Pro-Resolving Mediator Profiles And 5-Lipoxygenase Activity In Cerebrospinal Fluid Correlate with Disease Severity and Outcome in Adults with Tuberculous Meningitis

Tuberculous meningitis (TBM) is the most lethal form of tuberculosis infection, characterized by a dysregulated immune response that frequently leads to neurological injury and death despite the best available treatment. The mechanisms driving the inflammatory response in TBM are not well understood. To gain insights into these mechanisms we used a lipid mediator profiling approach to investigate the regulation of a novel group of host protective mediators, termed specialized pro-resolving mediators (SPM), in the cerebrospinal fluid (CSF) of adults with TBM enrolled into a randomised placebo-controlled trial of adjunctive aspirin treatment. We found distinct lipid mediator profiles with increasing disease severity, changes that were linked with an upregulation of inflammatory eicosanoids in patients with severe TBM and a decrease in the production of a number of 5-lipoxygenase (ALOX5)-derived SPM. CSF pro-resolving mediator concentrations were also associated with 80-day survival. In survivors, we found a significant increase in pro-resolving mediator concentrations, including the ALOX5-derived resolvin (Rv)T2, RvT4 and 15-epi-Lipoxin (LX)B4, compared to those who died. Aspirin administration increased the ratio of pro-resolving to pro-inflammatory mediators decreasing the concentrations of the prothrombic mediator TxA2, changes that were linked with early reductions in brain infarcts and deaths. Together, these findings identify a CSF SPM signature that is associated with disease severity and 80-day mortality in TBM. Furthermore, the therapeutic manipulation of the ratio between pro-resolving mediators and pro-inflammatory/thrombogenic mediators in the CSF, by aspirin for example, offers a novel treatment strategy to reduce the morbidity and mortality caused by TBM. Authors Summary Infections of the brain and the meninges by Mycobacterium tuberculosis (M. tb) lead to severe inflammation and are associated with poor outcomes. The mechanisms leading to this disease remain poorly defined. Herein, we investigated how M. tb infection regulates the concentrations of specialized pro-resolving mediators that are central in controlling the body’s ability to clear infections. In these investigations, we found that disease survival was linked with increased concentrations of a number of these protective molecules including resolvins and lipoxins. Treatment of M. tb-infected patients with aspirin decreased the production of the immunosuppressive and thrombogenic mediator thromboxane A2 improving the balance between protective and inflammatory molecules. Of note, these changes were linked with reduced disease severity and improved survival. Therefore, the present findings suggest a previously unappreciated role for pro-resolving mediators in TBM pathogenesis.

Mycobacterium tuberculosis (M. tb) is responsible for more deaths globally than any 69 other infectious disease. When it infects the brain and meninges to cause 70 tuberculous meningitis (TBM), which represents 1-5% of all forms of tuberculosis, it 71 either kills or severely disables around a half of all sufferers despite the best 72 available treatment (1). The pathogenesis of TBM is not well understood, but poor 73 outcomes have been linked to dysregulated intracerebral inflammation (2). Current 74 therapeutic approaches are aimed at killing M. tb infecting the brain or controlling the 75 inflammatory response (3,4). To date, the latter has primarily involved adjunctive 76 corticosteroid therapy and has been associated with increased survival, although 77 how corticosteroids modulate intra-cerebral inflammation to influence outcomes 78 remains uncertain (5).

79
It is now well established that under ideal conditions the body activates control both leukocyte trafficking and phenotype (7-11). These autacoids also threatening complication of TBM), and by enhancing the resolution of intra-cerebral 96 inflammation through the increased expression of SPMs (14). The trial found that in 97 patients with microbiologically confirmed TBM, aspirin was associated with reduced 98 brain infarcts and/or death in the first 60 days of treatment.

99
Little is known about the regulation of lipid mediators, and in particular SPM, 100 in the cerebrospinal fluid (CSF) during TBM. Thus, in the present studies we 101 investigated whether CSF lipid mediator concentrations were altered with increasing 102 disease severity. We found that concentrations of a number of SPM families were 103 reduced with increasing disease severity. This was linked with an upregulation of 104 inflammation-initiating eicosanoids, including prostaglandins and cystenyl 105 leukotrienes. Pre-treatment CSF lipid mediator concentrations were also found to be 106 predictive of outcome with distinct lipid mediator profiles obtained in those patients 107 that were alive at the end of the study versus those that died during the study.

108
Finally, we also found that in aspirin-treated patients there was a dose-dependent 109 alteration of the CSF lipid mediator profiles improving the balance between pro-110 resolving and pro-inflammatory/pro-thrombogenic mediators.   Table 2). In order 131 to gain insights into the relationship between CSF lipid mediator concentrations and 132 disease severity we first grouped the mediators by biological function assessing the 133 differences between pro-resolving mediators (lipoxins, resolvins, protectins and  Table 2). Results of this analysis demonstrated that with increasing 138 disease severity there was a decrease in overall pro-resolving lipid mediator 139 concentrations in the CSF of patients with TBM ( Figure 1A). This relationship was 140 coupled with a trend towards an increase in the concentrations of pro-inflammatory 7 resolution index, that is the ratio between the concentrations of pro-resolving 143 mediators and pro-inflammatory eicosanoids, (16) also demonstrate a reduction in 144 the resolution index, indicative of increased inflammation, in patients with the more 145 severe disease ( Figure 1C).

146
In order to gain further insights into the mediator pathways that were 147 differentially regulated between these patient groups we next interrogated the 148 biosynthetic pathways for each of the essential fatty acid metabolomes. This 149 demonstrated that in patients with an MRC score of 2 and 3, when compared with 150 patients with an MRC score of 1, there was a significant reduction in two pro-151 resolving mediator families, the n-3 docosapentaenoic acid-derived resolvins (RvDn-3 152 DPA) and the arachidonic acid-derived lipoxins (LX). This was coupled with a 153 significant increase in the pro-inflammatory arachidonic acid-derived prostaglandins 154 (PG) and leukotrienes (LT; Figure 1D).   Table 2). In addition, we also observed increased concentrations 176 of LTE4, the terminal product in the cysteinyl leukotriene biosynthetic metabolome 177 and that was recently found to be also bioactive (19), with increasing disease 178 severity, although the correlation was not statistically significant after correction for 179 multiple testing ( Figure 2C, Supplemental Table 2).

180
To further validate the potential utility of these mediators as biomarkers of 181 disease severity we conducted multiple regression analysis, using Least Absolute  Table 3).

186
Given the role that immune cells play in the biosynthesis of lipid mediators (7, 187 9), we next investigated whether disease severity influenced CSF leukocyte 188 numbers. Here we found that cell numbers were reduced in those with more severe 189 disease, from 416±53 cells/mm 3 in patients with an MRC score of 1, to 239±27 190 cells/mm 3 and 164±39 cells/mm 3 in patients with an MRC score of 2 and 3 191 respectively, although these changes were not statistically significant (p>0.05). We 192 next assessed whether there was an association between CSF leukocyte counts and 193 the concentrations for each of the identified LM families. With adjustment for multiple 194 testing, no significant correlations were found between these parameters (data not 195 shown), suggesting that the observed difference may be due to a differential 196 activation of the leukocyte population.    Table 5 and Figure 4B). Out of these mediators 15-epi-LXB4 was 229 found to be significantly reduced after multiple correction. This lipid mediator was 230 also a strong predictor of mortality in TBM as demonstrated by LASSO regression 231 analysis (Supplemental Table 6).

233
Aspirin administration upregulates CSF concentrations of select pro-resolving 234 mediators during TBM

235
We recently reported that treatment with aspirin, dexamethasone and anti-236 tuberculosis drugs was associated with reduction in new brain infarcts and deaths 237 within 60 days in patients with microbiologically confirmed TBM (14). Therefore, we 238 investigated the impact of aspirin on CSF lipid mediator pathways in this patient sub-239 population. We compared lipid mediator profiles obtained after 30 days of aspirin 240 (1000mg/day or 81mg/day) or placebo (Supplementary Table 7

251
To further evaluate the regulation of CSF lipid mediator concentrations by 252 aspirin we conducted OPLS-DA. This analysis demonstrated that while day 30 CSF 253 lipid mediator profiles between the 81mg and placebo groups were not markedly different, lipid mediator profiles from patients given 1000mg aspirin gave distinct 255 clusters to that from patients given placebo ( Figure 5D,E). This separation between 256 the two patient groups was linked with a differential regulation of 18 mediators from 257 all the four bioactive metabolomes, including RvT4 and TxB2, the inactive breakdown 258 product of the prothrombotic TxA2 (20)( Figure 5E). Statistical assessment of their 259 concentrations in CSF demonstrated that after 30 days TxB2 was reduced with both 260 81 mg and 1000mg of aspirin, reaching statistical significance in those given 1000mg 261 after correcting for multiple testing (Supplemental Table 8 ,9).

264
In the present study we investigated the regulation of CSF lipid mediator profiles 265 before and during the treatment of adults with TBM. We found that pre-treatment

289
Recent studies demonstrate that nitric oxide synthase promotes the S-nitrosylation of 290 COX-2 increasing its catalytic activity and upregulating the production of protective mediators including PGI2 (22) and RvT (7). On the other hand, the calcium-

311
Studies conducted by Vane and colleagues demonstrate that aspirin inhibits 312 the production of prostaglandins and thromboxane, a mechanism that is dependent 313 on the acetylation of COX-1 and COX-2 (27). Later investigations by Serhan and 314 colleagues found that acetylation of COX-2 also led to a switch in the catalytic activity 315 of the enzyme, from the production of PGG2 to the formation of epimeric forms of 316 resolvins, lipoxins and protectins (28,29). In the preset study we did not observe powered for this analysis or because at the interval tested, i.e. 30 days post initiation 320 of treatment, the biosynthetic pathways leading to the formation of these molecules 321 were downregulated. Future studies powered to interrogate this question will need to 322 establish whether daily regulation of high dose aspirin downregulates these 323 pathways in the CSF. Of note, we found high dose aspirin administration reduced 324 TxB2 concentrations and was linked with improved outcomes.

325
In summary, the present findings uncover novel mechanisms in the 326 pathophysiology of TBM that may have relevance to all forms of tuberculosis.

327
Disease severity is associated with an alteration in lipid mediator expression and a 328 dysregulation in the production of several SPM pathways; a mechanism that is also 329 linked with a poor prognosis. This dysregulation was at least in part restored by 330 aspirin, lending support to the hypothesis that TBM arises from a failure of the host to 331 engage resolution mechanisms and offering novel treatment strategies.

334
The patients in the current study were enrolled in a clinical trial of adjunctive aspirin, 335 the design and conduct of which have been previously described (14)

371
In the analysis of peptide-lipid conjugated mediators eluted in the methanol 372 fraction, an Agilent Poroshell 120 EC-C18 column (100 mm x 4.6 mm x 2.7 μm) was 373 kept at 50°C and mediators were eluted using a mobile phase consisting of methanol-

431
Loading plot interpretation identified the variables with the best discriminatory power 432 (Variable Importance in Projection greater than 1) that were associated with the 433 distinct intervals and contributed to the tight clusters observed in the Score plot.