New liposidomycin congeners produced by Streptomyces sp. TMPU-20A065, anti-Mycobacterium avium complex agents with therapeutic efficacy in a silkworm infection model

Three new liposidomycin congeners (1, 2, and 4), together with 14 known liposidomycins (3 and 5–17), were isolated from the culture broth of Streptomyces sp. TMPU-20A065 as anti-Mycobacterium avium complex agents. The structures of liposidomycins were elucidated by spectroscopic analyses, including NMR and MS. Compounds 1, 2, and 4 belong to type-I liposidomycin-containing sulfate groups and methylglutaric acid, each with a different acyl side chain in the structure. Compounds 1–17 exhibited in vitro anti-M. avium and M. intracellulare activities with MIC values ranging between 2.0 and 64 μg ml−1. Furthermore, 1–17 exerted potent therapeutic effects in an in vivo-mimic silkworm infection model with ED50 values ranging between 0.12 and 3.7 μg larva−1 g−1.


Introduction
Mycobacterium avium complex (MAC) infection, which is mainly caused by M. avium and M. intracellulare, is a nontuberculous mycobacterial pulmonary and intractable disorder, and the number of infected patients is increasing more than that of tuberculosis in developed countries [1,2].Symptoms are similar to those of tuberculosis, including weight loss, fever, fatigue, and night sweats [1], with slow progression and a less favorable prognosis.The first-line drug to treat MAC infections, clarithromycin (CAM), combines rifampicin and ethambutol.However, its therapeutic effect is insufficient and long-term administration for more than one year may be required, resulting in the emergence of drug-resistant bacteria, which is a problem [3].Although amikacin liposome inhalation suspension was newly approved as a treatment for MAC infection in 2018 [4], its application is limited to patients for whom conventional combination therapies are not sufficiently effective.Therefore, there remains an urgent need to develop new candidates for the treatment of MAC infections with novel skeletal structures and different mechanisms of action from those of existing drugs.To resolve this issue, we developed an in vivo-mimic silkworm infection model with MAC based on our experience [5][6][7][8][9][10][11][12][13] and used it in the first screening of anti-MAC agents from microbial resources.We identified potent anti-MAC compounds with therapeutic effects, three new liposidomycin congeners (1, 2, and 4) and 14 known liposidomycins (3 and 5-17), in a culture broth of Streptomyces sp.TMPU-20A065.The fermentation, isolation, structural elucidation, and in vitro and in vivo antimycobacterial activities of liposidomycins were reported herein.

General experimental procedures
UV spectra were recorded on a spectrophotometer (U-3310 UV-Visible spectrophotometer; Hitachi High-Technologies, Tokyo, Japan).IR spectra were recorded on a Fourier transform infrared spectrometer (FT/IR-4100; JASCO Co.,

Isolation
Anti-MAC activity-guided isolation was performed step by step, leading to the isolation of 1-17 from the culture broth of Streptomyces sp.

In vitro broth microdilution method
The broth microdilution method was performed according to a previously established method [12,14].Mycobacterial strains were grown for 2 to 7 days at 37 °C in Middlebrook 7H9 broth (Middlebrook 7H9 broth 1.04%, Tween 80 0.05%, bovine serum albumin 0.5%, glucose 0.2%, and NaCl 0.085%) up to approximately 1.0 × 10 9 CFU ml −1 .Culture suspensions were then diluted 500 times with the same fresh broth.The suspension (95 μl) was added to each well of a 96-well microplate with or without test samples (5 μl in MeOH).The microplate was incubated at 37 °C for 2 to 7 days.Turbidity was assessed by measuring absorbance at 550 nm with an absorption spectrometer.The MIC value was defined as the lowest concentration of the test compounds at which bacterial growth was inhibited by 90% of control growth (no compound).

In vivo-mimic silkworm infection model
The silkworm infection model was performed according to a previously established method [13].Fertilized silkworm eggs of Bombyx mori (Hu•Yo × Tukuba•Ne) were purchased from Ehime Sansyu (Ehime, Japan) and fed an artificial diet (Silk Mate 2 S; Nihon Nosan Kogyo, Kanagawa, Japan, and Silkmate; Katakura Industries, Tokyo, Japan) until the fourth-instar larval stage.Hatched silkworm larvae were raised by feeding an artificial diet containing antibiotics (Silk Mate 2 S, Nihon Nosan Kogyo, Kanagawa, Japan) in an incubator at 27 °C until the fourth molting stage.On the first day of the fifth-instar larval stage, silkworms were fed an antibiotic-free artificial diet (Silk Mate, Katakura Industries, Tokyo, Japan) until they weighed 2 g.On the second day, the M. avium or M. intracellulare suspension (2.5 × 10 7 CFU larva −1 g −1 in 50 µl Middlebrook 7H9 broth) was injected into the hemolymph of silkworm larvae (2.0 g, n = 5) using a disposable 1-mL syringe with a 27-G needle (TERUMO, Tokyo, Japan), followed by an injection of test samples (50 µl in saline or 10% DMSO) within 30 minutes.Infected silkworms were raised without feed at 37 °C, and their survival rate was measured for 96 hours after the sample injection.ED 50 values were defined as the amount of a sample required for a 50% survival rate normalized per 1 g of silkworm.

Assay for antibacterial activity
The broth microdilution method was performed according to the guidelines of CLSI document M07-A09 [15].Bacterial strains were grown overnight at 37 °C in Mueller-Hinton broth (Becton Dickinson, San Jose, CA, USA).Cultures were diluted with the same broth and adjusted to an optical density of 0.0548 at 550 nm (approximately 10 8 CFU ml −1 ).Culture suspensions were then diluted 3000 times with the same fresh broth.The suspension (95 μl) was added to each well of a 96-well microplate with or without test samples (5 μl in MeOH).The microplate was incubated at 37 °C for 24 hours.Turbidity was assessed by measuring absorbance at 550 nm with an absorption spectrometer.The MIC value was defined as the lowest concentration of the test compounds at which bacterial growth was inhibited by 90% of control growth (no compound).
Compound 1: Its molecular formula was elucidated as C 40 H 63 N 5 O 21 S based on HR-FAB-MS measurements (m/z 982.3837 [M + H] + , Δ+2.2 mmu).The 1 H and 13 C NMR spectra of 1 in CD 3 OD (Table 2) were similar to those of liposidomycin C-I (9).In comparisons with the 1 H and 13 C NMR spectra of 1 and those of 9, 1 appeared to possess the same basic structures as 9 containing 5'-substituted uridine, 5-amino-5-deoxyribose-2-sulfate, perhydro-1, 4-diazepine, and 3-methylglutaric acid.Moreover, the molecular formula of 1 was smaller than that of 9 (C 42 H 67 N 5 O 21 S) by C 2 H 4 (ethylene unit), suggesting that 1 possessed a shortened acyl side chain of 9 in a partial structure I. Furthermore, 1 H-1 H COSY correlations revealed the presence of a decanoic acid moiety in the partial structure I (Fig. 2).Therefore, the structure of 1 was confirmed by 2D NMR experiments, as shown in Fig. 2a, which fulfilled the molecular formula and degrees of unsaturation.
Compound 4: Its molecular formula was elucidated as C 42 H 65 N 5 O 21 S based on HR-FAB-MS measurements (m/z 1008.3962[M + H] + , Δ-0.9 mmu), indicating that 3 was bigger than 1 by C 2 H 2 .In comparisons with the 1 H and 13 C NMR spectra (CD 3 OD) of 4 and those of 1 (Table 2), two sp 2 methine signals, C-7a (δ C 128.6, δ H 5.31) and C-8a (δ C 130.3, δ H 5.36), were newly observed in 4, suggesting that 4 possessed a C 2 unit long monounsaturated fatty acid moiety.Moreover, a 1 H-1 H COSY analysis revealed the presence of 7-tetradecenoic acid in a partial structure I (Fig. 2b).Regarding the cis-trans configurations of the double bond, the 1 H-1 H coupling constants between H-7a and H-8a (J H7a-H8a = 11.4Hz) were elucidated by 1 H-1 H decoupling experiments, and appeared to be the cis form (Fig. 2c).The structure of 4 was confirmed by 2D NMR experiments, as shown in Fig. 2c, which fulfilled the molecular formula and degrees of unsaturation.The stereochemistries of 1, 2, and 4 were elucidated by ROESY experiments (Fig. S3, S9, S15, and S21) and 1 H-1 H coupling constants (Table 2), and the results obtained showed that the relative stereochemistries of 1, 2, and 4 were in good agreement with those of liposidomycin B-I (8) in the literature [17].In addition, the optical rotations of 1, 2, and 4 have the same value as that of 8 ([α] D 24 (c 0.4, H 2 O) = 17.3°) in the literature [16].

In vitro antimicrobial activity using the microdilution method
The MIC values of 1-17 against eight test microorganisms in the microdilution method are listed in Table 3. Compounds 1-17 exhibited anti-MAC activity against M. avium and M. intracellulare, with MIC values ranging between 2.0 and 64 μg ml −1 .Moreover, 5-7, 9-12, 15, and 17 exhibited weak antimycobacterial activity against M. smegmatis and/ or M. bovis BCG, with MIC values ranging between 16 and 64 μg ml −1 .On the other hand, 1-17 did not exhibit antibacterial activity, even at 64 µg ml −1 , against B. subtilis, S. aureus, E. coli, or P. aeruginosa.

In vivo-mimic anti-MAC activity using the silkworm infection model
Compounds 1-17, except for 15, were evaluated in the silkworm infection model with M. avium and M. intracellulare (n = 5), and their ED 50 values are summarized in Table 3.As an example, the therapeutic effects of 1 against M. avium and M. intracellulare are shown in Fig. 3a and b, respectively.All infected silkworms without a compound (control) died within 72 hours.When 1-17 were administered to silkworms infected with M. avium, potent therapeutic effects were confirmed in a dose-dependent manner with ED 50 values ranging between 0.12 and 2.7 μg larva −1 g −1 .Similarly, 1-17 exerted potent therapeutic effects in the silkworm infection model with M. intracellulare, with ED 50 values ranging between 0.44 and 3.7 μg larva −1 g −1 .Incidentally, 1-17 alone did not exhibit any toxicity towards silkworms at a dose of 32 µg larva −1 g −1 for 96 hours (data not shown).

Discussion
In the present study, three new liposidomycins congeners (1, 2, and 4), together with 14 known liposidomycins (3 and 5-17), were isolated from Streptomyces sp.TMPU-20A065 as selective anti-MAC compounds with therapeutic effects in the silkworm infection model.The in vivo-mimic silkworm infection model has several advantages as an alternative mammalian model, such as similar therapeutic effects and pharmacokinetics of antibiotics with the mouse model  and no ethical issues [20,21].Moreover, this model evaluates the therapeutic efficacy of test samples in a few days.Therefore, we have been adapting this model in the early stages of various screenings for antibiotics of microbial origin [5][6][7][8][9][10][11][12][13], which has resulted in the identification of new liposidomycin congeners in the present study.
Liposidomycins are classified as types I-IV based on the presence or absence of a 3-methylglutaric acid and sulfate group [18].Consequently, 1, 2, 4-6, 8, 9, 13, 14, and 16 belong to type I liposidomycins with both 3-methylglutaric acid and a sulfate group, while 3, 7, 10-12, 15, and 17 belong to type III liposidomycins with 3-methylglutaric acid and no sulfate group; therefore, new congeners 1, 2, and 4 are type-I liposidomycins with different acyl side chains in the partial structure I (Fig. 2).However, the production of types II and IV liposidomycins without 3-methylglutaric acid was not detected in the culture broth of Streptomyces sp.TMPU-20A065.
In 1985, liposidomycins were isolated from Streptomyces griseosporeus RK-1061 as uridyl liponucleoside antibiotics exhibiting antibacterial activity, including against M. phlei [16], and were found to inhibit phospho-Nacetyl muramyl pentapeptide transferase (MraY) in the peptidoglycan synthesis of E. coli [22].Caprazamycin B, which is also a uridyl liponucleoside acid antibiotic, was isolated from Streptomyces sp. in 2003 and exhibited potent antimycobacterial activity, including against multidrugresistant M. tuberculosis, through the inhibition of MraY without significant toxicity in mice [23].Furthermore, CPZEN-45 developed by structure-activity relationship studies on caprazamycin B [24] exhibited selective growth inhibitory activity against M. tuberculosis by inhibiting the phospho-N-acetylglucosaminyltransferase WecA [25], which plays a role in mycolyl-arabinogalactan biosynthesis.Therefore, CPZEN-45 is currently being developed as a drug candidate for anti-TB drugs with a new mechanism of action.Caprazamycin B and CPZEN-45 both exhibit in vitro anti-MAC activity [23,24,26]; however, their in vivo efficacy has not yet been clarified.
We herein demonstrated the in vivo efficacy of liposidomycins in the silkworm infection model with M. avium and M. intracellulare.The ED 50 values of each liposidomycin in the silkworm infection model were lower than their corresponding MIC values in the microdilution method.For example, the MIC value of 1 against M. avium was 64 µg ml −1 , whereas the ED 50 value was 0.26 µg larva −1 g −1 , indicating that in vivo activity was 250-fold stronger than in vitro activity.Several compounds are reported to show ED 50 values lower than the MIC values due to enhancement of the in vivo activity by host factors [27,28] or inhibition of antimicrobial activity by serum  albumin or other substances in the culture medium [29].Therefore, similar causes for the activity of liposidomycins are also possible and need to be clarified in the future.The MAC therapeutic agent, CAM, was previously reported to only be effective at high doses (50-200 mg g −1 ) in the mouse infection model [30], with similar results being obtained in the silkworm model (Table 3).The ED 50 values of CAM in the silkworm model with M. avium and M. intracellulare were 23 and 42 µg larva −1 g −1 , respectively, while the ED 50 values of 1-17 were markedly lower than those of CAM.Further studies are warranted to obtain more information on their in vivo activity in mice.Moreover, 1-17 exhibited selective anti-MAC activity without antimicrobial activity against Gram-positive bacteria.These experimental data show the potential of liposidomycins as candidate anti-MAC agents.However, the interpretation of this selective activity cannot be resolved by MraY inhibitory activity alone.Therefore, the possibility of other factors, such as targets other than MraY or selective transporters, is suggested.
In conclusion, we used an in vivo mimetic silkworm infection model to screen for anti-MAC agents and discovered liposidomycins with therapeutic effects.Although the target molecule of liposidomycins was previously reported to be MraY, it remains unclear whether this prominent in vivo anti-MAC activity is due to the direct inhibition of MraY or another mechanism of action.Therefore, future research on the anti-MAC activity of not only liposidomycins, but also other MraY inhibitors is needed to clarify this issue.

Table 2 (
MHz) and 1 H (600 MHz) spectra were taken on the JNM-ECZ600R/S1 (JEOL) in methanol-d 4 , and the solvent peaks were used as internal standards at 3.31 ppm for 1 H NMR and at 49.0 ppm for13C NMR a Signals are interchangeable within the same letter