly, incorporating TEC or TIC alone is insufficient to perturb HSC extravasation rates. Furthermore, we identified complex gene expressions from interactions between endothelial cells, TEC and TIC modulates the HSCs extravasation. Our results suggest that comprehensive signaling from the complex thymic microenvironment is crucial for thymus seeding and that our system will allow manipulation of these signals with the potential to increase thymocyte migration in a therapeutic setting.


| INTRODUCTION
The thymus is the primary lymphoid organ where hematopoietic progenitors are instructed to develop into T cells during thymopoiesis. Thymus seeding progenitors (TSP) are recruited from the pool of hematopoietic stem cells (HSC) in the fetal liver 1  The overall process of TSP entrance into the thymus has been poorly described in humans due to a lack of appropriate models, though extrapolation from mouse data suggests that chemokines secreted by TEC, such as CCL25 5,6 and CXCL12, 7,8 identified as ligands to CCR9 and CXCR4 receptors, respectively, as well as the expression of Pselectin will play a critical role in TSP migration and extravasation. 9 Transendothelial migration occurs most frequently within postcapillary venules 10 and may transpire in one of two ways. 11 In transcellular extravasation, a leukocyte passes through an endothelial cell, while the biomechanically favorable paracellular extravasation occurs between endothelial cell junctions. 12,13 The first early T-cell progenitors (ETP) are detected in the human fetal thymus as early as Week 5 14 though significant seeding is not seen until Week 8. 15 By Week 12 of embryonic development, mature T cells have begun to leave the thymus, continuing through development and postnatal life. 4 Crucial results from HSC transplantation 16,17 and gene therapies 18  CCR9, and CD10 (TSP1) and ITGβ7 and CD7 expression (TSP2) with the latter representing a "primed" state. 19 While ETP represent a very small fraction of the total hematopoietic cells within the thymus, they maintain a spectrum of lineage genes and can give rise to both myeloid and lymphoid cells including plasmacytoid dendritic cells, NK T cells, and innate lymphoid cells. 20,21 As the cross-talk between the developing thymocytes and the TEC is crucial for both T cell and epithelial development [22][23][24] and Notch signaling is critical for T lineage commitment, [25][26][27][28] the interactions between the TSP and other TSC, such as EC and TIC, may be similarly crucial for progenitor recruitment and commitment.
Significant efforts have been made to partly recapitulate the complexity of in vivo conditions using in vitro physiological systems. Recent advances in thymus reconstruction have allowed for long-term expansion of progenitors and functional reconstitution of the thymus. 29 However, the time scale and cellular number required make molecular insights difficult to interrogate in a systematic manner. Microfluidic cell culture devices facilitate 3D cell culture and allow tissue and organ-level functions to be assessed and recently have been utilized particularly well to mimic vascular formation and function. [30][31][32][33] High-throughput analysis with minimal cell number requirements, spatiotemporally high-resolution live microscopy, and retrieval of biological material are key advantages of microfluid-based 3D culture systems compared to bioreactors or organotypic culturing. Several in vitro models have been developed to investigate cellular extravasation in other contexts, that is, cancer [34][35][36][37][38][39][40] and immune cells extravasation. 41,42 The methods have so far focused on the innate extravasation potential of the cells themselves with less attention on the influence of the microenvironment in which this extravasation occurs. We have therefore sought to develop an assay that mimics the vasculature while also incorporating the thymic stroma to better understand the influence of thymus microenvironment on progenitor seeding and T-cell development.

| A 3D microfluidic device recapitulates HSC extravasation
To model HSC extravasation in the thymic environment, we employed three channel polydimethylsiloxane (PDMS) microdevices and studied HSC transendothelial migration (Figure 1a). The central channel of the device, which has a volume of 0.4 mm 3 , was filled with a fibrin hydrogel that is biochemically inert, has sufficient strength to support human umbilical vein endothelial cells (HUVEC) monolayer formation, and is sufficiently porous for HSC (pores > > 1 μm) 3D migration. 43 To recapitulate the lumen of a vessel, HUVEC were seeded into the left channel of the device to adhere to the fibrin in the center channel and the PDMS wafer as To demonstrate the quality of the endothelial barrier, function of the left channel was perfused with fluorescently labeled 40 kDa dextran ( Figure 1e). The dextran diffused slowly through the endothelial barrier, demonstrating a good degree of barrier function and ability to create chemical gradients within the device. While media, nutrients, and secreted factors were present throughout the device, any factor added into the right channel created concentration gradients over a span of a few hours allowing investigation into directed cell migration.
To assess extravasation rates, CD34 + HSC were seeded into the endothelial channel. After 24 h, some HSC were observed at the interface of the endothelial monolayer-fibrin gel (Figure 1f, inset) and some had migrated away from monolayer fully embedded within the fibrin gel (Figure 1f, white arrow) indicating different stages of extravasation. An example of an HSC passing through endothelial cells while squeezing itself is shown in (Figure 1f, inset) suggesting the HSC apply forces to transmigrate through tight endothelial barrier. 44,45 To quantify extravasation rates, the images were segmented at the endothelial border, with a counting area extending 200 μm from either side of the endothelial interface ( Figure S1a). The number of HSC, which had moved directly away from the endothelial interface into the gel channel, were counted and divided by the total number of HSC located within the counting area. Interestingly, assessment of the extravasation potential of CD34 + HSC of different origins (i.e., isolated from fetal liver, bone marrow, umbilical cord blood, and Lin À /CD3 À /CD4 À /CD8 À triple negative [TN] thymocytes) showed that these different HSC sources had similar intrinsic extravasation rates, with an average of 20% cells extravasated after 24 h ( Figure 1g). Moreover, the intrinsic extravasation rate of CD34 + HSC from fetal liver sourced at different weeks postconception (wpc) stayed at the same levels, despite the concordant differences in developmental stage (Figure 1h). (h) Extravasation rates of CD34+ HSC from fetal liver, cord blood, and bone marrow, as well as CD3 À /CD4 À /CD8 À triple negative thymocytes from thymus. (i) Extravasation rates of CD34+ HSC from fetal liver from 12, 15, 17, 22 wpc. Scale bars 100 μm unless otherwise indicated and nuclei stained with DAPI (blue)

| Cytokines and TSC increase HSC extravasation rates
Formation of a tight endothelial monolayer in microfluidic channels provides the opportunity to create cytokine gradients across endothelium. Furthermore, to understand the role that the thymus microenvironment plays in TSP recruitment, we took the advantage of in vitro expansion of human thymic epithelial (TEC) and interstitial (TIC) progenitors that demonstrated the capacity to attract HSC from bone marrow in vivo. 29 Therefore, we set out to apply a cytokine gradient and incorporate up to four different cell types within our microdevice ( Figure 2a). Twenty-four hours after formation of the endothelium, we seeded 120 K TSC previously expanded in culture into the right channel or added exogenous chemokines (100 ng/ml of CXCL12 and CCL25) to the media in the right channel. Twenty-four hours after seeding of TSC, $30 K CD34 + fetal liver HSC were injected into the right channel and their extravasation rates were assessed after an additional 24 h (Figure 2b). An average of 5439 HSC localized within 200 μm on either side of the endothelial border ( Figure S1c) and the experimental condition within the right channel did not affect HSC quantification ( Figure S1d). We found that the presence of exogenous chemokines or TSC significantly increased (4-and 3-fold, respectively) the number of extravasated HSC (Figure 2c). Consequently, the extravasation rate was also significantly increased when exogenous chemokines or TSC were present (Figure 2d).

| Impact of cellular origins
Next, we investigated whether and how the observed enhanced HSC extravasation under chemokines or TSC co-culture may depend on the different cell sources. Indeed, the source of primary stromal cells and their culture procedure may cause some variability in cellular behavior ( Figure 3a). To dissect the role of HSC source on this variability, we normalized extravasation rates for each condition with respect to its own control extravasation rates (i.e., the same HSC source). Notably, we found that regardless of the HSC source extravasation rates were significantly increased under the influence of chemokines and TSC (3-and 5-fold, respectively) ( Figure 3b).
We also explored the influence of the source of thymic stroma by testing the extravasation of HSC co-cultured with TSC that consists of mix culture of stromal TIC and TEC from different donors. While the cultivated TEC ( Figure S2a-c,f,g) and TIC ( Figure S2d-g) maintain their cellular identity in culture, the heterogenous nature of primary culture, and specifically the source of TSC (i.e., TIC and TEC from different donors) creates some degree of variability in the extravasation rates ( Figure 3c). Nonetheless, the rates were consistently higher compared to controls (i.e., HSC extravasation with no stromal cells).
Notably, incorporating TEC and TIC from different donors showed similar relative extravasation rates to those from the same donor ( Figure 3d).

| TEC and TIC act synergistically to recruit TSP
Motivated by the data suggesting that the presence of TSC increases HSC extravasation, we sought to expand the biological implications by CXCL12, used as one of the exogenous cytokines in the assay, was highly expressed in microdevices with TSC ( Figure 4g). This expression was driven primarily by the TIC rather than the TEC ( Figure S3e).
CCL5 (RANTES) was also significantly expressed in devices containing TSC or TEC alone (Figure 4h and Figure S3f). A proinflammatory chemokine, CCL5 has been reported to regulate intrathymic migration in mice. 49 To interrogate the role of CCL5 in HSC migration, TSC were pretreated with a CCR5 antagonist (Met-CCL5, 100 ng/ml) before seeding and Met-CCL5 was supplemented in the media of devices with TSC. Blocking CCR5/CCL5 signaling lead to decreased extravasation of HSC compared to the untreated condition (Figure 4i).

| DISCUSSION
The properties of human thymus seeding progenitors, including the signals that induce their migration, as well as lineage commitment instructions have begun to be unraveled 50 55 as well as microvascular endothelial cells 56 but interestingly is not expressed in mouse thymus. 57 As IL-8 has also been implicated in T-cell recruitment at inflammation sites, 58 it may also play a parallel role in precursor homing within the human thymus.
CD10, a marker for the earliest thymic seeding progenitors, 19 was significantly expressed in devices with TSC and TIC but not when only TEC were present, suggesting that TIC may play a crucial role in lineage commitment of thymocytes. As CD7 is a marker of lineage committed ETP, we were surprised to see expression in only 24 h, thus supporting that the thymic stroma may determine an inductive microenvironment for lymphoid lineages. Expression of these markers by HSC-derived cells in microdevices with TSC suggests that T lineage commitment may occur via indirect interaction between thymic stroma and the HSC. The signals, which likely lead to this early commitment, may stem from the TIC rather than the TEC. While cross-talk between TEC and thymocytes has been well described at least in mouse, 59 the role of the mesenchymal cells of the thymus is emerging 60 but has not been extensively investigated, highlighting the need to better understand this cellular compartment especially in human. In line with this, our data further support the important role of the nonepithelial stroma in thymus function.
While the intrathymic function of CCL5 (RANTES) is poorly understood, its expression has been implicated in myasthenia gravis, a thymoma-related autoimmune disorder. 61 On the contrary, the proinflammatory role of CCL5 (RANTES) has been well described in other contexts. 62 Given the high expression of CCL5 by cells seeded within the microdevice, we used the assay to assess the role of CCL5 in HSC migration. Inhibition of CCL5/CCR5 signaling by a CCL5 antagonist within the microdevice decreased HSC migration suggesting a previously undescribed role of CCL5 in thymus seeding. This also demonstrated the usefulness of the assay for therapeutic screening.
In summary, our results show that cooperation among different stromal populations of the thymus, including endothelial, epithelial, and interstitial cells, is crucial for TSP seeding and lymphoid commitment not only in vivo but also in vitro. Deciphering the signals that recruit progenitors to the thymus is important for both furthering our understanding of human thymus development, but also has clinical implications, including for reconstitution interventions following HSCT gene therapies. By identifying the signals, which direct thymus seeding, we could develop novel approaches to improve T-cell output and T-cell repertoire longevity postintervention in the clinic.

| CCR5 antagonist
Recombinant human CCL5/Met-RANTES (R&D Systems) was added to flasks with cultured TEC and TIC at 100 ng/ml 24 h before seeding microdevices. Sterile PBS was added to untreated flasks as a control.
The 100 ng/ml of CCL5/Met-RANTES was added to EGM2 media of experimental devices and assay continued as described above.

| Assessment of barrier function
Twenty-four hours or 72 h after seeding endothelial monolayer 40 kDa dextran labeled with TexasRed (Thermo Fisher) was resuspended in media and added to left channel of microfluidic device.
Images were taken every 5 min for 2 h, or ever minute for 10 min.
Dextran reaches right channel within 10 min and fully diffuses by 2 h.

| Quantification of extravasation
Extravasation assay images were compiled in FIJI using a Max Projection of the 20 μm Z-stacks. 63 Images were then imported into QuPath-0.2 64  Relative gene expression was normalized to two housekeeping genes (β-actin and TBP) and then benchmarked to control devices.

| Statistics
The exact sample size (n) for each experiment is given as a discrete number and demonstrated graphically. Statistical analysis was performed using one-way nonparametric ANOVA with post hoc analysis unless otherwise stated. Plots and graphs were generated with GraphPad Prism 9.

CONFLICT OF INTEREST
All authors declare that there are no conflicts of interest related to this work.