Erythromycin A (EMA), first isolated in 1952,1 is a clinically useful macrolide antibiotic, active against both Gram-positive bacteria and Gram-negative cocci and mycoplasmas. In the 1980s, EMA was found to have two other promising biological properties, gastrointestinal motor-stimulating activity and anti-inflammatory and/or immunomodulatory activity by Itoh et al.2 and Kudoh et al.,3 respectively. Based on these findings, our research group has been striving to create an ideal derivative, which exhibits a specific activity without showing antibacterial effect. We have already found a new motilide derivative (EM574, de-N-methyl-N-isopropyl-8,9-anhydroerythromycin A 6,9-hemiketal), which exhibited gastrointestinal motor-stimulating activity but showed no antibacterial activity.4, 5, 6, 7

As a result of its anti-inflammatory and/or immunomodulatory activity, EMA is an effective treatment for diffuse panbronchiolitis,3 chronic sinusitis and cystic fibrosis.8 Additionally, our group reported that EMA promotes monocyte to macrophage differentiation and inhibits proliferation of T cell in vitro.9, 10 Likewise, 14-membered macrolides are also well known to possess anti-inflammatory and/or immunomodulatory activity, which are represented by inhibition of inflammatory cytokine production.11, 12, 13 However, the detailed mode of action of EMA has not yet been determined. Macrolide derivatives with anti-inflammatory and/or immunomodulatory activity but lacking either antibacterial activity or gastrointestinal motor-stimulating activity would be extremely useful, as they would help avoid promotion of drug resistance, as well as help to minimize any adverse effects of EMA treatment.

In a previous paper,14 we reported our development of the novel 12-membered (8R,9S)-8,9-dihydro-6,9-epoxy-8,9-anhydropseudoerythromycin A (EM900), and analogues thereof, possessing the ability to promote monocyte to macrophage differentiation but lacking antibacterial activity. This provided us with a novel, acid-stable 12-membered macrolide chemical skeleton with anti-inflammatory and/or immunomodulatory activity, although no detailed structure–activity relationships had been clarified. From our initial study, we learned that neither disconnection of the side chain bearing from the C11 position nor the cladinose moiety has an important role in the anti-inflammatory and/or immunomodulatory effects. Consequently, we became interested in synthesis of analogues of the N,N-dimethylamino group on desosamine moiety, inspired by our experience with erythromycin derivatives with a potent motiline-like activity, especially motilide EM574.4 Generally, the N,N-dimethylamino group of EMA is an essential moiety for generating antibacterial activity. In contrast, in the case of motiline-like activity, modification of the N,N-dimethylamino group dramatically increases the activity.4 Taken together, we envisioned that modification of the N,N-dimethylamino group might be possible as a means of increasing anti-inflammatory and/or immunomodulatory activity.

In this communication, we report the elucidation of structure–activity relationships of the N,N-dimethylamino group of EM900, as well as those of the analogues, (8R,9S)-de(3′-N-methyl)-3′-N-(p-chlorobenzyl)-8,9-dihydro-6,9-epoxy-8,9-anhydropseudoerythromycin A (EM905) and (8R,9S)-de(3′-N, N-dimethylamino)-3′-morpholino-8,9-dihydro-6,9-epoxy-8,9-anhydropseudoerythromycin A (EM914). The two compounds, EM905 and EM914, were found to be effective for treatment in a mouse model of inflammatory bowel disease (IBD), even at considerably low doses, compared with the therapeutic sulfa drug, sulfasalazine. We propose that the ability to promote monocyte to macrophage differentiation might have potential as a reliable in vitro assay system to help identify and develop potent anti-inflammatory and/or immunomodulatory agents.

At the outset, removal of a methyl group from the nitrogen of desosamine via treatment of EM900 with I2, AcONa in MeOH,15 provided the de-N-methyl analogue EM901 in 79% yield (Table 1). Subsequent, de-N-methylation of EM901 using I2 and Na in MeOH15 afforded bis-de-N-methyl EM903 in 90% yield. With EM901 and EM903 in hand, we synthesized 24 analogues (Appendix), such as N-mono- and/or N,N-bis-substituted (for example, alkylated and cyclic alkylated) compounds to investigate structure–activity relationships focusing on this region. Reaction conditions (methods A–F) were as follows; (A) treatment of EM901 with alkyl halides (RI, RBr and RCl), and i-Pr2NEt in CHCl3 provided N-mono-alkylated compounds (entry 2–15); (B) treatment of EM901 or EM903 with aldehydes and NaBH(OAc)3 provided N-mono or N,N-bis-alkylated compounds (entry 18, 20, 21 and 23); (C) acetylation of EM901 provided EM960 (entry 16); (D) mesylation of EM901 provided EM961 (entry 17); (E) treatment with dibromoalkanes provided N,N-cyclic alkylated compounds (entry 22, 24–26); (F) removal of the Cbz group of EM965 was carried out under hydrogenation reaction condition to give EM966 (entry 26, step 2).

Table 1 Synthesis of N-substituted and N,N-bis-substituted analogues and their biological activities
Full size table

To investigate anti-inflammatory and/or immunomodulatory effects, we used the THP-1 assay system, which was modified according to the method of Keicho et al.,10 to test for promotion of differentiation of monocytic cells to macrophages. A THP-1 cell line, derived from a patient with monocytic leukemia, was supplied by the Japanese Cancer Research Resources Bank (Tokyo, Japan). THP-1 cells (1 × 105 per well in 0.5 ml) were placed into 48-well tissue culture microplates (IWAKI, Tokyo, Japan) and cultured in the presence of phorbol myristate acetate (PMA; 2 ng ml−1) or each macrolide compound (1–100 μM) alone, or both, for 4 days at 37 °C under 5% CO2 in humidified air. The number and viability of adherent cells was measured by colorimetric determination of MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) assay at 550 nm. The ED50 values were determined, which provided an evaluation of the promotion of monocyte to macrophage differentiation of each analogue, as compared with the result of EMA at 100 μM. Cytotoxicity (μM) was determined using cell-count reagent SF (Nacalai tesque, Tokyo, Japan) according to the manufacturer’s instructions.

As the bioactivity data of these analogues are summarized in Table 1, the ED50 of de-N-methyl and bis-de-N-methyl analogues EM901 and EM903 are similar compared with that of EM900 (ED50=17.1 μM). Of the N-benzyl analogues (N-mono-alkylated analogues), EM902 (ED50=14.5 μM) exhibited a similar effect to EM900 (ED50=17.1 μM), whereas the p-chlorobenzyl group (for example, EM905 (ED50=2.7 μM)) is five times more potent than EM900. Interestingly, the other p-substituted benzyl analogues (for example, CF3 (EM919), F (EM921), I (EM924), Br (EM920) and OMe (EM959); see entry 4–6, 9 and 15) dramatically decreased the anti-inflammatory and/or immunomodulatory effects (ED50>100 μM) except for the p-methylbenzyl group (EM958, ED50=4.8 μM). In addition, the o- or m-chloro benzyl analogues, EM922 (ED50=1.2 μM) and EM923 (ED50=2.7 μM), also increased the anti-inflammatory and/or immunomodulatory effects as did EM905. These results suggested that the chloro group has a remarkably important role in the anti-inflammatory and/or immunomodulatory effect. The N-alkylated analogues, such as propargyl (EM929), ethyl (EM933), i-propyl (EM940) and allyl (EM957) (entry 10–13, ED50>100 μM), showed no effect, indicating that a benzyl moiety might be necessary to develop the anti-inflammatory and/or immunomodulatory effect. Likewise, conversion to the acetamide moiety (entry 16, EM960, ED50>100 μM) completely removed any anti-inflammatory and/or immunomodulatory properties. In contrast, in terms of N,N-bis-alkylated analogues, the bis-de-N-methyl analogue EM903 slightly decreased the anti-inflammatory and/or immunomodulatory effects (ED50=27.6 μM). Conversely, the N,N-dibenzyl analogue (EM904) slightly increased anti-inflammatory and/or immunomodulatory effects, with cytotoxicity at 30 μM in THP-1 cells. Interestingly, N-mono-benzyl (EM912) without N-methyl group showed no activity (ED50>100 μM, cf. entry 2). Likewise, the p-chlorobenzyl group without the N-methyl group, EM928 (ED50>100 μM), completely lost anti-inflammatory and/or immunomodulatory effects. These findings suggested that the N-methyl group has an important part in creating the anti-inflammatory and/or immunomodulatory effect. Morpholine (EM914) and piperidine (EM955) analogues expressed moderate anti-inflammatory and/or immunomodulatory effects (ED50=52.8 μM for EM914 (Appendix) and 30.3 μM for EM955), whereas, the other three types of cyclic analogues, such as EM956, EM965 and EM966, completely lost the activity (ED50>100 μM).

To demonstrate the anti-inflamamatory and/or immunomodulatory effects in vivo in a rat model of IBD and to confirm that our in vitro assay system is efficient and reliable for investigating anti-inflammatory agents, we selected two types of analogue, EM905 and EM914, as model compounds. IBD is a chronic inflammatory disease, as are ulcerative colitis and Crohn’s disease. Treatment of IBD requires anti-inflammatory drugs and steroids, yet few anti-inflammatory agents are suitable for IBD therapy. Thus, there is a pressing need for new drugs to treat these anti-inflammatory diseases, especially compounds that possess a new skeleton and/or a new mode of action.

Crohn’s disease (Sprague–Dawley) rats (Charles River Laboratories, Japan) were fasted for 24 h with access to water ad libitum. On day 0, each rat was sedated by intraperitoneal injection of sodium pentobarbital (50 mg kg−1). Subsequently, 3.125 ml kg−1 of 2,4,6-trinitrobenzene sulfonic acid (Wako Pure Chemical Industries, Tokyo, Japan), 33.3 mg ml−1 in 33% ethanol/saline, was infused using a polyethylene tube, inserted through the rectum into the colon to a distance of 8 cm. The 2,4,6-trinitrobenzene sulfonic acid was retained in the colon for 60 min, after which the fluid was withdrawn. On day 2, we confirmed whether a colonic lesion had been generated in each rat by fecal occult blood test, and colitis-induced rats were divided into five groups for treatment as follows: group 1: 10 mg kg−1 EM905 treatment (n=14); group 2: 30 mg kg−1 EM905 treatment (n=15); group 3: 10 mg kg−1 EM914 treatment (n=15); group 4: 30 mg kg−1 EM914 treatment (n=13); and vehicle control (n=15). Test compounds were prepared by suspending in 0.5% sodium carboxymethyl cellulose (CMC-Na, Wako Pure Chemical Industries) at 2 or 6 mg ml−1 and orally administered twice daily by force with a microtube from day 2 to 7. On day 8, rats were killed, the colon was excised and opened longitudinally, rinsed with cold saline and colonic damage was evaluated according to a scale ranging 0–10.16 (Scale ranging: 0, normal appearance; 1, focal hyperhemia and slight thickening (no ulcers); 2, hyperhemia and prominent thickening (no ulcers); 3, ulceration with inflammation at one site; 4, ulceration with inflammation at two or more sites; 5, tissue damage extending >1 cm length; 6–10, area of tissue damage extending >2 cm length, the score being increased by 1 for each additional cm of involvement).

The data indicated the impact on experimental colitis, 2,4,6-trinitrobenzene sulfonic acid treatment inducing severe macroscopic inflammation in the colon after rectal administration, as assessed by the colonic damage score (4.27±0.38), that is, hyperemia, thickening of the bowel and the extent of ulceration. Treatment with EM905 and EM914 reduced the severity of the gross lesion score in a dose-dependent manner as shown in Figure 1. Both EM905 and EM914 at lower dose (10 mg kg−1) had little effect (EM905: 3.29±0.22 and EM914: 2.93±0.41) without statistical significance, whereas the higher dosage (30 mg kg−1) had a significant effect on the intensity of the inflammatory response (EM905: 2.67±0.40 and EM914: 2.69±0.33). The effect was approximately equal for the two compounds. Considering their biological and chemical properties, EM905 showed greater promotion of monocyte differentiation than EM914 (Table 1). However, EM905 is considerably more hydrophobic than EM914, so it was assumed that the bioavailability of EM905 was much lower than EM914 when administered orally to the rats.

Figure 1
figure 1

Effect of EM905 and EM914 with 6 days repeated oral administration on 2,4,6-trinitrobenzene sulfonic acid-induced colitis. Score means the following grade of colitis: 0, normal appearance; 1, focal hyperhemia and slight thickening; 2, hyperhemia and prominent thickening, no ulcers; 3, ulceration with inflammation at one site; 4, ulceration with inflammation at two or more site; 5, tissue damage extending >1 cm length; 6–10, area of damage extending >2 cm length, the score being increased by 1 for each additional cm of involvement. Test solutions were administered orally at the volume of 5 ml kg−1 for 6 days. N=number, *P<0.05 and **P<0.01.

Full size image

In conclusion, we have developed two analogues, EM905 and EM914, which exhibit similar beneficial impact on treatment of IBD in rats, dramatically decreasing dosages required when compared with therapeutic sulfa drugs such as sulfasalazine. The results suggested that the promotion of monocyte to macrophage differentiation may be a reliable and accurate assay to allow in vitro evaluation of potential and promising anti-inflammatory agents. The mode of action of these compounds is not yet fully elucidated and investigation of the EM900-binding protein is underway in our laboratory.