Radiation causes tissue damage by dysregulating inflammasome–gasdermin D signaling in both host and transplanted cells

Radiotherapy is a commonly used conditioning regimen for bone marrow transplantation (BMT). Cytotoxicity limits the use of this life-saving therapy, but the underlying mechanisms remain poorly defined. Here, we use the syngeneic mouse BMT model to test the hypothesis that lethal radiation damages tissues, thereby unleashing signals that indiscriminately activate the inflammasome pathways in host and transplanted cells. We find that a clinically relevant high dose of radiation causes severe damage to bones and the spleen through mechanisms involving the NLRP3 and AIM2 inflammasomes but not the NLRC4 inflammasome. Downstream, we demonstrate that gasdermin D (GSDMD), the common effector of the inflammasomes, is also activated by radiation. Remarkably, protection against the injury induced by deadly ionizing radiation occurs only when NLRP3, AIM2, or GSDMD is lost simultaneously in both the donor and host cell compartments. Thus, this study reveals a continuum of the actions of lethal radiation relayed by the inflammasome-GSDMD axis, initially affecting recipient cells and ultimately harming transplanted cells as they grow in the severely injured and toxic environment. This study also suggests that therapeutic targeting of inflammasome-GSDMD signaling has the potential to prevent the collateral effects of intense radiation regimens.


Introduction
Total body irradiation (TBI) is used in combination with other therapies as a conditioning regimen for the transplantation of bone marrow, cord blood, or hematopoietic stem cells (HSCs) for patients with hematological malignancies such as acute and chronic leukemia, lymphoma, and myelodysplastic syndrome [1][2][3]. TBI followed by HSC transplantation is also recommended for the treatment of nonmalignant blood disorders, including hemoglobinopathies, aplastic anemia, and immune deficiencies [3,4]. Thus, TBI-containing conditioning regimens are widely used in the clinic, yet they cause life-threatening injuries and side effects in multiple organs, including the intestine, lungs, kidney, brain, and spleen [2,[4][5][6][7]. The skeleton is also adversely affected by radiotherapy, as this procedure damages bone extracellular matrix, dysregulates the differentiation and activity of bone-forming cells (osteoblasts) and hematopoietic bone-resorbing cells (osteoclasts [OCs]), and destroys bone marrow hematopoiesis niches, events that, ultimately, cause growth retardation, osteoporosis, and higher fracture risk [8][9][10][11][12]. Thus, the increased survival rates of patients subjected to radiotherapy raise concerns for long-term skeletal complications, which can worsen other morbidities. For example, it was reported that 20% to 50% of geriatric patients (�65 years) with a hip fracture die within 1 year of the injury [13][14][15][16][17]. Therefore, understanding the mechanisms through which high doses of radiation induce bone toxicity may open avenues for tailored adjuvant therapies to improve the quality of life of the survivors.
TBI is routinely used in preclinical settings as a myeloablative treatment for allogeneic or syngeneic adoptive cell transfer (ACT). High doses of TBI cause massive cell death, thereby generating cues that activate inflammatory pathways, including the inflammasomes, in innate and adaptive immune cells in the allogeneic or syngeneic models. Consistent with this view, the inflammasomes, including those assembled by AIM2 and NLRP3, are implicated in radiation-associated tissue injury [18][19][20][21][22]. Thus, although the inflammasomes are a normal participant in immune responses and tissue repair, their hyperactivation is unhealthy. Yet the extent to which the activities of the inflammasomes in transplanted and/or host cells impact tissue outcomes following ACT in animals conditioned by TBI remains poorly understood.
The inflammasomes are intracellular protein complexes that process pro-interleukin-1β (pro-IL-1β) and pro-IL-18 into IL-1β and IL-18, respectively [23,24]. The inflammasomes also cleave gasdermin D (GSDMD) into N-terminal GSDMD (GSDMD Nt ) and C-terminal GSDMD (GSDMD Ct ) fragments; the GSDMD Nt moieties form pores at the plasma membrane through which IL-1β and IL-18 are secreted into the extracellular milieu [25][26][27][28]. Although these cytokines are efficiently released by live cells, sustained generation of GSDMD Nt causes cytolysis as the result of excessive pore formation, which compromises the integrity of the plasma membrane [29][30][31]. This form of cell demise, called pyroptosis, is proinflammatory as lysed cells uncontrollably release not only IL-1β and IL-18 but also danger-associated molecular patterns (DAMPs) such as ATP, high mobility group box 1 (HMGB1), and S100A9 proteins [32][33][34]; the release of these DAMPs results in the recruitment of immune cells and the perpetuation of inflammation.
In this study, we determined the impact of inflammasome sufficiency or insufficiency in donor and/or host cells on TBI-induced tissue injury in the syngeneic bone marrow transplantation (BMT) mouse model. We find that this procedure causes systemic inflammation and damage to several tissues, including bones and the spleen. These anomalies are significantly reduced when transplanted and recipient cells simultaneously lack NLRP3, AIM2, or GSDMD, but not NLRC4. These findings establish a novel concept whereby the inflammasome pathways mediate collateral effects of high dose of radiation not only in resident cells but also in transferred cells as the result of the toxic environment that this treatment creates in the host.

Radiation causes bone injury
The high prevalence of bone loss following radiotherapy [8][9][10][11][12] provided a strong rationale for assessing changes in this tissue in mice exposed to a high dose of radiation. To optimize the TBI/BMT protocol, we leveraged TRAP-tdTomato (tdT) reporter mice in which the Acp5 promoter drove tdT expression in OC precursors [35], which derived from HSCs [36]. Wild-type (WT; tdT -) mice were subjected to 9 Gy, a clinically relevant dose of radiation [1,11,12]. Because this radiation was lethal, irradiated animals were inoculated with bone marrow cells from tdT + mice to generate tdT + !tdTmice. Conversely, tdTcells were injected into irradiated tdT + mice to obtain tdT -!tdT + mice. Nonirradiated tdT + mice and tdTmice of the same sex and age served as positive and negative controls, respectively. All nontransplanted, irradiated mice died within 10 days, whereas >95% of transplanted mice overcome the lethality. Flow cytometry analysis revealed that approximately 6% and 2.5% of bone marrow cells were tdT + cells in nonirradiated tdT + mice and tdT + !tdTmice, respectively, 3 weeks post TBI/ BMT (S1A and S1B Fig). These values were higher than the 0.6% background fluorescence signal of tdTmice and tdT -!tdT + mice and consistent with the reported frequency of OC precursors in bone marrow [37][38][39]. Attempts to confirm the depletion of tdT + cells by tissue imaging were unsuccessful, likely because of the modest activity of the Acp5-tdT reporter. In any case, micro-computed tomography (μCT) analysis revealed that all WT!WT mice developed time-dependent loss of bone mass (bone volume/total volume [BV/TV]), associated with increased number and surface of OCs, pronounced adipogenesis, and decreased dynamic indices of bone formation (S1C-S1H Fig). Thus, lethal radiation efficiently depletes host OC precursors, which are replaced by donor counterparts. Although radiation adversely affects bone formation and resorption, this work focuses on the latter phase of bone turnover.

Radiation damages bones through the NLRP3 and AIM2 inflammasomes, and GSDMD, but not the NLRC4 inflammasome
Endogenous molecules such as ATP and uric acid released by dying cells are sensed as DAMPs by surrounding live innate cells, leading to the activation of the NLRP3 inflammasome [22,32,34,40]. The massive cell death that radiation caused to host tissues provided a strong rationale for investigating its impact on the NLRP3 inflammasome. WT!WT and Nlrp3 -/-! WT male mice exhibited similar bone loss compared to WT controls ( Thus, loss of NLRP3 simultaneously in donor and recipient cells is necessary to significantly reduce radiation-induced bone loss in host mice. Self-DNA [18][19][20][21] and nucleotide-derived metabolites [41] activate the AIM2 and NLRC4 inflammasomes, respectively, though the NLRP3 inflammasome is also activated by mitochondrial DNA [42][43][44]. Given the DNA-destabilizing actions of radiation, we determined the extent to which the AIM2 and NLRC4 inflammasomes contributed to radiation-mediated bone injury. Based on the similarity of the phenotype of Nlrp3 -/-!WT mice and WT!Nlrp3 -/mice, Aim2 -/or Nlrc4 -/but not WT mice were used as hosts. Although radiation caused bone loss and increased OC formation in WT!Aim2 -/mice compared to nonirradiated Aim2 -/mice, these responses were significantly attenuated in Aim2 -/-!Aim2 -/mice (Fig 1G-1J; S2E and S2F Fig). By contrast, osteolysis occurred in WT!Nlrc4 -/mice comparably to Nlrc4 -/-! Nlrc4 -/mice (Fig 1K-1N; S2G and S2H Fig). Collectively, these results show that TBI causes bone demise through the NLRP3 and AIM2 inflammasomes but not the NLRC4 inflammasome.
We also analyzed the expression and maturation of GSDMD in tissues harvested 2 and 7 days after TBI/BMT. Although the levels of GSDMD were consistently higher in the intestine, bone marrow, spleen, and liver of WT!WT mice compared to WT mice ( Noticeably, cleaved GSDMD was undetectable in the intestine 3 days after irradiation (Fig 3C), likely as the result of pyroptosis of certain cell populations caused by excessive generation of GSDMD Nt fragment. Collectively, these findings ruled out the scenario that the increase in GSDMD levels in WT!WT tissues relative to WT controls was simply the result of a higher number of cells in chimeric mice. Thus, radiation induces not only the expression of GSDMD but also its activation, responses that affect cell fates in tissue in a context-dependent manner.

GSDMD exerts OC lineage autonomous actions
The attenuated OC differentiation in Gsdmd -/-!Gsdmd -/mice may be the result of decreased secretion of inflammatory cytokines such as IL-1β, which promotes osteoclastogenesis through various mechanisms, including up-regulation of RANKL expression [36,49]. In addition, GSDMD may exert actions that are OC lineage autonomous. To test the latter scenario, we analyzed GSDMD expression during osteoclastogenesis. GSDMD mRNA and protein levels increased during in vitro OC differentiation of bone marrow-derived macrophages (BMDMs, Fig 4A-4C) and RAW 264.7 cells (S7A Fig) induced by RANKL. Although cleaved GSDMD was not detected during osteoclastogenesis, it was processed in OC cultures exposed to nigericin and lipopolysaccharide (LPS), though to a lesser extent compared to BMDMs (Fig 4C). Importantly, RANKL-driven OC formation was robust from WT BMDMs but impaired from Gsdmd -/cells (Fig 4D and 4E). More importantly, whereas baseline bone mass of non-littermate WT and Gsdmd -/male mice appeared comparable (Fig 2B and 2D), when littermates were used, bone mass was higher in Gsdmd -/mice compared to WT male controls (Fig 4F and  4G), a phenotype that correlated with OC parameters (Fig 4H and 4I). Notably, baseline indices of bone were comparable between WT and Gsdmd -/female mice (S3C and S3D Fig). Thus, whereas GSDMD is dispensable for bone homeostasis in female mice, it is crucial in physiological bone maintenance in male mice. In light of the OC-promoting actions of irradiation [50], our results suggest that irradiation may induce bone loss through nonautonomous and autonomous actions in immune cells and the OC cell lineage, respectively.

Discussion
The AIM2 and NLRP3 inflammasomes are rapidly activated in response to perturbing signals, acting to clear the perceived danger and restore tissue homeostasis [51,52]. The failure of the organism to rapidly dampen danger signals and fine-tune inflammasome overreactions can be detrimental to health. These signals can be induced by ionizing radiation, as this procedure causes genomic instability and cytoplasmic accumulation of DNA, which can be recognized by AIM2 among other sensors [53]. This procedure also causes cell death, a rich source of the  Fig 2B, 2D, 2F, 2G, 2I can be found in S1 Data. Data are mean ± SEM. � P < 0.05, �� P < 0.005, ��� P < 0.0005, ���� P < 0.0001. μCT, micro-computed tomography; BV/TV, bone volume/total volume; GSDMD, gasdermin D; H&E, hematoxylin and eosin; N.Oc/BS, OC number/bone surface; ns, not significant; OC, osteoclast; TBI, total body irradiation; WT, wild-type.
https://doi.org/10.1371/journal.pbio.3000807.g002 activators of the NLRP3 inflammasome such as ATP and uric acid [22,32,54,55]. Our results reinforce the view that radiation-inflammasome cascades are harmful, since the multiorgan damage induced by a high dose of radiation is profoundly attenuated when mice lack the inflammasome sensing components (NLRP3 or AIM2) in donor and recipient cells. Although a recent study suggests that focal irradiation promotes intestinal injury via the AIM2 inflammasome [18], here we show that TBI causes systemic tissue demise through the NLRP3 and AIM2 inflammasomes but not the NLRC4 inflammasome, actions that culminate in the activation of GSDMD. The residual tissue damage seen in some Gsdmd-deficient mice may be driven by GSDME, which is also expressed by hematopoietic cells, including macrophages, and mediates pyroptosis [56][57][58][59][60]. Another eloquent study shows that acute graft-versus-host disease (GvHD) is unaffected in Nlrp3 -/-!WT mice, consistent with our results; however, this work conflicts with ours, as it finds that GvHD is significantly delayed in WT!Nlrp3 -/mice [22]. Differences in the experimental designs and endpoint outcomes may account for the apparent discrepant observations. Our findings position GSDMD downstream of the NLRP3 and AIM2 inflammasomes in response to radiation-triggered signaling cascades. However, further studies are needed to determine whether or not NLRP3 and AIM2 are recruited to the same inflammasome complexes in irradiated cells as reported in murine cells exposed to cGAMP [61]. Since AIM2 senses DNA, it will also be interesting to determine the impact of deficiency in DNA repair on radiation-induced inflammasome activation. Poly(ADP-ribose) polymerases (PARPs), which are involved in DNA repair and affected by inflammasome signaling, are attractive candidates for such studies [62][63][64][65][66].
Serum levels of IL-1β correlate with the extent of bone loss in chimeric mice, findings that are consistent with the ability of this cytokine to regulate the expression of osteoclastogenic factors such as RANKL and to expand myeloid cell populations, some of which are OC precursors [36,67]. Serum levels of IL-1β in these mice also correlate with the magnitude of spleen enlargement, an outcome that is likely caused by the extensive EMH that occurs in this tissue as a homeostatic compensatory response to compromised hematopoiesis in the bone marrow. Consistent with skeletal outcomes, the increase in cytokine secretion and spleen weights are significantly reduced in chimeric mice fully lacking GSDMD. However, blood cell counts are unaffected in these mice compared to WT!Gsdmd -/mice, perhaps as the result of systemic inflammation in the latter mouse strain returning to baseline levels at the time of the sampling.
The increased levels of GSDMD during OC differentiation are unanticipated considering its pyroptotic actions. GSDMD is also unexpectedly not cleaved during osteoclastogenesis, yet its loss results in decreased OC differentiation both in vitro and in vivo. These results, in conjunction with those showing that LPS and nigericin induce the cleavage of this protein in OC cultures, rule out the scenario of defective inflammasome signaling in the OC lineage. However, the expression of GSDMD and maturation by terminally differentiated OCs still needs to be proven, given the cellular heterogeneity of OC cultures. Despite this knowledge gap, our findings revealing that GSDMD plays an important role in the differentiation of myeloid cells are novel because previous studies have focused on the function and regulation of this protein in inflammatory cells [25][26][27]68].
In summary, this study was driven by the hypothesis that the high sensitivity of the inflammasome pathways to homeostatic perturbations should, in theory, predispose these WT!WT, Gsdmd -/-!Gsdmd -/-, and WT!Gsdmd -/ mice. (C, D) Three-month-old WT male mice were left untreated or subjected to 9-Gy TBI (irradiated mice were not transplanted with bone marrow cells). Samples from 1-2 mice/group were collected 2 or 7 days after TBI/BMT (A, B) or 1, 2, and 3 days post TBI (C, D) and analyzed alongside control ("Cont") lysates by immunoblotting. (E, F) IL-1β and IL-18 levels in the serum from chimeric mice (3 weeks post TBI/BMT) and control mice. The data underlying this figure may be found in S1 Data and S2 Data. Data are mean ± SEM. � P < 0.05. BMT, bone marrow transplantation; GSDMD, gasdermin D; IL, interleukin; ns, not significant; TBI, total body irradiation; WT, wild-type.
https://doi.org/10.1371/journal.pbio.3000807.g003 multifunctional protein platforms to aberrant activation by excessive ionizing radiation signals. Additionally, the inflammasomes should be activated not only in the irradiated host cells but also in the donor cells as they engraft in the damaged tissues. We have validated this new concept by demonstrating that suppression of inflammasome-GSDMD signaling in both donor and recipient cells is required to achieve efficacy upon exposure to a high dose of radiation.

Bone microstructure and histomorphometry
The femurs were embedded in 2% agarose gel and scanned at 10 μm, 55 kVp, 145 μA, 8 W, 300-ms integration time using a μCT system (μCT 40; Scanco Medical AG, Zurich, Switzerland) as previously described [70,71]. The regions of interest (ROIs) were defined at 1 mm proximal to the end of the femoral growth plate.
For dynamic histomorphometry, mice were intraperitoneally injected with 10 mg/kg calcein green (Sigma-Aldrich, Cat# C0875, St. Louis, MO, USA) and 4 days later with 50 mg/kg alizarin red (Sigma-Aldrich, Cat# A3882). Mice were euthanized 2 days after the second injection. The left tibias and femurs were collected and fixed in 70% ethanol overnight, embedded in methyl methacrylate, and sectioned at 7-10 μm. Images were obtained using NanoZoomer. Measurements of dynamic bone histomorphometry were calculated from fluorochrome double labels at the endocortical surfaces as previously described [72].
For spleen histology, tissues were fixed in 10% formalin overnight, then embedded in paraffin followed by sectioning at 5 μm. After being mounted to positively charged glass slides, sections were dried and performed with hematoxylin-eosin (H&E) staining by standard methods.

OC differentiation and TRAP staining
Mice were euthanized and bone marrow was flushed out from the tibias and femurs. BMDMs were obtained by culturing bone marrow cells in media containing 10% CMG, a source of M-CSF, for up to 5 days in a 10-cm dish as previously described [70,73]. After removing the nonadherent cells by vigorous washes with PBS, adherent cells were plated at 5 to 10 × 10 3 /well in a 96-well plate in culture media containing 2% CMG and 50-100 ng/ml RANKL. Cells were maintained at 37˚C in a humidified atmosphere of 5% CO 2 , with media changed every other day. At the end of the culture period, the cells were rinsed with water and incubated with the TRAP staining solution (Sigma leukocyte acid phosphatase kit) at room temperature for 30 minutes. Multinucleated TRAP-positive cells with at least 3 nuclei were scored as OC under light microscopy.

RNA isolation and RT-qPCR
RNA was extracted from cells by using RNeasy Plus Mini Kit (Qiagen). cDNA was prepared using High-Capacity cDNA Reverse Transcription Kits (Applied Biosystems). Gene expression was detected by qPCR using SYBR Green (Applied Biosystems) according to the manufacture. The data were analyzed using the ΔΔCT method normalizing against cyclophilin B.

Measurements of IL-1β and IL-18 levels
For IL-1β measurements in bone marrow, flushed bone marrow was centrifuged, and the supernatants were collected as described previously. IL-1β levels were measured and quantified using Luminex kits (Minneapolis, MN, USA).

Statistical analysis
Statistical analysis was performed using Student t test, one-way ANOVA with Tukey's multiple-comparisons test, or two-way ANOVA with Tukey's multiple-comparisons test in Graph-Pad Prism7.