Ureteric stromal progenitors give rise to kidney inner cortical pericytes via an arterial mural cell intermediate

Abstract

During kidney formation, segmented epithelial tubules and blood vessels develop within a heterogeneous and progressively patterned stroma. By E18.5, the murine renal stroma exhibits several transcriptionally and spatially distinct populations, including specialized stromal cells associated with the vasculature, termed mural cells. However, the precise contributions of stromal progenitor lineages to this stromal heterogeneity, as well as the dynamics of renal mural cell investment, remain unclear. Previous studies have described a subset of stromal progenitors in the ureter that transiently expresses the transcription factor Tbx18 and gives rise to renal stromal cells, including vascular mural cells. Using pulse induction of Tbx18CreERT2 at different timepoints, we elucidate the contribution of this population to stromal patterning. We show that the Tbx18-lineage, when induced at E12.5, gives rise to arterial mural cells, without ever progressing through Foxd1+ cortical stromal progenitor state. These arterial mural cells are only transiently present along arteries during development, ultimately contributing instead to peritubular capillaries. When traced post-natally, the Tbx18-lineage gives rise to pericytes, which are enriched in S3-segment-associated, Cxcl14-enriched stroma in the inner cortex. We show that these pericytes arise directly from arterial mural cells seen earlier during development. These data help clarify a small portion of the complicated lineage relationships of renal stromal progenitors and their contribution to the kidney vascular-associated mural cells.

Introduction

The stromal compartment of many organs is largely understudied. In the kidney, the stroma was thought to function primarily as a supportive matrix for the epithelial and endothelial components of the functional nephron. These stromal cells, as well as the progenitors they arise from, have been shown to be critical inductive components that are essential to kidney development and function.^1,2 It is now clear that renal stromal cells are a highly heterogeneous population that contains numerous cell types, including cells that contribute to the vasculature.

Blood vessels are supported by different types of stromal mural cells.

Blood vessels in most tissues recruit and become ensheathed by stromally derived mural cells during organ and tissue formation. Mural cells consist of both vascular smooth muscle cells (vSMCs), around larger vessels, and pericytes around capillaries.^3–5 We and others recently showed that development of renal vascular mural cells, including the vSMCs that surround renal arteries, are required for proper patterning and branching of arteries.^6–8 Arteriogenesis and mural cell recruitment are tightly coordinated, beginning in the kidney at embryonic day 13.5 (E13.5). As small endothelial cell (EC)-lined vessels remodel into larger vessels (arteries), they recruit mural cells as a key step in arterial maturation. Interestingly, we demonstrated the presence of a heterogeneous and layered peri-arterial mural cell niche during kidney development.⁶ As renal arteries first take shape, they are surrounded by vSMCs that express smooth muscle actin (aSMA) as well as Neural glial antigen 2 (NG2). Embryonically, arteries are also surrounded by outer layers of non-muscle mural cells that express NG2, but not aSMA.⁶ These non-muscle mural cells are found only around arteries, while mural cells around smaller capillary vessels are termed pericytes. These cells have not been described previously in adult kidneys, where arterial mural cells primarily consist of vascular smooth muscle that express both aSMA and NG2.⁹ The origins of organ-specific mural cells, including vSMCs are incredibly diverse, arising not only from different embryonic locations but also different cell types, from mesenchyme to neural crest to hematopoietic lineages.^10–12 The origin of vSMCs in the kidney, as well as their lineage relation to non-muscle mural cells, is unknown.

Renal stromal cells originate from stromal progenitors in the cortex and ureter.

The specific origins of different renal stromal cell types, including vascular mural cells, remain only partially understood. Self-renewing stromal progenitors (SPs), which express the transcription factor Foxd1 and are found in the nephrogenic (NZ) at the periphery of the developing kidney cortex, have been shown to give rise to a majority of the renal stroma, including mural cells.¹³ We recently characterized the molecularly stratified nature of the developing Foxd1-derived renal stroma, forming distinct layers from most cortical (i.e., Foxd1-expressing stromal progenitors) to most medullary.² However, several questions regarding the origin of renal mural cells, especially arterial mural cells, remain.

The renal arterial tree initially takes shape as a branched structure at E13.5.^6,14 The largest first recognizable vessels are the interlobar arteries, which span the distance from the hilum to the corticomedullary junction. This contrasts with the epithelial and stromal components of the kidney, which initiate in the cortex.^6,15 Higher order, distal branching of the arterial tree during development occurs via remodeling of cortical endothelial cells, where they likely recruit mural cells from cortical Foxd1+ progenitors.¹⁶ However, as the initial arterial tree is primarily hilar and medullary, far from the cortex, it is possible that initial arterial mural cells arise from mural cells of the nearby dorsal aorta, or from another non-Foxd1 progenitor lineage.

Indeed, a second population of stromal progenitors is located around the developing ureteric epithelium. These cells, which express the transcription factor Tbx18, have also been shown to contribute the renal stroma, overlapping both spatially and in cell type (including mural cells) with the Foxd1 lineage.^17,18 Although the Foxd1 and Tbx18 lineages both give rise to multiple distinct types of stroma, the overlap in their lineages is unclear. Many studies addressing this issue were performed using constitutively active Foxd1eGFPCre and Tbx18Cre driver lines, and these studies suggested that a subset of Foxd1+ SPs near the hilum may arise from Tbx18+ cells early during nephrogenesis.^19,20 Therefore, the specific contributions of the Foxd1 cortical stromal progenitors that comprise each nephrogenic zone, versus the Tbx18+ ureteric progenitor population, has been difficult to disentangle. Nevertheless, it is `generally accepted in the field that Foxd1+ cortical SPs give rise to the majority of the renal stroma, with Tbx18+ ureteric SPs contributing to some of the medullary stroma and most of the ureteric stroma.¹⁹

Ureteric stromal progenitors may contribute to vascular mural cells of the kidney.

We recently described defects in kidneys in which netrin 1, whose expression is limited to the cortical Foxd1-expressing stromal progenitors, had been deleted.⁶ Although mutants showed impaired vascular smooth muscle recruitment, non-muscle mural cells remained invested around early arteries, suggesting a distinct origin that does not respond to loss of netrin 1.⁶ Based on the location of the maintained mural cells around the medullary arteries, we hypothesized that these cells might come from the Tbx18 lineage.

Here, we test the hypothesis that Tbx18-expressing ureteric SPs may contribute to developing kidney arterial mural cells. We used the inducible Tbx18CreERT2 driver line to address potential contributions in a spatiotemporal manner.²¹ Specifically, we interrogated whether Tbx18+ progenitors contributed to mural cell investment of arteries, which form at E13.5, after the Tbx18 and Foxd1 lineages diverge during early nephrogenesis.¹⁸ We show that Tbx18+ ureteric SPs, present during development, contribute to non-smooth muscle mural cells in the arterial niche during kidney development and pericytes when traced to post-natal stages. These pericytes are spatially restricted to the stroma surrounding the tubules of the inner cortex, corresponding specifically to the S3 segment of the proximal tubule. These results provide new insights into arterial and microvascular mural cell biology, as well as help clarify the contributions of different stromal progenitors found during early nephrogenesis.

Materials and Methods

Mouse Handling

Experimental protocols were approved by the Institutional Animal Care and Use Committee (IACUC) at UT Southwestern. All mouse embryos were used without sex identification. Postnatal mice were differentiated by sex. For embryonic studies (E13.5 - E18.5), timed pregnancies were induced and developmental stage of embryos was verified by various morphological landmarks. P0 is defined as the date of birth for postnatal studies. For postnatal studies, pups were fostered with foster litters aged P2-P3 after dissection at day E18.5-E19.5. For all lineage tracing studies, tamoxifen (dissolved in 90/10 corn oil/ethanol mixture) was given via oral gavage to the dam on embryonic day E10.5, E11.5, or E12.5 at a dose of 0.075 mg/kg. The following mouse lines were used in this study: Tbx18CreERT2 (RRID:IMSR_JAX:031520), Rosa26::LSL-tdTomato (RRID:IMSR_JAX:007914), and Rosa26::LSL-Confetti (RRID:IMSR_JAX:017492).

Immunofluorescence (IF) staining on tissue sections.

After dissection, tissue was fixed in 4% PFA overnight at 4°C. Paraffin embedding, sectioning, and immunofluorescence was performed as previously described.⁶ Briefly, 10μm sections were obtained onto SuperFrost Plus slides (Fisher Scientific). Slides were deparaffinized with xylene and rehydrated through an ethanol gradient. Slides were permeabilized with 0.3% Triton-X 100 in PBS for 10 minutes. Heat induced antigen retrieval in R-Buffer A (Electron Microscopy Sciences) for nuclear antigens or R-Buffer B (Electron Microscopy Sciences) for cytosolic or membrane-bound antigens was performed using a Retriever 2100 (Electron Microscopy Sciences). Slides were blocked in CAS-Block (Invitrogen) for at least one hour and incubated in primary antibody mixture overnight at 4° secondary antibody mixture for 1.5 hours at room temperature. After antibody binding, slides were incubated in a blood lysis solution (10mM CuSO₄, 50mM NH₄Cl, pH 5) for 15 minutes, followed by 5 minutes in de-ionized water. Slides were then mounted in DAPI-Fluoromount G (Southern Biotech) and imaged on Nikon CSU-W1 or Nikon CSU-W1 SoRa confocal microscopes at the Quantitative Light Microscopy Core at UTSW. Images taken of the IF sections were either single plane images or Maximum Intensity Projections (MIP)/Z-projection of the 10μm sections. Antibodies used are summarized in Table S1.

Whole mount immunofluorescence (WMIF) and microscopy

After dissection and fixation, tissue was embedded in 2.5–5% low melting point agarose (Invitrogen). 100μm sections were cut using a Leica vibratome. Floating sections were permeabilized with Triton X-100 in PBS (1% for tissue younger than P4 and 3% for P15 tissue) for 1.5 hours at room temperature. Sections were blocked in CAS-Block for 1 hour, and incubated in primary antibody mixture overnight at 4° with gentle nutation. Sections were washed in PBS (5× 1 hour), and were incubated in secondary antibody mixture overnight at 4°. Sections were washed in PBS (5× 30 min). Sections were placed on Superfrost Plus glass slides and cleared in a bubble of Rapiclear 1.47 (Cedarlane Labs) for at least 10 minutes. Slides were mounted in Rapiclear and imaged as in previous sections.

All samples were imaged using a Nikon CSU-W1 dual-camera inverted spinning-disc confocal microscope. Max intensity projection (Z. proj, or Z-projection) was used to compress multiple visual planes, or single visual planes were used for analysis.

RNAScope

Tissue preparation:

Kidneys were dissected as above and flash frozen in OCT (Tissue Tex). 10μm cryosections were obtained using a Leica cryostat on SuperFrost Plus glass slides (Fisher Scientific).

RNAscope:

Hiplex experiments were carried out using manufacturer protocol using the RNAScope Hiplex v2 kit (Advanced Cell Diagnostics). Sections were pretreated with ProIII for or 30 min at room temperature. Probe hybridization, amplification and fluorophore conjugation steps were carried out per manufacturer protocol. Slides were mounted in Prolong Gold mounting media (Invitrogen) and confocal imaging was performed as in the previous section. Fluorophore cleaving and re-conjugation was performed per manufacturer instruction. Immunofluorescence was performed after the final cleaving step as in the previous section, beginning with Triton X-100 permeabilization. Probes are tabulated in Table S1.

Statistical Analyses and visualizations

All data were plotted and graphs were generated in PRISM 10. No statistical methods were used to predetermine sample sizes. All embryonic experiments were carried out with at least n=3 embryos from two separate litters. Number of postnatal mice analyzed are as follows (out of 5 total litters): P3 – 3 mice (unknown sex, 2 litters); P6 – 1 mouse (unknown sex, one litter); P15 – 3 mice (2 female, 1 male, 2 litters); adult ages 2–7mo – 4 mice (3 male, 1 female, 2 litters). Due to different tissue processing requirements, not all tissue samples were analyzed for each experiment (RNAscope, WMIF, IF). No samples were excluded from analysis. Linear manipulations of images were carried out using ImageJ. Figures and models were made using Microsoft Powerpoint, BioRender.com and Adobe Illustrator.

Results

Tbx18 and Foxd1 stromal lineages in the kidney

To investigate the contribution of subsets of Tbx18-inducibleStromalLineage cells (Tbx18^iSL) to the early renal stroma, including vascular mural cells, we used lineage tracing experiments using the inducible Tbx18CreERT2 and the tdTomato reporter across different stages of kidney formation (Fig. 1A). We induced reporter expression at E10.5 using tamoxifen (tmx), termed here Tbx18^E10.5iSL, to assess stromal contribution at early nephrogenic stages. At E13.5, we observed arterial mural cells marked by NG2 (Neuron-glial antigen 2, also known as chondroitin-sulfate proteoglycan 4 or CSPG4) and associating with vessels, that were labeled by Tbx18^E10.5iSL cells, suggesting that the Tbx18+ SPs contribute to arterial mural cells (Fig. 1B). However, we also found that a significant portion of the cortical renal stroma was labeled, most likely due to the overlap with Foxd1+ SPs (Fig. 1B, arrow). Interestingly, Tbx18^E10.5iSL cells were more abundant on the cranial, adrenal-facing pole of the developing kidney (green asterisk) than on the posterior pole of the kidney (white asterisk), suggesting an asymmetry in Tbx18^E10.5iSL contribution along the antero-posterior axis of the kidney.

Figure 1. Tbx18-lineage overlaps with Foxd1-lineage stromal progenitors and derivatives when induced early in nephrogenesis.

(A) Schematic of Tbx18CreERT2; tdTomato lineage tracing experiments (Tbx18-tdT). Red arrowhead: timed tamoxifen (tmx) induction. Pink arrowhead: harvest of lineage traced kidneys. (B) Whole mount immunofluorescence (WMIF) of E13.5 Tbx18-tdT^E10.5iSPL kidney for NG2 and tdTomato. Green, white asterisks: cranial and caudal poles, respectively. White arrow: cortical tdTom+ cells. (B’) 20μm sub-stack of (B), focusing on one interlobar artery. (B”) NG2 only of (B’). (B”’) tdTomato only of (B’). (C) Schematic of overlap between Foxd1+ and Tbx18+ SPs in Tbx18-tdT^E10.5iSPL timed tracings (top), but not Tbx18-tdT^E12.5iSPL (bottom). Green, pink outlines: Cells expressing Foxd1 or Tbx18 respectively (stromal progenitors, SPs). (D) Immunofluorescent staining (IF) of E14.5 Tbx18-tdT^E10.5iSPL kidney section for tdTomato and FOXD1. (D’) Zoom view of nephrogenic zone (NZ). Dotted line: border of FOXD1+ stromal progenitors. (D”) FOXD1 only. (E) IF of E13.5 Tbx18-tdT^E11.5iSPL kidney section stained for tdTomato and FOXD1. (E’) Zoom view of NZ. (E”) FOXD1 only. (F) IF of E18.5 Tbx18-tdT^E11.5iSPL kidney section for tdTomato. (F’-F”) Zoom view of cortical areas. Scale bars: 100μm (B, D, E), 500μm (F). Number of embryos analyzed, represented by images: >3 from >2 litters (B, D, E, F).

Previous studies suggest that Tbx18 expression at E10.5 in this region precedes Foxd1 expression between E10.5 and E11.5, and thus that a portion of Foxd1+ SPs arise from an early population of Tbx18+ progenitors (Fig. 1C, model).¹⁸ It is therefore likely that the significant cortical stromal labeling might be due to co-labeling of Foxd1+ SPs in the NZ (and thus their derivatives). Indeed, Tbx18^E10.5iSL cells found in the NZ and expressed FOXD1 by IF at E14.5, especially in hilar (medial) and cranial regions of the kidney, (Fig. 1D’-D”, white dotted line), suggesting that some Foxd1+ SPs arise from cells that express Tbx18 at E10.5. Tbx18^E10.5iSL cells found within the kidney that did not express FOXD1 represented their derivatives (Fig. 1D’). Because this lineage also includes FOXD1+ cells throughout development, we could not conclude whether Tbx18^E10.5iSL cells labeled arose specifically from Tbx18+ SPs separate from Foxd1+ SPs.

We hypothesized that a population of Tbx18+ SPs in the ureteric mesenchyme might contribute to the mural cells independent of Foxd1+ contributions, and that inducing labeling after the lineages split would allow us to specifically identify these cells. When induced at E11.5 (Tbx18^E11.5iSL), the Tbx18 lineage still labels significant patches of cortical stroma between E13.5 and E18.5, including FOXD1+ SPs (Fig. 1A, E-F). We did note that the labeled cells were less heavily enriched on the cranial pole of the kidney. When we induced lineage tracing at E12.5 (Tbx18^E12.5iSL, Fig. 2A), by contrast, the Tbx18^E12.5iSL labeled significantly fewer FOXD1+ SPs in the NZ nearest the hilum at E13.5 (Fig. 2B-B”’, 2C). Thus, Tbx18^E12.5iSL were derived primarily from ureteric, Tbx18+ SPs, without co-labeling Foxd1+ SPs. Of note, Tbx18-lineage cells were found in FOXD1-negative cells surrounding the developing kidney (Fig. 2B’, white asterisks), suggesting that Tbx18 may migrate peripherally around the kidney. Altogether, however, the average number of tdTom/FOXD1 dual-positive cells were much reduced when induction was carried out a day later (Fig. 2C).

Figure 2. Tbx18-lineage overlaps significantly less with FOXD1+ stromal progenitors when induced at E12.5.

(A) Schematic of Tbx18CreERT2; tdTomato lineage tracing experiment with induction at E12.5 (Tbx18-tdT^E12.5iSPL). Red arrowhead: timed tamoxifen (tmx) induction. Pink arrowhead: harvest of lineage traced kidneys. (B) IF of E13.5 Tbx18CreERT2; tdTomato (Tbx18-tdT^E12.5iSPL) kidney section for tdTomato and FOXD1. (B’) tdTomato only. Asterisks: tdTomato+ cells outside of the developing kidney (FOXD1 negative). (B”) Zoom view of (B), showing a single NZ unit. (B”’) FOXD1 IF and DAPI only of (B”). UB: ureteric bud. NPC: nephron progenitor cell. SP: stromal progenitor. (C) Average number of tdTomato+/FOXD1 dual-positive stromal progenitors per nephrogenic zone in E13.5 kidney sections of Tbx18-tdT^E12.5iSPL versus Tbx18-tdT^E10.5iSPL or Tbx18-tdT^E11.5iSPL kidneys. n=5 sections from 3 embryos (E12.5), 7 sections from 4 embryos (E11.5/E10.5), p=0.0282. Scale bars: 50μm (B). Number of embryos analyzed, represented by images: >3 from >2 litters (B).

Tbx18 lineage traced cells associate with renal arteries embryonically

We next examined the contribution of Tbx18^E12.5iSL cells to the renal stroma using IF by analyzing tdTomato positive cells throughout the developing kidney. We found that Tbx18^E12.5iSL primarily contributed to the medullary stroma at E13.5 as expected (Fig. 3A-C). However, in addition to stroma surrounding the ureteric and medullary region (Fig. 3B, approximately outlined in red), a significant portion of labeled cells were found in periarterial niches (cells closely associating with vessels), visualized by co-staining tdTom and NG2, and localization along blood vessels (Fig. 3B-C). IF for tdTomato and NG2 showed that many Tbx18^E12.5iSL cells were arterial mural cells, expressing variable levels of NG2 (Fig. 3C-C””’, orange arrowheads: high NG2; cyan arrowheads: low NG2). We have previously described the expression patterns of NG2 and smooth muscle marker alpha smooth muscle actin (aSMA) in the developing renal arteries, showing that vSMCs (which co-express aSMA and NG2) are surrounded by non-muscle mural cells (that do not express aSMA) during development.⁶ The lineage relationship between these cells is not understood. Here, we found that most Tbx18^E12.5iSL cells at E13.5 did not express aSMA and were located primarily within outer layers of the non-muscle mural cells (Fig. 3C”’-C””). In addition, we found that Tbx18^E12.5iSL cells were only proliferative in regions around the ureter, but not the arteries or kidney cortex (data now shown). Together, these results suggest that outer, non-smooth muscle mural cells that surround early renal arteries derive from Tbx18-expressing stromal progenitors initially localized in the hilum.

Figure 3. Tbx18-tdT^E12.5iSPL (E12.5 induced stromal progenitor lineage) contributes to arterial non-muscle mural cells during nephrogenesis.

(A) WMIF of E13.5 Tbx18-tdT^E12.5iSPL kidney for NG2 and tdTomato. ia: interlobar arteries. (B) IF of E13.5 Tbx18-tdT^E12.5iSPL kidney section for NG2, aSMA and tdTomato. Dotted red line: medullary/ureteric region. (B’) Zoom of arterial mural cells showing NG2 and tdTomato IF only. (B”) Zoom of arterial mural cells with aSMA and tdTomato only. (C) IF as in (B), showing glancing section with interlobar artery (ia) visible. (C’) NG2 only of zoom of (C). White dotted line: outline of aSMA staining in (C”). Arrowheads: NG2-high (orange) and NG2-low (cyan) tdTomato+ arterial mural cells. (C”) aSMA only. (C”’) NG2 and tdTomato only. (C””) tdTomato and aSMA only. (C””’) tdTomato only. Scale bars: 100μm (A-C). Number of embryos analyzed, represented by images: >3 from >2 litters (A-C).

Contribution of Tbx18 lineage traced cells to newly forming arterioles at E18.5

Previous work has shown that NG2-only arterial mural cells only exist during development; mature arteries are only invested by NG2/aSMA dual positive vascular smooth muscle.^6,8,22 To determine whether Tbx18^E12.5iSL NG2-only arterial mural cells perdured in the periarterial niche, we performed longer term lineage tracing, labeling at E12.5 as above and assessing tdTomato+ cell localization along major arteries at E15.5 and E18.5.

At E15.5, we observed continued presence of Tbx18^E12.5iSL cells surrounding the interlobar arteries (ia, main renal arteries) (Fig. 4A-A’, B.i).¹⁵ We noted that these cells expressed low levels of NG2 and no aSMA (Fig. 4B.i-B.i’). Intriguingly, at this stage, Tbx18^E12.5iSL non-muscle mural cells were also found on smaller branches of interlobar arteries, which begin forming around E14.5 (Fig. 4B.ii-B.ii’).¹⁴ When traced to E18.5, Tbx18^E12.5SL cells were rarely associated with interlobar arteries (Fig. 5A-A’, 5A.i-A.i”). Instead, Tbx18^E12.5iSL non-muscle mural cells surrounded distal arterioles, where they continued to express NG2, but not aSMA (Fig. 5B-B”).

Figure 4. Tbx18-tdT^E12.5iSPL (E12.5 induced stromal progenitor lineage) contributes to arterial non-muscle mural cells of smaller arteries at E15.5.

(A) IF of E15.5 Tbx18-tdT^E12.5iSPL kidney section for tdTomato, NG2, and aSMA. ia: interlobar artery; art, arteriole. (A’) tdTomato and aSMA only. (B.i) Zoom of proximal interlobar artery, showing aSMA and tdTomato only. (B.i’) Zoom of proximal portion of interlobar artery, showing NG2 and tdTomato only. (B.ii) Zoom of distal portion of interlobar artery with smaller branching artery (arteriole), showing aSMA and tdTomato only. (B.ii’) Zoom of interlobar artery, showing NG2 and tdTomato only. White dotted lines: outline of aSMA signal. (B.i”, B.i”’, B.i”” and B.ii”, B.ii”’, B.ii””) NG2, aSMA and tdTomato channels only, respectively. Scale bars: 100μm (A). Number of embryos analyzed, represented by images: >3 from >2 litters (A).

Figure 5. Tbx18-tdT^E12.5iSPL non-muscle mural cells transiently occupy the peri-arterial niche during development.

(A) IF of E18.5 Tbx18-tdT^E12.5iSPL kidney section for tdTomato, NG2, and aSMA. (A’) tdTomato and aSMA only. (A.i) Zoom of proximal portion of interlobar artery, showing NG2 and tdTomato only. (A.i’) aSMA and tdTomato only. (A.i”) aSMA only. (B) IF as in (A), showing distal arterioles stained for NG2 and tdTomato only. (B’) aSMA and tdTomato only. (B”) aSMA only. (C) WMIF of P15 Tbx18-tdT^E12.5iSPL kidney for NG2, tdTomato and aSMA, showing interlobar artery (ia). Inset: distal arteriole (art) with tdTomato+ vascular smooth muscle cells. (C’-C”) Single z planes of (C), showing NG2 only. (C”’-C””) Single z planes of (C), showing aSMA only. (D) WMIF as in (C), showing smaller arteries. (E) Quantification of tdTomato+ mural cells per length of interlobar artery in E15.5 Tbx18-tdT^E12.5iSPL versus E18.5-P15 kidney sections. n=13 sections from 3 embryos from 2 litters (E15.5), n=9 sections from 4 embryos from 3 litters (E18.5-P15). Scale bars: 100μm (A, D), 200μm (C). Number of embryos analyzed, represented by images: >3 from >2 litters (A-B), 2 pups from 1 litter (C-D).

Together, these results suggest that Tbx18^E12i.5SL NG2-only arterial mural cells found around interlobar arteries at E13.5 do not remain invested along the same large arteries as development continues. Instead, Tbx18^E12.5iSL cells are observed invested along progressively smaller arteries over the course of development, where they continue to express NG2-only and not vascular smooth muscle marker aSMA.

The Tbx18 lineage contributes to inner cortical pericytes at postnatal stages

We next assessed the postnatal fate of Tbx18-lineage traced cells, including those that contributed to mural cells. We labeled cells at E12.5 and followed them until after birth, after fostering tdTomato+ pups at E18.5 (as tmx crosses the placental barrier and causes uterine dysfunction).²³ Similar to patterns observed at E18.5, we found that Tbx18^E12.5iSL cells were rare around mature interlobar arteries at P15, and that postnatal arteries did not exhibit associated tdTomato+/NG2+/aSMA- non-muscle mural cells, as they did in the earlier embryo (Fig. 5C-C””). Occasionally, aSMA+ Tbx18^E12.5iSL smooth muscle cells were observed around smaller arteries, indicating that a subset of the Tbx18^E12.5iSL lineage contributed to vascular smooth muscle (Fig. 5C””’, inset). However, these were rare and most large vessels examined at P15 showed little association with tdTomato+ cells (Fig. 5D). We quantified the average number of tdTomato+ arterial mural cells along interlobar arteries at E15.5 compared to multiple postnatal stages until P15. Interlobar arteries in E15.5 Tbx18^E12.5iSL kidneys had significantly more tdTomato+ arterial mural cells than later stages in development, suggesting that tdTomato+ cells only transiently occupy the interlobar arterial niche (Fig. 5E).

We then traced to adulthood and found that Tbx18^E12.5iSL cells were instead notably located in clusters along the outer medullary stripe and inner cortex, immediately deep to the arcuate artery/vein complex (Fig. 6A, B, arcuate veins – white asterisks, identified by large irregularly shaped vascular lumens in section of cortical areas).¹⁵ Labeled cells were infrequent in more cortical stroma. Outer medullary/inner cortical clusters contained Tbx18^E12.5iSL within the interstitium of proximal tubules labeled by LTL (Lotus tetragonolobus lectin), as well as dense clusters located just medullary to the LTL immunostaining border (Fig. 6B, LTL/medulla border indicated by a dotted line). Notably, Tbx18^E12.5iSL within the proximal tubule interstitium surrounded peritubular capillaries (PTCs) with fine cellular projections fully enwrapping microvascular lumens (Fig. 6C, 6C.i-C.i””). This cellular morphology consisted of long cellular filopodia and close physical association with EMCN+ small vessels (Fig. 6C.i-C.i””), suggesting these Tbx18-derived cells could represent a highly regionalized subset of pericytes. By contrast, Tbx18^E12.5iSL in the outer medulla were unassociated with EMCN+ capillaries (Fig. 6D).

Figure 6. Tbx18-tdT^E12.5iSPL cells give rise to peri-microvascular cells within the inner cortex.

(A) WMIF of adult (5-month-old female) Tbx18-tdT^E12.5iSPL kidney section for tdTomato and LTL. Asterisks: location of arcuate veins. (B) Zoom view of tdTomato only, showing corticomedullary junction. Dotted line: border of LTL staining (not shown to accentuate tdTom+ cells). (C) Zoom of inner cortical region around LTL staining, showing EMCN. (C.i) Zoom of single perivascular cell. (C.i’-C.i””) single z planes of (C.i). ortho view: Orthogonal view of (C.i). (D) Zoom of outer medullar region deep to LTL staining, showing EMCN (single z plane). Red arrowheads: tdTomato+ cells not associated with EMCN+ capillaries. Scale bars: 2000μm (A), 10μm (D). Number of embryos analyzed, represented by images: 2 mice from 1 litter (A-D).

The cortical stroma, including proximal tubule stroma, is believed to arise primarily from Foxd1+ SPs.¹³ Here, however, we show the striking localization of Tbx18^E12.5iSL cells, which contribute to the inner cortical stroma in adult kidneys (Fig. 6). We traced Tbx18^E12.5iSL cells to other perinatal and postnatal stages starting at E18.5 and fostering pups, due to the tamoxifen induction at E12.5 (Fig. 7A) in order to assess the consistency of this localization. Sagittal sections (defined as containing the long axis of the kidney and with both cortex and medulla present) taken from kidneys at E18.5, P3, P6 and P15 exhibited tdTomato+ cells localized to the corticomedullary junction, similar to adult kidneys (Fig. 7B-E). Many of these cells associated closely with capillary endothelial cells, resembling pericytes as in adult kidneys (Fig. 7F-G, 7F’-F””, 7G’-G””).

Figure 7. Tbx18-tdT^E12.5iSPL cells are localized to the corticomedullary junction through perinatal and postnatal stages.

(A) Schematic of Tbx18CreERT2; tdTomato lineage tracing experiment with induction at E12.5 (Tbx18-tdT^E12.5iSPL). Red arrowhead: timed tamoxifen (tmx) induction. Pink arrowhead: harvest of lineage traced kidneys. (B-D) IF of E18.5, P3, and P6 Tbx18-tdT^E12.5iSPL kidneys for tdTomato and aSMA. (E) WMIF of 100μm vibratome section of P15 Tbx18-tdT^E12.5iSPL kidney for tdTomato and NG2. White dotted line: edge of kidney. (F-G) 20μm Z projections of zoomed areas from (E) along corticomedullary axis with WMIF for EMCN and tdTomato. (F’-F””, G’-G””) single z planes of (F) and (G), showing individual tdTomato cells resembling pericytes enwrapping capillary vessels Scale bars: 100μm (B, C, D, F, G, 500μm (E). Number of mice analyzed, represented by images: >3 embryos from >2 litters (B), 2 mice from 2 litters (C, E), 1 mouse from 1 litter (F).

To assess whether Tbx18^E12.5iSL cells contributed to specific regionalized subsets of stroma, we used RNAscope to delineate kidney regions, by highlighting regionalized tubules and stroma. We performed RNAscope for Slc22a7, a marker of the S3 segment of the proximal tubule (found in the inner cortex), Cxcl14, a putative marker of outer medullary stroma identified in E18.5 single cell sequencing data (which we recently showed to be enriched in the cortex), and tdTomato on adult lineage traced kidneys (Fig. 8A-C).^2,24 We show that Cxcl14 is strongly expressed in both the outer medullary stroma, as well as the inner cortical stroma of adult kidneys (Fig. 8C). To identify Tbx18^E12.5iSL cells, we imaged at high resolution for cells enriched in tdTomato transcripts (Fig. 8D, yellow circles). Quantification of high resolution IF images showed that the majority of Tbx18^E12.5iSL cells associated with the S3 segment (bordering at least one Slc22a7+ tubule) (Fig 8D’, annotated numbers indicate segments contacting tdTomato+ cells represented by yellow circles, Fig. 8E-H showing zoomed images with one tdTomato+ cell, 8I). This demonstrated that embryonic Tbx18+ lineage traced cells invest around PTCs associated with the S3 segment within the tubules of the inner cortex. The majority of Tbx18^E12.5SL cells also expressed Cxcl14 (Fig. 8D”, annotations indicate relative levels of Cxcl14 expressed in tdTomato+ cells represented by yellow circles Fig. 8E”-H” showing zoomed images with one tdTomato+ cell, 8J,).

Figure 8. Tbx18-tdT^E12.5iSPL cells specifically contribute to S3 proximal tubule-associated stroma, which is enriched for Cxcl14 expression.

(A) IF for LTL and RNAscope for Slc22a7 transcripts on adult Tbx18-tdT^E12.5iSPL kidney section. (B) Same as (A), with localization of tdTomato+ nuclei highlighted by yellow circles. (C) RNAscope for Slc22a7 and Cxcl14 on section from (A). (D) Zoom view, showing tdTomato only. Yellow circles: tdTomato+ foci. (D’) LTL and Slc22a7 only. Yellow circles as in (D), annotated with number of Slc22a7+ tubules bordered. (D”) DAPI and Cxcl14 only. Yellow circles annotated with relative Cxcl14 expression levels. (E-H) Zoom views of RNAscope for Slc22a7 and tdTomato, showing single tdTomato+ foci. (E’-H’) tdTomato only. (E”-H”) Cxcl14 RNAscope on zoom views from (E-H). (I) Quantification of number of tdTomato+ cells associated with one or multiple Slc22a7+ S3 proximal tubule segments. (J) Quantification of number of tdTomato+ cells co-expressing Cxcl14. n=73 cells from 2 mice. Scale bars: 500μm (A-C). Number of mice analyzed, represented by images: 2 mice from 1 litter (A-H).

To determine whether Tbx18^E12.5iSL cells associated with capillaries were pericytes, we performed IF for PDGFRβ and PDGFRα, broad markers of mural cells and fibroblasts, respectively.^25,26 Pericytes are often identified by their tight association with microvascular capillaries and PDGFRβ expression, as pericyte-specific markers are rare and often not universal.^3,27 We found that Tbx18^E12.5iSL cells surrounding LTL-associated PTCs displayed strong PDGFRβ expression at P6, suggesting that they were indeed pericytes (Fig. 9A-B, white arrowheads). By contrast, tdTomato+ cells found medullary to LTL immunostaining did not make direct contact with endothelial cells as seen in adult kidneys and expressed less PDGFRβ (Fig. 9C). Additionally, these cells expressed high PDGFRα, suggesting they were fibroblasts, while pericytes in the inner cortex expressed very low levels of PDGFRα (Fig. 9D-F). Across all postnatal timepoints evaluated (E18.5-adult), an average of 54.9% of Tbx18^E12.5SL cells along the corticomedullary junction were localized around tubules and resembled pericytes on sagittal sections (7 sections from 7 mice across 4 stages, standard deviation 18.5%). These labeling patterns (summarized in Fig. 9G) were consistent across the corticomedullary junction.

Figure 9. Tbx18-tdT^E12.5iSPL cells become pericytes of the inner cortical microvasculature.

(A) IF of P6 Tbx18-tdT^E12.5iSPL kidney section for PDGFRb, LTL, EMCN, and tdTomato. (B) Zoom of inner cortex with LTL+ proximal tube. Arrowheads: tdTomato+/PDGFRb+ pericytes. Inset: zoom of single tdTomato+ cell (red-outlined arrowhead). (C) Zoom of outer medulla, deep to LTL+ proximal tube. (B’, C’) tdTomato and EMCN only. (B”, C”) PDGFRb and EMCN only. (D) IF of P6 Tbx18-tdT^E12.5iSPL kidney section for PDGFRa, LTL, EMCN, and tdTomato. (E) Zoom of inner cortex in D, with LTL+ proximal tubules. (F) Zoom of outer medulla in D, deep to LTL+ proximal tubules. (E’, F’) tdTom ato and EMCN only. (E”, F”) PDGFRa and EMCN only. Scale bars: 100μm (A, D), 20μm (B-C, E-F). Number of mice analyzed, represented by images: 4 mice from 2 litters (including P3, P6, and P15 stages) (A-F).

Inner cortical pericytes may derive from arterial non-muscle mural cells.

We speculated that mural cell lineages within the Tbx18^E12.5iSL at developmental and postnatal stages (arterial mural and pericyte, respectively) represented the same lineage over time. To test this hypothesis, we assessed Tbx18^E12.5iSL cells at E18.5 and P3 and looked near small arteries for evidence of a transition from arterial mural to pericyte identity. Indeed, we observed Tbx18^E12.5iSL cells extending cellular projections between NG2+ arteriolar mural cells and nearby capillary endothelium at both timepoints (Fig. 10A-B, arrowheads).

Figure 10. Tbx18-tdT^E12.5iSPL arterial mural cells leave the peri-arterial niche to become inner cortical pericytes.

(A) WMIF of P3 Tbx18-tdT^E12.5iSPL kidney showing NG2 and tdTomato. ia: interlobar artery. (A’) Single z plane of (A) with EMCN IF. (B)

The clustering of Tbx18^E12.5iSL cells within the inner cortical interstitium suggested a clonal expansion. To evaluate a possible lineage relationship between cells along arteries and those associated with capillaries at higher resolution, we assessed clones of Tbx18 lineage traced cells using the Confetti reporter and WMIF for GFP (an antibody which also binds to both YFP and CFP) and tdTomato (Fig. 10C). When traced from E12.5 to E18.5, we identified multicellular clones of Tbx18^E12.5iSL cells in the inner cortex (Fig. 10D-E). Notably, we found clones of labeled cells either in the inner cortex (associated with LTL, Fig. 10D) or the outer medulla (not associated with LTL, Fig. 10E), but not both (data not shown). These distinct clonal populations correspond to the locations of the pericytes and fibroblasts that we observed post-natally in Figs. 6, 8, and 9. In addition, clones were observed containing both arterial mural and capillary associated cells, suggesting that they share a common lineage of origin (Fig. 10F-F”’). These observations suggest that the Tbx18^E12.5iSL cells associated with LTL in the inner cortex do not arise from the same SPs as the Tbx18^E12.5iSL cells found in the outer medulla. In addition, they suggest that the Tbx18^E12.5iSL cells associated with capillaries and those associated with arterial mural cells arise from a common progenitor.

Together, these results demonstrate a previously unknown and novel lineage of stromal cells that arises from Tbx18+ stromal progenitors found surrounding the ureteric epithelium which contribute to transient arterial mural cells during development, as well as pericytes invested in the capillary vasculature of the inner cortex post-natally (Fig. 10G-I).

Discussion

In this study, we describe the lineage of a previously unknown population of renal mural cells arising from Tbx18+ stromal progenitors (SPs) found around the developing ureter. During early nephrogenesis, a subset of Tbx18-lineage cells contributes to non-smooth muscle mural cells in the peri-arterial niche, independent of Foxd1+ SPs found in the developing cortex. These cells transiently populate the arterial niche, initially only of proximal interlobar arteries and subsequently of distal arterioles as development continues. Intriguingly, when traced to post-natal stages, these Tbx18-lineage arterial mural cells give rise to pericytes that surround the microvascular peritubular capillaries (PTCs) of the inner cortex of the kidney. These results provide previously unknown insights into the patterning and origin of a subset of the adult kidney stroma, showing that Tbx18+ ureteric SPs contribute to the stroma of the inner cortex, a region presumed to be populated entirely by Foxd1+ SPs.

We recently characterized heterogeneous mural cell investment of renal arteries during development, demonstrating the presence of vascular smooth muscle cells (vSMCs, which express both aSMA and NG2) as well as non-muscle mural cells (which do not express aSMA and are found in outer layers of the niche) in the periarterial niche.⁶ Here, we show that a significant number of non-muscle mural cells of the initial arterial tree are derived from Tbx18+ stromal progenitors around the ureter.

We utilize the recently characterized inducible Tbx18CreERT2 to disentangle whether specific populations of mural cells arise from this population, while excluding overlap with Foxd1+ SPs seen in constitutive lineage tracing.²¹ By inducing reporter expression at different timepoints, we confirmed that overlap of the Tbx18 lineage with Foxd1+ SPs and their descendants occurred when induced at E10.5 and E11.5, but not at E12.5 or beyond (data not shown for later inductions, as Tbx18 expression becomes further restricted to ureteric smooth muscle).¹⁷ We note with interest that when labeled earlier, Tbx18-lineage reporter expression displays a spatial preference for the adrenal side of the kidney suggesting differential timing in Tbx18/Foxd1 expression along the cranial/caudal axis. Induction at E12.5 does not result in labeling of FOXD1+ cortical stromal progenitors, indicating that tracing from this timepoint labels stromal derivatives of ureteric progenitors specifically.

Notably, we expect that the derivatives of these ureteric progenitors likely still express Foxd1 at some point in development, as constitutive lineage tracing using the Foxd1eGFPCre shows that all renal stromal cells, including the arterial mural cells and pericytes we describe here, derive from a Foxd1-expressing lineage.¹³ By using an inducible Tbx18-driven reporter, whose expression appears to be more spatially restricted during development, we identify one population of cells that we believe to be derived from ureteric progenitors specifically, however, whether other pericytes, mural cells, and stromal cells in general are derived from ureteric progenitors is unknown. Definitive conclusions regarding the temporal dynamics of stromal contributions before E12.5 would require more sophisticated methods, such as dual Cre/Dre recombination strategies, to fully resolve the Foxd1 and Tbx18 lineages earlier in development.²⁸

Ureteric Tbx18 progenitors were previously thought to primarily contribute to stroma associated with the ureteric-derived collecting duct system.¹⁹ Here, we demonstrate novel contribution and dynamics of Tbx18 ureteric progenitors, that give rise to a regionalized subset of renal vascular mural cells throughout development. We show that early during arteriogenesis, outer layers of non-muscle mural cells are derived from Tbx18+ progenitors. In adult kidneys, Tbx18-lineage non-muscle mural cells are largely absent from arteries. Thus, the Tbx18-lineage in the peri-arterial niche is a strikingly transient cell state. In later stages of development (E15.5 and E18.5), we note that Tbx18-lineage cells appear on smaller branches of the arterial tree, suggesting an intriguing possibility that cells originally located on large arteries at E13.5 migrate along the arterial tree as it remodels, making their way to smaller vessels. However, we cannot formally exclude the possibility that initially labeled cells die, or that distal arterial mural cells are recruited from nearby labeled cells rather than migrating from larger arteries. Further studies into the migratory capacity of Tbx18-lineage cells are necessary to support this hypothesis.

Our study demonstrates that the final Tbx18-lineage in adulthood includes pericytes that surround PTCs in the inner cortex. The contribution of ureteric stromal progenitors to the proximal tubule stroma of this region is highly novel, as previously it has been assumed that all cortical renal stroma, including vascular mural cells, derives from the Foxd1+ progenitors.^29,30 The pericyte fate of Tbx18-lineage is not entirely unexpected as Tbx18-lineages have been reported to contribute to pericytes in multiple organs.²¹ Importantly, previous adult lineage tracing of the Tbx18CreERT2 (induced post-natally) did not track labeling in the kidney, as Tbx18 expression was absent from the kidney proper at this stage and limited to differentiated ureteric smooth muscle.²¹ Thus, the contribution to renal stroma shown here represents a long-lasting embryonic lineage arising from ureteric stromal progenitors found only in development.

What proportion of pericytes found in the adult kidney derive from ureteric progenitors is of interest, but is difficult to assess with current tools, especially due to overlap of Foxd1 and Tbx18 expression patterns. Importantly, identification of an expanded toolkit of useful markers for pericytes, currently starkly unavailable, will be needed to further untangle pericyte origins and heterogeneity in the growing and mature kidney. To date, less than a handful of non-specific markers combined with cellular anatomy and association with vessels are the only method to conclusively identify a pericyte.²⁷ In addition to pericytes in the inner cortex, we demonstrate that some Tbx18 lineage cells become PDGFRa+ fibroblasts found in the outer medulla. Notably, we see thin, elongated cells in the medullary region of the kidney labeled as early as E13.5, suggesting that these could be precursors to a regionalized subset of fibroblasts.

Our data supports a model whereby transient, arterial non-muscle mural cells migrate from proximal to distal arteries, and finally to PTCs during development, giving rise directly to inner cortical pericytes. Using fixed snapshots of Tbx18-lineage cells, we show that some mural cells make contact with both arterial mural cells as well as capillary endothelium. In addition, we observe clones of Tbx18-lineage cells containing both arterial-associated and capillary-associated cells. Together, these experiments provide strong evidence that non-muscle mural cells comprise a progenitor population with restricted potential. Given these snapshots, and the absence of E12.5-pulse lineage traced cells on the larger vessels of late embryonic or postnatal vessels, we speculate that this population contributes directly to pericytes of the inner cortex by ‘jumping’ off of the distal arterial niche into nearby developing capillary architecture. This is a highly novel model of pericyte recruitment which has not been shown in other model systems; however, more sophisticated live imaging or loss of function experiments would be required to demonstrate this incontrovertibly. Live imaging would require being carried out in live animals (intravital imaging) as we have previously shown that the renal vasculature regresses rapidly in cultured organs.³¹ Notably, pericytes and arterial mural cells have been linked in the opposite direction, where pericytes contribute to vascular smooth muscle cells during normal as well as pathologic development.^22,32 The mechanisms which govern this progenitor migration, including spatially restricted endothelial-stromal crosstalk, will be of great interest for future study.

The question remains as to why these cells would travel along arteries in order to arrive at their final location in the inner cortex. It could be that Tbx18-derived non-muscle mural cells are temporarily housed along arteries to allow the medullary stroma to form from Foxd1-lineage cortical stroma as well. As the medulla begins developing during mid-gestation, tubules, vessels, and Foxd1-derived stroma arise from the cortex.^2,19,33–35 The final interlobar and arcuate arteries form internal loops that broadly straddle the inner cortex.^6,15,36 Perhaps by using the arteries as intermediary location during early nephrogenesis, Tbx18-derived cells are able to deposit spatiotemporally into the inner cortex, while the rest of the cortex and medullary stroma are derived from Foxd1+ progenitors. Non-muscle mural cells could also play a functional role in arteriogenesis, but loss of function or ablation studies would be required to test this hypothesis.

Why a portion of the inner cortical stroma would be derived from ureteric progenitors, rather than Foxd1+ cortical progenitors as previously expected, is unclear. Foxd1-expressing cells likely give rise to most stromal cells of the kidney, however our study, and that of Kispert and colleagues¹⁸, demonstrate that ureteric progenitors also contribute to both smooth muscle and fibroblasts far from the ureter. It is possible that the inner cortical stroma needs to be functionally distinct from outer cortical, as well as medullary stroma. In fact, the inner cortical stroma contains cells capable of producing erythropoietin (Epo) in response to hypoxic or anemic conditions.^37–39 These cells have been described as having similarities to both pericytes and fibroblasts, and additionally, have been shown to be enriched for Cxcl14 expression.⁴⁰ Whether Tbx18-lineage cells are capable of producing Epo would be of major translational interest. Additionally, whether Foxd1-derived and Tbx18-derived stromal populations can compensate for each other in the context of targeted disruption (such as in the loss of cortical netrin 1⁶) is unknown.

Altogether, these results provide novel insights into the spatial organization and origin of renal stromal cells, as well as demonstrate previously uncharacterized mural cell biology linking the arterial and capillary perivascular niche. We demonstrate a previously unknown lineage of mural cells that arises from Tbx18+ stromal progenitors found surrounding the ureteric epithelium. These progenitors contribute to transient arterial mural cells during development, which serve as progenitors of regionalized pericytes in the inner kidney cortex. These results are likely to have far-reaching implications for understanding the dynamics and stabilization of heterogeneous organ vascular beds, including the kidney, providing new insights that could inform future regenerative and translational studies.

Resource Availability

Materials availability:

This study did not generate any new unique reagents.