Nociceptors protect sickle cell disease mice from vaso-occlusive episodes and chronic organ damage

Sickle cell disease (SCD), caused by a single missense mutation in the β-globin gene, affects ∼100,000 patients in the United States and millions worldwide (Piel et al., 2017;Ware et al., 2017). Mutated β-globin polymerizes under deoxygenation, leading to sickle-shaped RBCs, which exhibit increased adherence to other blood cells or to the endothelium and are prone to undergo premature clearance and hemolysis (Sundd et al., 2019). The RBC alterations lead to a chronic inflammatory state, resulting in ischemic tissue damage, manifested by severe pain and organ failure. Acute vaso-occlusive episodes (VOEs) and chronic organ damage are the hallmarks of SCD, although it is not clear how the latter is related to the former. The pathophysiology of VOE and chronic organ damage is complex and involves the interplay of altered blood rheology, endothelial activation, and the secretion of inflammatory cytokines enabling leukocyte adhesion and activation (Zhang et al., 2016). Intravital microscopy analysis of the SCD mouse microcirculation has revealed that RBCs interact with adherent leukocytes in the inflamed vasculature (Turhan et al., 2002). The accumulation of activated neutrophils interacting with RBCs progressively reduces blood flow and produces venular occlusions (Chiang et al., 2007; Hidalgo et al., 2009). Further studies suggest that the aged neutrophil subset (marked by low CD62L and high CXCR4 expression; Casanova-Acebes et al., 2013;Van Eeden et al., 1997) is the most active leukocyte subset driving VOE and possibly contributing to chronic organ damage in a humanized SCD mice model (Zhang et al., 2015).

Pain is the most frequent reason for hospital visits among SCD patients (Aich et al., 2019; Ballas et al., 2012; Piel et al., 2017; Ware et al., 2017). While acute and chronic pain may coexist in SCD patients, their mechanisms may partially overlap (Aich et al., 2019; Ballas et al., 2012). Recent studies using a humanized SCD mouse model have revealed that hyperalgesia in SCD was caused by substance P, prostaglandin E₂, and chemokine receptor 2 (Aich et al., 2019; Khasabova et al., 2019; Sadler et al., 2018; Vincent et al., 2013). The pain in SCD (and other conditions) is largely sensed by the transient receptor potential vanilloid 1 (TRPV1; Hillery et al., 2011), which confers sensitivity to heat (Basbaum et al., 2009) and is activated by its natural ligand, capsaicin (Caterina et al., 1997).

Nociceptive nerves are present in every tissue in the body and are excited when thermal, mechanical, or chemical stimuli reach the noxious range. The sensory nervous system has recently emerged as a critical regulator of the immune response (Baral et al., 2018; Chiu et al., 2013; Cohen et al., 2019; Godinho-Silva et al., 2019; Nagashima et al., 2019; Oetjen et al., 2017; Riol-Blanco et al., 2014;Talbot et al., 2015). In particular, sensory nerves were reported to regulate neutrophils, group 2 innate lymphoid cells, and T cells via calcitonin gene–related peptide (CGRP) released from nerve terminals in both skin and lung (Baral et al., 2018; Cohen et al., 2019; Nagashima et al., 2019; Pinho-Ribeiro et al., 2018; Wallrapp et al., 2019). Because pain sensation is the major clinical manifestation of SCD, we asked whether, and if so how, pain affects the inflammatory response that plays a key role in SCD.

To study the effect of pain in SCD, we bred Trpv1-Cre mice with transgenic mice in which the diphtheria toxin fragment A can be conditionally expressed (iDTA) to obtain Trpv1-Cre; iDTA animals, leading to the elimination of TRPV1-expressing nociceptors. To generate nociceptor-deficient and control SCD mice, we transplanted bone marrow nucleated cells from Berkeley SCD mice into lethally irradiatedTrpv1-Cre; iDTA (SCD → Trpv1-Cre; iDTA) and iDTA control (SCD → iDTA) animals (Fig. 1 A). Donor SCD engraftment in the bone marrow of Trpv1-Cre; iDTA and iDTA recipients was similar (>97%; not depicted). To confirm the efficiency of the deletion, we subjected recipient mice to a tail flick assay in hot water and found that control SCD → iDTA mice lifted their tails within 3 s, whereas the SCD → Trpv1-Cre; iDTA animals kept their tails in the hot water for >10 s (10 s was used as cutoff to avoid excessive damage; Fig. 1 B). These results indicate that nociceptors were efficiently depleted in SCD → Trpv1-Cre; iDTA mice.

To determine the role of somatosensory nerves in VOE, we compared the VOE responses between SCD → Trpv1-Cre; iDTA mice and SCD → iDTA controls using the TNFα-induced VOE model (Turhan et al., 2002; Zhang et al., 2015). Surprisingly, we observed an exaggerated inflammatory response in SCD →Trpv1-Cre; iDTA mice, manifested by significant elevations of leukocyte adhesion, reduced rolling fraction, and increased rate of interactions between RBC and leukocytes by intravital microscopy (Fig. 1, C–E). In addition, the mean centerline RBC velocities, blood flow rate, and wall shear rate were all significantly reduced in SCD →Trpv1-Cre; iDTA mice compared with SCD → iDTA control animals, suggesting a more severe vaso-occlusion in nociceptor-depleted animals (Fig. 1 F and Table S1). SCD → Trpv1-Cre; iDTA mice also succumbed to the acute vaso-occlusive challenge significantly faster than SCD → iDTA animals (median survival time, 237 and 177 min, respectively;Fig. 1 G). Although we found no significant difference in the severity of anemia between SCD → Trpv1-Cre; iDTA and SCD → iDTA mice (Fig. S1, A–C), the number of circulating aged neutrophils, which were previously suggested to drive VOE in humanized SCD mice (Zhang et al., 2015), was significantly elevated in the SCD →Trpv1-Cre; iDTA mice (Fig. 1, H and I). These data suggest that the nociceptive pathway protects SCD mice from VOE.

CGRP, released from activated sensory nerve terminals, is a potent vasodilator that has also been suggested to regulate inflammation (Baral et al., 2018; Nagashima et al., 2019; Wallrapp et al., 2019). To investigate whether this neurotransmitter mediated the protective effect of sensory nerve, we administered CGRP or vehicle using a slow-releasing micropump and analyzed these mice 6 d after the initiation of treatment (Fig. 2 A). Interestingly, the inflammatory response was reduced in CGRP-treated SCD mice, as evidenced by significant reduction in leukocyte adhesion, increased rolling fraction, and reduced rates of interactions between RBC and leukocytes (Fig. 2, B–D), suggesting that CGRP inhibited leukocyte activation. In addition, the mean centerline RBC velocities, blood flow rate, and wall shear rate were all significantly improved in CGRP-treated SCD mice compared with vehicle-treated SCD mice (Fig. 2 E and Table S2). Moreover, the survival time of CGRP-treated SCD mice was greatly prolonged compared with control mice (Fig. 2 F), suggesting that CGRP released from activated somatosensory nerves mediated the protective effect of nociceptors against VOE. Hematologic analysis showed no difference in the degree of anemia between vehicle-treated and CGRP-treated SCD mice (Fig. S1, D and E). FACS analysis of the peripheral blood cells revealed significant reductions of aged neutrophil numbers in CGRP-treated SCD mice (Fig. 2, G and H). As activation of the CGRP receptor results in Gα_s-mediated activation of adenylate cyclase (AC), causing accumulation of cAMP (Russell et al., 2014), we tested whether stimulation of AC with forskolin, an AC activator, could mimic the effect of CGRP in protecting SCD mice from VOE. Interestingly, forskolin, like CGRP, protected SCD mice from VOE, as shown by the reduced inflammatory response (Fig. S2, A and B), significant improvement of blood flow (Fig. S2 C andTable S3), and prolonged survival time (Fig. S2 D). Nociceptors thus inhibit VOE through CGRP.

To get more mechanistic insights into how CGRP protects the SCD mice from VOE, we generated SCD mice in which nonhematopoietic cells are deficient in CGRP receptors. Receptor activity–modifying protein 1 (RAMP1), together with calcitonin receptor-like receptor (CALCRL), form the CGRP receptor and mediate its downstream signaling (McLatchie et al., 1998). We transplanted bone marrow cells from SCD mice into lethally irradiatedRamp1^+/+ andRamp1^−/− mice to generate SCD → Ramp1^+/+ and SCD →Ramp1^−/− animals, respectively (Fig. 2 I). We then challenged these mice with VOE and found thatRamp1 deletion in nonhematopoietic cells had no effect on the outcome of VOE (Fig. 2, I–M; and Table S4), suggesting that CGRP exerts its protective effects by acting on hematopoietic cells. Neutrophils, especially aged neutrophils, drive VOE in SCD mice (Chiang et al., 2007; Hidalgo et al., 2009; Turhan et al., 2002;Zhang et al., 2016). Our results suggest that CGRP reduces aged neutrophil numbers and activity (Fig. 2, G and H), which is consistent with recent studies in the context of aStreptococcus pyogenes infection (Pinho-Ribeiro et al., 2018). Taken together, these results support a model in which the activation of nociceptors in SCD mice leads to the release of CGRP, which inhibits aged neutrophil accumulation and subsequent VOE.

As the chili pepper extract capsaicin is a natural ligand of TRPV1 (Caterina et al., 1997), we hypothesized that low-grade activation of nociceptors resulting from spicy food ingestion might mitigate VOE severity in SCD mice. To test this idea, we fed SCD mice for 6 wk with a diet containing capsaicin (100 ppm) and analyzed their response to VOE challenge (Fig. 3 A). The capsaicin-containing diet had no effects on the consumption of food or water (Fig. S3, A and B), body weight (Fig. S3 C), or the degree of anemia of SCD mice (Fig. S3, D and E). Intravital microscopy analyses revealed that the addition of capsaicin in the diet significantly reduced the number of adherent white blood cells (WBCs) and WBC–RBC heterotypic interactions while significantly increasing the number of rolling leukocytes compared with standard diet (Fig. 3, B–D). In addition, the mean centerline RBC velocities, blood flow rate, and wall shear rate were all significantly improved in SCD mice that consumed spicy food compared with controls, suggesting that capsaicin can mitigate acute vaso-occlusion (Fig. 3 E and Table S5). We also found that the survival time after the VOE challenge was significantly prolonged in mice fed with capsaicin-containing diet compared with standard chow (Fig. 3 F). Interestingly, the capsaicin diet, like CGRP, significantly reduced the number of aged neutrophils in the circulation (Fig. 3, G and H). These results suggest that a diet containing a TRPV1 agonist dramatically mitigates vaso-occlusive events.

Chronic organ damage is another major hallmark of SCD and the key cause of mortality. Although chronic organ damage might result from repeated cycles of VOE and ischemia-reperfusion injury (Nath and Hebbel, 2015), whether it is the case is not certain. For example, SCD mice lacking the integrin α_Mβ₂ are protected from acute VOE, but no impact on organ damage was observed (Chen et al., 2016). Given the improved response to VOE challenge by dietary spicy food, we tested whether chronic capsaicin consumption could mitigate liver and kidney damage (Fig. 4 A). We found that a long-term (4-mo) capsaicin diet also reduced the number of circulating aged neutrophils (Fig. 4, B and C). Strikingly, the liver injury, as measured by elevated aspartate transaminase (AST) and alanine transaminase (ALT) enzyme levels in plasma, was significantly improved in SCD mice fed for 4 mo on capsaicin-containing chow compared with standard chow (Fig. 4, D and E). Kidney failure is among the most devastating events in SCD patients (Nath and Hebbel, 2015). Compared with SCD mice born with the disease, SCD animals generated by bone marrow transplantation (BMT) develop damage to the kidney at a later stage than to the liver (Nasimuzzaman et al., 2019). Accordingly, we continued the treatment for 14 mo to evaluate the impact on kidney function and anemia. We did not observe any change in the degree of anemia (Fig. S3, F and G) or obvious behavioral abnormalities after long-term treatment of SCD mice (Video 1 and Video 2). Analysis of kidney function by measuring urine osmolality after overnight water deprivation revealed that BMT SCD mice developed significant kidney damage 14 mo after BMT, to a level similar to that of age-matched parental SCD mice (Fig. 4 F). Interestingly, the capsaicin diet markedly improved kidney function in BMT SCD mice (Fig. 4 F). We confirmed the beneficial effect by histological analysis of kidneys and livers from standard or capsaicin diet–treated SCD mice (Fig. 4, G and H). Capsaicin is used topically to treat pain from a wide range of chronic conditions (Derry et al., 2009). To test whether long-term consumption of a capsaicin diet altered pain perception, we analyzed the sensitivity of SCD mice to a heat stimulus using the Hargreaves test (Hargreaves et al., 1988). Our results showed that the capsaicin diet did not change the response of SCD mice to a noxious heat stimulus, as shown by similar paw withdrawal latency between standard or capsaicin diet–treated SCD mice (Fig. S3 H). These data thus demonstrate the protective effects of capsaicin diet in the development of liver and kidney damage in humanized SCD mice, possibly through CGRP release, suggesting that manipulation of the nociceptive pathway by diet is a promising method to preserve target organs without affecting nociception.

Under normal circumstances, pain is an evolutionarily conserved protective mechanism to alarm the host and avoid noxious stimuli. Excessive pain, however, can be detrimental to human health and a source of disease manifestation in SCD. Although much effort has been dedicated to alleviate pain in SCD, our results indicate, surprisingly, that depletion of nociceptors exacerbates VOE, and that, conversely, the activation of nociceptors using a capsaicin-containing diet protects humanized SCD mice from both VOE and organ damage. These data echo the evolutionarily conserved benefit of pain pathways and argue that current pain management methods, counterintuitively, may predispose SCD patients to VOE and organ damage. This hypothesis is in line with observations suggesting that SCD patients on chronic opioid treatment may have poorer functional outcomes and more hospitalization than the control cohort (Carroll et al., 2016). Mechanistically, our results suggest that it is not the pain perception, but rather CGRP release, that mediates the protective effect of nociceptors. Thus, drugs reducing pain perception without affecting CGRP release would be most suitable for pain management. CGRP administration or induction of cAMP by forskolin treatment, or a diet containing the TRPV1 ligand capsaicin, all reduced excessive inflammation during tissue injury. Our results point to an inherent protective mechanism of nociceptors, and shed new light on potential novel disease-modifying targets to treat SCD.

Tg[Hu-miniLCRα1^Gγ^Aγδβ^S]Hba^−/− Hbb^−/− mice (Berkeley sickle cell mice) are bred in the laboratory and have been used in previous studies (Hidalgo et al., 2009; Turhan et al., 2002; Zhang et al., 2015). B6.129-Trpv1tm1(cre)Bbm/J (Trpv1-Cre) and B6.129P2-Gt(ROSA)26Sortm1(DTA)Lky/J (iDTA) mice were purchased from the Jackson Laboratory. Ramp1^−/− mice were a gift from Dr. Kathleen M. Caron (University of North Carolina, Chapel Hill, NC; Li et al., 2014), and littermate controls were used in our experiments. C57BL/6 mice were purchased from Charles River Laboratories. All mice were housed in specific pathogen–free conditions and fed with autoclaved food. Experimental procedures were approved by the Animal Care and Use Committee of Albert Einstein College of Medicine.

SCD mouse cohorts were generated by transplantation of bone marrow cells from Berkeley SCD mice into lethally irradiated 6–8-wk-old C57BL/6 mice as previously described (Hidalgo et al., 2009; Turhan et al., 2002;Zhang et al., 2015). These BMT SCD mice were analyzed for reconstitution efficiency ≥3 mo after transplantation. Fully reconstituted mice (>97%) were used for experiments and are referred to here as SCD mice; Berkeley SCD mice are referred to as genetic SCD mice.

Cells were surface stained in PEB buffer (PBS supplemented with 0.5% BSA and 2 mM EDTA) for 30 min at 4°C in the dark. Multiparametric flow cytometric analyses were performed on a LSRII equipped with FACS Diva 6.1 software (BD Biosciences) and analyzed with FlowJo software (TreeStar). Dead cells were excluded by forward and side scatter and DAPI (Sigma-Aldrich) staining. Neutrophils were gated by Gr-1^hi CD115^lo SSC^hi. Aged neutrophils were gated by CD62L^low CXCR4^high within the neutrophil gate. Fluorophore-conjugated antibodies against mouse Gr-1 (RB6-8C5), CD115 (AFS98), and CXCR4 (2B11) were purchased from eBioscience. Antibody specific to CD62L (MEL-14) was purchased from BD Biosciences.

Blood was collected in 0.5 M EDTA, pH 8.0, through the retroorbital sinus. WBC count, RBC count, and hemoglobin concentration were determined using an ADVIA 120 Hematology System (Siemens).

Experimental procedures and data analyses were performed as previously described (Turhan et al., 2002; Zhang et al., 2015). Briefly, male SCD mice were treated i.p. with 0.5 μg TNFα (R&D Systems) and anesthetized 2 h later with an i.p. injection of a mixture of 2% chloralose (Sigma-Aldrich) and 10% urethane (Sigma-Aldrich) in PBS. Tracheal intubation was performed to ensure normal respiration after anesthesia. The cremaster muscle was gently exteriorized, mounted onto a microscopic stage, and superfused with Ringer’s solution (pH 7.4, 37°C) aerated with a mixture of 95% N₂ and 5% CO₂. Postcapillary venules (18–25-μm diameter) were recorded using a custom-designed upright microscope equipped with a 60× water-immersion objective and a charge-coupled device video camera (Hamamatsu). The adhesion fraction was quantified as the number of leukocytes remaining stationary for >20 s within a 100-μm venular segment. In this model, >90% of adherent leukocytes are neutrophils, based on previous studies (Chiang et al., 2007). WBC–RBC interactions were defined as the associations between an RBC and an adherent leukocyte for ≥1 s and quantified as the number of interactions within a 100-μm vessel segment/min per adherent leukocyte. Rolling cells along venules within 1 min were quantified. Venular diameters were measured using a video caliper, and centerline red cell velocity (V_RBC) for each recorded venule was measured using an optical Doppler velocimeter (Texas A&M). Blood flow rate (Q) was calculated as Q =V_mean × πd²/4, where d is venule diameter, and centerline velocity (V_mean) is estimated as V_RBC/1.6. Survival times, defined as the time from TNFα injection until death, were recorded.

Plasma levels of ALT and AST were measured on a Beckman Coulter AU480 Chemistry Analyzer in the Biomarker Analytic Research Core at Albert Einstein College of Medicine.

Organs were excised and fixed in 10% neutral buffered formalin for 48–72 h and routinely processed for paraffin embedding. Samples for histopathology diagnostics were sectioned to a thickness of 5 μm and stained with H&E. Images were taken using a P250 slide scanner in the Analytical Imaging Facility at Albert Einstein College of Medicine.

Mice were deprived of water for 16 h before urine collection. Urine osmolality was measured with a Model 3320 Osmometer according to the manufacturer’s protocol (Advanced Instrument).

100 μl of PBS or CGRP solution (0.3125 mg/ml; Tocris Bioscience) was injected into the ALZET microosmotic pumps (model 1007D). The micropumps were left in PBS at 37°C overnight for equilibration and then implanted subcutaneously into the SCD mice.

The customized capsaicin-containing diet (100 ppm; capsaicin purchased from Sigma-Aldrich) and standard diets were generated by Tekland Diets at Evigo.

Forskolin (Cayman Chemical) was given i.p. at 2 mg/kg twice a day for 5 d, and control mice were treated with vehicle control (DMSO) in PBS.

Thermal nociceptive responses were determined by measuring the latency to paw withdrawal using the Hargreaves apparatus (IITC Plantar Analgesia Meter; IITC Life Science). This assay was done blinded to different treatments. Briefly, male SCD mice were placed in a clear plastic cage and allowed to acclimatize for 1 h. The glass base temperature was 30°C (± 1°C). The aiming light (no thermal stimulus) was used to direct the source to the plantar surface of the paw. Stimuli of a given intensity, designated active intensity as percentage of maximum intensity of the apparatus, were presented until there were three or four unequivocal withdrawal responses to the stimulus. A withdrawal response was defined as a rapid, reflexive lifting or moving of the stimulated paw. Voluntary movement or ambiguous responses were not recorded. Stimuli were presented to alternating hind paws, with 5–10 min between stimulus presentations. The mean latency to withdraw the paw is presented and analyzed. This test was performed at the Rodent Behavior Core at Albert Einstein College of Medicine.

Results are presented as means ± SEM. Statistical analyses were performed with Prism 7 and 8 (GraphPad Software). Unpaired two-sided Student’st tests (two-tailed) were used to compare two groups. One-way ANOVA was used for multiple group comparisons using Tukey’s multiple comparisons test provided in Prism. Mantel–Cox test (log-rank test) was used to compare survival curves.

Fig. S1 provides additional information for Figs. 1 and 2 and shows the degree of anemia in SCD mice with depletion of nociceptors or CGRP administration.Fig. S2 relates to Fig. 2 and shows the effect of forskolin on VOE. Fig. S3 relates to Figs. 3 and 4 and shows the water/food consumption, body weight, and degree of anemia in SCD mice after short-term or long-term capsaicin-diet or standard-diet treatment, and the nociception of SCD mice after capsaicin-diet or standard-diet treatment. Table S1, Table S2, Table S3, Table S4, and Table S5 provide additional hemodynamic parameter data for Figs. 1–,3 and Fig. S2.Video 1 and Video 2 relate to Fig. 4 and show the behavior of SCD mice treated with the standard diet and capsaicin diet, respectively.

We thank Colette Prophete, George Amatuni, Xizhe Wang, and Peng Guo for technical assistance and members of the Frenette laboratory for helpful discussions, in particular Dr. Xin Gaofor helpful advice. We thank Dr. Kathleen M. Caron (University of North Carolina, Chapel Hill, NC) for providingRamp1^−/− mice.

This work was supported by National Institutes of Health R0 grants (HL069438, DK056638, DK116312, and DK112976 to P.S. Frenette). The Biomarker Analytic Research Core is supported by an Institute for Clinical and Translational Research Einstein-Montefiore Clinical and Translational Science Award grant (5UL1TR002556) and an Einstein–Mount Sinai Diabetes Research Center grant (P30-DK020541). The Analytical Imaging Facility was supported by a National Institutes of Health Shared Instrumentation Grant (1S10OD019961). The Rodent Behavior Core was supported by a Rose F. Kennedy Intellectual and Developmental Disabilities Research Center grant (P30 HD071593).

Author contributions: C. Xu conceived and designed the research, performed experiments, collected and analyzed data, and wrote the manuscript. M. Gulinello performed the Hargreaves test. P.S. Frenette designed and supervised the research and wrote the manuscript.