Metastatic infiltration of nervous tissue and periosteal nerve sprouting in multiple myeloma induced bone pain

Multiple myeloma (MM) is a neoplasia of B plasma cells that often induces bone pain. However, the mechanisms underlying myeloma-induced bone pain (MIBP) are mostly unknown. Using a syngeneic MM mouse model, we show that periosteal nerve sprouting of calcitonin-gene related protein (CGRP+) and growth associated protein 43 (GAP43+) fibres occurs concurrent to the onset of nociception and its blockade provides transient pain relief. MM patient samples also showed increased periosteal innervation. Mechanistically, we investigated MM induced gene expression changes in the dorsal root ganglia (DRG) innervating the MM-bearing bone and found alterations in pathways associated with cell cycle, immune response and neuronal signalling. The MM transcriptional signature was consistent with metastatic MM infiltration to the DRG, a never-before described feature of the disease that we further demonstrated histologically. In the DRG, MM cells caused loss of vascularization and neuronal injury, which may contribute to late-stage MIBP. Interestingly, the transcriptional signature of a MM patient was consistent with MM cell infiltration to the DRG. Overall, our results suggest that MM induces a plethora of peripheral nervous system alterations that may contribute to the failure of current analgesics and suggest neuroprotective drugs as appropriate strategies to treat early onset MIBP. Significance statement Multiple myeloma is a painful bone marrow cancer that significantly impairs the quality of life of the patients. Analgesic therapies for myeloma-induced bone pain (MIBP) are limited and often ineffective, and the mechanisms of MIBP remain unknown. In this manuscript, we describe cancer-induced periosteal nerve sprouting in a mouse model of MIBP, where we also encounter metastasis to the dorsal root ganglia (DRG), a never-before described feature of the disease. Concomitant to myeloma infiltration, the lumbar DRGs presented blood vessel damage and transcriptional alterations, which may mediate MIBP. Explorative studies on human tissue support our preclinical findings. Understanding the mechanisms of MIBP is crucial to develop targeted analgesic with better efficacy and fewer side effects for this patient population.


Intrafemoral inoculation of 5TGM1-GFP cells induces nociception
To understand the time course of neuronal changes leading to MIBP and the mechanisms involved, 92 we conducted a time course study where C57BL/KaLwRijHsd mice were transplanted with 5TGM1-  Our previous studies revealed that 5TGM1 inoculation induced complete bone marrow denervation 144 at the end stages of the model, leading us to speculate that tumour-induced nerve injury contributes 145 to MIBP. To further evaluate the temporal effect of 5TGM1 cell inoculation on the bone marrow 146 microenvironment, we performed immunohistological analyses of sensory (calcitonin-gene related 147 peptide, CGRP + ) and sympathetic (tyrosine hydroxylase, TH + ) nerve fibres. We observed that already 148 in MM D17 femurs, TH + and CGRP + fibres were not detectable in the bone marrow, which had been 149 colonized by 5TGM1-GFP + cells ( Figure 3E, G), suggesting that tumour-induced nerve injury in the 150 bone marrow is not a main player in MIBP. To confirm that the absence of marrow innervation was 151 not a result of technical difficulties, we identified both TH + and CGRP + nerve fibres in bones of the 152 sham D17 mice ( Figure 3B-D) and in the periosteum of MM D17 (Fig 3F, H). 153 We next sought to examine the effect of intrafemoral 5TGM1-GFP inoculation on periosteal 154 innervation. The periosteum is the bone compartment with the highest nerve density (12, 13) and 155 alterations to periosteal nerve fibre innervation have been described as a feature of bone pain (14,156 15), including cancer-induced bone pain (16, 17). Importantly, we found a significant increase in the To investigate the human relevance of our findings, we next examined whether periosteal infiltration 169 of MM cells in patients is associated with nerve sprouting. In formalin-fixed, paraffin-embedded 170 trephine iliac crest bone biopsies from 13 newly diagnosed MM (NDMM) patients, we performed a 171 multiplex immunostaining for CD138 + MM cells, CD34 + blood vessels and the pan-neuronal marker 172 PGP9.5 ( Figure 3U). Our quantification showed that the median periosteal nerve density in NDMM 173 patients was 3.736 profiles/mm 2 , ranging from 0 to 82.869 profiles/mm 2 ( Figure 3V); this is in 174 contrast with reports of periosteal nerve density in non-cancerous patients showing a median of 0.077 175 profiles/mm 2 (range: 0.02-0.68 profiles/mm 2 ) (18). Moreover, we found a significant increase in 176 periosteal nerve density in NDMM displaying periosteal infiltration of CD138 + cells (Fig 3W), 177 suggesting a direct role for MM cells in promoting nerve sprouting. Periosteal nerve density in 178 NDMM patients was positively correlated with age ( Figure 3X; R 2 =0.33; p<0.05) but independent of 179 tumour burden, which was assessed as percentage bone marrow clonality ( Figure 3Y) and 180 paraproteinemia ( Figure 3Z). Moreover, periosteal nerve density was independent of sex and IgG 181 type (data not shown). This is, to our knowledge, the first evidence of MM-induced alterations to 182 bone innervation in MM patients and altogether our data suggest that periosteal nerve sprouting may 183 play a role in MIBP. CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in  CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in Next, we tested the mechanistic role of periosteal nerve sprouting on MIBP using a therapeutic anti-207 netrin-1 blocking antibody (NP137). Netrin-1 is an axon guidance molecule known to play a pivotal 208 role in neurogenesis through binding to its canonical receptors UNC5 homolog (UNC5H) and deleted CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 30, 2022. ; https://doi.org/10.1101/2022.12.29.522199 doi: bioRxiv preprint tumour burden, as assessed by terminal splenomegaly ( Figure 4D). To evaluate whether the transient 218 analgesic effect was a consequence of decreased osteolysis, we performed µCT analyses of endpoint 219 sham VEH , MM VEH and MM NP137 femurs. Our results demonstrated that netrin-1 blockage does not 220 revert the BV/TV reduction induced by MM ( Figure 4E). Moreover, the structural parameters of 221 trabecular bone and bone mineral density were unchanged between MM VEH and MMN P137 femurs 222 (data not shown). To confirm the capacity of NP137 treatment to block periosteal nerve sprouting, 223 we assessed the presence of CGRP + nerve fibres in the femoral periosteum of the treated mice. Our 224 evaluation confirmed that NP137 treatment inhibits nerve sprouting, as MM NP137 mice showed 225 CGRP + periosteal nerve density similar to sham, while MM VEH ipsilateral femurs showed significant 226 CGRP + nerve sprouting ( Figure 4F, G); likewise, periosteal infiltration of MM cells was decreased 227 in antibody-treated MM mice ( Figure 4H). Anatomical presence of a microneuroma was observed in 228 25% of the vehicle-treated MM mice ( Figure 4F), a feature of the disease never before described but 229 that is consistent with animal models of solid bone cancers such as prostate (24) and breast (16) bone 230 metastases or osteosarcoma (17, 25). The concentration of serum NP137 was significantly higher in 231 serum of sham NP137 mice compared to MM 137 mice, potentially due to higher antibody clearance by 232 the tumour (Supplementary Figure S2). Systemic antibody treatment by itself in sham mice had no 233 effect on behaviour, bone structural parameters or neuronal innervation (data not shown).

234
. CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in     CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 30, 2022. ; https://doi.org/10.1101/2022.12.29.522199 doi: bioRxiv preprint S5). Likewise, no changes in glial fibrillary acidic protein (GFAP) staining were observed, suggesting 294 that astrocytosis is not a main feature of MIBP (Supplementary Figure S5). 295 296 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 30, 2022. ; https://doi.org/10.1101/2022.12.29.522199 doi: bioRxiv preprint Friedman´s two-way test followed by Wilcoxon´s two-sample test and ****p<0.0001 by unpaired, CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 30, 2022. ; https://doi.org/10.1101/2022.12.29.522199 doi: bioRxiv preprint controls ( Figure 6B, C). Thus, our data indicate that MM has the capacity to metastasize to the 323 peripheral nervous system, which occurs concomitantly to development of nociception. 324 Next, we examined the integrity of DRG vascularization and potential neuronal damage through 325 immunofluorescent staining of the ipsilateral DRG of sham and MM mice 17 or 24 days after cell 326 inoculation. Our analyses of CD31 + blood vessels ( Figure 6D) revealed a significant decrease in blood 327 vessel length ( Figure 6E, H, K) and density ( Figure 6F     CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 30, 2022. ; https://doi.org/10.1101/2022.12.29.522199 doi: bioRxiv preprint allowing the passage of MM cells into the DRG) or is a consequence of it, remains to be elucidated.     CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 30, 2022. ; https://doi.org/10.1101/2022.12.29.522199 doi: bioRxiv preprint confirmation of loss of pedal reflex, the mouse was placed on a heating pad and an incision (<1 cm) 554 was made above the right anterior knee upon its shaving and disinfection with 70% ethanol. The   CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 30, 2022. ; https://doi.org/10.1101/2022.12.29.522199 doi: bioRxiv preprint Quantification was performed by counting the number of CD31 + blood vessels per unit volume (area 632 x thickness) (52-54), where blood vessels were identified as CD31 + and 2-10 μm in diameter; CD31 + 633 branched blood vessels were counted as one single vessel (55).. DRG sections were initially scanned 634 at low magnification (10×) to identify the areas with the highest density of CD31 + blood vessels and 635 then a confocal image was obtained at 40× magnification. At least three pictures per animal were 636 taken. An extended depth of focus processing was performed on all Z-stack files for each image and 637 total number and length of CD31 + blood vessels within cell body-rich areas (56) were determined 638 using Image J (NIH). Then this area was multiplied by the thickness of the section (16 µm).

Immunohistochemistry in femurs 640
Following µCT, femurs were decalcified in 10% EDTA for two weeks at 4°C prior to cryoprotection 641 in 30% sucrose and 4°C storage; decalcification was monitored using a portable x-ray equipment CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 30, 2022. ; https://doi.org/10.1101/2022.12.29.522199 doi: bioRxiv preprint 5 min (1:20,000, Sigma Aldrich; #D21490), washed, dehydrated through an alcohol gradient and 657 xylene, and sealed with DPX mounting medium (Slide mounting medium; Sigma-Aldrich; #06522).

Quantification of the density of nerve fibers on femoral periosteum and neuroma identification 659
For quantifications of CGRP + and GAP43 + nerve fibers density, at least three sections of each mouse 660 femur were analyzed. The areas with the highest density of nerve fibers at the metaphyseal periosteum 661 were identified using a 10x objective. In all cases, these areas were located within 0.5 mm-1 mm 662 distance from the distal femoral growth plate. Subsequently, a confocal image was obtained using

669
For the identification of neuroma-like structures, the following three criteria was followed 1) 670 disordered mass of blind-ending axons (CGRP + ) that had an interlacing and/or whirling morphology, 671 2) structure with a size of more than 10 individual axons that is at least 20 μm thick and 70 μm long, 672 and finally 3) a structure that is never observed in the periosteum of normal bone (57, 58). CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 30, 2022. ; https://doi.org/10.1101/2022.12.29.522199 doi: bioRxiv preprint performed as described for our MM transcriptomic data but ranking the genes by - CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 30, 2022. ; https://doi.org/10.1101/2022.12.29.522199 doi: bioRxiv preprint genes") were uniquely mapped to probesets on the chip. We assigned ranks to genes by sorting them CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in   CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 30, 2022. ; https://doi.org/10.1101/2022.12.29.522199 doi: bioRxiv preprint dorsal horn and calculating the corrected total cell fluorescence (CTCF) as previously described by CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 30, 2022. ; https://doi.org/10.1101/2022.12.29.522199 doi: bioRxiv preprint Supplementary Table S1 994 Procedure description

Randomization
• Time-course experiment: mice were randomized into sham or MM according to their baseline burrowing performance and baseline weight.
• Transcriptomics experiment: mice were randomized into sham or MM according to weight.
Randomization was performed by listing all mice from lower to higher baseline burrowing capacity or weight and stratifying in two groups. Average baseline burrowing capacity or weight was then calculated and mice were inter-exchanged among groups until the averages were equal ± 10%.

Allocation concealment
To minimize bias and cage, mice from different groups (i.e. MM and sham) were caged together. No physical distinctions among groups were available at mouse or cage level. The allocation code was kept in a close, secure file until the end of the experiment.

Blinding
All behavioural experiments were conducted by an investigator blinded to the experimental group.
Exclusion criteria • In all experiments: mice displaying surgical complications.
• In the transcriptomics experiment: mice with RNA RIN < 8.0 Reporting of excluded animals • Time-course experiment: • 2 mice due to surgical complications.
• 10 mice due to cancer contamination in vehicle solution.

995
996 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in  . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 30, 2022. ; https://doi.org/10.1101/2022.12.29.522199 doi: bioRxiv preprint