Evaluation of IL-1 blockade as a host-directed therapy for tuberculosis in mice and macaques

In 2017, there were over 550,000 estimated new cases of multi-drug/rifampicin resistant tuberculosis (MDR/RR-TB), emphasizing a need for new treatment strategies. Linezolid (LZD) is a potent antibiotic for antibiotic-resistant Gram-positive infections and is an effective treatment for TB. However, extended LZD use can lead to LZD-associated host toxicities, most commonly bone marrow suppression. LZD toxicities may be mediated by IL-1, a pathway important for early immunity during M. tuberculosis infection that later contributes to pathology. We hypothesized LZD efficacy could be enhanced by modulation of IL-1 pathway to reduce BM toxicity and TB associated-inflammation. We used two animal models of TB to test our hypothesis, mice and cynomolgus macaques. Antagonizing IL-1 in chronically-infected mice reduced lung neutrophil numbers and partially restored the erythroid progenitor populations that are depleted by LZD. In macaques, we found no conclusive evidence of BM suppression associated with LZD, indicating our treatment time may have been short enough to avoid the toxicities observed in humans. Though treatment was only 1 month, the majority of granulomas were sterilized with reduced inflammation (assessed by PET/CT) in animals treated with both LZD and IL-1 receptor antagonist (IL-1Rn). However, overall lung inflammation was significantly reduced in macaques treated with both IL-1Rn and LZD, compared to LZD alone. Importantly, IL-1Rn administration did not noticeably impair the host response against Mtb or LZD efficacy in either animal model. Together, our data support that inhibition of IL-1 in combination with LZD has potential to be an effective HDT for TB. Author summary Host-directed therapies (HDTs) are a potential option in combating drug resistant TB as they can circumvent bacterial drug-resistance by targeting host responses rather than the pathogen. Here we designed an HDT to target the IL-1 pathway, an inflammatory immune response that is both critical and detrimental to TB disease outcome. We combined IL-1Rn, an IL-1R antagonist, with linezolid (LZD) which is an effective antibiotic for drug-resistant M. tuberculosis. Extended treatment causes severe host-toxicities that might be mediated in part by the IL-1 pathway. Our goals were to enhance LZD efficacy by negating LZD-host toxicities and to reduce adverse inflammation caused by TB. In mice, IL-1Rn effectively reduced inflammatory signatures associated with TB and reversed linezolid-induced bone marrow suppression, the most common toxicity. In cynomolgus macaques, inflammation, as assessed by PET/CT, was reduced by the combination of IL-1Rn and LZD therapy, compared to LZD alone. In contrast to mice, we did not observe bone marrow suppression in macaques, highlighting the importance of both models when assessing prospective therapies. These data show the potential of IL-1Rn as a therapy for TB and support LZD as an effective antibiotic against drug-resistant TB.


82
Since HDT would be used as part of a multi-drug regimen, targeting mechanisms that increase 83 drug exposure or decrease toxicity could also be envisioned. While some HDT strategies hold 84 promise, very few have been rigorously tested in pre-clinical models (4).

86
Interleukin-1 (IL-1) has been implicated in TB disease and inflammation, making it a possible 87 target of HDT. This cytokine plays an important yet complicated role in TB disease progression.

88
The susceptibility of mice lacking critical mediators of IL-1 signaling indicates that some initial 89 production of IL-1b upon Mtb infection is essential for priming downstream immune responses 90 necessary for disease control (5-8). In contrast, IL-1 is also responsible for the accumulation of 91 disease-promoting neutrophils in chronically-infected susceptible mice, and genetic variants that 92 result in higher IL-1b production are associated with increased disease severity and neutrophil 93 accumulation in humans (9-11). Given that HDT is designed to be administered to chronically-94 infected patients during treatment, when persistent IL-1 production appears to play a more

106
This damage acts on the NOD-like receptor family, pyrin domain containing 3 (NLRP3) protein 107 that has been shown to be necessary for LZD-mediated bone marrow suppression in mice (16).

108
NLRP3, in conjunction with caspase-1 and ASC, forms an inflammasome complex, which 109 cleaves a number of substrates resulting in cell death and/or the release of active of IL-1b.

5
While the importance of NLRP3 in bone marrow suppression is clear, the relative roles of 111 inflammasome activation and IL-1 signaling remain uncertain.

113
Based on these observations, inhibiting the IL-1 pathway by HDT could serve two purposes: 114 first, to alleviate LZD-associated host toxicity, and second, to reduce the pathology associated 115 with the IL-1 pathway during TB disease. Due to the pro-inflammatory nature of the IL-1 116 pathway, strict regulatory mechanisms exist within the host to quell this pathway. IL-1 receptor 117 antagonist (IL-1Rn) is a protein produced constitutively at low levels that can increase in 118 response to a variety of cytokine signals. IL-1Rn serves as a decoy ligand for the IL-1R1, 119 blocking signal transduction and subduing activation of subsequent pro-inflammatory pathways 120 (17). Anakinra is an FDA-approved recombinant IL-1Rn that is used to treat rheumatoid 121 arthritis. As there are no FDA approved drugs to inhibit inflammasome activation, inhibition of 122 the IL-1 pathway with biologics, such as Anakinra, is the only currently feasible strategy to 123 modulate this pathway (18).

125
We hypothesized that the combination of Anakinra (herein referred to as IL-1Rn) with LZD for 126 treatment of active TB disease would reduce LZD-associated toxicities and host inflammation 127 resulting in a more efficacious therapy. To test this concept, we employed two established TB 128 animal models to assess differing aspects of host responses to LZD and IL-1Rn.

135
Together, our data further verify LZD as an efficacious antibiotic in both mice and macaques 6 and indicate that addition of IL-1Rn may accelerate the resolution of inflammation and partially 137 alleviate LZD-mediated BM suppression.

IL-1 receptor blockade reduces inflammation in mouse models of TB disease. 141
Given the complex role played by IL-1 during TB, we initially sought to determine the effect of 142 inhibiting this cytokine during established TB disease in mice. These studies compared two IL-1 143 antagonists, a blocking antibody to the murine IL-1 receptor (aIL-1R1) and the human IL-1 144 receptor antagonist, Anakinra (IL-1Rn), each delivered between days 14 and 28 post infection.

145
Two mouse strains were employed to assess the effects of these treatments in animals with 146 different amounts of IL-1 activity. In relatively resistant C57BL/6 animals, mature IL-1 147 production is controlled during chronic disease, whereas unregulated IL-1 drives inflammatory 148 disease in Mtb-infected Nos2 -/mice (9, 10).

150
While the human IL-1Rn had little effect in these models, the aIL-1R1 treatment significantly 151 reduced PMN numbers in the lungs of both resistant and susceptible mouse strains, and 152 reversed the weight loss observed in Nos2 -/mice ( Fig. 1A-B). While neither regimen 153 significantly altered the lung bacterial burden, both IL-1Rn and aIL-1R1 treatment reduced 154 bacterial burden in the spleens of Nos2 -/mice ( Fig. 1C-D). This generally beneficial effect of 155 aIL-1R1 treatment was consistent with qualitatively improved histopathological disease (Fig.   156   1E). By no metric did aIL-1R1 treatment exacerbate established disease in either mouse strain.

7
The IL-1R1 blocking antibody aIL-1R1 was next tested in combination with LZD to determine 161 whether the efficacy or toxicity of the antibiotic was altered. C3HeB/FeJ mice, which are 162 relatively susceptible to Mtb and develop histopathological lesions that more closely resemble 163 human disease, were used for these studies. Mice with established disease were treated 164 between days 28 and 46 post-infection with vehicle alone, LZD, aIL-1R1 or a combination of the 165 two. As previously reported, LZD was effective in this model, reducing lung neutrophil numbers, 166 bacterial burden and weight loss (Fig. 2 A-D) (23). The addition of aIL-1R1 to this regimen 167 further reduced lung neutrophil numbers, and did not significantly alter the antimicrobial activity 168 of LZD. As IL-1b production in response to both Mtb infection and LZD treatment depends 169 largely on the NLRP3 inflammasome, we also investigated a regimen in which LZD and a small 170 molecule NLRP3 inhibitor (MCC950) was administered between days 56 and 77 post-infection.

171
As observed with aIL-1R1, the addition of MCC950 reduced PMN numbers in the lung, relative 172 to LZD alone, and did not significantly alter the rate of bacterial killing (Fig. 2 E-G). Using a 173 more rapid treatment protocol and C57BL/6 mice with genetic deficiencies in Caspase 1 or 174 NLRP3, we confirmed that the antimicrobial activity of LZD was unaffected by inflammasome

188
While aIL-1R1 also significantly increased the number of erythroid precursors in the BM, the 189 suppression of LZD toxicity was less pronounced at this site.       ). In the current study, we did not have untreated control macaques, so we provide total 240 thoracic CFU for 3 similarly infected historical controls (untreated) for reference; these data 241 were excluded from statistical analyses. This supports that high dose LZD as a single drug is 242 effective at killing bacteria even in a short (4 week) regimen. Bacterial burden in CFU+ thoracic 243 lymph nodes was similar between the groups though trended towards lower CFU in LZD+IL-244 1Rn treated macaques (p=0.0965) ( Fig 5B). We previously reported that a 2-month lower dose 245 regimen of LZD could reduce bacterial burden compared to untreated controls, with ~80% 246 sterilized granulomas (14). Since the bacterial burden in macaques with active TB can vary 247 substantially, we estimated the pre-treatment total thoracic CFU from the total lung FDG by PET 248 CT as previously described (20), and compared the actual bacterial burden post-treatment 249 against the estimated pre-treatment value. All macaques, regardless of treatment group, had 250 lower total thoracic CFU at necropsy compared to the estimated total thoracic CFU prior to 251 treatment initiation (Fig. 5C). Our data indicate that while IL-1Rn did not significantly enhance 252 bacterial killing, it did not impair LZD-mediated bacterial clearance.

Lack of LZD-induced bone marrow suppression in macaques 255
In humans, LZD is associated with host toxicities during extended treatment periods of > 4 256 weeks (25). To determine whether bone marrow suppression occurred during the 4-week high 257 dose LZD therapy and whether IL-1Rn could modulate observable host toxicities, we isolated

IL-1 blockade modulates granuloma specific responses and healing dynamics 302
To determine whether IL-1Rn modulated immune responses at the site of infection, we chose at 303 random 5 granulomas per animal (25 per treatment group) and performed a multi-plex analysis 304 of granuloma supernatants (Fig. 8A). There were no significant differences in IL-1b, IL-1RA, or 13 IL-18 levels, which are associated with the IL-1 pathway. IL-2 and IL-17 are correlated with 306 protective immune responses during TB (26); there was a trend for higher IL-2 levels in LZD+IL-307 1Rn treatment and a statistically significant increase in IL-17a in LZD+IL-1Rn treated animals. Fibrosis is modulated by the IL-1 pathway, however while fibrosis is associated with 371 granuloma healing, lung fibrosis can cause secondary complications after TB disease 372 resolution (2) . We have shown in previous studies that drug therapy for TB induces 373 fibrotic healing in granulomas (31, 32). Therefore, we assessed whether IL-1Rn was 374 associated with changes in granuloma pathology. While we did not observe a 375 significant difference in the frequency of fibrotic versus non-fibrotic granulomas 376 (p=0.1531), there was a significant decrease in necrotizing granulomas when IL-1Rn 377 was added to LZD therapy. Thus, although there is no synergistic effect between IL-378 1Rn and LZD in promoting fibrosis associated with drug clearance, the reduction in 379 neutrophils could skew granuloma resolution towards a non-necrotizing, fibrotic lesion. 380

381
We also assessed LZD-associated bone marrow suppression and reversal with IL-1Rn 382 therapy. We designed our HDT to match the current FDA guidelines for LZD and IL-383 1Rn schedules, which resulted in a lack of observable bone marrow suppression in 384 macaques. In mice however, LZD-induced bone marrow suppression was reduced with 385 the addition of IL-1R1 antagonists, supporting our initial hypothesis. The role of IL-1 386 signaling is consistent with the ability of IL-1 to suppress erythropoiesis in mice by 387 reducing the number of progenitors (33). The remaining deficit in erythropoiesis during 388 IL-1 blockade could reflect either incomplete inhibition by IL-1Rn or an independent role 389 of inflammasome activation. Optimizing this effect will require further work to 390 understand the relative roles of IL-1 signaling and inflammasome activation. 391

392
Our data support and extend our previous data in macaques that LZD has excellent 393 efficacy against TB, even as a single-drug given for only 4-weeks, providing additional 394 support for LZD as an antimicrobial for MDR/XDR-TB cases (12). IL-1Rn therapy in 395 conjunction with LZD was successful in reducing TB-associated inflammation with no 396 negative effects on Mtb clearance, however additional studies addressing long-term 397 effects on immune responses and TB disease resolution are needed. Anakinra (IL-1Rn) 398 is already FDA-approved for adult and pediatric use in other inflammatory disorders; our 399 data provide pre-clinical evidence that IL-1Rn could be a potential therapy for cases of 400 severe TB to quell excessive inflammation and improve standard therapies. Our study 401 highlights the potential of HDTs for TB but also the necessity of assessment in 402 translational models prior to implementation in human trials.