Ontogeny of Stromal Organizer Cells during Lymph Node Development

Mice

BALB/c (H-2^d), C57BL/6 (H-2^b), LtβR^−/− (H-2^b) (13) (C57BL/6 background), and Rorγ^−/− (H-2^b) (14) (C57BL/6 background) mice were bred and maintained under specific pathogen-free conditions in the Biomedical Services Unit at the University of Birmingham (Birmingham, U.K.) according to Home Office and local ethics committee regulations. Day of vaginal plug detection was designated as E0.

Isolation of LN stromal and CD45⁺ cells and mLN organ culture

iLNs and mLNs were isolated at the indicated embryonic stages by microdissection and were disaggregated by incubation in 2.5 mg/ml collagenase/dispase (R&D Systems, Minneapolis, MN) and 100 µg/ml DNase I (Sigma-Aldrich, St. Louis, MO) in RF10 media at 37°C to obtain single-cell suspensions. In some experiments, freshly isolated mLNs were explanted in fetal organ culture prepared as described (15) and treated with the agonistic anti-LTβR Ab (clone 4H8) at 2 µg/ml for 3 d (16, 17).

Abs and flow cytometry

The following Abs were used for flow cytometry: CD45.2-FITC clone 104, ICAM-1–PE clone YN1/1.7.4, VCAM-1–APC clone 429, MAdCAM-1–biotin clone MECA-367, VEGFR-3–biotin clone AFLA4, PDGFR-α–APC clone APA5, (eBioscience, San Diego, CA), and gp38/podoplanin 8.1.1 supernatant (hamster; kind gift from A. Farr, University of Washington, Seattle, WA). Biotinylated Abs were detected using streptavidin-PECy5.5 (eBioscience). gp38/podoplanin Ab was detected using goat anti-hamster– FITC (Southern Biotechnology Associates, Birmingham, AL). All stainings were made in PBS plus 0.1% BSA, 10% mouse serum, and 10% rat serum. Four-color flow cytometric analysis was performed using an FACSCalibur (BD Biosciences, San Jose, CA) with forward/side scatter gates set to exclude nonviable cells. Gates were applied appropriately according to the expression profile of the molecules being assayed. Data were analyzed with FlowJo software (Tree Star, Ashland, OR).

Cell sorting, RT-PCR, real-time RT-PCR

Isolation of the different CD45⁻ ICAM-1/VCAM-1 mLN stromal cell populations from wild-type (WT) and LtβR^−/− embryos of the indicated ages was performed by MoFlow (DakoCytomation, Carpinteria, CA) cell sorting from disaggregated LN cell suspensions and then snap frozen for RT-PCR.

High-purity cDNA was generated from purified mRNA using µMacs One-Step cDNA synthesis kit, according to the manufacturer’s instructions (Miltenyi Biotec, Auburn, CA). Real-time PCR was performed using IQ SYBR Green Supermix (Bio-Rad, Hercules, CA) with primers specific for β-actin, Ccl21, Ccl19, Cxcl13, Il-7, LtβR, RelB, RankL, TnfrI, and Mmp9. PCRs were conducted in triplicate in 10 µl volumes containing 200 nM primers. After an initial denaturation step (95°C for 10 min), cycling was performed at 95°C for 15 s, 60°C for 20 s, and 72°C for 5 s (45 cycles). Specific amplification was verified by melt curve analysis and also by fractionation of PCR products on a 2% agarose gel that were identified by fragment size (data not shown). mRNA transcript levels of each gene were analyzed using Applied Biosystem’s SDS software (Applied Biosystems, Foster City, CA) by setting thresholds determining the cycle number at which the threshold was reached (C_t). The C_t of the β-actin was subtracted from the C_t of the target gene, and the relative amount was calculated as 2^−ΔC_t. Means of triplicate reactions (multiply 1000-fold) were represented, and data shown are representative of at least two separate cell-sorting experiments. Primer sequences are shown in Supplemental Fig. 1.

Immunofluorescence stainings

For tissue-section stainings, iLNs and mLNs were isolated and embedded in OCT compound (Tissue Tek, Torrance, CA), then frozen in liquid nitrogen. Six-micrometer sections of tissue were cut and fixed in acetone. Abs used were gp38/podoplanin 8.1.1, ICAM-1–FITC clone YN1/1.7.4, PDGFRα clone APA5, CD44-FITC clone IM7, receptor activator for NF-κB ligand (RANKL)-biotin clone IK22/5, MAdCAM-biotin clone MECA-367 (eBioscience), collagen type I (Chemicon International, Temecula, CA), laminin α5 (rat; gift of A. Zapata, Complutense University, Madrid, Spain) perlecan (rat; gift of Z. Lokmic, Bernard O’Brien Institute of Microsurgery, Melbourne, Australia), CCL21 (R&DSystems), Lyve-1 (rabbit polyclonal; Abcam, Cambridge, MA), fibronectin clone FN-3E2 (mouse IgM; Sigma-Aldrich), ER-TR7 supernatant (Biogenesis, Poole, U.K.). CD4 was directly conjugated using the Alexa Fluor 647 mAb Labeling Kit (Invitrogen, Carlsbad,CA). ICAM-1–FITC-conjugated Ab was detected using rabbit anti-FITC (Sigma-Aldrich), then goat anti-rabbit IgG-FITC (Jackson ImmunoResearch Laboratories, West Grove, PA). Anti-PDGFRα was detected using anti-rat IgG-Alexa 594 (Molecular Probes, Eugene, OR). Lyve-1 Ab was detected using goat anti-rabbit IgG-FITC (Jackson ImmunoResearch Laboratories). Collagen type I Ab was detected with goat anti-rabbit IgG-biotin (DakoCytomation). CCL21 Ab was detected using donkey anti-goat FITC (Jackson ImmunoResearch Laboratories). gp38/podoplanin and fibronectin Abs were detected using goat anti-hamster biotin (Cambridge Bioscience, Cambridge, U.K.). Biotinylated Abs were detected using streptavidin-Alexa Fluor 555 or 488 (Molecular Probes). Sections were mounted using Vectashield mounting medium (Vector Laboratories, Burlingame, CA).

Image acquisition and analysis of confocal images

For tissue-section stainings, confocal images were acquired using a Zeiss LSM 510 laser scanning confocal head with a Zeiss Axio Imager Z1 microscope (Zeiss, Oberkochen, Germany). Digital images were recorded in four separately scanned channels with no overlap in detection of emissions from the respective fluorochromes. Confocal micrographs were stored as digital arrays of 2048 × 2048 pixels with 8-bit sensitivity; detectors were routinely set so that intensities in each channel spanned the 0–255 scale optimally.

Inguinal LN anlagen formed by budding endothelial cells surrounded by mesenchymal cells

Over the past few years, the structure of the developing LN anlagen in the mouse embryo had been studied by assessing the recruitment and presence of LTi cells in sections of whole mouse embryos (11, 18–20). However, the iLN anlage forms a very distinct structure that can be dissected as early as E13. Our previous studies have shown that E15 iLNs contain very few CD45⁺ cells and no CD4⁺IL7Rα⁺ LTi cells (12). Therefore, experimental approaches relying on the presence of LTi cells for the identification of LN anlagen do not allow for early detection and analysis of LN primordium. For an in-depth analysis of LN anlage development prior to LTi cell colonization, we examined sections of dissected E13 and E15 iLNs.

At E13, the primitive iLN anlage had a spherical shape formed by a bud of tightly packed gp38/podoplanin⁺ cells surrounded by layers of fibroblast-like cells that expressed the component of the reticular network ER-TR7 (Fig. 1A). We further characterized these two cellular compartments for the expression of endothelial and mesenchymal markers on E15 iLNs sections by immunofluorescence staining and confocal imaging.

The bud of gp38/podoplanin⁺ cells expressed the endothelial markers ICAM-1 and CCL21 and the extracellular matrix proteins collagen type I, and laminin α5 (Fig. 1B, left columns). Perlecan staining was restricted to a thin layer at the interface between the endothelial and the mesenchymal cells, indicating the presence of an intact basement membrane produced by the former.

The fibroblast-like cells that formed the outer layers of the LN primordium expressed several molecules characteristic of mesenchymal cells, such as PDGFRα, the extracellular matrix protein fibronectin, ER-TR7, and the hyaluronic acid receptor CD44 (Fig. 1B, right columns).

Staining of iLN sections demonstrated that ER-TR7⁺ stromal cells and the gp38/podoplanin⁺ lymphatic endothelial cell compartment remained separated up to E16, and then ER-TR7⁺ cells appeared to invade the endothelial core to form the proper internal compartments of the anlage (Fig. 2, Supplemental Fig. 2). Following that process, a large number of mesenchymal cells at E17 expressed both ER-TR7 and gp38/podoplanin and could be distinguished from the single-positive cells (Supplemental Fig. 2).

Based on the above observations, we envision the structure of the E13–15 iLN anlagen comprising two distinct cells types, a bud of lymphatic endothelial cells forming the lymph sacs, surrounded by a basement membrane and layers of mesenchymal cells.

Ongoing remodeling of the iLN anlage structure between E15 and E17

To further define the organ remodeling taking place between E15 and E17 in iLNs, we tracked the changes occurring in the mesenchymal and endothelial compartments by analyzing E17 iLNs, when LTi cells arrived to the anlagen. Immunofluorescence staining of molecules expressed in endothelial and mesenchymal cells were performed in E17 iLN sections and compared with the stainings of these organs at E15.

The endothelial cells that formed a central bud at E15 still form the core of the anlagen at E17 (Fig 2, circled by a white line), but now contained perlecan⁺ laminin α5⁺ structures that form the LN vasculature (Fig. 2A, 2D).

The mesenchymal cells formed the outer structure of the LN anlage, similar to what was observed at E15. However, at E17, mesenchymal cells expressing fibronectin and PDGFRα started to migrate to the endothelial core, as shown in the central region delimitating the endothelium (Fig. 2C–H). CD44 staining (Fig. 2I) highlighted the structure of the mesenchymal compartment as well as hematopoietic cells (round cells) that have entered the LN anlagen. Fibronectin also started to be expressed by cells of the vasculature (Fig. 2C, 2E, arrowheads). Taken together, these observations show the dynamic changes occurring between E15 and E17 in the iLN anlagen involving both the endothelium and the mesenchyme. For the endothelium, the most dramatic change consisted of the remodeling of the whole vasculature and for the mesenchyme of the invasion of the central endothelial compartment. We further characterize the endothelial and mesenchymal cell populations to better understand their contribution to LN anlagen formation.

Concomitant emergence of stromal organizer cells during remodeling of the iLN structure

The LN organizer cells have been previously characterized by their coexpression of ICAM-1 and VCAM-1 on mLNs, and the stainings suggested a differentiation of the organizer cells from mesenchymal cells (18, 21, 22). We assessed by FACS analysis the presence of these cells at different stages of embryonic iLN development. At E15, we were able to clearly identify the mesenchymal and endothelial cell compartments described in Fig. 1: 1) the mesenchymal cell precursors that were negative for ICAM-1 andVCAM-1(I⁻V⁻) and PDGFRα⁺ that correlated with the fibroblast-like cells of the surrounding mesenchyme in Fig. 1B; and 2) the ICAM-1 singlepositive (I⁺V⁻) endothelial cells that also expressed gp38/podoplanin and MAdCAM-1 and were VEGFR3⁻, similar to the central endothelium shown in Fig. 1B (Fig. 3A, top panels). Interestingly, we identified a third cell population that expressed intermediate levels of ICAM-1 and VCAM-1 (I^intV^int) that were PDGFRα⁺ and gp38/podoplanin⁺, suggesting that they derived from the I⁻V⁻ mesenchymal cells. At E17, the mature stromal organizer cells were identified by their high expression of ICAM-1 and VCAM-1 (I^highV^high) as well as MAdCAM-1, a late differentiation marker. Mature organizer cells also expressed PDGFRα and gp38/podoplanin but not VEGFR3, suggesting that their progenitors were the I^intV^int cell population.

The I⁺V⁻ endothelial cell population was negative for both VEGFR-3 and the lymphatic marker Lyve-1 at E15 but became positive for both molecules at E17, indicating an ongoing differentiation process toward lymphatic endothelial cells (Fig. 3A, bottom panels, and data not shown). These results indicate that mesenchymal cells start to mature before the lymphatic endothelium is fully differentiated.

To summarize, vascular/lymphatic endothelial cells could be defined as I^+V− gp38/podoplanin⁺ VEGFR3⁺ PDGFRα⁻, whereas the mesenchymal cell compartment encompassed the I⁻V⁻, I^intV^int, and I^highV^high MAdCAM-1⁺ gp38/podoplanin⁺ PDGFRα⁺ VEGFR3⁻ cells.

Stromal cells of the iLN and mLN followed a similar pattern of development

Blocking of LTβR in pregnant female mice has shown that mLNs are the first to develop in the embryos and are followed by formation of LNs from head to tail (1, 23, 24). Several reports have indicated the different developmental requirements for mLN and iLN formation (25–27). A recent report has also shown differences in the stromal cell subsets in these organs in newborn mice (18), and we wanted to investigate whether those differences were already present in embryos and whether mLN stromal cells expressed similar molecules than their iLN counterparts. Indeed, E15 mLN showed an increased frequency of the I^intV^int cell population, explained by the earlier development of the mLN anlage compared with iLN. However, the same three compartments were still identifiable: 1) mesenchymal cells that were I⁻V⁻; 2) I^intV^int cells; and 3) I⁺V⁻ endothelial cells that also expressed MAdCAM-1 (Fig. 2B, top panel). At E17, the mature stromal organizer cells I^highV^high were present expressing MAdCAM-1, as well as the I⁻V⁻ mesenchymal and the I^intV^int cells. However, the I⁺V⁻ endothelial cells were not clearly identifiable in the mLNs at this stage (Fig. 3B).

Cellular localization and expression of different molecules were assessed by immunofluorescence staining in E18 iLNs and mLNs. Both organs showed the presence of CD4⁺IL-7Rα⁺ LTi cells as well as stromal organizer cells that expressed RANKL and MAdCAM-1 (Fig. 4). Some lymphatic endothelial cells in the developing subcapsular sinus surrounding the E18 mLNs and iLNs coexpressed MAdCAM-1 and Lyve-1, showing no difference in the presence and structure of the lymphatic endothelium between iLN and mLN (Fig. 4, middle panels). Thus, the absence of an I⁺V⁻ endothelial cell population in E17 mLNs compared with iLNs might be due to the stripping of the mLN capsule during dissection of the organs prior to FACS analysis (Fig. 3A, 3B).

Gradient of gene expression of stromal organizer markers during maturation of the mesenchyme

To confirm the sequence of maturation of the mesenchyme, we assessed the mRNA levels of several genes that are characteristic of stromal cells. The small number of cells present in embryonic iLNs prevented a more in-depth analysis. To have a significant number of stromal cells for these assays, CD45⁻ cell populations (I⁻V⁻, I^intV^int, and I^highV^high) were isolated from E18 mLNs, and the gene expression profile was analyzed by quantitative PCR (qPCR).

The expression of most of the genes analyzed was nil in I⁻V⁻ cells with the exception of Tnf-R1 and Ccl21. All CD45⁻ cells expressed LtβR, but only the differentiating I^intV^int and I^highV^high cells expressed RelB. The I^highV^high stromal organizer cells expressed the highest levels of homeostatic chemokines (Ccl21, Ccl19, and Cxcl13) (18), RankL and Il-7, confirming their potential to attract and stimulate LTi cells (Fig. 5). The I^intV^int and I^highV^high cells expressed Mmp9, which has been shown to increase cellular invasiveness and mobility, indicating a role in tissue remodeling by these cell subsets. Finally, we noticed that all the markers studied showed a gradient of expression, with the I^highV^high cells showing the highest levels, followed by I^intV^int and I⁻V⁻ that reflected the stage of maturation of the stroma (Fig. 5).

Absence of LtβR or LTi cells resulted in a block of stromal cell maturation at the I^intV^int stage

LN development is strictly dependent on LTi cells and LTβR signaling, as adult Rorγ^−/−, LtβR^−/−, and Ltα^−/− mice lack all LNs (13, 14, 26, 28, 29). We assessed whether LTβR signaling and LTi cells were required for the initial step of stromal cell maturation or for full progression of these cells to the I^highV^high stage. FACS analysis of stromal cell populations from E15 iLN and mLN anlagen from WT, LtβR^−/−, and Rorγ ^−/− embryos showed the emergence of the I^intV^int cells in all strains (Fig. 6A). However, at E17, the iLNs were difficult to find in both LtβR^−/− and Rorγ^−/− embryos, and what remained had the appearance of fibrous structures (data not shown). FACS analysis of the few iLNs isolated from these mutants confirmed the absence of the I^highV^high mature stromal organizer cells in both strains (Fig. 6A).

We next investigated whether mLNs were similarly affected in their development as iLNs in these mutants. In contrast to the low numbers of iLNs found in E17 LtβR^−/− or Rorγ ^−/− mice, mLNs were present in normal numbers in embryos from these strains. However, FACS analysis of stromal cells revealed a similar block on the progression from I^intV^int to I^highV^high mature organizer cells (Fig. 6A).

Recruitment of CD45⁺ cells to iLNs and mLNs between E15 and E17 was impaired in LtβR^−/− and Rorγ^−/− embryos. For example, in WT iLNs, we observed an 18-fold increase in the recruitment of CD45⁺ cells between E15 and E17 and only a 1.6- and a 6-fold increase in iLNs from LtβR^−/− and Rorγ^−/− embryos, respectively (Fig. 6A).

Our analyses showed that LTβR engagement and the presence of LTi cells were required for the appearance of the I^highV^high mature organizer cells but not for progression of mesenchymal cells from I⁻V⁻ toward the I^intV^int compartment.

LTβR-induced maturation of stromal cells from I^intV^int to I^highV^high

We next tested whether the I^intV^int stromal cells were the precursors of the I^highV^high organizer cells. E14 mLNs and E15 iLNs were stimulated in organ cultures with an agonistic αLTβR Ab and the ICAM-1 and VCAM-1 expression profiles of the CD45⁻ cell populations were assessed by FACS analysis. As shown in Fig. 6B, a significant increase in the frequency of I^highV^high cells was detected. Of note, E14 mLNs and E15 iLNs contained very few if any I^highV^high cells, suggesting that LTβR stimulation induced either those few cells to proliferate or the I^intV^int cells to mature, becoming I^highV^high organizer cells. Cell counts prior to and after LTβR stimulation in organ cultures demonstrated no increase in cell numbers (data not shown). These results strongly suggest that the I^intV^int cell population gives origin to I^highV^high organizer cells and that signaling through the LTβR was necessary and sufficient to induce the maturation.

LTβR is not fully required for the recruitment retention of LTi cells to mLNs

The accumulation of LTi cells in the LN anlagen may result from active recruitment of these cells or proliferation of LTi cells or their precursors. A previous report showed that proliferation of hematopoietic cells was very rare in LN anlagen, indicating that recruitment of LTi cells and their precursors might be the dominant mechanism (19). IL-7Rα and CXCL13 (ligand of CXCR5 expressed by LTi cells) have been shown to play an important role in the recruitment of LTi cells, and IL-7 is thought to be an early localization signal, as late blockade of this cytokine had minor effects on LTi cell migration and LN development (19).

Cell sorting of mLN cells from E15 WT and LtβR^−/− embryos and qPCR were used to assess the effect of LTβR deficiency in stromal cell production of chemokines and survival factors for LTi cells. Cxcl13 and Il-7 were expressed at similar levels by the I^intV^int cells in both WT and LtβR^−/− cells. We also noted that Ccl21 was mainly expressed by the I⁺V⁻ endothelial cells, and its expression was reduced but still present in LtβR^−/− E15 mLN cells (Fig. 7A) (2).

The presence of LTi cells was assessed in E17 iLNs and mLNs from WT and LtβR^−/− embryos. A very low percentage of both CD45⁺ cells and LTi cells was present in LtβR^−/− iLNs when compared with WT (Fig. 7B). Interestingly, whereas the percentage of CD45⁺ cells was reduced in LtβR^−/− compared with WT E17 mLNs, a similar frequency of LTi cells was found in both strains, corroborating previous findings in E16.5 Ltα^−/− embryos (19). Staining of sections of E16 mLNs from WT and LtβR^−/− embryos confirmed the presence of LTi cells in both strains and showed a significant reduction in MAdCAM-1 and undetectable levels of RANKL in the latter (Fig. 7C).

These results showed that although overall colonization by CD45⁺ cells is reduced in the mLNs of LtβR^−/− embryos, LTi cells are still being recruited, suggesting that LTβR signals are not fully required. However, MAdCAM-1 and RANKL levels are severely reduced in these organs, showing their dependence on the LTβR pathway. Based on the results shown above we propose a specific role for LTβR signaling in stromal cells during maturation (Fig. 8, see Discussion).