Hepatic leukemia factor-expressing paraxial mesoderm cells contribute to the developing brain vasculature

ABSTRACT Recent genetic lineage tracing studies reveal heterogeneous origins of vascular endothelial cells and pericytes in the developing brain vasculature, despite classical experimental evidence for a mesodermal origin. Here we provide evidence through a genetic lineage tracing experiment that cephalic paraxial mesodermal cells give rise to endothelial cells and pericytes in the developing mouse brain. We show that Hepatic leukemia factor (Hlf) is transiently expressed by cephalic paraxial mesenchyme at embryonic day (E) 8.0-9.0 and the genetically marked E8.0 Hlf-expressing cells mainly contribute to the developing brain vasculature. Interestingly, the genetically marked E10.5 Hlf-expressing cells, which have been previously reported to contain embryonic hematopoietic stem cells, fail to contribute to the vascular cells. Combined, our genetic lineage tracing data demonstrate that a transient expression of Hlf marks a cephalic paraxial mesenchyme contributing to the developing brain vasculature. This article has an associated First Person interview with the first author of the paper.


INTRODUCTION
During vertebrate development, the segmented paraxial mesoderm of the somite gives rise to different mesodermal derivatives including vascular cells. The quail-chicken chimera and labeling system demonstrated that vascular progenitors derive from the somite and contribute to the trunk (Pardanaud et al., 1996;Pouget et al., 2006;Sato et al., 2008) and limb vasculature (Huang et al., 2003;Kardon et al., 2002). Genetic lineage tracing experiments with the Cre-loxP system clearly demonstrated the contribution of the somatic vascular progenitors into the trunk and limb vasculature: paraxial mesoderm-specific Cre lines such as Meox1-Cre, Pax3-Cre, and Myf5-Cre showed that paraxial mesodermal cells contribute to the trunk and limb vasculature (Hutcheson et al., 2009;Mayeuf-Louchart et al., 2014;Stone and Stainier, 2019;Wasteson et al., 2008). What induces paraxial mesodermal cells to differentiate into tissue-specific vascular cells remains to be investigated.
Previous studies with the quail-chicken chimera and labeling system demonstrated that the cephalic paraxial mesoderm has an angiogenic potential and contributes to the head and neck vasculature (Couly et al., 1995). In the developing brain and spinal cord of the central nervous system (CNS), the cephalic paraxial mesodermderived angioblasts are assembled to form perineural vascular plexus (PNVP) around the neural tube (Hogan et al., 2004;Jukkola et al., 2005). Subsequently, sprouting vessels from the PNVP invade the CNS tissues and extend the branches from the plexus towards the ventricle (Gupta et al., 2021;Paredes et al., 2018;Tata et al., 2015). In addition to the mesodermal origin, recent genetic lineage tracing studies with the Cre-loxP system demonstrated that erythro-myeloid progenitors (EMPs) contribute to brain endothelial cells (Plein et al., 2018), although contradictory data were reported (Feng et al., 2020;Palis and Yoder, 2020). The cephalic neural crest cells penetrate and differentiate into pericytes in the forebrain vasculature (Korn et al., 2002;Reyahi et al., 2015;Yamanishi et al., 2012). Genetic lineage tracing studies have revealed heterogeneous origins of CNS vascular cells from distinct populations, but CNS vascular cells derived via a classical pathway of mesodermal differentiation into endothelial cells and pericytes were not examined.
Here, we studied the developmental timing of angiogenic cephalic paraxial mesenchyme in the developing vasculature of the CNS and various tissues using the Cre-loxP-based lineage tracing system. First, we found a unique expression of hepatic leukemia factor (Hlf), the proline and acid rich basic region leucine zipper (Par-bZip) transcription factor, in the cephalic paraxial mesenchyme at embryonic day (E) 8.5. Second, the lineage tracing experiments using Hlf-Cre ERT2 ; ROSA-LSL-tdTomato embryos, with tamoxifen administration at E8.0, revealed that a transient expression of Hlf marked angiogenic cephalic mesenchyme, which mainly contributes to vascular cells in the developing CNS tissues at E15.5. Interestingly, a transient expression of Hlf at E10.5 failed to mark vascular cells at E15.5, suggesting that Hlf marks an angiogenic paraxial mesenchyme subpopulation from an early stage in vascular development.

Hlf-expressing cells in the cephalic mesoderm but not yolk sac
Previous studies revealed that Hlf expression marks embryonic hematopoietic stem cell (HSC) precursors within the dorsal aorta of the aorta-gonad-mesonephros (AGM) region at E10.5 and maturing HSCs in the fetal liver between E11.5 and E14.5 (Yokomizo et al., 2019). Moreover, Hlf expression does not mark EMPs within the yolk sac or endothelial cells of the yolk sac, AGM, or fetal liver (Yokomizo et al., 2019). These observations were supported by the recent report by Tang et al. (2021). Compared with the unique Hlf expression in embryonic HSCs from E10.5 to E14.5, Hlf expression in non-hematopoietic cells remains elusive. To address this, we initially analyzed the published single cell RNA-sequence data set of E8.0 and E8.5 mouse embryo (Pijuan-Sala et al., 2019). At these stages, most hematopoietic progenitors emerge in the yolk sac and migrate to intra-embryonic organs via circulation (Gomez Perdiguero et al., 2015;Hoeffel et al., 2015;Lux et al., 2008). Indeed, the clusters of blood progenitors and erythrocytes express RUNX family transcription factor 1 (Runx1), which is known to be expressed by hematopoietic progenitors and HSCs (Utz et al., 2020), while few blood progenitors and erythrocytes express Hlf ( Fig. S1A-F). In contrast, Hlf expression is highly enriched in the paraxial mesoderm in both E8.0 and E8.5 embryos (Fig. S1A-F). To confirm the expression pattern of Hlf in E8.5 embryos, we first performed RNA whole-mount in situ hybridization chain reaction (HCR). Hlf expression was clearly detectable in the cephalic region, where the angiogenic cephalic mesenchyme was found in the quailchicken chimera and labeling experiment (Couly et al., 1995)  Combined, these data suggest that Hlf-expressing cells at E8.5 represent a mesenchyme subpopulation in the cephalic region.

Transient expression of Hlf marked cephalic mesenchyme
We next examined differentiation potentials of Hlf-expressing cephalic mesenchymal cells using a tamoxifen-induced Hlf-Cre ERT2 knock-in mice in combination with a Cre-mediated ROSA-LSL-tdTomato reporter mice ( Fig. 2A). Like Hlf-tdTomato knock-in reporter mice (Yokomizo et al., 2019), Hlf-Cre ERT2 mice were generated by inserting a T2A-Cre ERT2 gene fusion before the endogenous stop codon within exon 4 ( Fig. 2A). We first induced tdTomato expression in Hlf-Cre ERT2 ; ROSA-LSL-tdTomato embryos at E8.0 and used high-resolution whole-mount imaging to analyze the distribution of tdTomato-expressing Hlf-lineage cells at E9.0 ( Fig. 2A). Double RNA whole-mount in situ HCR using probes to Hlf and tdTomato revealed that Hlf and tdTomato signals have been largely overlapping in the cephalic region ( Fig. S2I-P). Consistent with the analysis of Hlf-tdTomato reporter embryos at E8.5, we found the majority of Hlf-lineage cells were detectable in the cephalic region ( Fig. 2B-G; Fig. S2I-P): a few Hlf-lineage cells were found in the rostral trunk region (Fig. 2B-G). Hlf-lineage cells were not detectable in the head, trunk, and yolk sac vasculature at E9.0 (Fig. 2C,D,F,G,H,I and J, head and trunk; K, L, and M, yolk sac). These data suggest that most Hlf-expressing cells at E8.0-9.0 represent cephalic mesenchymal cells.
No contribution of Hlf-expressing cephalic mesenchymal cells to the muscular tissues We next examined a contribution of Hlf-lineage cells to muscular tissues in the embryonic head such as tongue in the oral cavity (Fig. 4A). Given that tdTomato-expressing Hlf-lineage cells were found in PECAM-1 + vasculature in the tongue of Hlf-Cre ERT2 ; ROSA-LSL-tdTomato embryos at E15.5, Hlf-lineage cells were not detectable in muscle progenitors which express Desmin, a muscle-specific intermediate filament (Fig. 4B-D). Likewise, tdTomato-expressing Runx1-lineage cells were found in PECAM-1 + vasculature but not in Desmin + muscle progenitors in the tongue of E15.5 Runx1-Cre ERT2 ; ROSA-LSL-tdTomato embryos (Fig. S4A-

RESEARCH ARTICLE
Biology Open (2022)  C). These data suggest that Hlf-expressing cells at E8.0 are a unique subset of cephalic mesenchymal cells which mainly contribute to vascular cells.

Contribution of Hlf-expressing cells in the rostral trunk to the tissue vasculature
Since a few Hlf-lineage cells were found in the rostral trunk region, we next examined whether Hlf-lineage cells also contribute to the vasculature of the lung, liver, and heart at E15.5 (Fig. 4E,F). Immunostaining analysis showed a significant contribution of tdTomato-expressing Hlf-lineage cells into the vasculature of lung (5.5±1.2% of blood vessels were positive for tdTomato), but not liver (0.95±0.44%) and heart (2.2±1.5%) in Hlf-Cre ERT2 ; ROSA-LSL-tdTomato embryos at E15.5 ( Fig. 4G-O,R). These data suggest that Hlf-expressing cells in the rostral trunk at E8.0 are angiogenic mesenchymal cells. Note that some Hlf-expressing cells in the trunk at E8.0 also contribute to non-vascular cells such as cardiomyocytes in the heart (Fig. 4P-Q). Meanwhile, tdTomato-expressing Runx1lineage cells largely contribute to the vasculature of liver (33.7±3.5% of blood vessels were positive for tdTomato) and heart (12.2±4.7%), but not lung (1.1±0.2%) in E15.5 Runx1-Cre ERT2 ; ROSA-LSL-tdTomato embryos ( Fig. S4D-L,M). Since Runx1 is expressed by embryonic endothelial cells at E8.0, Runx1expressing endothelial cells can be a source of the liver vasculature.

No contribution of Hlf-expressing cephalic mesenchymal cells to tissue-localized macrophages
Having established that cephalic mesenchyme contains hemogenic endothelial cells (Gama Sosa et al., 2021;Li et al., 2012), we next examined the differentiation potential of Hlf-and Runx1-lineage cells into tissue-localized macrophages. No significant contribution of Hlf-lineage cells was found in F4/80 + macrophages in brain, liver, lung, and heart in Hlf-Cre ERT2 ; ROSA-LSL-tdTomato embryos at E15.5 ( Fig. S5A-L ROSA-LSL-tdTomato embryos ( Fig. S5M-X,Y). These data suggest that the differentiation potential of Hlf-expressing mesenchymal cells at E8.0 appears to be restricted to vascular cells, while Runx1expressing cells at E8.0 can differentiate into both vascular cells and hematopoietic cells including tissue-localized macrophages.

Hlf-expressing cells at E10.5 are devoid of the angiogenic potential
Previous studies demonstrated that Hlf expression marks hematopoietic clusters in the AGM but not EMPs in the yolk sac, and endothelial cells of the yolk sac, AGM, and fetal liver at E10.5 (Yokomizo et al., 2019), but the studies did not examine what Hlf expression marks in the head. We first examined whether Hlfexpressing cells at E10.5 have an angiogenic potential. We induced tdTomato expression in Hlf-Cre ERT2 ; ROSA-LSL-tdTomato embryos at E10.5 and analyzed the distribution of tdTomatoexpressing Hlf-lineage cells at E15.5 (Fig. S6A). Given that brain microglia are mainly derived from yolk-sac EMPs but not AGM hematopoietic progenitors (Hoeffel et al., 2015), Hlf-lineage cells hardly contributed to microglia in the brain parenchymal region ( Fig. S6B-J,L): a few Hlf-lineage cells contributed to F4/80 + meningeal macrophages (Fig. S6C,F, open arrowheads). Likewise, Hlf-lineage cells were not detectable in the developing brain vasculature (Fig. S6B-J,K). The observation that Hlf-expressing cells at E8.0 are angiogenic but the cells at E10.5 are devoid of the angiogenic potential suggest that a transient expression of Hlf marks angiogenic mesenchymal cells.

DISCUSSION
With the genetic lineage tracing experiments using Hlf-Cre ERT2 ; ROSA-LSL-tdTomato mouse model, the results presented here identify angiogenic cephalic and rostral trunk mesenchymal cells. Previous genetic lineage tracing experiments using paraxial mesoderm-specific Cre lines demonstrated that paraxial mesoderm contributes to trunk and limb vasculature, but these experiments did not show the developmental timing of the paraxial mesoderm commitment into vascular lineage. Given that Hlf expression marks vascular cells but not muscle cells in the head and neck regions, Hlfexpressing cephalic mesenchymal cells at E8.0 are mainly committed to vascular lineage (Fig. 5).
Given that the recombination efficiency of Hlf-Cre ERT2 is high, based on the observation that the expression of tdTomato reporter is present in the vast majority of Hlf-expressing cells in Hlf-Cre ERT2 ; ROSA-LSL-tdTomato embryos at E9.0 (24 h after tamoxifen administration) (Fig. S2I-P), the Hlf-lineage contribution in the brain vascular cells at E15.5 is relatively low (Fig. 3V, the contribution of Hlf-lineage cells in the whole-brain vasculature; Fig. 4R, the contribution of Hlf-lineage cells in the vasculature of hindbrain, midbrain, and diencephalon). One potential explanation is that non-Hlf-lineage cells are the major contributors to the brain vascular cells and Hlf-lineage cells merge at some point to contribute to the brain vascular cell populations. Indeed, PECAM-1 + endothelial cells in the nascent vascular plexus around the open neural tube (the future PNVP) are negative for tdTomato in Hlf-Cre ERT2 ; ROSA-LSL-tdTomato embryos at E9.0 (Fig. 2D), suggesting that Hlf-lineage cells marked at E8.0 do not contribute to these endothelial cells. In this scenario, Hlf marks a subset of angiogenic cephalic mesenchyme cells. Another Fig. 3. Hlf-expressing cephalic mesenchymal cells contribute to the brain vasculature. (A) Generation of Hlf-Cre ERT2 ; ROSA-LSL-tdTomato embryos. Tamoxifen was administered by oral gavage at E8.0 and embryos were harvested at E15.5 for analysis. (B) A representative immunofluorescent image of sagittal sections of E15.5 head stained with anti-PECAM-1 antibody (white). The yellow boxes indicate hindbrain, midbrain, and diencephalon in the brain parenchyma. (C-K) Representative immunofluorescent images of E15.5 Hlf-Cre ERT2 ; ROSA-LSL-tdTomato brain sections stained with antibodies to tdTomato (C-H, red) together with the pericyte marker PDGFRß (C-H, green) and the endothelial cell marker PECAM-1 (C-H, blue; I-K, white). The boxed regions in (C-E) are magnified in (F-K). Arrowheads indicate tdTomato + vascular cells. Note that quantification of tdTomato + blood vessels in the hindbrain, midbrain, and diencephalon is shown in Fig. 4R. Scale bars: 100 µm (C-E) and 20 µm (F-K). n=3. (L-U) High-resolution images of E15.5 Hlf-Cre ERT2 ; ROSA-LSL-tdTomato brain sections stained with antibodies to tdTomato (red) together with PDGFRß (L, N, Q, and S, green; P and U, white) and PECAM-1 (L, M, Q, and R, blue; O and T, white). Arrows indicate tdTomato + /PECAM-1 + endothelial cells; open arrows indicate tdTomato + /PDGFRß + pericytes. Scale bars: 10 µm. n=3. (V) Representative flow cytometry data analyzing E15.5 Hlf-Cre ERT2 ; ROSA-LSL-tdTomato brain. Non-hematopoietic cells are negative for the pan-hematopoietic marker CD45 and erythrocyte marker Ter119. Endothelial cells are positive for PECAM-1 and pericytes are positive for PDGFRß. Positive gates were defined by the non-staining controls. Percentages of tdTomato + cells in the fractions of endothelial cells (CD45 − /Ter119 − /PECAM-1 + /PDGFRß − ) and pericytes (CD45 − /Ter119 − / PECAM-1 − /PDGFRß + ) were calculated. n=3. explanation is that Hlf-expressing cells earlier than E8.0 may initiate the first wave of brain vascularization including PNVP: these Hlf-expressing cells turn off the Hlf expression by E8.0, so these cells are not genetically marked in Hlf-Cre ERT2 ; ROSA-LSL-tdTomato embryos with tamoxifen administration at E8.0. In this scenario, angiogenic cephalic mesenchyme cells are heterogeneous and transiently express Hlf at different developmental time points.
Although the transient expression of Hlf marks an angiogenic mesenchyme, it is not clear whether Hlf is required for the differentiation of mesenchymal cells into vascular cells. Previous studies demonstrated that Hlf homozygous mutants are morphologically normal and fertile (Gachon et al., 2004). Given that Hlf is specifically expressed in HSCs in the AGM and fetal liver (Yokomizo et al., 2019) and the adult bone marrow (Komorowska et al., 2017;Wahlestedt et al., 2017), Hlf is dispensable for HSC generation. Likewise, given that Hlf marks a classical mesodermal origin of CNS vascular cells, Hlf appears not to be required for CNS vascular development. It is possible that other Par-bZip transcription factors (Dbp and Tef ), which share a similar DNA binding motif with Hlf, may compensate for loss of Hlf in the vascular development. What controls Hlf expression may provide an insight in understanding the transcriptional machinery of mesodermal differentiation into vascular cells.

Mice
All animal procedures were approved by the Animal Care and Use Committee of Kumamoto University; the National Heart, Lung, and Blood Institute (NHLBI) Animal Care and Use Committee in accordance with NIH research guidelines for the care and use of laboratory animals. The following mice were used in this study: Hlf-tdTomato knock-in mice (Yokomizo et al., 2019), Runx1-Cre ERT2 transgenic mice (Matsuo et al., 2017), and ROSA-LSL-tdTomato mice (Madisen et al., 2010). The targeting strategy of Hlf-Cre ERT2 is the same as the one of Hlf-tdTomato reporter mice as described previously (Yokomizo et al., 2019). Hlf-Cre ERT2 knock-in mice were generated in the Kumamoto University by inserting a T2A-Cre ERT2 gene fusion before the endogenous stop codon within exon 4. The targeting construct was knocked into the locus using the CRISPR/Cas9 method. The detailed procedure will be reported elsewhere (Yokomizo et al., Nature in press). The Cre-mediated excision was induced by administrating 2.5 mg tamoxifen (T-5648, Sigma-Aldrich) by oral gavage at embryonic days (E) 8.0 or 10.5 and embryos were harvested at E9.0 or E15.5. Hlf-Cre ERT2 ; ROSA-LSL-tdTomato or Runx1-Cre ERT2 ; ROSA-LSL-tdTomato double heterozygous embryos were used for all experiments.
either Alexa-488-, Alexa-568-, or Alexa-647 conjugated secondary antibodies (Jackson ImmunoResearch or Thermo Fisher Scientific, 1:250) were used. All confocal microscopy was carried out on a Leica TCS SP5 confocal (Leica). Area of tdTomato positive blood vessels and macrophages were quantified using ImageJ (NIH). The percentage of tdTomato-positive blood vessels was based on the area of tdTomato-positive blood vessels within the area of PECAM-1-positive blood vessels ( Fig. 4R; Figs S4M, S6K). Likewise, the percentage of tdTomato-positive macrophages was based on the area of tdTomato-positive macrophages within the area of F4/ 80-positive macrophages (Figs S5Y, S6L).