E-Selectin Orchestrates IL-1β–Dependent Neuroinflammation via NLRP3 in Vincristine-Induced Neuropathy

Reviewer #1 (Public review):

This paper looks at the effect of vincristine-induced peripheral neuropathy (VIND), a common effect of cancer therapy. The authors performed in vivo experiments in mice by injecting them with vincristine sulphate i.p. (+/- various inhibitors or antibodies) or E-selectin intraplantar (i.pl.), and in vitro experiments using dorsal root ganglia (DRG) neurons and bone marrow derived macrophages (BMDMs).

Inhibition of E-selectin with antibodies or genetic depletion reduced the accumulation of F4/80+ macrophages in the DRG and sciatic nerves (located beside the spine) after vincristine administration, and attenuated the mechanical hypersensitivity (paw withdrawal).

The authors went on to perform spatial transcriptomics on isolated DRG neurons and found some pathways changed. E-selectin injected directly intraplantar (i.pl.) mimicked the effect of vincristine administration on the mechanical hypersensitivity. Whereas chlodronate depletion of myeloid cells reduced these changes in the E-selectin model. Using LPS priming before vincristine in BMDMs in vitro, the authors demonstrate an increase in many cytokines, including IL-1beta (typically associated with the formation of an inflammasome) and elevated p-NFkB. Finally, treatment of mice with anakinra (which neutralizes IL-1beta) also attenuated the E-selection-induced reduction in mechanical hypersensitivity when injected i.pl.

The authors address an important aspect that after cancer therapy, there can be peripheral nerve damage that has lasting consequences for patients, although the precise mechanism is unknown. The authors delineate that E-selectin has an important role in the mouse model, where depletion or inhibition attenuated the negative effect of vincristine (i.p.) on mechanical hypersensitivity (paw withdrawal). Administration of E-selectin into the foot (i.pl.) also mimicked the changes observed in the vincristine-treated mice. There seems to be a role for macrophages, as they were associated with DRGs in vivo, and depletion attenuated motor deficits in the E-selectin injection model.

However, I am not convinced by the data supporting some of the conclusions drawn by the authors, particularly on the role of the NLRP3 inflammasome in their in vivo model.

Main points:

(1) The initial experimental paradigm looks at the DRG neurons, which are located by the spine, and from there the foot pad is examined in subsequent experiments. It would be relevant to show whether the foot pad is altered in the vincristine-treated mice and whether the infiltration of myeloid cells that was demonstrated at DRGs is also observed in the foot in the vincristine model. Otherwise, the mechanism being investigated in the vincristine model, which might have similar functional results (paw withdrawal), but the mechanism behind both could be completely different.

(2) The rationale of performing spatial sequencing on DRG neurons isolated from vincristine mice is unclear. It is likely that more information could have been obtained from looking at sections from these animals, and there would be a better link to the experiments on BMDMs which follow afterwards. Indeed, the spatial data does not seem to play a key role in the study. There is not a clear link between it (which was carried out on DRGs) and the later focus on macrophages and indeed the NLRP3 inflammasome.

(3) The authors suggest that the E-selectin is enhancing NFkB-induced priming of the NLRP3 inflammasome. LPS+vincristine increased IL-1b release from BMDMs in vitro, which was elevated in the presence of E-selectin. E-selectin also increased ASC speck formation by approx. 20% in vitro. The ASC speck formation in vitro was blocked by MCC950, a specific NLRP3 inhibitor, but the authors went on to use anakinra in vivo using the E-selectin i.pl. model. It is really unclear why the switch to anakinra occurred for the in vivo work, as blocking IL-1b is central to many inflammatory pathways, not just NLRP3. Use of MCC950 would have been more appropriate to demonstrate that negative effects on mechanical function are mediated by the NLRP3 inflammasome. As there were no readouts of NLRP3 inflammasome activity measured in any of the mice in vivo (e.g. local ASC specks, IL-1b release, western blot of typical inflammasome components such as IL-1b, caspase-1, ASC or gasdermin D) either at the DRG site or the foot, we cannot say that the cell culture data mimics or models the in vivo conditions at this time.

(4) Additionally, the reliance on the E-selectin administration models for the second half of the paper is curious. It would have been relevant to test whether the immune-modulating inhibitors could also attenuate the vincristine-induced effects on mechanism hypersensitivity, to better link the E-selectin model with the vincristine one.

(5) Details are missing from the figure legends and the methods. The concentrations of compounds used in cell culture and exposure times are not clear.

Reviewer #2 (Public review):

Summary:

Using antibody treatments, genetic models and in vitro studies, the authors convincingly show that E-Selectin is a driver of VIPN.

Strengths:

In vivo studies are robust. Antibody studies as well as the inflammasome-related work are very well done.

Weaknesses:

(1) The spatial transcriptomics data need to be improved in terms of visualization.

The authors should plot genes that define the cell types in the sequencing dataset, and also show the cellular map on the cut, not only UMAPs. Can the authors not use the n=3 samples per condition to perform some statistical analyses? While the CellChat analyses are informative, they are difficult to read. Fold change over control when comparing conditions and pathways may be a better way of visualization.

(2) In Figure 1B, the % DAB is not easy to understand. It would be better to do the staining also via IF, and then maybe to count nuclei. The corresponding figures shown in the supplemental figure are not convincing. If F4/80 is not working well, Iba1 could be an alternative.

(3) The authors should define the background of all the mice used. Have they been back-crossed to B6j mice? One cannot compare C57BL6J mice with full knockouts if they are not littermate controls. Especially immunological responses are completely dependent on the background of mice. See e.g. PMID: 40568896.

(4) Could the authors comment on the role of ICAM2? Why was this not tested as well?