TLR4 signaling drives tissue inflammation, Claudin-5 internalization, and vascular barrier breakdown in a mouse model of neonatal meningitis

Reviewer #1 (Public review):

Summary:

In this manuscript, Seegren and colleagues demonstrate that in a mouse model of neonatal E. coli meningitis, loss of toll-like receptor 4 (TLR4) in VE-cadherin+ endothelial cells and a subset of meningeal fibroblasts leads to a marked decrease in transcriptional dysregulation across multiple leptomeningeal cell types, a decrease in vascular permeability, and a decrease in macrophage abundance. In contrast, loss of macrophage TLR4 had less pronounced effects. Using cultured wildtype and TLR4-knockout endothelial cells, the authors further demonstrate that TLR4 signaling leads to reversible internalization of the tight junction protein claudin-5, establishing a potential mechanism of increased vascular permeability. Authors also show that claudin-5 internalization is independent of NF-κB. Finally, the authors use RNA-sequencing of wildtype and TLR4-knockout endothelial cells to define the TLR4-dependent cell-autonomous transcriptional response to E. coli.

Comments on revised version.

The authors have considerably improved and strengthened the work through the addition of new experimental data, new data analyses, and modifications to their interpretation. Notably, the authors used additional Cre-reporter mice to clarify that Cdh5-CreER is active in endothelial cells and some meningeal fibroblasts, and thus revised nomenclature and interpretation to acknowledge that the Tlr4fl/-;Cdh5-CreER cKO (Tlr4-VEKO) is not exclusively endothelial. The authors also demonstrated that Tlr4-VEKO does not affect peripheral E.coli burden, but acknowledge that changes to periphery-derived signals (e.g., cytokines) may contribute to observed leptomeningeal phenotypes.

The authors added PCA plots to show similarity in gene expression shifts across biological replicates (mice). This provides support for the claim that Tlr4-VEKO attenuates infection-associated transcriptional changes. With respect to differential expression analysis, I agree with authors that characteristics of individual cells (e.g. heterogeneity) are of interest. I remain concerned, however, that the formal differential analysis strategy appears to consider cells as independent experimental units, which they are not because a single cell cannot be randomly assigned to an experimental group (control or cKO, uninfected or infected). The mouse is the correct experimental unit for a comparison across these groups because it can be randomized. I appreciate that many of the gene expression changes appear consistent across mice (e.g. Figure 1 - Figure supplement 7) and that there are clear infection- and genotype-associated phenotypes in other assays. I would simply caution that the authors' analysis strategy likely leads to a larger number of type I errors (false positives) than is generally accepted; a mixed (hierarchical) model or pseudo-bulk approach would be more appropriate for future studies.

Reviewer #2 (Public review):

Summary:

The authors use a postnatal mouse model of E. coli bacterial meningitis and a mouse brain endothelioma cell line combined with cell type specific gene deletion to study the function of endothelial TLR4, a cell surface receptor that recognizes gram positive bacterial wall components, in the local leptomeningeal (LPM) response with a focus on endothelial barrier breakdown mediated by TLR4. Single cell transcriptional profiling and imaging studies using wholemount preps of the LPM support that LPM endothelial, CD206+ local macrophage and LPM fibroblast and arachnoid barrier cell inflammatory response and is abrogated in endothelial specific KO of TLR4, pointing to a role for endothelial TLR4 in local LPM response. Culture studies using Bend3.1 cells (a mouse brain endothelioma cell line) support a direct role for TLR4 in the bacteria-mediated inflammatory response and in internalization of Cldn5 via the endosomal-lysosomal pathway, resulting in loss of barrier integrity

Strengths:

The local LPM cell response in meningitis and the role of specific LPM cells in inflammation and CNS barrier breakdown has not been extensively studied, despite ample evidence for primary immune response in the meninges in human patients and in animal models. The authors employ a robust, multi-model approach using both in vivo and in vitro models with cell-type specific knockout to study the function of TLR4 in brain endothelial cell response. The authors nicely combine functional barrier assays with IF for junctional localization in their experimental design and they delve into potential mechanisms of Cldn5 internalization using markers of endosomal-lysomal pathway localization. The authors also describe a new type of barrier assay using a streptavidin-coated plates upon which barrier forming cell cultures can be plated, this could be a very useful alternative or complement to other size-selective barrier assays and presumably could work for other barrier forming cell types, like epithelial cells.

Comments on revised version.

In their revision, the authors addressed prior noted weaknesses with new data and analysis. They now show that TLR4-VE-cad cKO mice have a largely similar disease progression as control mice, including increased bacterial burden in the LPM and brain. This underscores that that the reduced vascular leakage and blunted inflammatory response is due to loss of TLR4 response to bacteria on VE-cad recombined cells and not because the mice are protected from meningitis. The authors also performed additional experiments to show that Cldn5 internalization via the endosomal-lysosomal pathway is independent of NFKB signaling. The authors also added in important discussion points about how their results fit into the broader literature on TLR4 in BBB endothelial cell junctional protein localization and prior work on meningitis in global TLR4.

Reviewer #3 (Public review):

Summary:

This study investigates the molecular underpinnings of immune responses in the leptomeninges in neonatal bacterial meningitis. Bacterial meningitis is a major disease burden, particularly for neonates, and it has previously been noted that the meningeal immune environment in infants is permissive to opportunistic infection (Kim et al., Sci Immunol, 2023). There is less known about the contribution of the stromal compartment to meningeal immune responses. Seegren et al. interrogate the role of leptomeningeal endothelium in host defense in E. coli infected neonatal mice using mouse genetic tools to delete the LPS receptor Tlr4 from either endothelial cells/stromal cells (using Cdh5-CreER) or myeloid cells (using LysM-Cre). The authors use snRNAseq, cleared cortical mounts, and in vitro work to define the impact of E. coli infection on leptomeningeal endothelial cells. This study uses a range of innovative techniques to probe the role of the stromal compartment in meningitis. With additional experiments to confirm the specificity of their Cre models, this strengthens the interpretation of the study significantly. The only major weakness is the inability to confirm TLR4 knockout in myeloid cells.

Strengths:

This study makes excellent use of cleared cortical mounts to examine the biology of the leptomeninges, in particular, changes to the endothelium, with unprecedented detail. In combination with high-quality sequencing data provide new insights into the impact of meningitis on the leptomeninges. The data presented by the authors is of very high quality.

The authors have also done substantial work to address my two major comments regarding 1) the specificity of their Cre systems and 2) peripheral impacts of the interventions.

(1) The authors identified and acknowledged some impacts in the leptomeningeal stroma (the relatively high level of recombination in ECs vs FBs presumably reflects a single low dose being given, where other groups have done more aggressive tamoxifen regimens that drive recombination in FBs as well). Given the incomplete recombination in the leptomeningeal FBs, I agree with their conclusion that it is probably endothelial driven. Acknowledging the contributions of other myeloid cells with the L. The Cre-NLS experiments with nuclear markers provided excellent data and had beautiful staining.

(2) The authors did not observe differences in bacterial burden in peripheral organs in either CKO model, suggesting that CNS impacts are not downstream of peripheral bacterial control.

Weaknesses:

(1) The inducible Cre lines used by the authors target peripheral tissues as well as CNS tissues. Although this is mollified by the lack of impact on peripheral disease burden.

(2) The authors were not able to confirm TLR4 knockout in myeloid cells, and this caveat is acknowledged. The lack of response in TLR4 VEKO mice strongly suggests successful conditional knockout.

(3) The cell line model (bEnd.3) is a relatively low fidelity model of BBB endothelial cells. The authors acknowledge this, and it is likely that endothelial cell responses to LPS are highly conserved.

(4) It is perhaps not surprising that Tlr4 is required for meningitis responses with E. coli. However, it is unclear if these findings can be generalised to other, more common, meningitis infections (streptococcal/pneumococcal).

Author response:

The following is the authors’ response to the original reviews.

Public Reviews:

Reviewer #1 (Public review):

Summary:

In this manuscript, Seegren and colleagues demonstrate that in a mouse model of neonatal E. coli meningitis, loss of endothelial toll-like receptor 4 (TLR4) leads to a marked decrease in transcriptional dysregulation across multiple leptomeningeal cell types, a decrease in vascular permeability, and a decrease in macrophage abundance. In contrast, loss of macrophage TLR4 had less pronounced effects. Using cultured wild-type and TLR4knockout endothelial cells, the authors further demonstrate that TLR4-NF-κB signaling leads to reversible internalization of the tight junction protein claudin-5, establishing a potential mechanism of increased vascular permeability. Finally, the authors use RNA sequencing of wild-type and TLR4-knockout endothelial cells to define the TLR4dependent cell-autonomous transcriptional response to E. coli.

Strengths:

(1) The authors address an important, well-motivated hypothesis related to the cellular and molecular mechanisms of leptomeningeal inflammation.

(2) The authors use model systems (mouse conditional knockouts and cultured endothelial cells) that are appropriate to address their hypotheses. The data are of high quality.

Weaknesses:

(1) The authors perform single-nucleus RNA-seq on dissected leptomeninges from control and E. coli-infected mice across three genotypes (WT, Tlr4MKO, and Tlr4ECKO). A major discovery from this experiment, as summarized by the authors, is: "Tlr4ECKO mice exhibited a global attenuation of infection-induced transcriptional responses across all major leptomeningeal cell types, as judged by the positions of cell clusters in the UMAP." This conclusion could be considerably strengthened by improving the qualitative and quantitative analysis.

Thank you for this comment. We agree that the UMAP-based interpretation would benefit from additional qualitative and quantitative support. We have expanded the snRNA-seq analysis with additional images and supplemental figures (Figure 1 – figure supplement 3, Figure 1 – figure supplement 5, and Figure 1 – figure supplement 6). The first and third of these new supplemental figures show dot plots for each major leptomeningeal cell type, for each genotype, for the two experimental conditions (infected vs. uninfected), and for individual genes in three immune-related gene sets (NF-kB and TNF-α, JAK-STAT, and IFN-ɣ), providing a more explicit comparison of infection-induced transcriptional responses across genotypes. The second of these new supplemental figure shows principal component analysis (PCA) of the individual snRNA-seq datasets (one mouse per dataset) for each genotype and experimental condition, demonstrating that the observed transcriptional shifts are consistent across biological replicates. Finally, Figure 1 – figure supplement 7, which was included in the original submission, shows changes in the most up- and down-regulated genes (based on adjusted p-value or fold change) in endothelial and myeloid cells across individual mice and genotypes/conditions, further supporting the genotype-dependent effects at the level of individual animals.

(2) The authors interpret E. coli infection-induced increases in leptomeningeal sulfo-NHSbiotin as evidence of compromised BBB integrity (i.e., extravasation from the vasculature) (Results, page 7), but another possible route in this context is sulfo-NHS-biotin entry from the dura across a compromised arachnoid barrier. The complete rescue in Tlr4ECKOs is strongly suggestive that the vascular route dominates, but it would strengthen the work if the authors could assess arachnoid barrier fidelity (e.g. via immunohistochemistry). At a minimum, authors should mention that the sulfo-NHS-biotin signal in this context may represent both vascular and arachnoid barrier extravasation.

Thank you for this comment. We agree that our data cannot rule out leakage across the arachnoid barrier during infection. While the rescue observed in Cdh5-CreER; Tlr4CKO (Tlr4^VEKO) mice strongly supports a dominant vascular contribution, we acknowledge that the sulfo-NHS-biotin signal may reflect permeability at both the vascular and arachnoid barriers. We do not think there is a clear way to directly test this possibility functionally, since the arachnoid barrier appears intact by confocal microscopy. Subtle differences in barrier cell morphology might be detectable by electron microscopy, but this would not definitively address whether infection permits molecular passage across the arachnoid barrier. We have followed the reviewer’s suggestion and revised the Results section to reflect this interpretation. Specifically, we added the following: “The simplest interpretation of these data is that the site of sulfo-NHS biotin leakage is primarily vascular. However, we cannot exclude some contribution from increased arachnoid barrier permeability.”

(3) The authors state that "deletion of TLR4 prevented both NF-κB nuclear translocation and Cldn5 internalization in response to E. coli (Figure 4A-D)" (Results, page 9). In Figures 4C and D, however, there is no indicator of a statistical test directly comparing the two genotypes. A comparison of within-genotype P-values should not be used to support a genotype difference (PMID: 34726155).

Thank you for pointing out this omission. We have updated the figures so that the between-genotype p-values are shown for those panels (including this panel) that had not previously shown them.

(4) In the first paragraph of the Results, the authors summarize the meningeal layers as (1) pia, (2) subarachnoid space, (3) arachnoid, and (4) dura, and then state "The second and third layers constitute the leptomeninges." This definition of leptomeninges seems to omit the pia, which is widely considered part of the leptomeninges (PMID: 37776854).

Thank you for pointing out this error, which has now been corrected.

(5) The Cdh5-CreER/+;Tlr4 fl/- mouse lacks TLR4 in all endothelial cells (i.e., in peripheral organs as well as CNS/leptomeninges), and, as the authors note, the periphery is exposed to E. coli. It would be helpful if the authors could comment in the Discussion on the possibility that peripheral effects (e.g., peripheral endothelial cytokine production, changes to blood composition as a result of changes to peripheral endothelial permeability) may contribute to the observed leptomeningeal phenotypes.

Thank you for raising this point. We agree that peripheral responses could contribute to the observed leptomeningeal phenotypes in this model. We have added two sentences to the second paragraph of the Discussion to address this: “We note that these experiments do not distinguish between local vs. distal anatomic sources of LPS or downstream effector molecules, such as cytokines, that activate the leptomeningeal inflammatory response (Huang et al., 2021). Histologic observations of RFP-expressing E. coli in the brain, liver, and lungs, together with positive blood cultures, indicate substantial systemic dissemination in this model. Thus, the inflammatory responses of leptomeningeal cells likely reflect exposure to bacterial products and inflammatory mediators derived from both local meningeal and peripheral sources.”

Reviewer #2 (Public review):

Summary:

The authors use a postnatal mouse model of E. coli bacterial meningitis and a mouse brain endothelioma cell line combined with cell-type-specific gene deletion to study the function of endothelial TLR4, a cell surface receptor that recognizes gram positive bacterial wall components, in the local leptomeningeal (LPM) response with a focus on endothelial barrier breakdown mediated by TLR4. Single-cell transcriptional profiling and imaging studies using whole-mount preps of the LPM support that LPM endothelial, CD206+ local macrophage and LPM fibroblast and arachnoid barrier cell inflammatory response and is abrogated in endothelial-specific KO of TLR4, pointing to a role for endothelial TLR4 in local LPM response. Culture studies using Bend3.1 cells (a mouse brain endothelioma cell line) support a direct role for TLR4 in the bacteria-mediated inflammatory response and in internalization of Cldn5 via the endosomal-lysosomal pathway, resulting in loss of barrier integrity

Strengths:

The local LPM cell response in meningitis and the role of specific LPM cells in inflammation and CNS barrier breakdown have not been extensively studied, despite ample evidence for primary immune response in the meninges in human patients and in animal models. The authors employ a robust, multi-model approach using both in vivo and in vitro models with cell-type-specific knockout to study the function of TLR4 in brain endothelial cell response. The authors nicely combine functional barrier assays with IF for junctional localization in their experimental design, and they delve into potential mechanisms of Cldn5 internalization using markers of endosomal-lysosomal pathway localization. The authors also describe a new type of barrier assay using a streptavidin-coated plate upon which barrier-forming cell cultures can be placted, this could be a very useful alternative or complement to other size-selective barrier assays and presumably could work for other barrier forming cells types, likely epithelial cells.

Weaknesses:

(1) There are no measures of bacterial burden in peripheral organs, blood, in the LPM or brain in the TLR4 endothelial cKO mice. Lack of TLR4 in endothelial cells could prevent bacterial 'access' into the LPM and brain, essentially preventing meningitis and leading to a lack of inflammatory responses in the LPM-located cells simply because there is no bacteria present. Bacteremia may also be reduced, as might inflammatory responses in peripheral organs with TLR4-deficient peripheral endothelium. Bacterial counts and inflammatory measures in peripheral organs and blood are important to better understand the mechanism(s) underlying the reduced inflammatory profile in LPM cells and no LPM endothelial breakdown in the Tlr4 endothelial cKO mice. In other words, does deleting TLR4 in EC protect against the development of meningitis by somehow blocking bacteria access to the LPM (this would be supported by low or no CFU counts in infected Tlr4 endothelial cKO) or is it what the authors appear to propose in Figure 1J that TLF4 in EC is the only cell responding to the bacteria to trigger the immune cascade in the LPM? More data is needed to resolve this, as this is a major claim of the paper.

Thank you for this comment. We agree that it is important to distinguish whether the reduced inflammatory response in Cdh5-CreER; Tlr4CKO (Tlr4^VEKO) mice reflects altered bacterial burden versus altered host sensing. We have fleshed out these issues by conducting the following comparisons between infected and uninfected WT and infected and uninfected Cdh5-CreER; Tlr4CKO mice: (1) quantifying E. coli in the blood of infected mice by counting colonies on agar plates; (2) quantifying E. coli in the brain by measuring red fluorescent protein (RFP) signal (the infecting E. coli carry an RFP-expression plasmid); (3) histologically surveying liver and lung for RFP+ E. coli; (4) monitoring the weights of infected and uninfected mice. These data are presented in Figure 2 – figure supplement 4 and in the Results section, and they can be summarized as follows. (1) E. coli is consistently detectable in the blood, brain, and peripheral organs of infected mice and is not detectable in control mice; (2) there are no statistically significant differences between infected WT and infected Cdh5-CreER; Tlr4floxed mice in E. coli burden; (3) infected mice of both genotypes stop gaining weight between the time of infection (P5) and the time of sacrifice 24 hours later (P6). Our conclusion is that loss of TLR4 in endothelial cells and in a subset of other non-myeloid leptomeningeal cells does not alter the overall clinical course of the infection despite changes in leptomeningeal gene expression and vascular permeability.

(2) The authors look at the underlying cortical response (cerebral vasculature for ICAM and immune cells) but do not use markers that could identify microglia (Iba1), the primary resident immune cell (CD206 is not useful, at this stage, in perivascular macrophages that are extremely sparse in the postnatal brain). This would be important to better study the impact on CNS resident immune cell morphological activation.

Thank you for this comment. In response, we have analyzed Iba1 staining in the cortex in infected vs. uninfected mice. This is shown in Figure 2 – figure supplement 3. These data demonstrate a several-fold increase in Iba1 immunostaining in infected compared to uninfected cortex, consistent with increased microglial activation in response to infection. There is no statistically significant difference between infected WT and infected Cdh5-CreER; Tlr4CKO mice in Iba1 staining in cortex.

(3) The authors suggest that Cldn5 junctional localization is selectively disrupted upon bacterial exposure, mediated by TLR4 - they suggest this based on studying PECAM, GLUT1, ZO-1 and B-catenin (all normally junction or cell surface located in cultured Bend3.1) in relationship to Cldn5 localization (normally high) - it is possibly these are also impact by bacteria exposure (maybe through different mechanisms?) - a better measure would be to use the similar cyto/PM measure they do for Cldn5 in Fig. 4D and to evaluate this or to use intensity measurements.

Thank you for this comment. As the reviewer noted, the analysis of Cldn5 localization with vs. without E. coli exposure and in WT vs. Tlr4KO bEnd.3 cells (shown in Figure 4B and D) – uses Cell Trace to partition the image into cytoplasmic vs. plasma membrane territories. For the analyses in Figure 5, we wanted to compare the localization (and potentially re-localization) behaviors of a variety of subcellular markers with the localization and re-localization of Cldn5 following E. coli exposure. By directly measuring the % overlap of the two immunostains, we get that data. We note that the goal of this analysis is to assess relative co-localization with Cldn5 rather than absolute subcellular partitioning of each marker. While this analysis could have been extended to include independent quantification of the subcellular localization of each of those other markers with respect to cytoplasmic vs. plasma membrane territories, it is clear by visual inspection of Figure 5A-C that beta-catenin, ZO-1, and PECAM1 remain plasma membrane-associated with E. coli exposure, and GLUT1 goes from the part of the plasma membrane not involved in cell-cell contact without E coli exposure to cytoplasmic with E. coli exposure (as judged by the appearance of a nuclear “shadow” after E. coli exposure). Thus, we do not believe that additional cytoplasmic vs. plasma membrane quantification for these markers would alter the interpretation. The main reason that we did not extend this analysis to include independent quantification of the subcellular localization of each of those other markers with respect to cytoplasmic vs. plasma membrane territories is because that would introduce the Cell Trace localization as an additional variable.

(4) The discussion could benefit from delving more into the prior literature on E coli mediated breakdown of junctions in cultured human microvascular brain endothelial cell model and critical host-pathogen interactions of the bacteria with ECs (PMID: 14593586), and how this might involve TLR4.

Thank you for this comment. Two paragraphs addressing the prior literature have now been added to the discussion.

(5) It would be important to discuss how their results relate to earlier studies on TLR4-/- and TLR2-/- global knockout mice and protection vs vulnerability to development of meningitis (see PMCID: PMC3524395) - this paper showed that TLR4 global KO mice have increased susceptibility to die from meningitis and have much higher CFU counts in the CNS. In this manuscript and their prior work (Wang et al., 2023), this group shown that both global TLR4-/- mutants and their EC-specific KO have reduced barrier permeability, but we don't have any information about CFU or susceptibility to death from meningitis in their models.

Thank you for these comments. The model we use – subcutaneous injection of E. coli (a clinical isolate from an infant with meningitis) at postnatal day (P)5 – results in the death of the infected mouse within 2 days (shown in Figure 1 – figure supplement 3 in Wang et al. 2023). Our analyses of infected mice were conducted 24 hours after infection. As noted in the reply to comment #1, in the revised manuscript we present a clinical assessment of WT vs. Cdh5-CreER; Tlr4CKO mice 24 hours after infection based on (1) a quantitative microscopic analysis of E. coli burden in the brain (visualized based on RFP fluorescence in the E. coli used here), (2) quantifying CFUs in blood and (3) mouse weights at P5 and P6, a sensitive indicator of overall health since this is a time when mice are normally gaining weight rapidly (~25% weight gain per day). These data (shown in Figure 2 figure supplement 4) indicate that bacterial burden and disease severity are similar between genotypes in our model. In Wang et al., 2023, we did not conduct a quantitative clinical assessment of WT vs. Tlr4-/- mice following infection, but by visual inspection, infected WT and Tlr4-/- mice appeared to have similar downhill clinical trajectories. We have expanded the Discussion to relate these findings to prior studies of global TLR4 and TLR2 knockout mice, noting that differences in experimental models and the distinction between global versus VECadCreER-specific deletion may account for the differing outcomes reported.

Comment on the paper listed by the reviewer (PMCID: PMC3524395).

The cited study demonstrates that global TLR4 deficiency leads to increased bacterial burden and mortality, indicating an essential role for TLR4 in host defense and bacterial clearance. In our study of Cdh5-CreER; Tlr4CKO mice, bacterial burden and disease severity at 24 hours post-infection are similar between WT and Cdh5-CreER; Tlr4CKO mice, indicating that Cdh5-CreER; Tlr4CKO does not alter the clinical course at this time point. This difference is noted in the Discussion section.

Reviewer #3 (Public review):

Summary:

This study investigates the molecular underpinnings of immune responses in the leptomeninges in neonatal bacterial meningitis. Bacterial meningitis is a major disease burden, particularly for neonates, and it has previously been noted that the meningeal immune environment in infants is permissive to opportunistic infection (Kim et al., Sci Immunol, 2023). There is less known about the contribution of the stromal compartment to meningeal immune responses. Seegren et al. interrogate the role of leptomeningeal endothelium in host defence in E. coli infected neonatal mice using mouse genetic tools to delete the LPS receptor Tlr4 from either endothelial cells (using Cdh5-CreER) or macrophages (using LysM-Cre). The authors use snRNAseq, cleared cortical mounts, and in vitro work to define the impact of E. coli infection on leptomeningeal endothelial cells. This study uses a range of innovative techniques to probe the role of the stromal compartment in meningitis.

Strengths:

This study makes excellent use of cleared cortical mounts to examine the biology of the leptomeninges, in particular, changes to the endothelium, with unprecedented detail. In combination with high-quality sequencing data provide new insights into the impact of meningitis on the leptomeninges. The data presented by the authors is of very high quality.

Weaknesses:

The weaknesses of the study were in terms of interpretation and perhaps study design.

(1) Most importantly, the authors need to provide additional validation of their conditional knockout models. The authors need to confirm that the Cdh5-CreER does not impact leptomeningeal fibroblasts and to confirm gene deletion in macrophages.

We are very grateful for this critique. After several years of using the Cdh5-CreER line in other parts of the CNS, where its expression is endothelial-specific, we applied it to the meninges without realizing that its specificity is broader in that tissue. Our initial analysis with a Cre reporter line that uses a membrane tdTomato appeared to confirm endothelial-specific recombination in the meninges. Following receipt of the reviews of this manuscript, we repeated this analysis with two Cre reporter lines that use a nuclearlocalized GFP, and we immunostained for each of several transcription factors to assess various meningeal cell types and quantified GFP co-localization (Figure 1 – figure supplements 1 and 2). This quantitative Cre reporter analysis shows CreER expression from the Cdh5-CreER transgene in all or nearly all endothelial cells and in a subset (~20%) of dural border cells and/or leptomeningeal fibroblasts, but not in myeloid cells. Additionally, our snRNA-seq analysis of Cdh5 transcripts shows expression in endothelial cells, dural border cells, and leptomeningeal fibroblasts, but not in myeloid cells (Figure 1– figure supplement 4), which agrees with several recent publications (Mapunda et al., 2023; Pietilä et al., 2023; Smyth et al., 2024). Thus, our initial interpretation that the phenotypes in the Cdh5-CreER; Tlr4floxed mouse were a consequence of recombination exclusively in endothelial cells was not quite correct. The Results section of the revised manuscript includes an expanded description of Cre and CreER expression specificity analysis, with supporting data in Figure 1 – figure supplements 1 and 2. Throughout the text of the revised manuscript, we are careful to note that the Cdh5-CreER; Tlr4floxed mouse has Tlr4 deletion in a subset of dural border cells and leptomeningeal fibroblasts. To reflect this fuller understanding of the specificity of Cdh5-CreER, we have changed the name of the Cdh5-CreER; Tlr4floxed mice in the text and figures from TLR4ECKO (“endothelial cell KO”) to TLR4VEKO (“VE-cadherin CreER KO”).

(2) The authors could also strengthen the paper by providing data on the impact of these conditional knockout models on the course of meningitis and bacterial burden.

Thank you for this comment. We agree that these additional analyses strengthen the manuscript. We have fleshed out these issues by conducting the following comparisons between infected and uninfected WT and infected and uninfected Cdh5-CreER; Tlr4floxed mice: (1) quantifying E. coli in the blood of infected mice by counting colonies on agar plates; (2) quantifying E. coli in the brain by measuring the red fluorescent protein (RFP) signal (the infecting E. coli carry an RFP-expression plasmid); (3) histologically surveying liver and lung for RFP+ E. coli; (4) monitoring the weights of infected and uninfected mice. These data are presented in Figure 2 – figure supplement 4 and in the Results section, and they can be summarized as follows. (1) E. coli is consistently detectable in the blood, brain, and peripheral organs in infected mice and is not detectable in control mice; (2) there are no statistically significant differences in bacterial burden between infected WT and infected Cdh5-CreER; Tlr4floxed mice; (3) infected mice of both genotypes stop gaining weight between the time of infection (P5) and 24 hours later at the time of sacrifice (P6). Our conclusion is that loss of TLR4 in endothelial cells and in other non-myeloid cells in the leptomeninges does not alter the overall clinical course of the infection despite changes in leptomeningeal gene expression and vascular permeability.

(3) Finally, it is perhaps not surprising that Tlr4 is required for meningitis responses with E. coli. However, it is unclear if these findings can be generalised to other, more common, meningitis infections (streptococcal/pneumococcal).

At present, it is an open question whether TLR4 plays as a large a role in meningitis caused by other gram-negative bacteria and whether TLR2 plays a similarly large role in meningitis caused by gram-positive bacteria. In the Discussion, the last two sentences under “Limitations of the study” summarize this point: “Finally, the present study focused on E. coli K1, the dominant Gram-negative neonatal pathogen. Future work could assess TLR signaling in response to other bacterial pathogens, such as Group B Streptococcus.”

(4) There are additional minor issues; for instance, the arachnoid fibroblast 2 population appears to closely resemble dural border cells.

Thank you for this comment. That is correct, and we have changed the nomenclature to “dural border cells”.

(5) The cell line model (bEnd.3) is a relatively low-fidelity model of BBB endothelial cells, and this should be acknowledged.

Thank you for this comment. That is correct. Despite being brain-derived, bEnd.3 cells have lost many BBB-specific attributes. Their responses might best be considered as generic endothelial responses rather than brain-specific endothelial responses. This is now stated in the Results section: “Although they are brain-derived, bEnd.3 cells lack many BBB-specific attributes and, therefore, they likely exhibit generalized endothelial responses rather than brain-specific responses to bacterial exposure.”

With these caveats, it is difficult to be certain that the endothelium alone is the driver of meningeal immune responses in meningitis, and what the impact of these is.

We agree with this critique. As noted above, the expression of Cdh5-CreER in essentially all endothelial cells and in a subset of dural border cells and leptomeningeal fibroblasts means that the comparison of TLR4 CKO with Cdh5-CreER vs. Lyz2-Cre is assessing phenotypes driven by TLR4 signaling in endothelial plus a subset of other non-myeloid cells vs. TLR4 signaling in myeloid cells. We have revised the text to reflect this more precise understanding of Cdh5-CreER specificity.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

(1) Transcriptomic analysis: The analysis and display of the single-nucleus RNA-seq data should be improved. The authors could perform a more granular, unbiased clustering of each cell class in the combined dataset and then compare the proportion of each experimental group (genotype x control/infected) in each cluster. At present, it appears the differentially-expressed genes (DEGs) shown in Figure 1 were identified using the Seurat FindMarkers function with default parameters (Methods). This considers each cell as an independent experimental unit and is therefore not appropriate for a comparison of control versus infected groups (see e.g., PMID 34584091, 35880426. The authors should implement a statistical analysis strategy that considers true biological replicates (mice, as shown in Supplementary File 1).

We do not fully agree with this critique. We agree that biological replication at the level of individual mice is important for interpreting these data, but within each mouse, the characteristics of individual cells is also of interest, including the degree of heterogeneity, the sample size for a given cell cluster, and the statistical significance of any observed changes in transcript abundance. As requested, we have prepared a new supplemental figure (Figure 1 – figure supplement 5) showing a principal component analysis of the scRNA-seq data for each mouse (one mouse was used for each snRNA-seq dataset) and for each of the principal leptomeningeal cell types. This analysis shows, for example, that the three infected Cdh5-Cre; Tlr4flox/- mice have transcriptomes for each of the six cell clusters that are very similar to the transcriptomes of the two uninfected WT and the two uninfected Cdh5-CreER; Tlr4flox/- mice. Thus, the genotype- and condition-dependent effects are consistent across biological replicates. At the most granular level, Figure 1 – figure supplement 7, which was part of the original submission, shows for the most up- and down-regulated genes (based on adjusted p-value or based on fold-change) in endothelial cells and in myeloid cells how individual transcript abundances change for each mouse and for each genotype/condition.

(2) The authors use immunohistochemistry to assess claudin-5 "disorganization and redistribution" (Results, pages 7-8 and Figures 3A-B). They state that "Tlr4ECKO mice showed minimal changes in the distribution of Cldn5, implying that cell autonomous endothelial TLR4 signaling regulates tight-junction organization." It is not clear, however, that the quantified parameter (Cldn5+ area relative to total area) would be an accurate readout of claudin-5 organization/distribution (i.e., subcellular localization) as it would also be sensitive to claudin-5 expression, vascular density, and vessel diameter. The authors use a similar assessment of ZO-1 to suggest that changes to claudin-5 are not due to a "generalized disassembly of TJs" and could also use this to argue that the above potential confounds (vascular density, vessel diameter) do not change, but the data in Figure 3 - Figure Supplement 1B, lower panel, show that infection does cause an increase in ZO-1 area relative to total area (P = 0.0004). Thus, the statement in the results "Zonula Occludens-1 (ZO-1) [...] remained unchanged during infection (Figure 3 - figure supplement 1)" is not accurate. The authors should revise this section to ensure their conclusions are aligned with the presented data.

Thank you for this comment. The reviewer is correct that the Cldn5 area measurement is unable to deconvolve the various factors that might contribute to it (vessel density and diameter, and Cldn5 distribution). This part has been rewritten. “Consistent with prior findings (Wang et al., 2023), both WT and Tlr4^MKO mice showed an increase in the area occupied by Cldn5 in the leptomeninges following infection, likely referable to both increased vessel diameter and a redistribution of Cldn5 within ECs (Figure 3A-B; Figure 3 – figure supplement 1C).”

The reviewer is also correct about our initial description of the ZO-1 data. What we meant to write and what the revised manuscript now shows is: “The area occupied by Zonula Occludens-1 (ZO-1), a tight junction scaffold protein, showed a modest but statistically significant increase in WT leptomeningeal vessels but no significant change in Tlr4^VEKO leptomeningeal vessels during infection (Figure 3 – figure supplement 1A and B).”

(3) In the Methods, under Mouse Models and E. coli Infection, the authors state, "The Cdh5-CreER line (Monvoisin et al., 2023) was the same line used in Wang et al. (2023)." However, there is no mention of Cdh5-CreER in Wang et al. (2023). Could authors please clarify? Also, because this appears to be an inducible Cre, the authors must include details on the dose and timing of tamoxifen or 4-OHT used in this study.

Thank you for catching that error. We meant to reference Wang et al (2025), not Wang et al (2023). [Wang et al (2025) is: Wang Y, Rattner A, Li Z, Smallwood PM, Nathans J. (2025) Vascular endothelial-specific loss of TGF-beta signaling as a model for choroidal neovascularization and central nervous system vascular inflammation. Elife 14:RP107018.] This has now been corrected.

We have now included the details related to 4HT injection in the Methods section “Mouse Models and E. coli Infection”. These are intraperitoneal injection at P2 with 40- 50 µL of 2 mg/ml 4HT.

(4) The legend for Figure 1A is "Schematic of the leptomeninges", but the figure shows the entire brain-skull interface, including underlying cortex, leptomeninges, dura, and skull.

Thank you. Corrected.

(5) Page 7, typo: "In the brain, CD206+ cell were too sparse ..." Should be "cells".

Thank you. Corrected.

(6) Page 14, typo: "... could represents a double-edged ..." Should be "represent".

Thank you. Corrected.

Reviewer #2 (Recommendations for the authors):

(1) Perform CFU counts from LPM, dura, brain, peripheral organs (liver) in infected v mock mice from control v TLR4 EC-cKO.

Thank you for this comment, with which we agree. We have addressed this by quantifying bacterial burden and assessing disease severity in WT and Cdh5CreER; Tlr4floxed mice. Specifically, we performed CFU measurements in blood, monitored mouse weights at P5 and P6, and histologically surveyed the E. coli-RFP signal (i.e., E. coli burden) in brain, liver, and lung. These analyses show that bacterial burden and disease progression are comparable between WT and Cdh5-CreER; Tlr4floxed mice at 24 hours post-infection. These data are presented in Figure 2 – figure supplement 4 and described in the Results.

(2) Lyz2Cre/+ is used to delete TLR4 from macrophages, but recombination efficiency (in LPM BAMs) is described as only partial, suggesting that TLR4-response in LPM BAMs (and potentially macrophages in the dura) is at least partially intact. It undercuts conclusions that can be made using this line.

Thank you for this comment. We have conducted a more detailed analysis of Lyz2^Cre specificity by immunostaining for multiple markers and quantifying the results (Figure 1 – figure supplement 2). We now think that the more cursory analysis in the original submission was inaccurate. The more in-depth analysis shows that Lyz2^Cre directed Cre-recombination with 90-100% efficiency in CD206+ cells and with 50-70% efficiency in ASC+ and PU.1+ cells, the range depending on whether tdTomato or GFP colocalization was being scored (Figure 1 – figure supplement 2). The Results section text now states: “In the text that follows, we will refer to Lyz2^Cre-recombined cells simply as “myeloid cells”, although they should be understood as CD206+ myeloid cells.”

Also, as noted in the reply to comment 4 below, a direct analysis of Tlr4 recombination efficiency is technically challenging due to the low abundance of TLR4 and the failure, in our hands, of commercial anti-TLR4 antibodies to produce clear immunostaining. We have added a comment in the results section noting that we do not have a measure of the efficiency of recombination of the floxed Tlr4 target in vivo: “The low abundance of TLR4 and the limitations of commercial anti-TLR4 antibodies precluded a direct immunohistochemical assessment of TLR4 loss in Tlr4^VEKO and Tlr4^MKO mice.”

(3) Inflammatory responses [qPCR] from peripheral organs and also physiological measures in the pups [weight post-infection, time to moribund or death curves] in control v TLR4 EC-cKO and TLR4 mac-cKO.

Thank you for this comment. We have not conducted a qPCR analysis of inflammatory gene expression in peripheral organs because (1) the dramatic upregulation of these transcripts in the leptomeninges, (2) the presence of E. coli in blood and peripheral organs, and (3) the clinical assessment (cessation of weight gain) all predict that such an analysis would reveal a large up-regulation of inflammatory gene expression throughout the body. More specifically, we have conducted the following comparisons between infected and uninfected WT and infected and uninfected Cdh5-CreER; Tlr4floxed mice: (1) quantifying E. coli in the blood of infected mice by counting colonies on agar plates; (2) quantifying E. coli in the brain by measuring the red fluorescent protein (RFP) signal (the infecting E. coli carry an RFP-expression plasmid); (3) histologically surveying liver and lung for RFP+ E. coli; (4) monitoring the weights of infected and uninfected mice. These data are presented in Figure 2 – figure supplement 4 and in the Results section, and they can be summarized as follows. (1) E. coli is consistently detectable in the blood, brain, and peripheral organs in infected mice and is not detectable in control mice; (2) there are no statistically significant differences between infected WT and infected Cdh5CreER; Tlr4floxed mice; (3) infected mice of both genotypes stop gaining weight between the time of infection (P5) and 24 hours later at the time of sacrifice (P6). Our conclusion is that loss of TLR4 in endothelial cells and in a subset of other non-myeloid cells in the leptomeninges does not alter the overall clinical course of the infection despite changes in leptomeningeal gene expression and vascular permeability.

(4) The conditional macrophage line is problematic due to the partial recombination. I question the utility of including this unless they can come up with a way resolve the response of recombined TLR4 macrophages vs ones that are not (could they use the single cell data to pick this a part? Are TLR4-null cells and TLR4 'wt' cells transcriptionally similar in the infected condition, suggesting TLR4 is not doing much in the macs, potentially due to alternate TLRs?). There are good BAM Cre lines that have been described [Lyve1-cre would be good for LPM BAMS, the other is Pf4-cre, see https://pmc.ncbi.nlm.nih.gov/articles/PMC7375817/ - just as an FYI for the future].

Thank you for this comment. As noted in the reply to point 2 (above), we have conducted a more in-depth analysis of Lyz2^Cre specificity by immunostaining for multiple markers and quantifying the results (Figure 1 – figure supplement 2). We now think that the more cursory analysis in the original submission was inaccurate. The more in-depth analysis shows that Lyz2^Cre directed Cre-recombination with 90-100% efficiency in CD206+ cells and with 50-70% efficiency in ASC+ and PU.1+ cells, the range depending on whether tdTomato or GFP colocalization was being scored (Figure 1 – figure supplement 2). The text now states: “In the text that follows, we will refer to Lyz2^Cre-recombined cells simply as “myeloid cells”, although they should be understood as CD206+ myeloid cells.”

We agree that, based on Figure 6 in the cited paper [McKinsey et al (2020) A new genetic strategy for targeting microglia in development and disease eLife 9:e54590], the Pf4-Cre line may be superior to the Lyz2^Cre line that we used for recombination in leptomeningeal myeloid cells. Unfortunately, we missed this paper in our literature searches, probably because it focuses on a microglial CreER line, P2ry12-CreER, and the Pf4-Cre line is not mentioned in the title or abstract. Our decision to use the Lyz2^Cre line was based on an extensive comparison among myeloid Cre lines showing that Lyz2^Cre was the most efficient [Abram CL, Roberge GL, Hu Y, Lowell CA. 2014. Comparative analysis of the efficiency and specificity of myeloid-Cre deleting strains using ROSA-EYFP reporter mice. J Immunol Methods 408:89-100.] However, the Abram et al study did not look at the leptomeninges. Regarding the efficiency of recombination of the floxed Tlr4 target, a direct analysis is technically challenging due to the low abundance of TLR4 and the failure, in our hands, of commercial anti-TLR4 antibodies to produce clear immunostaining. We have added a comment in the results section noting that we do not have a measure of the efficiency of recombination of the floxed Tlr4 target: “The low abundance of TLR4 and the limitations of commercial anti-TLR4 antibodies precluded a direct immunohistochemical assessment of TLR4 loss in Tlr4^VEKO and Tlr4^MKO mice.”

(5) Figure 1 - Figure Supplement 2 - the authors nicely break down the pathway response [NFKB and TNF] in EC and macs, it would be great to have similar information for the fibroblasts (in the main figure or the supplement). Does their inflammatory response show a similar pattern?

Thank you for this suggestion. We have now done that analysis and present it in Figure 1 – figure supplement 3. For completeness, we also performed the same type of analyses for JAK-STAT signaling and IFN-gamma response and these are shown in Figure 1 – figure supplement 6. The principal conclusion is that across all major leptomeningeal cell types, the Cdh5-CreER; Tlr4floxed samples (i.e., Tlr4 KO’d in non-myeloid cells) show much reduced transcriptome changes with infection.

(6) What is ASC and Cd206 quantification measuring, and how does this relate to 'activation' - is this the intensity of signal or a morphological change? What is the precedence for using ASC (citations)? In their prior work, they showed no change in CD206 number, so a significant increase upon infection here, it's confusing exactly what is being studied. Also, loss of Lyve1 is a well-accepted measure of activation that they have previously used, adding that it could be helpful. This is not a major issue since they have robust data that the macrophages are not transcriptionally activated. Clarification of what exactly is being measured would be sufficient (in the text).

CD206 immunostaining, which reveals myeloid cell morphology, shows that, with E. coli infection, myeloid cells convert from a more compact morphology to a more expanded morphology. This is now explained more fully in the Results section.

Regarding ASC, changes in the state of ASC aggregation and ASC subcellular localization have been used by others to monitor immune cell responses to inflammatory signals (Sester et al., 2016; Franklin et al., 2018). While this change in subcellular localization may explain part of the increase in immunostained area in myeloid cells in the infected mice (Figure 2D), the increase in the area of ASC immunostaining largely reflects a shift of myeloid cells from a compact to a more extended morphology. This is now explained more fully in the Results section. We have also added two references (Sester et al., 2016; Franklin et al., 2018) that described how ASC distribution changes with inflammation.

Regarding LYVE1, we observe a decrease in LYVE1 transcript abundance in myeloid cells with infection, as predicted. Given the large amount of other data that document myeloid activation with infection, we have elected not to include this.

(7) The authors suggest the internalization of Cldn5 is not due to NFKB downstream signaling that includes transcriptional mechanisms because it happens as early as 1 hour, prior to NFKB localization to the nucleus. However, a lot of their experiments, including on endosomal-lysosomal protein co-localization are done at 4 hours, when their RNAseq data show robust NFKB-mediated gene upregulation and (though not tested) potentially protein production of factors that can act back on the cells, including to impact endo-lysosomal processing. Without studies at earlier timepoints post-bacteria exposure, separating these two mechanisms is difficult.

Thank you for this comment. We have explored this question by looking at Cldn5 internalization in bEnd.3 cells at 1 hour after E. coli exposure, and the data clearly show that internalization occurs within 1 hour. Additionally, we have conducted this experiment in the presence of 1 uM ACHP, an IKK inhibitor that blocks NF-кB migration to the nucleus. ACHP treatment shows no effect on the rapid internalization of Cldn5, implying a mechanism independent of NF-кB control of gene expression. These data are shown in a new figure (Figure 6) in the revised manuscript.

(8) Figure 2 - CD206 are quite sparse however, Iba1 would work well to look at microglial activation.

Thank you for this suggestion, which we have followed. To assess microglial activation, we have immunostained for Iba1 and quantified the data. These are now included in Figure 2 – figure supplement 3. The data show that there is an increase in Iba1 immunostaining following E. coli infection in both WT and Cdh5-CreER; Tlr4floxed mice, with more in the former than the latter, but the difference is not statistically significant.

(9) Suggest performing the LAMP+ co-localization experiment at <1hr, prior to NFKB nuclear localization and transcriptional changes. This would better support it, this is (or is not) independent of the NFKB. Could also test this with an NFKB inhibitor, do they still see the CLDN5 internalization when NFKB is blocked?

Thank you for these suggestions. We have done both of these analyses, and the results are presented in Figure 6. The results show that (1) Cldn5 is internalized within 1 hour and (2) its internalization is independent of NF-кB signaling inhibition by 1 uM ACHP. Since ACHP treatment shows no effect on the rapid internalization of Cldn5, that implies a mechanism independent of NF-кB control for gene expression.

Reviewer #3 (Recommendations for the authors):

Major points

(1) The most important caveat is that the Cdh5-CreER model is known to recombine in leptomeningeal fibroblasts (10.1038/s41586-023-06993-7, 10.1101/2025.05.13.653681), and Cdh5 expression in these populations is now well described (10.1038/s41467-02341580-4, 10.1016/j.neuron.2023.09.002). Although the authors did not observe recombination in their reporter (details of the tamoxifen injection protocol should be provided), it is imperative to validate the specificity of their model to Tlr4 in endothelial cells, leveraging their sequencing data and providing additional IHC or ISH to confirm this. Alternatively, Tlr4 could be deleted in a more specific model, e.g., the Pdgfb-iCreERT2 or Slco1c1-CreERT2. It is also important to do the same with the LysM model, to confirm that the lack of impact of macrophage Tlr4 is not due to failure to delete the gene. This is again important to the interpretation of the study, since the authors propose that the endothelium, specifically, is the driver of the meningitis response.

We are very grateful for this critique. After several years of using the Cdh5-CreER line in other parts of the CNS, where its expression is endothelial-specific, we applied it to the meninges without realizing that its specificity is broader in that tissue. Our initial analysis with a Cre reporter line that uses a membrane tdTomato appeared to confirm endothelial-specific recombination in the meninges. Following receipt of the reviews of this manuscript, we repeated this analysis with two Cre reporter lines that use a nuclear-localised GFP, and we immunostained for each of several transcription factors to assess various meningeal cell types and quantified GFP co-localization (Figure 1 – figure supplements 1 and 2). This quantitative Cre reporter analysis shows CreER expression from the Cdh5-CreER transgene in all or nearly all endothelial cells and in a subset (~20%) of dural border cells and/or leptomeningeal fibroblasts, but not in myeloid cells. Additionally, our snRNA-seq analysis of Cdh5 transcripts shows expression in endothelial cells, dural border cells, and leptomeningeal fibroblasts, but not in myeloid cells (Figure 1– figure supplement 4), which agrees with several recent publications (Mapunda et al., 2023; Pietilä et al., 2023; Smyth et al., 2024). Thus, our initial interpretation that the phenotypes in the Cdh5-CreER; Tlr4floxed mouse were a consequence of recombination exclusively in endothelial cells was not quite right. The Results section of the revised manuscript has an expanded description of Cre and CreER expression specificity analysis, with supporting data in Figure 1 – figure supplements 1 and 2. Throughout the text of the revised manuscript, we are careful to note that the Cdh5-CreER; Tlr4floxed mouse has Tlr4 deletion in a subset of dural border cells and leptomeningeal fibroblasts. To reflect this fuller understanding of the specificity of Cdh5-CreER, we have changed the name of the Cdh5-CreER; Tlr4floxed mice in the text and figures from TLR4ECKO (“endothelial cell KO”) to TLR4VEKO (“VE-cadherin CreER KO”).

We have also conducted a more detailed analysis of Lyz2^Cre specificity by immunostaining for multiple markers and quantifying the results (Figure 1 – figure supplement 2). We now think that the more cursory analysis in the original submission was inaccurate. The more in-depth analysis shows that Lyz2^Cre-directed Cre-recombination with 90-100% efficiency in CD206+ cells and with 50-70% efficiency in ASC+ and PU.1+ cells, the range depending on whether tdTomato or GFP colocalization was being scored (Figure 1 – figure supplement 2). The text in the Results section now states: “In the text that follows, we will refer to Lyz2^Cre-recombined cells simply as “myeloid cells”, although they should be understood as CD206+ myeloid cells.”

Regarding the efficiency of recombination of the floxed Tlr4 target, a direct analysis is technically challenging due to the low abundance of TLR4 and the failure, in our hands, of commercial anti-TLR4 antibodies to produce clear immunostaining. The phenotype of Cdh5-CreER; Tlr4floxed mice – a dramatically reduced infection-associated transcriptional response – argues that the floxed Tlr4 target was recombined at appreciable efficiency in those mice (Figure 1D and 1E). For Lyz2^Cre; Tlr4floxed mice the principal phenotype is an up-regulation of infection-associated transcripts in a subset of dural border cells in the absence of infection; the transcriptional response to infection was largely unaffected in all leptomeningeal cell types (Figure 1D and 1E). We have added a comment in the results section noting that we do not have a measure of the efficiency of recombination of the floxed Tlr4 target in vivo: “The low abundance of TLR4 and the limitations of commercial anti-TLR4 antibodies precluded a direct immunohistochemical assessment of TLR4 loss in Tlr4VEKO and Tlr4MKO mice.”

(2) The authors did not examine the consequences of Tlr4 cKO on the course of meningitis or bacterial burden. Knowing the impact of this would strengthen the paper and allow us to determine if the endothelial responses are helpful or harmful in meningitis progression.

For the revised manuscript, we have conducted the following comparisons between infected and uninfected WT and infected and uninfected Cdh5-CreER; Tlr4floxed mice: (1) quantifying E. coli in the blood of infected mice by counting colonies on agar plates; (2) quantifying E. coli in the brain by measuring the red fluorescent protein (RFP) signal (the infecting E. coli carry an RFP-expression plasmid); (3) histologically surveying liver and lung for RFP+ E. coli; (4) monitoring the weights of infected and uninfected mice. These data are presented in Figure 2 – figure supplement 4 and in the Results section, and they can be summarized as follows. (1) E. coli is consistently detectable in the blood, brain, and peripheral organs in infected mice and is not detectable in control mice; (2) there are no statistically significant differences between infected WT and infected Cdh5-CreER; Tlr4floxed mice; (3) infected mice of both genotypes stop gaining weight between the time of infection (P5) and 24 hours later at the time of sacrifice (P6). Our conclusion is that loss of TLR4 in endothelial cells and in a subset of other non-myeloid cells in the leptomeninges does not alter the overall clinical course of the infection despite changes in leptomeningeal gene expression and vascular permeability.

(3) TLR4 is a known receptor for LPS. It is unsurprising (especially in the in vitro experiments) that Tlr4 knockout reduces NF-kB signalling and other downstream changes to endothelial cells. Furthermore, it is uncertain if the infection was left to continue, similar changes to the endothelium would nonetheless occur through other mediators such as IL1B and TNFa.

We agree that it makes logical sense that Tlr4 KO decreases NF-кB signaling. The interesting next question is: what are the mechanistic underpinnings of the responses that are downstream of TLR4 and NF-кB? The cell culture experiments with WT vs. Tlr4KO bEnd.3 cells identify one set of cell biological responses related to Cldn5 and junctional integrity, and the NF-кB inhibition experiment (Figure 6) implies that rapid internalization of Cldn5 occurs in the absence of NF-кB mediated transcriptional changes. Regarding the possibility that other mediators such as IL1B or TNFα might, at least partially, make up for the lack of TLR4 signaling later in the infection, that is an open question at present.

(3) The arachnoid fibroblast 2 cluster should be renamed to dural border cells based on their high expression of Slc4a10, Adamtsl3, Tmeff2, etc which are all highly enriched in dural border cells. I suspect this cluster is also highly enriched for Slc47a1, probably the most specific marker for these cells (10.1038/s41586-023-06993-7, 10.1016/j.neuron.2023.09.002).

Thank you for this comment. The reviewer is correct. These are dural border cells and they express Slc47a1, as seen in a new supplemental Figure 1 – figure supplement 4, which shows UMAP plots for many leptomeningeal cell type-specific genes. We have updated our cell cluster assignment to align with the assignments in Pietilä et al (2023).

(4) It would be helpful to provide higher resolution images of Cldn5 in the leptomeningeal mounts. At the current resolution, it is difficult to tell if there is a similar internalisation/disruption phenotype to what is observed in vitro. Notably, this finding is similar to another recent publication on Cldn5 recycling (in the context of stroke) (10.1186/s40478-025-02125-6).

Higher resolution images of Cldn5 in leptomeningeal vessels without or with E. coli infection are now shown in Figure 3 - figure supplement 1C. There is a visual impression of greater area occupied by Cldn5, which is confirmed by quantification (Figure 3A and B). This effect appears to be due to both an average increase in vessel diameter and a redistribution of some of the Cldn5 away from plasma membrane junctions. Thank you for pointing out the interesting and relevant Cottarelli et al (2025) paper, which we had not read. This is now referenced.

Minor points

(1) Typo: prominant should be spelled prominent.

Thank you for catching that one. It is now corrected.

(2) Strictly speaking, the arachnoid layer is not epithelial (despite Cdh1 expression). They are fibroblasts that acquire barrier-forming properties.

Thank you for that comment. That appears to be the consensus view, and we will go along with it.

(3) Notably, LyzM Cre will also recombine in other myeloid populations, so I wouldn't describe it as a macrophage.

Thank you for this comment. We agree, and we have therefore changed the text and figure labels from “macrophage” to “myeloid”.

(4) It is interesting and notable that ICAM1 expression is observed in nonendothelial populations, in the IHC, too, perhaps.

We agree. ICAM1 may be a broader marker/mediator of inflammation than is generally recognized.

(5) In F1B, your labels on the right image to the arachnoid barrier and pial surface are presumably meant to refer to the image on the left with DPP4 and laminin labelling? The subarachnoid should be between the laminin and DPP4 layers (although it will be collapsed in your preparations).

Thank you for catching this error. The vertical bars were sized erroneously, and the labels were also placed erroneously. These have now been corrected.

(6) I would reference the papers that defined leptomeningeal cell type markers (10.1038/s41586-023-06993-7, 10.1016/j.neuron.2023.09.002) when you define your cell types.

Thank you. We have done that, and we have updated our cell cluster assignment to align with the assignments in Pietilä et al (2023).

(7) I would change references to the subarachnoid space in your figures to the leptomeninges (which include the SAS, but extend either side of it).

Thank you. The labels have been changed to “leptomeninges”.

(8) In Figure 2 - Supplement 1A, it looks like the populations are mislabelled.

Thank you. This has been corrected to be consistent with the assignments in Figure 1B