Loss of meningeal lymphatic endothelial cells results in formation of a newly forming vascular network with scavenging function
Katharina Uphoff, Andreas van Impel, Lars Muhl, Martin Stehling, Christer Betsholtz, Stefan Schulte-Merker
TL;DR
Deleting specific brain lymphatic cells in zebrafish leads to a new vascular network that may help remove waste from the brain.
Contribution
A new protocol for ablating brain lymphatic endothelial cells in zebrafish reveals compensatory vascular network formation.
Findings
Ablated brain lymphatic endothelial cells do not regenerate in zebrafish.
Loss of these cells disrupts meningeal blood vessels and arachnoid structure.
A new vascular network forms with potential scavenging functions.
Abstract
For an extended period of time, the meninges were believed to be devoid of lymphatic vessels. The recent discovery of lymphatic vessels in the dura mater of humans, mice and zebrafish, and the identification and characterization of scavenging lymphatic endothelial cells in zebrafish leptomeninges have contributed to a better understanding of waste removal routes from the brain. Zebrafish brain lymphatic endothelial cells (BLECs), in particular, have a high propensity for particulate waste removal from the cerebrospinal fluid. To further understand the function of BLECs, we have established a protocol for specifically ablating this cell type during embryonic stages. We show that BLECs do not recover upon complete depletion, even at adult stages. The absence of BLECs results in detrimental effects on the meningeal blood vasculature and on arachnoid mater structure. We further report that,…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7- —Universität Münsterhttp://dx.doi.org/10.13039/501100004869
- —Karolinska Institutethttp://dx.doi.org/10.13039/501100004047
- —Deutsche Forschungsgemeinschafthttp://dx.doi.org/10.13039/501100001659
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsCerebrospinal fluid and hydrocephalus · Spinal Dysraphism and Malformations · Vestibular and auditory disorders
INTRODUCTION
The lymphatic system is present in almost all organs and provides essential functions for tissue fluid homeostasis, absorption of dietary fats and immune surveillance (González-Loyola and Petrova, 2021). A noticeable exception is the brain parenchyma, which is devoid of lymphatic vessels (Petrova and Koh, 2018). How the brain, with its high metabolic activity, is cleared of waste has long been under investigation (Chachaj et al., 2023; Proulx, 2021). The existence of lymphatic vessels in the dura mater of the meninges, which build a protective physical layer and envelop the vertebrate central nervous system (CNS), was only recently (re-)discovered in mice, humans and other vertebrates (Absinta et al., 2017; Aspelund et al., 2015; Castranova et al., 2021; Louveau et al., 2015).
The zebrafish lymphatic system shares morphological, functional and molecular characteristics with lymphatics found in higher vertebrates (Hogan et al., 2009; Küchler et al., 2006; Yaniv et al., 2006). This also pertains to lymphatic vessels in the meninges, which are in immediate proximity to the brain. Specifically, zebrafish intracranial meningeal lymphatics constitute a network of lymphatic vessels situated immediately beneath the skull and distal to meningeal blood vessels (Castranova et al., 2021). They adhere to the skull cap upon removal, as they are located in the dura mater, as they are in mice. Intracranial lymphatics are distinct from the blood vasculature, drain interstitial fluid from the brain and contain neutrophils that transmigrate through the lymphatic endothelium (Castranova et al., 2021). Upon removing the skull cap of adult zebrafish, another lymphatic cell population that is attached to the brain surface becomes evident. These cells express key lymphatic markers such as prospero homeobox protein 1 (Prox1a), vascular endothelial growth factor receptor-3 (Vegfr3), lymphatic vessel hyaluronan receptor 1 (Lyve1b), pattern-recognition receptor mannose c-type lectin receptor 1 (Mrc1a) and friend leukemia integrated 1a (Fli1a). They were independently discovered and termed either mural lymphatic endothelial cells (muLECs) (Bower et al., 2017), fluorescent granular epithelial cells (FGPs) (Galanternik et al., 2017) or brain lymphatic endothelial cells (BLECs) (van Lessen et al., 2017). Here, these cells are referred to as BLECs.
BLECs transdifferentiate from the venous endothelium of the choroidal vascular plexus (CVP) behind the eye at 56 hpf and migrate bilaterally along the mesencephalic vein (MsV) (Bower et al., 2017; Galanternik et al., 2017; van Lessen et al., 2017). They remain as single cells in symmetric loops distal to the optical tectum (TeO) of zebrafish larvae, without entering the parenchyma. The cells subsequently expand over the fore-, mid- and hindbrain and at 3 weeks of age cover the entire brain surface, with the exception of the dorsal forebrain (Bower et al., 2017; Galanternik et al., 2017; van Lessen et al., 2017). Analogous to other lymphatic structures, their development depends on the Vegfc/Vegfr3 pathway. BLECs are closely associated with meningeal blood vessels but are, in contrast to pericytes, not embedded in their basement membrane (van Lessen et al., 2017). They contain large lysosomal autofluorescent vesicles and have a functional role in the clearance of substrate molecules via clathrin-mediated endocytosis in the range of 10-150 kDa (Huisman et al., 2021; Uphoff et al., 2024; van Lessen et al., 2017). While microglia play a central role during waste removal in the brain (Platt et al., 1998), BLECs have been shown to have a higher endocytotic capacity and to be more efficient in taking up small substances, while failing to remove large particles such as bacteria (Huisman et al., 2021). Given their nature as single cells, BLECs, in contrast to normal lymphatic vessels, are unable to transport fluids and particles back to the venous blood system. However, BLECs represent a scavenging endothelial cell (SEC)-like cell population that, together with microglia, ensures tissue homeostasis in the zebrafish CNS alongside other brain-cleansing mechanisms such as cerebrospinal fluid (CSF) drainage via intracranial lymphatic vessels (Castranova et al., 2021), through the uptake of excess particles (Huisman et al., 2021; van Lessen et al., 2017). Furthermore, BLECs affect developmental angiogenesis of the meningeal blood vasculature (Bower et al., 2017) and laser ablation of these cells in one hemisphere of a zebrafish embryonic brain results in reduced numbers of blood endothelial cells (BECs) in the meningeal blood vasculature. Moreover, vegfc; vegfd double heterozygous embryos with fewer BLECs that are asymmetrically distributed in the brain hemispheres show blood endothelial asymmetry in the meningeal blood vasculature. Consistent with this, BLECs express pro-angiogenic factors such as vegfaa and vegfab in BLECs (Bower et al., 2017). Therefore, BLECs constitute a functionally important cell population in the zebrafish leptomeninges and conduct many different tasks relevant for waste removal and maintenance of the meningeal blood vasculature.
Progress towards understanding the physiological role of BLECs has been hampered by the inability to ablate these cells quantitatively. Here, we report a BLEC ablation model, and find that, in the absence of BLECs, the meningeal blood vasculature and the arachnoid mater are strongly affected. We further show that, in the absence of functional BLECs, a distinct vessel network with features of lymphatic vasculature develops, and that this network can exert scavenging functions, thus possibly providing a compensatory response to the absence of BLECs.
RESULTS
BLECs have only recently been discovered, and their full spectrum of functions requires further analysis. To gain insight into the functional relevance of these cells for the meninges and the brain, their specific ablation represents an appealing approach, but has not been possible due to their high regenerative potential. Targeting BLECs via the NTR (Curado et al., 2007) or the Kid (Labbaf et al., 2022) system is hampered by the lack of specific promoters that could be used to express the related constructs exclusively in BLECs, but not in other endothelial beds. Laser ablation, a possible alternative method, results in efficient regeneration of BLECs, most likely because not all cells are removed (Bower et al., 2017). We therefore aimed to develop an alternative approach. We have shown previously that BLECs represent a SEC population that endocytoses substances in a highly efficient manner, including antibody conjugates (Huisman et al., 2021). In an attempt to capitalize on this scavenging function, we employed the antibody-drug-conjugate (ADC) trastuzumab-DM1 (T-DM1), which is normally used for treatment of HER2-positive breast cancer (Lewis Phillips et al., 2008). Upon binding of T-DM1 to HER2, the receptor is internalized. The stable, non-reducible thioether linker (MCC) of the ADC releases DM1 only as a result of proteolytic degradation in lysosomes (Barok et al., 2014; Lewis Phillips et al., 2008), whereupon DM1 enters the cytoplasm and binds microtubule plus ends (Akhmanova and Steinmetz, 2015). High concentrations of DM1 then result in depolymerization of microtubules (Kovtun and Goldmacher, 2007), leading to mitotic arrest, mitotic catastrophe, disruption of intracellular trafficking and cell apoptosis (Lopus, 2011). Due to these properties, T-DM1 appeared to be a suitable ADC for targeted ablation of BLECs.
The cytotoxic antibody trastuzumab emtansine ablates BLECs after injection
To establish the ablation system and test whether T-DM1 leads to cell death after injection, either 0.2 ng or 2 ng of T-DM1 was injected into zygotes at the one-cell stage. This led to an instantaneous halt in development, without any further cell divisions, when checked at 2.5 h post fertilization (hpf) (Fig. 1A). At 24 hpf, all injected zygotes were dead. Next, we injected the ADC into the CSF in the zebrafish head at 3 days post-fertilization (dpf) (Fig. 1B) and tested concentrations of 1 ng, 5 ng and 10 ng. When counting the number of BLECs at 5 dpf, we found a significant reduction in T-DM1-injected embryos compared to control-injected embryos (Fig. 1D-H). However, concentrations of 5 ng and 10 ng resulted in severe systemic effects, as most injected larvae developed pronounced edema and became apoptotic (Fig. 1I-L). More-detailed analysis showed that the facial lymphatic vessel network, including the otholithic lymphatic vessel (OLV), was compromised in these embryos (Fig. 1I′-L′), thereby explaining the observed edema formation. Thus, these results indicated that at least some parts of the facial lymphatic network are also targeted by the injected T-DM1. When using an IgG-Alexa-647 antibody as a traceable substrate, we noticed that BLECs take up more dye compared to the facial lymphatics when injected into the CSF (Fig. S1A,B). Taking this into account, we tried to optimize the amount of injected T-DM1 so that most treated embryos would be devoid of BLECs but would still have unaffected facial lymphatics. While injection of 2 ng T-DM1 resulted in facial lymphatic defects and a high mortality rate, injection of 1 ng T-DM1 into the CSF gave rise to the highest proportion of embryos with ablated BLECs but intact facial lymphatics (Fig. 1J,J′), and this amount was therefore used in all subsequent experiments.
*Injection of T-DM1 interrupts the development of zygotes and ablates BLECs in zebrafish embryos. (A) Injection of T-DM1 into zygotes leads to a halt of embryonic development at the one-cell stage, as cell division is suppressed [IgG-A647, n=179 (100%) zygotes with normal development; 2 ng T-DM1, n=129 (100%) zygotes with halt in development; 0.2 ng T-DM1, n=5 (2,67%) zygotes with normal development and n=182 zygotes (97.32%) with halt in development]. (B) Schematic representation of a zebrafish embryo (dorsal view) showing the injection site into the CSF and the arrangement of dorsal BLECs distal to the TeO. (C) Schematic lateral view of a zebrafish embryo depicting BLECs on the dorsal (light green) and ventral (dark green) side of the brain and facial lymphatics in red. (D-G) Maximum projections of fli:nucGFP; lyve1:dsRed-positive embryos at 5 dpf that were injected with IgG-Alexa-647 (D), or with 1 ng (E), 5 ng (F) or 10 ng (G) T-DM1 at 3 dpf. (H) The injection of T-DM1 leads to a significant reduction in the number of dorsal BLECs. n=10 embryos per group. IgG-Alexa-647 versus 1 ng T-DM1 ***P.adj=0.000966; IgG-Alexa-647 versus 5 ng T-DM1 ***P.adj=0.00087; IgG-Alexa-647 versus 10 ng T-DM1 **P.adj=0.0009 (Mann–Whitney U-test; data are mean±s.d.). (I-L′) Bright-field images (dorsal views, I-L) and maximum projections of lyve1:dsRed-positive embryos (lateral views, I′-L′) at 120 hpf that were injected with IgG-Alexa-647 as control (I,I′), or with 1 ng (J,J′), 5 ng (K,K′) or 10 ng T-DM1 (L,L′), showing that high concentrations of T-DM1 (5 ng and 10 ng) not only ablate BLECs (green asterisks) but also affect facial lymphatic structures (red asterisks), resulting in edema formation at 5 dpf (red arrows in bright-field images). 1 ng T-DM1 ablates BLECs but does not affect the OLV (red arrows). CSF, cerebrospinal fluid; FCLV, facial collecting lymphatic vessel; hpf, hours post-fertilization; LAA, lymphatic branchial arches; LFL, lateral facial lymphatics; MFL, medial facial lymphatic; s.d., standard deviation; TeO, tectum opticum; OLV, otolithic lymphatic vessel.
Since injection of 1 ng T-DM1 at 3 dpf did not always lead to a complete absence of BLECs in all treated embryos, larvae were screened for normal facial lymphatics, the presence of an inflated swim bladder and the number of remaining BLECs at 5 dpf. The embryos were subsequently sorted into three groups depending on the number of dorsally or ventrally remaining BLECs: (1) 0 remaining BLECs; (2) 1-4 remaining BLECs located on the ventral side; and (3) 1-4 BLECs distributed over both the dorsal and ventral side of the brain. We differentiated between dorsal and ventral BLECs (Fig. 1C), as it is not yet known which role the BLECs on the ventral side of the brain fulfill. The sorting was verified at 8 dpf to rule out the loss of additional BLECs after the initial grouping of embryos. Furthermore, we injected embryos with IgG-Alexa-647 as a control.
In control-injected embryos, BLECs were found to be arranged as single cells in symmetric loops distal to the TeO (Fig. 2A-B′; white arrowhead). They proliferated and migrated into the inner sites of these loops and covered the TeO after 3 weeks of age. BLECs that were located at the midbrain-hindbrain boundary spread out caudally and covered the hindbrain after 3 weeks of age (Fig. 2C-D′; white arrowhead). Intracranial/dural lymphatics (Fig. 2B,B′; red arrowhead) emerged from a branch of the OLV and elongated over time to form three sprouts that grew rostrally along the lateral edge of the TeO, medially along the boundary of the TeO and hindbrain, and caudally along the lateral edge of the cerebellum, where they became thicker and connected to the contralateral branch at the caudal edge of the hindbrain (Fig. 2C-D′; red arrowhead), similar to what has been shown previously (Castranova et al., 2021). T-DM1-injected fish, which are completely devoid of all BLECs at 8 dpf, did not show recovery of BLECs, as they were still absent at 2, 3 and 4 weeks post-fertilization (wpf) (Fig. 2E-G′; white asterisk). The intracranial lymphatics (Fig. 2E-G′; red arrowhead) developed normally and were not affected by the lack of BLECs.
T-DM1-injected embryos lacking all BLECs at 8 dpf fail to recover this cell population at later stages while showing no overt effects on intracranial lymphatics development. (A) Schematic representations of the dorsal head region of a zebrafish at 2 wpf (left) and 4 wpf (right), showing BLECs (green) and intracranial lymphatics (red), and indicating the zoomed-in areas shown in B-D and E-G. (B-G′) Maximum confocal projections of either 2, 3 or 4-week-old larvae, with mrc1a:mCitrine- and lyve1:dsRed-positive lymphatic endothelial cells [dorsal (B-G) and lateral (B′-G′) views]. Embryos were injected with either IgG-Alexa-647 (control) or T-DM1 at 3 dpf. In control-injected larvae (B-D′), BLECs are arranged as single cells in symmetric loops dorsal to the TeO (white arrowheads). Intracranial lymphatics (red arrowheads) emerge from a branch of the otolithic lymphatic vessel. BLECs cover the whole TeO after 3 weeks of age. In larvae injected with 1 ng T-DM1 (E-G′) at 3 dpf, BLECs are ablated (white asterisks). Intracranial lymphatics (red arrowheads) are unaffected. Different larvae were used for each stage. (H) Schematic representation of an extracted zebrafish brain (dorsal view). During dissection of the brain, the intracranial/dural lymphatics are removed. The dotted line indicates the region of the dissected brains shown in I-T. (I-T) Confocal images showing maximum projections of 8 dpf, 3 wpf, and 4 wpf larvae with mrc1a:mCitrine-positive lymphatic endothelial cells and kdrl: mCherry-positive blood vessels. T-DM1-injected embryos were grouped according to the number of BLECs at 8 dpf (I,L,O,R). Ventrally located BLECs are not visible in these images. Dorsally located remaining BLECs are highlighted (white arrow). Fish with 0 BLECs (L-N) or with 1-4 ventral BLECs (O-Q) did not regenerate BLECs, whereas several fish with 1-4 remaining dorsal BLECs (S) fully regenerated the BLEC population. For each stage, different larvae were used. FB, forebrain; HB, hind brain; TeO, tectum opticum; dpf, days post-fertilization; wpf, weeks post-fertilization.
When examining T-DM1-injected embryos that only had 1-4 ventrally located BLECs, we also never observed a recovery of this cell population (Fig. 2O-Q). The group in which not all dorsal BLECs had been ablated (1-4 ventrally and dorsally distributed BLECs), however, gave rise to fish in which the BLEC population recovered from the initial ablation within 3 weeks (Fig. 2S, Fig. S2 blue star), while in others this regeneration was not evident at all (Fig. 2T, Fig. S2). Therefore, we never observed a regeneration of BLECs whenever all dorsally located BLECs had been completely ablated at 8 dpf (Fig. 2M,N,P,Q and Fig. S2), suggesting that this regeneration might depend on an increased proliferation of remaining dorsal BLECs rather than a differentiation of new BLECs from venous endothelium.
Ablation of BLECs leads to the formation of newly forming vascular structures and meningeal blood vessel defects
Analyzing the brain of fish from the 0 BLEC group at the age of 6 months revealed the presence of a previously unreported vessel network in the leptomeninges that was not detectable in the control group (Fig. 3A-E′ and Fig. S3). When examining several fish from the 0 BLEC group, we observed that not all brains were covered with this vessel network on the dorsal side of the brain (Fig. 3D). In contrast, the ventral sides always exhibited coverage with this additional endothelial structure (Fig. 3D′,E′ and Fig. S3A). Using different transgene combinations, we found that network cells expressed mrc1a, flt4 (Fig. 3F,G), lyve1b (Fig. 7G), and the tight junction protein 1a (zo-1; tjp1a) (Fig. 3H′,I′), while being negative for kdrl and flt1 (Figs 1, 3 and 7), thus indicating that the network had lymphatic characteristics. Network development was also evident in all fish from the other two T-DM1-injected groups, with the exception of fish in which the BLEC population recovered (fish derived from the group with 1-4 remaining dorsal+ventral BLECs at 8 dpf) (Fig. S2, yellow star). The network was detected as early as 5 wpf, but the progression of its development varied between individuals. Taken together, our results suggest that this ectopic network develops in all adult fish lacking the BLEC population and that this structure emerges from the ventral side of the brain (Fig. 3D-E′). It populates the same leptomeningeal layer as BLECs do and is likewise not covering the dorsal forebrain (Movies 1 and 2 and Fig. S3B).
Complete ablation of BLECs results in the formation of an mrc1a-, flt4-, tjp1a-positive vessel network and a disrupted arachnoid mater. (A,B) Schematic drawing of the dorsal (A) and ventral (B) head region of an adult zebrafish, indicating the brain areas shown in C-E and C′-E′. (C-E′) Confocal images of 6-month-old mrc1a:mCitrine;kdrl:mCherry-positive zebrafish. (C,C′) In control fish, BLECs (green) are in close contact with the meningeal blood vessels of the dorsal and ventral sides of the brain. (D-E′) T-DM1-injected fish without BLECs show no regeneration of the population but develop a mrc1a-positive endothelial network, which emerges from the ventral side of the brain. In some fish, the network also covers the dorsal side of the brain (E,E′); in others, it remains exclusively ventral (D,D′). Note that the displayed brains underwent identical treatment prior to imaging (see Fig. 6, where samples from C and E are shown with IgG-Alexa-647 uptake indicated in Fig. 6D-G′). (F,G) Confocal images of a 6-month-old flt4:mCitrine;flt1:tdTomato-positive zebrafish (dorsal view). In control fish (F), BLECs are located close to meningeal blood vessels. The ectopic network that develops upon complete BLEC ablation is also positive for flt4:mCitrine (G). (H-I′) Confocal images of 6-month-old mrc1a:mCitrine;tjp1a:tdTomato-positive zebrafish brain (zoomed-in). In control fish (H,H′), BLECs with auto-fluorescent vacuoles are close to tjp1a-positive endothelial cells of the meningeal blood vessels (red arrowheads). The large cells of the arachnoid mater are also positive for tjp1a (green arrowhead). Endothelial cells forming the additional vascular network in BLEC-depleted fish are tjp1a positive (I,I′, red arrows), as is the case for meningeal blood vessels. The arachnoid mater and meningeal blood vessels appear disrupted in T-DM1-injected fish.
BLECs are located in the leptomeninges, which consist of the pia mater and the arachnoid mater. The arachnoid mater consists of large barrier cells with strong tight junctions. By employing a novel tjp1a knock-in reporter line (see Materials and Methods), we found that BLECs did not express this marker (Fig. 3H,H′), as expected due to their single-cell organization. However, tight junctions of the arachnoid mater (Fig. 3H,H′, green arrowhead), as well as the tight junctions of the meningeal blood vessels (Fig. 3H,H′, red arrowhead), were clearly visible in control-injected fish, and BLECs were found to be located distal to the large arachnoid mater cells. Endothelial cells of the meningeal blood vessels (Fig. 3I,I′, red arrowhead), as well as the endothelial cells of this additional vascular network (Fig. 3I,I′, red arrow) in BLEC-depleted fish, displayed Tjp1a-positive tight junctions. Interestingly, the arachnoid mater appeared to be disrupted in the T-DM1-injected fish that developed the vessel network. We also noticed that at least part of the ectopic network formed a lumen (Fig. 3I). Of note, we observed that, in rare cases, parts of the vessel network contained blood cells (data not shown), while the vast majority of analyzed brains did not show blood within the ectopic vessel network.
Since BLECs affect angiogenesis of the meningeal blood vasculature, and as they express pro-angiogenic factors such as vegfaa and vegfab (Bower et al., 2017), we wondered whether fish lacking BLECs have meningeal blood vasculature defects. In 6 to 12-month-old fish lacking BLECs but showing an ectopic vessel network extending to the dorsal side of the brain, the blood vasculature did not grow in an organized manner, as found in control fish (Fig. 4A-C). Analysis of meningeal blood vessels showed that the percentage of the area covered by blood vessels was reduced by 34%. Furthermore, these fish had 50% fewer vessel junctions and branches, with an increased branch length and vessel thickness compared to control fish (Fig. 4D). Younger fish (1-3 months) without BLECs and without a network that reached the dorsal side did not display such severe damage in the dorsal meningeal blood vessels (Fig. 4E-G). However, the number of branches and junctions decreased slightly, but not significantly (Fig. 4H), by 14%, suggesting that, either with age or upon the appearance of the endothelial network on the dorsal side, the defects in meningeal blood vessels become exacerbated.
*The meningeal blood vasculature in BLEC-depleted fish shows a massive reduction in blood vessel coverage and in the number of vessel branches and junctions, while the average vessel thickness and branch length are increased. (A) Schematic representation of a zebrafish head (dorsal view). Red rectangles mark the analyzed areas (0.166 mm2) per fish. (B,C) Maximum projection of the dorsal TeO (tectum opticum) of 12-month-old mrc1a:mCitrine;kdrl:mCherry-positive zebrafish with either BLECs (control, B) or with a dorsally extending network after BLEC ablation (C). The number of blood vessels is decreased in fish with a dorsal network. (D) Quantitative analysis of blood vessels from 6- to 12-month-old fish, normalized to the control of each age. The area covered by blood vessels is significantly reduced (*P=0.044, Man-Whitney U-test) by 34% in fish with a dorsal network after BLEC ablation. In addition, the number of vessel branches (*P=0.044, Mann–Whitney U-test) and junctions (*P=0.044, Mann–Whitney U-test) is reduced. The average branch length (34%, **P=0.007, unpaired Student's t-test) and the average vessel thickness (14%, P=0.019, unpaired Student's t-test) are increased in fish with a dorsal network. (E-G) Schematic representation (E) and maximum projections (F,G) of the dorsal TeO of 5-week-old mrc1a:mCitrine;kdrl:mCherry-positive zebrafish with BLECs (control, F) or with ablated BLECs but lacking the vessel network on the dorsal side of the brain (G). The number of blood vessels, the average branch length and the average vessel thickness do not differ in 1- to 3-month-old fish. However, there is a slight decrease in the number of branches (14%, P=0.052, unpaired Student's t-test) and junctions (14%, P=0.069, unpaired Student's t-test) compared to control fish. Results are presented as mean±s.d.
Endothelial cells forming the ectopic network show lymphatic characteristics and a scavenging function
Next, we aimed to obtain further insights into the cellular identity of the cells forming the newly discovered network. We performed single-cell RNA (scRNA) sequencing (Smart-Seq3) analysis utilizing the meninges of three control and three T-DM1-injected fish, which displayed a network that reached the dorsal side of the brain, at 1 year of age (Fig. 5A). mrc1a:mCitrine, kdrl:mCherry and double-positive cells were FACS sorted into one ‘control’ and one ‘T-DM1 treated’ 384-well plate. Due to the presence of autofluorescent vacuoles in BLECs (Galanternik et al., 2017), also visible in Fig. 3H′, we expected to detect a low (compared to BECs) mCherry-signal within the kdrl:mCherry-negative BLECs, and indeed found that the mCitrine-positive cell population was also moderately positive for red fluorescence (Fig. 5A). This was consistent with the observation that, in the T-DM1-treated group, the amount of mrc1a:mCitrine;kdrl:mCherry double*-*positive cells was low in comparison to the *mrc1a-*positive cells. After initial quality control filtering, 525 single-cell transcriptomes (244 control and 281 treated) remained for analysis (Fig. 5B). Based on the expression profile, we identified eight different clusters (Fig. 5C, Fig. S4A,B). Fig. 5D,E show which cells were derived from which condition, also indicating their transgene expression during FACS sorting. Cluster 7 reflects the BLEC population (expressing map7d2a, osr2, meis2a, mafa, lyve1b, stab2, stab1, flt4, mrc1a, loxl1, b3gnt7, exoc6, osbpl10b and ctsla). Expectedly, the BLECs cluster was composed exclusively of control cells (Fig. 5B, red dots), and the cluster contained cells from the mCitrine and mCherry double-positive population (Fig. 5D). Cluster 1, reflecting a heterogeneous EC cluster [ehd2b, ptprb, kdr, cdh6, igfbp7, rgcc, sox7, jam2a, edn1, gpr182 (ackr5), cavin2b, fabp4a, cav1, krt15, podxl, vwa1, polr2b, loxl1, cxcl12a and cldn5b] was mostly composed of cells from the T-DM1-treated plate (Fig. 5C, blue dots). In addition, the immune and antigen-presenting clusters [cluster 0: mhc2dab, mhc2daa, cd74a, il1b, cebpb, ptpn6 and lcp1; and cluster 2: c1qb, mhc2b (mhc2dgb), mhc2dhb, marco, cd74b and apoc1; Fig. S4B] could be divided in control and treated cells, with the treated cell cluster being much larger compared to the control cluster (Figs 5B, 7A). Subsequent subclustering of the endothelial population in cluster 1 revealed two sub-populations, one reflecting blood endothelial cells (sub-clusters 2) and one reflecting endothelial cells, which we allocated to the additional network (sub-clusters 1 and 3) (Fig. 5F and Fig. S4C).
Endothelial cells forming the additional network in BLEC-depleted fish represent a distinct endothelial cell population with lymphatic vessel features. (A) Confocal images of a control brain with BLECs (mrc1a:mCitrine) and meningeal blood vessels (kdrl:mCherry), and a brain from a T-DM1-injected fish with the endothelial network (mrc1a:mCitrine) and meningeal blood vessels (kdrl:mCherry). Meninges of control and T-DM1-treated fish were FACS sorted into two different plates based on their mCitrine and mCherry signals (four groups: negative cells, mCit+ cells, mCherry+ cells and double positive mCit+;mCherry+ cells). Control plate: 203,627 live cells; T-DM1 plate: 286,098 live cells. (B,C) The scRNA sequencing data were analyzed with the Seurat pipeline, revealing eight clusters. The BLECs cluster is exclusively composed of cells from the control plate, whereas the network/BEC cluster is mainly composed of cells from the T-DM1-treated plate. (D,E) Distribution of the FACS-sorted cells with respect to their transgene expression and origin: control (D) and T-DM1-treated (E). (F) Cluster 1 subanalysis. Location of cells from pagoda2 algorithm sub-clustering in the original UMAP space. Analysis with Seurat pipeline (20 dim). Differentially expressed (DE) gene analysis of the sub-clusters 1 and 3 (network) versus sub-cluster 2 (BECs) (log2FC>1). (G) DE gene analysis of sub-clusters 1 and 3 (network) versus cluster 7 (BLECs) (log2FC>1). (H) Dot plot indicating expression levels of selected BLEC, LEC and BEC markers (y-axis) in BECs, network cells and BLECs. Color scale represents average expression; point size is relative to the percentage of cells expressing the gene. (I) Analysis of network cells (sub-clusters 1 and 3) in terms of their BEC and LEC identity shows that cells in the network cluster exhibit significantly higher expression of genes typically found in LECs. A list of marker genes indicative of either the BEC or LEC phenotype was generated based on a dataset from Grimm et al. (2023). The network cells were profiled by applying these gene lists, and the per-cell index is shown (see Materials and Methods for details). A value of 1 indicates expression of all analyzed BEC or LEC marker genes in a given cell. (J) Differentially expressed genes in network cells versus BLECs. (K) Differentially expressed genes in network cells versus meningeal BECs.
Comparing the expression profile of network cells to the BECs cluster demonstrated that network cells expressed lymphatic markers like mrc1a, stab1, flt4, stab2 and dab2, which were low and/or absent in cells from meningeal blood vessels. Blood vascular markers such as flt1 and kdrl, in turn, were expressed by BECs, but expression was low or absent in network endothelial cells (Fig. 5F,K). Furthermore, network cells expressed cdh6, whereas BECs expressed cdh5 (Fig. 5H,K), consistent with previous RNA sequencing data establishing cdh5 as a marker for blood ECs and cdh6 as a marker for lymphatic endothelial cells (Grimm et al., 2023). In addition, when the expression profile of network cells was compared to the BLEC population from control brains, many BEC markers (fabp4b, icn, s100a10b, cav1, rgcc and plscr3b) were upregulated in network ECs compared to BLECs (Fig. 5G,J). The BLEC population, however, shows higher expression of markers such as lyve1b, osr2 and mafa when compared to the network (Fig. 5H). Comparing the expression profile of the network to known (embryonic) LEC and BEC markers, the network cluster expressed significantly more markers that are typically expressed by LECs (Fig. 5I). Venn diagram analysis (Fig. S4D) revealed that BLECs and network cells both expressed scavenging receptors such as mrc1a, stab1, stab2 and dab2, as well as genes that are known to be active in lysosomes and cytoplasmic vesicles, such as hyal2a, hyal2b, ap1b1, hexb, ctsla, lgmn and ctsk. Most of these markers are not expressed or only at low levels by BECs (Fig. 5H, Fig. S4E). Furthermore, network cells expressed many junctional markers, consistent with the notion that these cells form vascular structures and do not remain as single cells, as BLECs do (Fig. S4D). When comparing the network cells to BECs, Gene Ontology (GO) analysis revealed that, in BECs, genes associated with ‘organic transport’ and ‘transmembrane transport’ are enriched, whereas in network cells, the GO terms ‘lymphangiogenesis’ and ‘lymph vessel morphogenesis/development’, as well as ‘receptor-mediated endocytosis’, are more prominent (Fig. S4F). When comparing the network to BLECs, we only see an enrichment of genes that are associated with the GO term ‘translation’ in the network (data not shown). Thus, the scRNA sequencing analysis confirmed that network cells constitute a previously unreported cell population distinct from both BECs and BLECs, which cannot be found in the leptomeninges of control-injected zebrafish.
Given the aforementioned expression of several scavenging receptors and the enriched GO term ‘receptor-mediated endocytosis’, we wondered whether this network reflects a compensatory structure that is formed in response to the lack of BLEC activity in the leptomeninges. To address this possibility, we performed injections of a known BLEC substrate into the brain parenchyma, which had been shown to be efficiently endocytosed by BLECs during embryonic stages (Huisman et al., 2021). Injections of IgG-Alexa-647 in control fish indeed revealed efficient uptake of the dye into small vesicles distinct from the previously reported large autofluorescent vacuoles in BLECs (Galanternik et al., 2017) (Fig. 6A-E). In T-DM1-treated fish, endothelial cells within the network also exhibited the capacity to endocytose injected dye (Fig. 6F-G′), confirming the scRNA sequencing results on a functional level and suggesting a possible compensatory function of this vascular structure.
Cells of the vascular network are capable of endocytosing injected IgG-Alexa-647, similar to BLECs in control brains. (A-B′) Maximum projection showing BLECs (green) closely associated with meningeal blood vessels (red) at the dorsal tectum opticum (TeO) of a mrc1a:mCitrine; kdrl:mCherry-positive brain dissected from a 6-month-old uninjected control fish (A,A′) or IgG-Alexa-647-injected fish (B,B′). The white arrows mark a large autofluorescent vacuole of a BLEC. The arrowheads highlight small vesicles strongly positive for IgG-Alexa-647 (B,B′), which are not visible in uninjected controls (A,A′). (C) Schematic dorsal view of the head region of an adult zebrafish, indicating the zoomed-in area shown in D and F. (D) Maximum projection showing the dorsal TeO of a mrc1a:mCitrine;kdrl:mCherry-positive brain dissected from a 6-month-old control fish that was injected intracranially with IgG-Alexa-647, showing BLECs (green) and meningeal blood vessels (red). The rectangle indicates the zoomed-in area in E and E′. (E,E′) BLECs take up IgG-Alexa-647 in small vesicles (arrowheads). Large autofluorescent vacuoles are marked with arrows. (F) Maximum projection showing the dorsal TeO of a mrc1a:mCitrine;kdrl:mCherry-positive brain dissected from a 6-month-old fish with ablated BLECs. The fish was injected intracranially with IgG-Alexa-647. The rectangle indicates the zoomed-in area in G and G′. (G,G′) The ectopic network takes up IgG-Alexa-647 intracellularly (arrowheads). Large autofluorescent vacuoles, as observed in BLECs, are not visible in these cells. The fish were imaged at a maximum of 24 h post-injection.
Chronic inflammatory response and increased numbers of microglia in brains showing the ectopic vascular network
In the embryonic brain, microglia and BLECs work in tandem and are both important for removing different substance classes from the CSF and the brain parenchyma (Huisman et al., 2021). We therefore addressed whether the lack of BLECs would have an impact on macrophages based on our scRNA sequencing data. Clusters 0 and 2 of the Seurat analysis both represented myeloid cell clusters, most likely composed of dendritic cells or macrophages. Since cluster 2 was almost exclusively composed of cells from the T-DM1-treated plate, analysis of differentially expressed genes in these clusters was conducted (Fig. 7A,B). Cluster 0 was composed of cells from the control and the T-DM1-treated plate, and represents cells with an early cellular response to stress or stimulation (positive for expression of ier2a, ier2b, junba, junbb, fosab, egr1, egr3, nr4a1, nr4a3, dusp1, dusp2, mapk11, ddit3 and gadd45bb) characterized by transcription factors and signaling regulators that mark the initial phase of an immune response. Cluster 2 was composed of cells from the T-DM1-treated plate, with an increased expression of immune mediators such as chemokines (ccl34b.8, ccl39.1, ccl39.3, tnfsf10, cxcl8b.1 and gbp1), immune receptors (marco and mrc1a), activated macrophages (cd68 and cd63), increased antigen presentation (Mhc genes), as well as lysosomal/catabolic enzymes (ctsk, ctsc, ctsz, lgmn, npc2.1 and npc2.2). Thus, the gene signature showed strong enrichment in the expression of genes that are typical of a chronic inflammatory response. To verify whether we could detect changes in macrophages in the T-DM1-treated fish, we utilized the mpeg:GFP transgene that marks microglia. Analyzing the area that was covered by microglia revealed that in fish with no BLECs and a ventral network only, the number of microglia was not significantly increased on the dorsal side of the brain (P=0.151, Fig. 7C,D,E,H), suggesting that microglia might not compensate for the loss of BLECs, at least not by proliferation. However, the number of microglia was significantly increased on the ventral side where the network was present (P=0.0015, Fig. 7C,F,G,I). Furthermore, during brain dissection, control brains appeared white, whereas brains in which the network extended up to the dorsal surface appeared darker. Some of those brains exhibited hemorrhaging, which was never observed in control brains. In addition, comparing the relative brain weight of fish with BLECs and fish with a network revealed a significant increase (P=0.0325) from 1.6% to 3% in the latter group (Fig. 7J). Importantly, the described inflammatory response does not seem to reflect a direct effect of the T-DM1 injections: T-DM1-injected embryos show neither increased expression levels of the proinflammatory cytokine IL1b at 5 dpf compared to the control-injected group (quantified via RT-PCR, Fig. S5D) nor an increase in the number of microglia at 8 dpf (Fig. S5A-C).
*After ablation of BLECs, immune cells show gene signatures indicating a chronic inflammatory response, and the number of microglia is increased in areas covered by the endothelial network. (A) Seurat cluster analysis of clusters 0 (red) and 2 (green), indicating the origin of the cells (control plate, red; T-DM1 plate, blue). (B) Volcano plot of differentially expressed genes. Cluster 0, which is composed of cells from the control and T-DM1 plate, represents dendritic cells with an initial inflammatory response, whereas cluster 2, which is composed of cells from the ‘T-DM1-treated’ plate, represents dendritic cells with a chronic inflammatory response. (C) Schematic overview of the dorsal and ventral head region of a zebrafish, indicating the zoomed-in areas in D,E (left) and F,G (right). (D-G) Maximum projections showing the dorsal (D,E) or ventral (F,G) TeO of lyve1:dsRed;mpeg:GFP-positive brains dissected from 1.5-year-old control (D,F) or T-DM1-injected (E,G) fish. Rectangles indicate the zoomed-in areas of the green channel shown on the right. Control fish have BLECs (red) and microglia (green) (D,F), whereas, in T-DM1-injected fish, BLECs are absent (E,G). The ectopic endothelial network is lyve1 positive, and the number of microglia is increased in T-DM1-injected fish (G). (H) Quantification of the percentage area that was covered with microglia at the dorsal brain surface [n=5 fish per group; P=0.151 (ns, not significant) Mann–Whitney U-test]. (I) Quantification of the percentage area that was covered with microglia at the ventral brain surface (n=5 fish per group; **P=0.00151, unpaired Student's t-test). (J) Dissected brains of T-DM1-treated fish with the ectopic vessel network had a higher relative brain weight compared to control brains (n=3 brains per group, P=0.0325, unpaired Student's t-test). ns, not significant.
DISCUSSION
This study aimed to functionally analyze the role of BLECs by ablating these cells and monitoring possible consequences for the brain. The ADC that we utilized to ablate BLECs consists of a humanized anti-HER2-IgG1 antibody covalently linked to the chemotherapy agent DM1 and is used in human patients to deliver potent chemotherapy to HER2-overexpressing metastatic breast cancer cells. Previous studies have shown that tumor cells with normal HER2 expression and healthy cells are unaffected by T-DM1 (Lewis Phillips et al., 2008). This can be explained by its mode of action, as T-DM1 binds specifically to the HER2 receptor, which is overexpressed on HER2-positive cancer cells in human patients (Slamon et al., 1989). Healthy cells have fewer HER2 on their surface (Rubin and Yarden, 2001) and consequently take up less T-DM1, resulting in DM1 concentrations below the cytotoxic threshold (Hunter et al., 2020; Lewis Phillips et al., 2008). While Her2 is also expressed in zebrafish, protein alignment of the human (Q9UK79) and zebrafish (Q90464) Her2 amino acid sequence reveals only 22.89% identity on the amino acid level, indicating that the monoclonal antibody most likely does not bind to the zebrafish protein and is not predicted to affect cells lacking general scavenging functions for CSF proteins, which is in line with our observations on treated embryos.
High concentrations (5 ng, 10 ng) of T-DM1 injected into embryos at 3 dpf ablate BLECs, but also cause edema formation around the eyes and the gut in most of the treated embryos. Since LECs of the facial lymphatic network are also ablated after T-DM1 injection, the question arose of whether facial lymphatic endothelial cells could also fulfill a scavenging function. Indeed, upon injection of fluorescent-labeled IgG into the CSF, facial lymphatics showed intracellular internalization, supporting the notion that they can also exert a scavenging function. This is supported by the fact that lymphatic vessels not only take up fluid intraluminally but also transport substances through vesicular or transcellular pathways across the LECs (Triacca et al., 2017). The zebrafish facial lymphatic system performs an essential role in the regulation of fluid homeostasis and becomes functional at around 4-5 dpf. Mutations in genes affecting facial lymphatic structures, such as itga9, lead to edema around the gut and the eyes (Shin et al., 2019), mimicking the effects we observed here. Injection of a lower concentration (1 ng) of T-DM1 into the CSF results in BLEC ablation, but does not negatively affect the development of intracranial or facial lymphatic structures in most embryos.
Employing this new model, we found that BLECs can regenerate and reconstitute near-normal BLEC numbers if some cells remain on the dorsal TeO due to incomplete ablation, a notion consistent with laser ablation of only some BLECs (Bower et al., 2017). However, our results highlight that after complete ablation of the dorsal BLEC population, this cell population does not regenerate and is also absent in adult stages, indicating that ablated BLECs are not replaced by newly differentiating cells. There was no change in outcome regardless of whether or not a low number of mrc1a-positive cells were left on the ventral side of the brain. Only if at least 1-3 BLECs remain dorsally will they be able to proliferate and repopulate the TeO at later stages of the development.
Further analysis revealed that fish lacking BLECs develop a mrc1a-, flt4-, tjp1a-, lyve1b-positive, but kdrl- and flt1-negative vessel network, suggesting lymphatic characteristics of this endothelial cell population. The network first appeared on the ventral side of the brain, also extending to the dorsal side in older fish. Adult zebrafish in which the ectopic vessel network also reached the dorsal side of the brain showed defects in the meningeal blood vasculature, with 50% fewer branches and junctions, and increased average branch length and vessel thickness. Theoretically, it is possible that T-DM1 directly affects the meningeal vasculature, but we consider this unlikely for a number of reasons. First, T-DM1 has an anti-angiogenic effect within PT-xenograft primary tumors (Lange et al., 2011). However, during cancer therapies, the ADC is injected directly into the blood vascular system, whereas the injection in zebrafish was performed into the CSF. Thus, T-DM1 does not come into direct contact with BECs in the treated embryos. Second, while expression of zebrafish Her2 in meningeal blood vessels has not been shown, T-DM1 is known to affect only cells with high HER2 expression levels; cells with normal HER2 expression are not impaired (Lewis Phillips et al., 2008). Additionally, the active metabolite of T-DM1 is lysine-MCC-DM1. These maytansinoid-linker-lysine derivatives have been tested as pure compounds in vitro, and it was reported that their cytotoxic effect on surrounding tissue is limited, probably because diffusion across the plasma membrane into neighboring cells is prevented due to their hydrophilicity and charge (Kovtun and Goldmacher, 2007). Therefore, bystander effects, which describe the ability to kill other cells in the vicinity and not only cells that have taken up the ADC, are not caused by T-DM1 (Kovtun and Goldmacher, 2007; Kovtun et al., 2006). In conclusion, it appears unlikely that T-DM1 is directly responsible for defects in meningeal blood vessels.
Previously, it has been shown that BLECs express pro-angiogenic factors such as vegfaa and vegfab (Bower et al., 2017; Galanternik et al., 2017). Thus, it is likely that important signaling molecules cannot be provided in the absence of BLECs, which would impact the surrounding vasculature. However, in our study, younger fish between 1 and 3 months of age without any BLECs, in which the network did not reach the dorsal side of the TeO, had slightly, but not significantly, fewer branches (P=0.0516) and junctions (P=0.0691) in the meningeal blood vasculature. Thus, the meningeal blood vessel defects correlate either with age or with the development of the newly discovered network. The disruption of the meningeal blood vessels could also affect the integrity of the blood-brain barrier, which would result in an entry point for pathogens and adverse substances into the brain parenchyma (Rua and McGavern, 2018). Furthermore, the brain is the organ with the highest oxygen consumption, consuming up to 25% of the body's energy (Chen and Zhang, 2021). Abnormal blood vessels in the meninges might therefore interfere with oxygen distribution in the brain, which would lead to further pathological effects and a worsening of the phenotype.
BLECs are located distal to the arachnoid mater of the meninges, which consists of two overlapping layers connected by adherens junctions, Tjp1a-positive tight junctions and tricellular junctions to regulate the transport of molecules across this meningeal layer in mice (Betsholtz et al., 2024). In fish without BLECs, the lymphatic vessel network was located in the same leptomeningeal layer as the BLECs in control fish. However, the arachnoid mater barrier seemed disrupted. Impairment of the meningeal barrier has a serious impact on the homeostasis of the brain, as disrupted meninges provide another entry route for pathogens and substances from the outer layers into the parenchyma, leading to an inflammatory response. Furthermore, acute or chronic immune responses in the meningeal layers can result in neurological dysfunctions due to the penetration of the infection into the brain parenchyma (Rua and McGavern, 2018).
Using different imaging approaches, including tissue clearing and light sheet microscopy, we tried to identify connection points between the network and the endogenous vasculature. However, although we occasionally noticed blood cells in lumenized parts of the network, we could not identify such a connection site. This might be due to the fact that our approaches are hampered by the need to isolate the brain for imaging, and hence connections to distally positioned vessels would therefore be mechanically disrupted. Further studies are therefore required to identify the local origin of this ectopic network and its connection to the endogenous vasculature.
Transcriptome analysis of the network utilizing scRNA sequencing revealed that the network is composed of a distinct endothelial cell population that is not present in the leptomeninges of control fish. The network shares features of BLECs, both molecularly by expressing scavenging receptors, as well as functionally, by endocytosing substances injected into the brain. Thus, the endothelial cells of the network can compensate on a functional level for the absence of BLECs, at least to some extent. In line with this assumption, the network was evident only in cases where the BLEC population was not able to recover. We therefore conclude that the network develops only upon permanent loss of BLEC function.
Importantly, network cells express the lymphatic marker cdh6, whereas BECs express the venous blood marker cdh5, strongly suggesting that the network has more lymphatic than venous characteristics. In line with these conclusions, a comparison of the genes expressed by network cells to published datasets shows that the network has significantly more characteristics of LECs than BECs. However, since there are, to our knowledge, no datasets from adult fish available that contain enough LECs, we had to use a dataset from embryonic stages. We cannot rule out that the network is of a hybrid identity, since LECs and venous BECs share many markers. In fact, a lot of lymphatic markers are also expressed by venous endothelial cells during embryonic stages. It would be important to compare the expression signature of the network to the meningeal lymphatics in the dura mater in the future.
The RNA sequencing analysis suggested that in T-DM1-treated fish, a chronic immune response is activated, and further analysis indeed showed an increased number of microglia in areas where the network is present. Furthermore, brains with the compensatory network have a higher relative weight compared to brains with BLECs, which is also a sign of inflammatory processes and indicates an imbalance in brain homeostasis. In control fish, an initial or early immune response is activated. As we detected neither increased expression of Il1b at 5 dpf, nor an increased amount of microglia at 8 dpf, the injection of T-DM1 does not immediately result in an inflammatory response. The question is, which initial process is responsible for the chronic immune response in T-DM1-treated brains? There are a number of options, such as loss of BLECs, the altered meningeal blood vessels, the disruption of the arachnoid mater, the increased amount of microglia or a specific property of cells within the network itself. We hypothesize that the loss of BLECs leads to malformations of meningeal blood vessels, which leads to an imbalance in homeostasis, causing a chronic inflammatory response and the development of the compensatory network. Whether the network also forms under other pathophysiological conditions, such as brain edema or disruption of the meningeal blood vessels in adult fish due to hemorrhage, stroke or imbalanced homeostasis, is an interesting question that remains to be answered in the future. We have evidence that, in fish with brain edema formation, a comparable network can form, but further analysis is required to reveal the underlying mechanisms. Furthermore, it would be interesting to analyze whether mice are capable of developing such a vessel network. However, it is not firmly established whether mammals have BLECs, even though cells with comparable features have been described (Shibata-Germanos et al., 2020). Further analysis is required to identify the analogous cell populations in mammalian leptomeninges. Our ablation model in zebrafish can be instrumental not only for better understanding of the physiological role of BLECs but also for assessing the molecular mechanism driving the development of the ectopic leptomenigeal vascular network described here, and for better understanding the pathological processes in the brain and meninges.
MATERIALS AND METHODS
Zebrafish strains
Research on animals was conducted according to the guidelines of the animal ethics committees of the University of Münster, Germany. Zebrafish strains were maintained under standard husbandry conditions according to FELASA guidelines (Aleström et al., 2020). Adult zebrafish were kept at 26°C and embryos at 28.5°C with a 14 h light and 10 h dark cycle. In this study, the following transgenic and mutant lines were used: Tg(kdr-l:HRAS-mCherry-CAAX)^s916^, referred to as kdrl:mcherry (Hogan et al., 2009); Tg(lyve1:dsRed2)^nz101^ (Okuda et al., 2012); Tg(flt4:mCitrine)^hu7135^ (van Impel et al., 2014); Tg(flt1enh:tdTomato)^hu5333^ (Bussmann and Schulte-Merker, 2011); Tg(mpeg1:EGFP)^gl22^ (Ellett et al., 2011); Tg(fli1a:eGFP)^y1^ (Lawson and Weinstein, 2002); Tg(mrc1a:mCitrine)^mu409^; and Casper (White et al., 2008). The TgKI(tjp1a-tdTomato)^mu427^ line was generated as described previously (Levic et al., 2021). A 522 amino acid long tjp1a fragment (ENSDART00000148347.5) (tjp1a_Intron27_28+Ex28_fwd, ACTATAGTGAGTCGTATTACAGTTTCGATGACCACAGGGTCGAAA; tjp1a_Intron27_28+Ex28_rev, CCTTGCTCACGAAATGGTCAATAAGCACAG) was cloned in pCS2+ containing the fluorescent protein coding sequence of td:Tomato and two ubb poly-adenylation sequences. The gRNA target sites used in this study were: tjp1a_target_site1, TGCGAATAGGGGTTGGTAAT; and tjp1a_target_site2, GAGTTTCGATGACCACAGGG.
Cerebroventricular microinjection
Injections were carried out using a pneumatic PicoPump with glass capillary needles (Science Products, GB100TF-10) generated by a Micropipette Puller (Shutter Instruments, P-1000). For injections into the brain ventricles of 3- to 5-day-old larvae, the embryos were anesthetized with 0.0168% tricaine (MS-222, Sigma, A5040) and embedded (dorsal side up) in 1.5% low-melting point agarose (ThermoFisher, 16520100) dissolved in E3-Medium [5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl_2_ and 0.33 mM MgSO_4_ (pH 7.2)] containing MS-222. The needle was inserted into the brain at a sloped angle, and care was taken not to penetrate too deeply into the brain tissue. The total injection volume for each injection was 1 nl per bolus. After injection, the embryos were put in fresh E3-Medium and transferred to the nursery facility in order to raise them. Alternatively, they were embedded in 0.7% agarose for confocal imaging. All embryos were routinely checked.
Juvenile and adult fish were anesthetized with 0.0168% tricaine (MS-222, Sigma, A5040) and embedded (dorsal side up) in a mold made of agarose to secure the fish. Using the sharp edge of a fine needle with a short bevel (30G) (305178, BD Biosciences), the cranial bone was penetrated. This incision generates a slit through the skull with a diameter of ∼200 µm without damaging the brain underneath (Jurisch-Yaksi et al., 2020). With a glass capillary needle, up to 2 µl of the IgG-Alexa647 was injected. After a maximum of 24 h post-injection and strict monitoring, the fish was euthanized and analyzed. All procedures were approved by the local animal care committee [Landesamt für Natur, Umwelt und Verbraucherschutz (LANUV) Nordrhein-Westfalen: 81-02.04.2023.A145] and the University of Münster.
Dyes and other substances used for injection
For injection, the following fluorescent dyes and concentrations were used: Alexa-Fluor-647 cross-adsorbed secondary antibody (2 mg/ml, ThermoFisher, A21447); and Kadcyla (Trastuzumab Estansine, Roche), kindly provided by the Cytostatic Department of the Universitätsklinikum Münster (UKM).
Brain dissection
Juvenile and adult fish were euthanized and brains were dissected manually in phosphate-buffered saline (PBS). The youngest possible stage for dissection was at about 3 weeks of age. First, fine scissors were used to remove both eyes, in order to avoid tearing off the forebrain. Subsequently, a lateral cut was carried out through both sides of the corner of the mouth to the caudal end of the head. Forceps were used to lift the top of the skull (calvaria) from caudal to rostral. The exposed brain was freed from tissue with forceps and placed in fresh PBS. The time between euthanizing the fish and imaging the dissected brain was kept to a minimum (about 15 min).
Imaging
To inhibit pigment formation, embryos were treated with 1-phenyl-2-thiourea (PTU, Sigma-Aldrich, P7629) initiated before 24 hpf. Up to the age of 5 dpf, zebrafish larvae were anesthetized with MS-222 and mounted in an imaging dish in 0.7% low-melting-point agarose supplemented with 0.0168% tricaine (Sigma A-5040). Brains of larvae, juveniles and adult fish in a Casper background older than 120 hpf were embedded in agarose for imaging. Dissected brains were placed in an imaging dish containing PBS. Images were taken with a Leica SP8 microscope using 10× dry, 20× dry or 40× water immersion objectives. Z-stacks were acquired at 3 μm to 5 μm increments. The confocal stacks were flattened by maximum projection and processed using the software Fiji-ImageJ version 1.52 f. Figures were assembled using Adobe Illustrator.
For the analysis of meningeal blood vessels, a macro was used in Fiji-ImageJ using the plug-ins RidgeDetection 1.4.1 (Wagner et al., 2017), Tubeness (Sato et al., 1998) and AnalyzeSkeleton (Arganda-Carreras et al., 2010) (see supplementary Materials and Methods).
Immunostaining, tissue clearing and light sheet microscopy
After dissection, brains were fixed in 4% paraformaldehyde (PFA) (Sigma 158127) at 4°C for 8-12 h and washed four times for 30 min with 1× PBS containing 0.3% Triton X-100 (Tx, Sigma-Aldrich; X100). Brains were permeabilized with 1% Tx/5% dimethyl sulfoxide (DMSO; Sigma-Aldrich, 472301) in 1× PBS for 24 h in the dark at room temperature and rinsed with 1× PBS-0.3% Tx four times for 30 min. Samples were blocked in 1× PBS containing 0.3% Tx, 10% donkey serum, 1% bovine serum albumin (Sigma-Aldrich; A7906) and 0.04% w/v sodium azide overnight at room temperature in the dark. Brains were incubated with primary antibodies [chicken anti-GFP (Abcam, ab13970; 1:200) and rabbit anti-RFP (Abcam, ab62341; 1:200) in blocking solution for 7 days and rinsed afterwards four times in 1× PBS-Tx 0.3% for 1 h each. Brains were incubated with the secondary antibodies [donkey anti-chicken-488 (Thermo Scientific, A78948; 1:500) and donkey anti-goat-568 (Thermo Scientific, A10042; 1:500)] diluted 1:500 in 1× PBS-0.3% Tx containing 0.04% sodium azide for 7 days and rinsed in 1× PBS-0.3% Tx four times for 1 h. Each brain was embedded in 1% low-melting agarose in a well of a 24-well plate. Using a scalpel, the agarose was trimmed into cubes, slightly larger than the embedded brains. For dehydration, trimmed blocks were transferred into glass tubes and put in increasing concentrations of 50% and 70% methanol for at least 1 h each and incubated in >99.5% methanol for 4 h. The methanol was replaced, and brains were incubated for another 3 h. For clearing, brains were put in a 1:1 mixture of >99.5% methanol and BABB [1:2 ratio of benzyl alcohol (Sigma-Aldrich, 402834): benzyl benzoate (Sigma-Aldrich, B6630)] for 1 day. Afterwards, samples were put in 100% BABB, and BABB was replaced after 6 h. Brains were imaged using LaVision UltraMicroscope II SuperPlan with a 4× objective and images were analyzed with Arivis Vision4D.
Statistical analysis and data visualization
Datasets were tested for normality (Shapiro–Wilk) and equal variance. P-values of datasets with normal distribution were determined using an unpaired Student's t-test. In cases where data values did not show normal distribution, a Mann–Whitney U-test was performed instead. For multiple comparisons, the P-values were adjusted with Bonferroni correction. Results are presented as mean±s.d. Statistical tests were performed using R software (Version 4.4.2). For visualizations, the R package ggplot2 was used. P>0.05 was considered not significant (*P≤0.05, **P≤0.01, ***P≤0.001, ****P≤0.0001).
RNA isolation, cDNA synthesis and qPCR
Total RNAs were extracted using RNeasy Mini Kit (Qiagen, 74104). cDNA synthesis reactions and qPCR were performed as described in the manufacturer's instructions [M-MLV RT (H–), Point Mutant (Promega, M5301); SYBR™ Green Universal Master Mix (Applied Biosystems, 4309155)] and normalized against the expression of eef1a1l1 as a housekeeping gene. The following primers were used for qPCR: eef1a1l1_fwd, CCTCTTTCTGTTACCTGGCAAA; eef1a1l1_rev, GAGTCGACGTGGCCAATAA; il1b_fwd, GCCTGTGTGTTTGGGAATCT; il1b_rev, TGATAAACCAACCGGGACA (Shiau et al., 2013).
Smart-Seq3
For Smart-Seq3 analysis, three brains per group were dissected as described and put in a buffer containing 0.5% Trypsin, 1 mM EDTA and 10 U DNaseI (Promega) for 10 min at 30°C and shaken at 600 rpm every 3 min until the BLECs were detached from the brain. This was controlled under a fluorescent stereo microscope. Brains were removed, and the cells were centrifuged at 400 g for 5 min. The pellet was washed and resuspended in PBS containing 3% FCS and then filtered through a 40 µm nylon cell strainer. Cell sorting and analysis was performed on a FACSAria Fusion cell sorter (BD Biosciences) with FACSDiva 8.0.2 software using a 100 µm nozzle. Flow cytometric data was analyzed using FlowJo software (version 10.10.0, BD Biosciences). Single cells were sorted for mCitrine and mCherry into 384-well plates, containing lysis buffer (3 µl per well: 0.15% Triton X-100, 0.5 U/ml RNase inhibitor, 0.5 mM dNTP, 1 µM Smart-Seq3 oligo-dT primer, 5% PEG and ERCC at 1:4×10^4^ dilution), and stored at −80°C until further processing.
Single-cell libraries were prepared using the SmartSeq3 protocol, as previously published (Hagemann-Jensen et al., 2020). In brief, after oligo-dT priming, mRNA was reverse transcribed to yield cDNA using template switching and Maxima H-minus reverse transcriptase (ThermoFisher Scientific). The full cDNA was then amplified using PCR and quality controlled (QC) using a TapeStation 4200 (Agilent Biotechnologies). Samples that passed QC were then fragmented, tagged (tagmented) using Tn5 transposase and individually indexed. The indexed libraries were sequenced on a NovaSeq 6000 PE150 sequencer (Illumina) at the National Genomics Infrastructure (NGI) Sweden, SciLifeLab.
The sequences obtained from the pooled samples were de-multiplexed yielding single-cell fastq files that quality-trimmed using Trim Galore and UMI were extracted using the UMI-tools package. Reads were then mapped to the Danio rerio reference genome GRCz11 using STAR and duplicates were removed using UMI-tools dedup based on identified UMIs. Further processing, including generation of per cell and gene raw-count expression matrices using the featureCounts package, has been described previously (Hagemann-Jensen et al., 2020; Wang et al., 2020). The annotation of the ENSEMBLE identifiers was carried out using the org.Dr.db.eg package (version 3.14.0) in R-software (version 4.1.1, R core team Vienna Austria). The annotated raw counts were loaded into R-software, and the Seurat package (version 4.3.0) (Satija et al., 2015) was used for organization and basic processing. Cells with low quality were filtered out using the thresholds: ≤25,000 counts library size, detected genes ≤1000 or ≥12,000 (doublets), ≥10% mitochondrial genes, ≥10% ERCC counts, resulting in a dataset of 525 single-cell transcriptomes. Genes with low expression in ≤3 cells with a detection limit of 20 counts and/or ≤300 cumulative counts per gene were removed from the dataset before further processing. The Seurat pipeline was applied to calculate general attributes [normalization, PCA (n=50), UMAP] of the dataset. Cluster identification was carried out using the FindClusters() function with the resolution set to res=0.8, resulting in eight distinguishable clusters. Differential gene expression analysis was carried out using the FindMarkers() function, with MAST test (version 1.20.0) option for P-value adjustment. For gene qualification, the thresholds adjusted P-value ≤0.05, log2 fold-change ≥1 and expression in ≥30% (pct) of cells were applied. For data visualization in UMAP, DimPlot() functions from the Seurat package were applied. For data visualization in heatmaps, the pheatmap (version: 1.0.12) R-software package was used. For the sub-analysis of cells in cluster 1, the pagoda2 (https://github.com/kharchenkolab/pagoda2) R-software package was used for PCA (n=50), and cluster identification (k=30, distance=“cosine”). The pagoda2 results were organized within the Seurat object, and embedded functions from the Seurat R-software package were used for visualization as described above.
Genes were ranked by log2 fold change values and subjected to gene set enrichment analysis (GSEA) for GO biological processes using the R-software package clusterProfiler (https://bioconductor.org/packages/clusterProfiler/) with P-value cutoff=0.05. For Danio rerio, the org.Dr.eg.db database (https://bioconductor.org/packages/org.Dr.eg.db/) was used.
The per-cell index for either BEC or LEC indicative expression markers was calculated using gene lists generated using wild-type cells of the Grimm et al. (2023) dataset. In brief, scRNA sequencing data was retrieved from GEO (GSE188341) and reprocessed using R-software and the Seurat package as described above. Cells indicated with wild-type origin were used to obtain gene lists representative of either the BEC or LEC phenotype. Cell clusters containing BEC or LEC were identified by canonical marker gene expression, and differential gene expression between these clusters was performed using the FindMarkers() function as described above. Genes with adjusted P-value ≤0.05, log2 fold-change ≥1, and expression in ≥30% of cells were selected for BEC and LEC gene lists, respectively.
For index calculation, first the expression value of each gene was normalized to its highest value across all network cells. The sum of gene expression values for each cell was calculated and divided by the number of marker genes included in the BEC (71 genes) or LEC (60 genes) signature, resulting in the per-cell index for either BEC or LEC phenotype.
Supplementary Material
10.1242/develop.204988_sup1Supplementary information
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Absinta, M., Ha, S.-K., Nair, G., Sati, P., Luciano, N. J., Palisoc, M., Louveau, A., Zaghloul, K. A., Pittaluga, S., Kipnis, J. et al. (2017). Human and nonhuman primate meninges harbor lymphatic vessels that can be visualized noninvasively by MRI. e Life 6, e 29738. 10.7554/e Life.2973828971799 PMC 5626482 · doi ↗ · pubmed ↗
- 2Akhmanova, A. and Steinmetz, M. O. (2015). Control of microtubule organization and dynamics: two ends in the limelight. Nat. Rev. Mol. Cell Biol. 16, 711-726. 10.1038/nrm 408426562752 · doi ↗ · pubmed ↗
- 3Aleström, P., D'angelo, L., Midtlyng, P. J., Schorderet, D. F., Schulte-Merker, S., Sohm, F. and Warner, S. (2020). Zebrafish: housing and husbandry recommendations. Lab. Anim. 54, 213-224. 10.1177/002367721986903731510859 PMC 7301644 · doi ↗ · pubmed ↗
- 4Arganda-Carreras, I., Fernández-González, R., Muñoz-Barrutia, A. and Ortiz-De-Solorzano, C. (2010). 3D reconstruction of histological sections: application to mammary gland tissue. Microsc. Res. Tech. 73, 1019-1029. 10.1002/jemt.2082920232465 · doi ↗ · pubmed ↗
- 5Aspelund, A., Antila, S., Proulx, S. T., Karlsen, T. V., Karaman, S., Detmar, M., Wiig, H. and Alitalo, K. (2015). A dural lymphatic vascular system that drains brain interstitial fluid and macromolecules. J. Exp. Med. 212, 991-999. 10.1084/jem.2014229026077718 PMC 4493418 · doi ↗ · pubmed ↗
- 6Barok, M., Joensuu, H. and Isola, J. (2014). Trastuzumab emtansine: Mechanisms of action and drug resistance. Breast Cancer Res. 16, 209. 10.1186/bcr 362124887180 PMC 4058749 · doi ↗ · pubmed ↗
- 7Betsholtz, C., Engelhardt, B., Koh, G. Y., Mc Donald, D. M., Proulx, S. T. and Siegenthaler, J. (2024). Advances and controversies in meningeal biology. Nat. Neurosci. 27, 2056-2072. 10.1038/s 41593-024-01701-839333784 PMC 11862877 · doi ↗ · pubmed ↗
- 8Bower, N. I., Koltowska, K., Pichol-Thievend, C., Virshup, I., Paterson, S., Lagendijk, A. K., Wang, W., Lindsey, B. W., Bent, S. J., Baek, S. et al. (2017). Mural lymphatic endothelial cells regulate meningeal angiogenesis in the zebrafish. Nat. Neurosci. 20, 774-783. 10.1038/nn.455828459441 · doi ↗ · pubmed ↗