Macrolide derivatives reduce proinflammatory macrophage activation and macrophage‐mediated neurotoxicity

Summary Introduction Azithromycin (AZM) and other macrolide antibiotics are applied as immunomodulatory treatments for CNS disorders. The immunomodulatory and antibiotic properties of AZM are purportedly independent. Aims To improve the efficacy and reduce antibiotic resistance risk of AZM‐based therapies, we evaluated the immunomodulatory and neuroprotective properties of novel AZM derivatives. We semisynthetically prepared derivatives by altering sugar moieties established as important for inhibiting bacterial protein synthesis. Bone marrow‐derived macrophages (BMDMs) were stimulated in vitro with proinflammatory, M1, stimuli (LPS + INF‐gamma) with and without derivative costimulation. Pro‐ and anti‐inflammatory cytokine production, IL‐12 and IL‐10, respectively, was quantified using ELISA. Neuron culture treatment with BMDM supernatant was used to assess derivative neuroprotective potential. Results Azithromycin and some derivatives increased IL‐10 and reduced IL‐12 production of M1 macrophages. IL‐10/IL‐12 cytokine shifts closely correlated with the ability of AZM and derivatives to mitigate macrophage neurotoxicity. Conclusions Sugar moieties that bind bacterial ribosomal complexes can be modified in a manner that retains AZM immunomodulation and neuroprotection. Since the effects of BMDMs in vitro are predictive of CNS macrophage responses, our results open new therapeutic avenues for managing maladaptive CNS inflammation and support utilization of IL‐10/12 cytokine profiles as indicators of macrophage polarization and neurotoxicity.

chemokines and contribute to continued cell death and a persistent inflammatory microenvironment within the injured spinal cord. 1,2 In contrast, alternatively activated macrophages (M2) release anti-inflammatory cytokines and facilitate tissue repair. 2 Increasingly, clinicians and researcher are testing the therapeutic potential of drugs that polarize macrophage activation toward reparative phenotypes in a variety of CNS disorders.
Macrolide antibiotics are a class of natural products consisting of a highly substituted macrocyclic 14-, 15-, or 16-membered lactone ring. Azithromycin (AZM) is a 15-membered, second generation, synthetic derivative of erythromycin with improved pharmacokinetic properties and a broad antimicrobial spectrum. 3 Azithromycin is well tolerated and commonly prescribed. Moreover, AZM becomes highly concentrated in macrophages and other phagocytes. 4,5 Across a variety of inflammatory conditions, AZM attenuates proinflammatory cytokine production by macrophages and other immune cells. 6 Azithromycin and other macrolide antibiotics are now being tested as immunomodulatory agents for CNS disorders.
Specifically, we and others observed immunomodulatory effects and improved recovery with AZM treatment in spinal cord injury, stroke, and retinal ischemia/reperfusion injury. [7][8][9][10][11][12] The neuroprotective properties of AZM in these models are associated with direct effects on macrophages. [7][8][9] We have shown that in vitro application of AZM to proinflammatory M1 bone marrow-derived macrophages (BMDMs) dampens the release of proinflammatory cytokines, increases M2-associated anti-inflammatory cytokines, and reduces the neurotoxicity of M1 macrophage-conditioned medium. 7 In efforts to improve efficacy and/or reduce the risk of increasing antibiotic resistance, researchers are evaluating the immunomodulatory potential of AZM derivatives and other macrolide derivatives with the goal of separating the antibiotic from immunomodulatory properties. As a result, some macrolide derivatives have been shown to retain immunomodulatory properties in models of lung inflammation, inflammatory bowel diseases, arthritis, and skin inflammation. [13][14][15][16][17][18] The ability of macrolide derivatives to reduce macrophage-mediated neurotoxicity, however, is unknown. With the increased use of AZM as an immunomodulatory agent for macrophage-mediated neurotoxicity in CNS pathologies, our goal in the present study was to determine whether macrolide derivatives retain neuroprotective properties. Using a semisynthetic approach to target modification of the sugar moieties of AZM, we generated a small library of derivatives, some of which lacked the cladinose found in the parent. We then tested the cytokine profiles and neurotoxicity of M1-stimulated BMDMs treated with derivatives and observed that unique derivatives reduce M1-macrophage activation and subsequent neuron death.
Previously we determined that the effect of BMDMs in vitro is predictive of macrophage responses in the injured CNS 8,19,20 ; therefore, the results of the current study open new therapeutic avenues for the management of maladaptive inflammation in CNS disorders.

| Semisynthesis of AZM derivatives
Like other macrolides, AZM inhibits bacterial protein synthesis by binding to the 50S subunit of ribosome and thus interfering with the growth of the polypeptide chain. 21,22 Specifically, the sugar moieties are known to play an essential role in establishing binding interactions with the bacterial ribosomal assembly. We used a semisynthetic approach to create derivatives with targeted alterations in these bacterial binding sugar residues. The AZM derivatives that were synthesized and structurally confirmed by spectroscopic analysis are listed in Table 1. See Supporting information for mass and H NMR spectrum profiles. AZM 1: The removal of the cladinose of AZM followed a previously described procedure. 23  AZM4: AZM1 (4.00 mmol) was dissolved into 100 mL anhydrous ethyl acetate. Acetic anhydride (3.67 mL, 40.0 mmol) was added, and the solution was stirred for 2 hours at room temperature. The mixture was washed twice with NaHCO 3 (5% in water, 30 mL), and the organic phase was dried under vacuum. The product was purified by flash column chromatography with silica gel (DCM/MeOH) to give AZM2 (3.22 mmol). AZM2 (2.0 mmol) was dissolved into 40 mL anhydrous acetone solution on ice, and 2.0 mL Jones reagent was added dropwise. After 30 minutes, 10 mL of methanol and 10 mL saturated (d, J = 6.6 Hz, 3H), 0.87 (dd, J = 8.4, 6.5 Hz, 3H). 13

| Antibiotic activity
The minimum inhibitory concentration (MIC) displays the relative antibiotic potency of AZM and each derivative against a common strain of bacteria (Staphylococcus aureus). The protocol used for the determination of the MIC was as previously described with minor modifications. 24

| Statistical analyses
Results are expressed as mean ± standard deviation (SD) and analyzed using GraphPad Prism 6.0 (GraphPad Software). Data were compared by one-way analysis of variance (ANOVA) among groups followed by F I G U R E 1 Altering the antibiotic properties of azithromycin does not decrease macrophage viability. Bone marrow-derived macrophages (BMDMs) were isolated from adult mice and were treated with AZM, AZM1, AZM4, AZM5, and AZM7 at concentrations of 1, 5, 25, and 125 μmol/L for 24 h Cell viability was measured by using MTT assay. AZM or AZM derivatives exhibited no cytotoxicity at any tested concentration as compared to unstimulated, nontreated BMDM control (dotted line). Moreover, AZM and AZM derivatives at 25 and/or 125 μmol/L significantly increased proliferation of BMDMs as compared to unstimulated controls at *P < 0.05, **P < 0.01, ***P < 0.001. Data are mean ±SD and representative of three independent biological replicate experiments Dunnett's or Holmes-Sidak multiple comparison tests. Differences were determined to be statistically significant at P value ≤0.05.

| Assessment of macrophage viability
We chose to use primary bone marrow-derived macrophages (BMDMs) for our studies as BMDM responses in vitro are predictive of CNS macrophage response in vivo. When stimulated with LPS+IFN-gamma), BMDMs model proinflammatory macrophages found in neuropathologies. 2,8,19 No doses of AZM or its derivatives were toxic to BMDMs when applied directly to the cells for 24 hours (Figure 1). Interestingly, this prolonged stimulation of BMDMs with high doses of AZM and its derivatives resulted in increased readouts on the MTT assay indicative of increased BMDM proliferation or increased metabolic activity (Figure 1).
This effect was not as robust after 6 hours of stimulation ( Figure S8), and therefore, a 6-hr stimulation time point was used for subsequent assays.

| Macrophage IL-12/IL-10 levels with derivatives
The  Figure 2). In addition, the 25 μmol/L stimulation with derivative 5 significantly increased production of IL-10 relative to M1 (P < 0.001; Figure 2). RT-PCR analyses of select genes associated with M1 or M2 macrophage phenotypes (ie, IL-6, IL-1b, TNF-a, and TGF-b) demonstrated similar immunomodulatory effects between AZM and derivatives 4 and 7 ( Figure S9). Collectively, these data demonstrate that altering the bacterial binding residues of AZM does not reduce its immunomodulatory properties with AZM5 having similar dose-response properties as the parent compound.

| Macrophage-mediated neurotoxicity with derivatives
We previously reported that M1 supernatant, that is, macrophage- Contrastingly, AZM1 did not significantly decrease M1 MCM-induced neurotoxicity at any tested concentrations.

| D ISCUSS I ON
In this study, we demonstrate the retention of immunomodulatory activity in AZM derivatives with altered sugar moieties using ourin vitro model of macrophage CNS inflammation. This model accurately predicts the macrophage/microglial response in the injured CNS. 7,8,19,20 Specifically, we demonstrate that these derivatives, Further, this relatively unexplored therapeutic approach could likely be exploited more effectively with continued optimization of therapeutics such as AZM and related macrolides. In particular, one major obstacle in the clinical development of anti-inflammatory macrolide antibiotics, such as AZM, is the concern that increased use of these drugs for their secondary anti-inflammatory effects may inadvertently promote bacterial resistance to this antibiotic in the treatment of a variety of infections. In the spinal cord injury patient population, for example, AZM is the antibiotic of choice for treating recurrent respiratory infections and pneumonia, 32 a leading cause of death following spinal cord injury; thus, antibiotic resistance is a major concern. Fortunately, recent studies have indicated that macrolides modified to remove their antibacterial activity retain beneficial anti-inflammatory effects in models of inflammatory skin disorders and chronic lung diseases. 13,16 Collectively, we demonstrate that AZM derivatives with altered sugar moieties retain immunomodulatory properties. We did not, however, observe uniform immunomodulatory and neu- roprotective properties with all derivatives tested. AZM7, which had the most extensive chemical modifications (diacetylation and the removal of the cladinose moiety), invoked modest immunomodulatory effects exclusively at the highest concentration tested. Interestingly, however, AZM7 did not retain any antistaphylococcal activity. Similarly, AZM4, which also lacks the cladinose that is replaced by a carbonyl, had modest yet significant immunomodulatory effects with minimal antistaphylococcal activity.
Although AZM1 at the concentration of 125 μmol/L significantly increases IL-10 level, it has no effect in reducing IL-12 production.
Interestingly, AZM1 also induced significant but small changes, relative to AZM, in BMDM metabolic activity at this high dose.
AZM5, which was the only derivative tested without the cladinose removed, closely mimicked or slightly exceeded AZM's activity at all concentrations tested including the BMDM MTT assay.
Unfortunately, the acetylation of both sugars in AZM5 did not abolish the antistaphylococcal activity as desired. This may sug- it remains unknown whether these findings remain valid in vivo or whether derivatives of AZM retain the same mechanism of action. While complicated, continued work in these areas is essential as it could lead to new therapeutics, such as the compounds described here, or provide novel therapeutic targets for future drug development.
One interesting finding in the current study was the fairly pronounced increase in macrophage viability when treated with the highest dose of AZM or derivative for 24 hours (widely used time point for measuring drug toxicity in vitro). This assay measures the conversion of tetrazolium dye MTT 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide to formazan by NAD(P)H-dependent cellular oxidoreductase enzymes, and this measure of metabolic activity is routinely used to quantify changes in cell number or vitality. How AZM and its derivatives induce this effect at high concentrations, and how this may alter inflammatory activities remains unclear, however, it is unlikely that this is directly related to our observed shifts in cytokine profiles and neurotoxicity across drug concentrations. If the observed increases in IL-10 following AZM stimulations were simply a result of cellular proliferation, then IL-12 would also be expected to rise, instead, however, IL-12 levels fell dramatically. Similarly, measurements of cellular proliferation/metabolism at the 6-hour time point, when we measure IL-10/12 levels, displayed more modest increases in proliferation/ metabolism and are thus less likely to influence our cytokine profiles.
In conclusion, we have identified AZM derivatives that retain key immunomodulatory functions in our in vitro model of CNS inflammation. While the antiinfective properties of the derivatives were associated with neuroprotection, we also observed that some derivatives with greatly reduced antiinfective characteristics retained neuroprotective and anti-inflammatory functions. Although a limited sample size of derivatives was created and tested, this indicates that the antibiotic properties of AZM may not be required for immunomodulatory-mediated neuroprotection. With continued development, these compounds could become viable clinical neuroprotectants and immunomodulatory treatments for neuropathologies. Additionally, given the usage of AZM's anti-inflammatory properties across disciplines, these drugs hold great potential in treating a wide variety of inflammation-based human disorders.

ACK N OWLED G M ENTS
This work was supported by the University of Kentucky Igniting Research Collaborations Pilot Grant Program to JCG and SGVL.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.