Sensory innervation in porous endplates by Netrin-1 from osteoclasts mediates PGE2-induced spinal hypersensitivity in mice

Spinal pain is a major clinical problem, however, its origins and underlying mechanisms remain unclear. Here we report that in mice, osteoclasts induce sensory innervation in the porous endplates which contributes to spinal hypersensitivity in mice. Sensory innervation of the porous areas of sclerotic endplates in mice was confirmed. Lumbar spine instability (LSI), or aging, induces spinal hypersensitivity in mice. In these conditions, we show that there are elevated levels of PGE2 which activate sensory nerves, leading to sodium influx through Nav 1.8 channels. We show that knockout of PGE2 receptor 4 in sensory nerves significantly reduces spinal hypersensitivity. Inhibition of osteoclast formation by knockout Rankl in the osteocytes significantly inhibits LSI-induced porosity of endplates, sensory innervation, and spinal hypersensitivity. Knockout of Netrin-1 in osteoclasts abrogates sensory innervation into porous endplates and spinal hypersensitivity. These findings suggest that osteoclast-initiated porosity of endplates and sensory innervation are potential therapeutic targets for spinal pain.


Introduction
The first 4 sentences should be condensed. Line 55 The authors can only state that disc degeneration is not painful in some patients -they do not know if it is true for all. Line 58 -Tell us straight away what the relationship is between the Modic changes and LBP. Please re-read the introduction in full as grammatical errors are apparent throughout.
Behavioural testing for mechanical allodynia -why were only 2 forces used?
Statistical analysis: No sample size is reported in the abstract. Can the authors confirm potential bias due to lost, missing or excluded data? Can the authors include a statement about outliers? Ensure that the ARRIVE checklist has been adhered to. The experimental unit is not always clear. Were randomisation methods used to assign animals to experimental conditions? Blinding? Each experimental group should have a clear section relating to HOW each experiment was carried out (i.e. drug dose or method of euthanasia). What methods were used to allocate animals to experimental groups?
Reviewer #2 (Remarks to the Author): Low back pain (LBP) is one of the most common chronic pain conditions that causes significant social and economic burden, and degenerative disc is considered as the major source of LBP. In this report, Ni et al demonstrated that porous endplate, rather than the disc, is a major source of LBP. Specifically, they showed increased osteoclast activity in endplates of lumbar spine instability (LSI) mouse model and in aged mice, which induces CGRP (+) neuron innervation of porous endplates. The porous endplates produce PGE2 to activate CGRP (+) nerve fibers by PGE2 receptor EP4, and eventually leads to sodium influx through Nav1.8 channel. Inhibition of EP4 from DRG sensory neurons, inhibition of osteoclast formation by knocking out Rank1, or preventing sensory innervation into porous endplates by knocking out Nestrin-1 from osteoclasts greatly attenuate pain behavior induced by LSI. This study provides a new mechanism of LBP, and potentially has significant impact in the field. I only have a few moderate concerns and some minor concerns about this paper.
Moderate concerns: 1. Although vocalization threshold in response to force applied can be considered as a pain behavior of LBP, the authors should clarify that all the testes of spontaneous activities are not specific pain behaviors. Lumbar spine instability itself, with damage of the spine, can potentially reduce spontaneous activities without pain. Similarly, aged mice can have reduced spontaneous activities without pain.
2. It is hard to understand why LSI mice can have hind paw mechanical hypersensitivity, which is usually the result from hind paw local inflammation or nerve injury. In LSI model, the injury is far away from hind paw, and without local inflammation or nerve injury (with negative straight leg raising test), the authors should discuss how hind paw mechanical hypersensitivity can develop in LSI model.
3. In pathological study of human samples, the subjects with LBP are significantly older than subjects without LBP (Supplementary Table). The authors should clarify this key difference in main test. Moreover, because the age has significant impact on endplate pathology, as demonstrated in mouse study, the authors should tone down their claim that the high endplate scores is linked to history of frequent LBP.
Minor concerns: 1. Although porous endplates might be one of the major sources of LBP, it is unclear it is the "primary" source of LBP, as claimed in line 53.  10. The function of angiogenesis is unclear in LBP and therefore seems to be irrelevant to the main goal of this study. It might be better to move these results to supplementary data.
Reviewer #3 (Remarks to the Author): Low back pain (LBP) associated with Degenerative disc disease is a serious clinical problem with no effective treatment. The underlying mechanism of LBP is still largely unknown. This manuscript attempts to study the cellular and molecular mechanisms of LBP using two mouse models: aging and lumbar spine instability surgery. Analysis of sclerotic endplates in these models revealed that PGE2 activates sensory nerves to induce spinal pain behavior and that osteoclasts play a primary role in promoting sensory innervation though secreting Netrin-1. To my knowledge, this is the first study proposing that remodeling of endplate via endochondral ossification is the major cause of LBP. Therefore, it represents an important breakthrough in delineating the mechanisms of LBP. However, the quality of data, especially the imaging data, somewhat diminish the enthusiasm for this m anuscript. Major comments: 1. Most imaging data are shown at a high magnification. It would be better to show at least some of them at a low magnification in order to orientate the readers and to display internal controls. For example, in Fig. 1A, the image should include some disc area and some vertebral bone area. The authors proposed that after LSI endplates become a bone like structure containing osteoblasts, osteoclasts, and bone marrow. Since osteoclast is the major cause for pain, it would be interesting to see whether sclerotic endplates have more osteoclasts than vertebral bone. The same applies to  Fig. 5J, L: images seem to be at a low resolution and not focused. 5. Fig. 5K: Western blot is a better way to demonstrate changes of p-PKA and p-CREB in cell culture than immunofluorescence. 6. Fig. 7A: does Rankl CKO affect the bony structure in endplates without surgery? Again, showing the underlying vertebral bone will serve as a positive control that CKO does have osteopetrosis phenotype. 5. Two mechanisms are proposed to explain LBP: PGE2/EP4/nerve and osteoclast/Netrin-1/nerve. Are they interconnected? Are osteoclasts the major source of PGE2 in endplates? From Fig. 5B, it seems that most bone marrow cells are PGE2 and Cox2 positive. Minor comments: 1. Type H vessels, as proposed by Adams group in 2014, refer to CD31highEmcnHigh vessels. All bone marrow capillaries express CD31 and Emcn. Therefore, the authors cannot call their CD31+Emcn+ cells as type H vessels. 2. Osteoclast number should be normalized against tissue area.

Introduction
The first 4 sentences should be condensed.

Response:
We appreciate the reviewer's suggestion. The first 4 sentences have been condensed in the revised manuscript.
Low back pain (LBP) is one of the most common chronic pain conditions that causes significant social and economic burden, and the degenerative disc is considered as the major source of LBP. In this report, Ni et al demonstrated that porous endplate, rather than the disc, is a major source of LBP. Specifically, they showed increased osteoclast activity in endplates of lumbar spine instability (LSI) mouse model and in aged mice, which induces CGRP (+) neuron innervation of porous endplates. The porous endplates produce PGE2 to activate CGRP (+) nerve fibers by PGE2 receptor EP4, and eventually leads to sodium influx through Nav1.8 channel. Inhibition of EP4 from DRG sensory neurons, inhibition of osteoclast formation by knocking out Rank1, or preventing sensory innervation into porous endplates by knocking out Netrin-1 from osteoclasts greatly attenuate pain behavior induced by LSI. This study provides a new mechanism of LBP, and potentially has significant impact in the field. I only have a few moderate concerns and some minor concerns about this paper.

Response:
We are encouraged by the reviewer's insightful and accurate comments.
Moderate concerns: 1. Although vocalization threshold in response to force applied can be considered as a pain behavior of LBP, the authors should clarify that all the testes of spontaneous activities are not specific pain behaviors. Lumbar spine instability itself, with damage of the spine, can potentially reduce spontaneous activities without pain. Similarly, aged mice can have reduced spontaneous activities without pain. . We conducted tests of spontaneous activity to indicate the potential effect of spinal pain. We agree with the reviewer's comment about these tests are not specific for spinal pain. We have clarified that we monitored spontaneous activity to indicate the potential effect of spinal pain, but all the tests of spontaneous activity are not specific pain behaviors in the Results part.

Response
2. It is hard to understand why LSI mice can have hind paw mechanical hypersensitivity, which is usually the result from hind paw local inflammation or nerve injury. In LSI model, the injury is far away from hind paw, and without local inflammation or nerve injury (with negative straight leg raising test), the authors should discuss how hind paw mechanical hypersensitivity can develop in LSI model.

Response:
We appreciate the reviewer's important suggestions. The sign of the hind paw mechanical allodynia is considered as the secondary indicator of spinal pain-associated behaviors. Several studies reported that the hind paw mechanical allodynia develops in low back pain animal models as the secondary hypersensitivity (ref 58-61)). Among these works of literature, one study about the lumbar facet joint osteoarthritis-induced spinal pain excluded the local inflammation or nerve injury (with negative straight leg raising test) (ref 61). One study demonstrated that the mouse sciatic nerve predominantly origins from the L3 and L4 DRG by injecting retrograde labeling in the hind paw (ref 62). Our retrograde tracing data demonstrated that L3 DRG is also the partial origin of sensory nerves in the endplates of L4/5 in LSI mice (Fig. 4B, 4C). In addition, the dorsal horn of spinal cord receives inputs from several segmental DRGs (ref 63, 64). The major monosynaptic input for the dorsal horn neurons in L4 segment is from the L4-L6 DRGs, the dorsal horn neurons in L3 segment is from the L2-L5 DRGs (ref 65). These anatomical features might be the basis of the hind paw mechanical hypersensitivity in the LSI model, indicating the development of the referred pain. Referred pain is pain perceived at a location other than the site of the painful origin. It is caused by the nociceptive dorsal horn and brain stem neurons that receive convergent inputs from various tissues. Clinically, low back pain can be accompanied by referred pain in the lower extremities in patients without sciatica. We have included it in the Discussion part.
3. In pathological study of human samples, the subjects with LBP are significantly older than subjects without LBP (Supplementary Table). The authors should clarify this key difference in the main test. Moreover, because the age has significant impact on endplate pathology, as demonstrated in mouse study, the authors should tone down that the high endplate scores is linked to history of frequent LBP.

Response:
We appreciate the reviewer's important suggestion. We have clarified the difference of age between patients with LBP and patients without LBP in the main text. Also, we toned down the statement that high endplate scores are linked to a history of frequent LBP. This sentence was rewritten as 'The increased endplate scores were also observed in patients with a history of frequent LBP (Supplementary Figure 4B). However, the patients with the history of frequent LBP are older than the ones without the history of frequent LBP (Supplementary Table  1)' in the revised manuscript.
Minor concerns: 1. Although porous endplates might be one of the major sources of LBP, it is unclear it is the "primary" source of LBP, as claimed in line 53.

Response:
We appreciate the reviewer's valuable suggestion. We deleted this word in the revised manuscript.

Reviewer #3 (Remarks to the Author):
Low back pain (LBP) associated with Degenerative disc disease is a serious clinical problem with no effective treatment. The underlying mechanism of LBP is still largely unknown. This manuscript attempts to study the cellular and molecular mechanisms of LBP using two mouse models: aging and lumbar spine instability surgery. Analysis of sclerotic endplates in these models revealed that PGE2 activates sensory nerves to induce spinal pain behavior and that osteoclasts play a primary role in promoting sensory innervation though secreting Netrin-1. To my knowledge, this is the first study proposing that remodeling of endplate via endochondral ossification is the major cause of LBP. Therefore, it represents an important breakthrough in delineating the mechanisms of LBP. However, the quality of data, especially the imaging data, somewhat diminish the enthusiasm for this manuscript.

Response:
We appreciate the reviewer's overall insightful, constructive, and accurate comments. We have improved the quality of data in the revised manuscript.
Major comments: 1. Most imaging data are shown at a high magnification. It would be better to show at least some of them at a low magnification in order to orientate the readers and to display internal controls. For example, in Fig. 1A, the image should include some disc area and some vertebral bone area. The authors proposed that after LSI endplates become a bone like structure containing osteoblasts, osteoclasts, and bone marrow. Since osteoclast is the major cause for pain, it would be interesting to see whether sclerotic endplates have more osteoclasts than vertebral bone. The same applies to Fig. 5B. For uCT images, only sclerotic endplates were shown. How about the underlying vertebral trabecular bone? Is there any change there?

Response:
We appreciate the reviewer's important suggestion. We have replaced the images in Fig. 2A and Fig. 5B with low magnification images to show some disc area and some vertebral bone area as internal controls in the revised manuscript. For the μCT data in Fig. 3 Figure. 2). Also, the trabecular BV/TV, Tb.N, trabecular bone thickness (Tb.Th) increased and trabecular bone separation distribution (Tb. Sp) decreased significantly in RANKL -/mice relative to RANKL f/f mice. However, there is no significant difference in these parameters between sham surgery group and LSI surgery group in RANKL -/mice or RANKL f/f mice (supplementary fig. 7).
2. Some images, such as Fig. 2F and 4B, are totally black on paper. When zoomed in on the computer screen, it appears that

Response:
We appreciate the reviewer's important suggestion and apologize for the inappropriate representative images. Because Fig. 2F were overexposed, the green signals were background signals. For Fig. 4B LSI L3, there were red signals of Dil in the original image. However, the image quality in the merged PDF is poor leading to the loss of information. We have replaced these images with more clear and representative ones in the revised manuscript.
3. Some images do not match with their quantification data. Fig. 4D L2-L top panel: I do see many Dil+ only cells but quantification in E says 100% Dil+ cells are CGRP+. The label for red square in E should be Dil+CGRP+ cells %. Fig. 5E and F also do not match. There were very few yellow cells in all 4 images.

Response:
We appreciate the reviewer's important suggestion and apologize for the inappropriate representative images. Dli + signals are co-localized with CGRP+ in the previous Fig. 4D L2-L top panel. But the image quality got poor in the merged PDF, and the positive signals became unclear. We have replaced the inappropriate images in Fig. 4D and Fig. 5E with more precise and representative images in the revised manuscript. For the quantification in Fig.  4E, we showed the percentage of Dil + CGRP + cells in the total Dil + cells.
4. Fig. 5J, L: images seem to be at a low resolution and not focused.

Response:
We appreciate the reviewer's important suggestion. The images in Fig. 5J, L were selected from time-lapse photos, which might reduce the image resolution. Besides, the image quality got poor in the merged PDF. We have replaced and optimized the images in Fig. 5J, L in the revised manuscript. 5. Fig. 5K: Western blot is a better way to demonstrate changes of p-PKA and p-CREB in cell culture than immunofluorescence.

Response:
We appreciate the reviewer's valuable suggestion. To validate the changes of pPKA and pCREB, we did western blot to examine the level of pPKA and pCREB in the primary sensory neurons after treatment in the revised manuscript. Western blot demonstrated that the PKA inhibitor or EP4 -/significantly reduced PGE2-induced increase in the level of pPKA and PCREB ( Fig. 6D and 6E in the revised manuscript). These western blot data further validated the immunofluorescence data. 6. Fig. 7A: does Rankl CKO affect the bony structure in endplates without surgery? Again, showing the underlying vertebral bone will serve as a positive control that CKO does have osteopetrosis phenotype. Response: We appreciate the reviewer's valuable suggestion. From the μCT analysis and safranin O/fast green staining, RANKL CKO did not significantly affect the structure of endplate without surgery. To demonstrate the osteopetrosis phenotype in RANKL CKO mice, the uCT analysis of the L5 vertebrae was conducted in the revised manuscript (Supplementary Figure. 7). The uCT results demonstrated that the trabecular BV/TV, Tb.N, and Tb.Th increased and trabecular bone separation distribution (Tb.Sp) decreased significantly in the L5 vertebrae of RANKL CKO mice relative to RANKL f/f mice. 7. Two mechanisms are proposed to explain LBP: PGE2/EP4/nerve and osteoclast/Netrin-1/nerve. Are they interconnected? Are osteoclasts the major source of PGE2 in endplates? From  Fig. 5B, it seems that most bone marrow cells are PGE2 and Cox2 positive.

Response:
We appreciate the reviewer's thoughtful comment. In our manuscript, the osteoclastsecreted Netrin-1 attracts the growth of sensory nerves into the porous endplates. The PGE2 accumulated in the porous endplates stimulates the newly innervated nerves through its receptor EP4 and causes spinal pain. The immunostaining results in Fig. 5B showed that many bone marrow cells in porous endplates are COX2 and PGE2 positive. Some studies stated that the osteoclast and osteoblast are the potential sources of PGE2 (Blood. 2005 Aug 15;106(4):1240-5. Nat Commun. 2019 Jan 14;10(1):181.). To show the potential source of PGE2 in the porous endplates, we conducted the co-immunostaining for COX2 with F4/80, COX2 with TRAP, COX2 with osteocalcin (OCN). The results of immunofluorescent staining demonstrated that the COX2 + cells are co-localized with F4/80 + , some OCN + , and a few TRAP + cells in the porous endplates (Supplementary Figure. 5). These data showed that several types of cells in the porous endplates might be the source of PGE2 which accumulates in the porous endplates and stimulates the sensory nerves.
Minor comments: 1. Type H vessels, as proposed by Adams group in 2014, refer to CD31highEmcnHigh vessels. All bone marrow capillaries express CD31 and Emcn. Therefore, the authors cannot call their CD31+Emcn+ cells as type H vessels.

Response:
We appreciate the reviewer's important comment. We deleted the claim of type H vessels in the revised manuscript. We claimed the blood vessels in porous endplate as CD31 + Emcn + vessels.
2. Osteoclast number should be normalized against tissue area.

Response:
We appreciate the reviewer's important suggestion. We have normalized osteoclast number against tissue area in the revised manuscript.