Meridianin D Analogues Display Antibiofilm Activity against MRSA and Increase Colistin Efficacy in Gram-Negative Bacteria

Abstract

In the last 30 years, development of new classes of antibiotics has slowed, increasing the necessity for new options to treat multidrug resistant bacterial infections. Development of antibiotic adjuvants that increase the effectiveness of currently available antibiotics is a promising alternative approach to classical antibiotic development. Reports of the ability of the natural product meridianin D to modulate bacterial behavior have been rare. Herein, we describe the ability of meridianin D to inhibit biofilm formation of methicillin-resistant Staphylococcus aureus (MRSA) and to increase the potency of colistin against colistin-resistant and sensitive Gram-negative bacteria. Analogues were identified that are capable of inhibiting and dispersing MRSA biofilms and lowering the colistin MIC to below the CLSI breakpoint against Acinetobacter baumannii, Klebsiella pneumoniae, and Escherichia coli.

Antibiotic resistance is quickly becoming one of the largest threats to human health. Without the development of new strategies to defeat antibiotic resistance, it is predicted that 10 million people will die from multidrug resistant (MDR) bacterial infections by 2050.¹ Compounding the problem, resistant isolates to the two classes of antibiotics introduced most recently in the clinic, cyclic lipopeptides and oxazolidinones, were observed within five years of clinical use.² An alternative strategy to new antibiotics is to develop adjuvants that intercept the pathways responsible for resistance to clinically relevant antibiotics. Bacteria are capable of avoiding antibiotic treatment in many ways, including biofilm formation. Biofilms are highly organized surface-associated communities that are encased in an extra-cellular polymeric substance (EPS). Bacteria within a biofilm are upwards of 1000-fold more resistant to antibiotic treatment and reach a higher cell density than their planktonic counterparts, increasing the chances of horizontal gene transfer.^3,4 In many cases, bacteria also evade antibiotic treatment by acquiring resistance elements in small gene vectors. One example is the spread of the plasmid-borne mobile colistin resistance-1 (mcr-1)(5,6) gene, which likely evolved from overuse of colistin as a food additive in animal husbandry. Selective pressure has also been applied by the resurgence of colistin treatment clinically as an antibiotic of last resort in MDR Gram-negative bacterial infections.⁷ The emergence of these colistin resistant strains brings us closer to a postantibiotic world, exemplified when a MDR strain of Klebsiella pneumoniae was observed clinically in 2016 that was resistant to all clinically available antibiotics.⁸

Natural products from marine sponges have long been a rich source of molecules that display a myriad of biological activities. The meridianins are one such example of a family of structurally related marine natural products. These secondary metabolites were first reported in 1998 after being isolated from the marine invertebrate Aplidium meridianum near the South Georgia Islands.⁹ The family of meridianins and their respective derivatives have shown diverse biological activities including kinase inhibition,^10,11 adipogenesis inhibition,¹² antitumor activity,¹³ and antimalarial activity.¹⁴ Reports of antibacterial activity of these compounds, however, have been scarce and fragmentary. These reports have been limited to antimicrobial activity against Staphylococcus aureus,¹⁴Mycobacterium tuberculosis,¹⁵ and an unidentified sympatric marine Antarctic bacterium.¹⁶ Despite these limited reports, we posited that the meridianins and their analogues would possess the ability to control bacterial behavior based on the shared structural features of the meridianins with the desformylflustrabromine (dFBr),¹⁷ 2-aminopyrimidine (2-AP),¹⁸ and oroidin^19,20 analogues that were previously shown by our group to possess antibiofilm activity against methicillin-resistant S. aureus (MRSA) (2 and 3) and the ability to lower the colistin minimum inhibitory concentration (MIC) against Gram-negative bacterial pathogens carrying the mcr-1 plasmid (4) (Figure 1). Herein, we report the antibiofilm and antibiotic activity of analogues based on the meridianin scaffold against S. aureus. Moreover, we report the ability of meridianin analogues to lower the MIC of colistin against both colistin-sensitive and -resistant strains of Gram-negative bacteria.

Figure 1.

Structures of meridianin D (compound 1), desformylflustramine analogues (compound 2), 2-AP analogues (compound 3), and oroidin analogue (compound 4) previously shown to control bacterial behavior.

To begin the structure–activity relationship (SAR) study of the meridianin molecules, we synthesized meridianin D 1 to establish its biological activity. Attempts to synthesize meridianin D following the procedure described by Jiang et al.²¹ were unsuccessful as problems described by Simon et al.²² were encountered. Therefore, we applied the synthetic approach outlined by Bredereck to access meridianin D and analogues (Scheme 1A).^11,22−24 To begin, commercially available substituted indole derivatives were acylated at the 3-position using acetyl chloride and tin chloride in toluene to yield compounds 6a–l. The indolic nitrogen of compounds 6a–l was subsequently protected using p-toluenesulfonyl chloride (TsCl), triethylamine, and 4-dimethylaminopyrimidine (DMAP) in DCM. Next, the enaminone derivatives were prepared by reacting compounds 7a–l with DMF/dimethylformamide-dimethylacetal (DMF-DMA) at 110 °C for 4 h. Cyclization and deprotection of the enaminone in 2-methoxyethanol using potassium carbonate and guanidine hydrochloride or commercially available substituted guanidine derivatives yielded compounds 1 and 9a–o.

Scheme 1. Synthetic Route to Compounds 1, 9a–o (A), and 13a–n (B).

Reagents and conditions: (a) acetyl chloride, SnCl₄, toluene, 0 °C to rt, 2 h; (b) p-toluenesulfonyl chloride, triethylamine, 4-dimethylaminopyridine, DCM, rt, 16 h; (c) DMF-DMA, 110 °C, 3 h; (d) guanidine derivative, K₂CO₃, 2-methoxyethanol, reflux, 16 h; (e) MeOH/HCl; (f) dimethyl sulfate or R₂X, 50% NaOH, DCM, 16 h; (g) 1-bromo-2-methylpropane, K₂CO₃, acetone, reflux, 16 h; (h) pyrrolidine, DMA, 80 °C, 1 h; (i) DMF, 110 °C, 4 h.

Compounds were first assessed for their ability to inhibit MRSA biofilm formation. Meridianin D (1) returned an IC₅₀ value of 87.4 ± 4.0 μM (Table 1), where the IC₅₀ value is defined as the concentration at which a compound inhibits 50% of biofilm formation. This result confirmed our hypothesis that the meridianin natural products would be capable of inhibiting MRSA biofilm formation. The 4-, 5-, and 7-bromo analogues were assayed to investigate the effect that the substitution of the bromine atom had on the compound’s antibiofilm activity. The 4-bromo (compound 9k) and 7-bromo (compound 9j) analogues displayed reduced antibiofilm activity, with IC₅₀ values of >100 and 99.8 ± 15.2 μM, respectively. The 5-bromo analogue, compound 9a, displayed increased biofilm inhibitory activity with an IC₅₀ value of 17.9 ± 2.2 μM.

Table 1. IC₅₀ and MIC Values for Compounds 1, 9a–g, and 13a–f^a.

compd	R1	R2	R3	MIC (μM)	IC50 (μM)
1	6-Br	H	H	>200	87.4 ± 4.0
9a	5-Br	H	H	200	17.9 ± 2.2
9b	5-Br	Bn	H	6.25	9.62 ± 1.4
9c	5-Br	Me	Me	100	23.4 ± 1.8
9d	5-Br	Et	H	50	24.6 ± 0.7
9e	5-Br	Me	H	100	28.9 ± 2.3
9f	6-I	H	H	>200	42.5 ± 8.1
9g	5-I	H	H	200	49.3 ± 5.1
13a	6-Br	n-Pr	H	200	34.3 ± 6.8
13b	5-Br	n-Bu	H	50	38.1 ± 2.8
13c	6-Br	Me	H	200	59.6 ± 2.3
13d	6-Br	Et	H	200	64.0 ± 8.8
13e	5-Br	Et	H	200	66.2 ± 7.6
13f	5-Br	Pr	H	100	69.3 ± 3.6

^aAll values displayed against MRSA 43300. Full biofilm inhibition results can be found in the Supporting Information.

After identifying the 5-bromo and 6-bromo analogues as the most active derivatives, various substitutions on the indole ring were prepared while preserving the 2-AP ring to probe the promiscuity of the indole substitution. Of these analogues, the debromo analogue 9o, 5-methyl 9n, 5-fluoro 9l, and 6-fluoro 9m analogues all displayed no antibiofilm activity with IC₅₀ values of >100 μM. The 6-chloro analogue 8i displayed comparable activity to meridianin D, while the 5-chloro analogue 9h showed reduced activity compared to the 5-bromo analogue (8b). Finally, the 6-iodo analogue 9f displayed increased activity compared to meridianin D, with an IC₅₀ of 42.5 ± 8.1 μM, and the 5-iodo analogue 9g displayed reduced activity compared to the 5-bromo analogue (9a), with an IC₅₀ of 49.3 ± 5.1 μM.

Upon identifying compound 9a as the most active compound from this series, the effect of alkylation of the exocyclic amine of the 2-AP ring on the antibiofilm activity was investigated. Cyclization with a substituted guanidine in place of guanidine hydrochloride in the final step of the synthesis yielded methyl (compound 9e), ethyl (compound 9d), dimethyl (compound 9c), and benzyl (compound 9b) analogues. Methyl (9e) and ethyl (9d) substitutions on the 2-AP ring decreased activity, delivering compounds with IC₅₀s of 28.9 ± 2.3 and 24.6 ± 0.7 μM, respectively. These substitutions also made these analogues more toxic to planktonic bacterial growth with MICs of 100 μM for the methyl and 50 μM for the ethyl, compared to 200 μM for the unsubstituted parent. Dimethyl substitution (9c) of the 2-AP was still less active than the unsubstituted 2-AP but showed similar activity compared to the mono methyl substituted analogue, with an IC₅₀ of 23.4 ± 1.8 μM. Placement of a benzyl group at the exocyclic amine of the 2-AP ring, compound 9b, significantly increased the activity of the compound, lowering the IC₅₀ to 9.62 ± 1.4 μM. The reduction in IC₅₀ for the benzyl analogue was coupled with a significant increase in toxicity, reducing the MIC from 200 μM for compound 9a to 6.25 μM for the benzyl substituted analogue 9b. Observing a significant decrease in MIC, the antibiotic activity of compound 9b was explored against a small panel of Gram-positive pathogens. The benzylated analogue returned an MIC value of 6.25 μM (2.60 μg/mL) against two additional S. aureus isolates and an MIC of 25 μM (10.4 μg/mL) against a strain of vancomycin resistant Enterococcus faecium (VRE). No significant difference in compound MIC value was observed when compounds 9b, 9d, 9n, and 9o were tested in the presence or absence of 0.01% triton X-100 against MRSA 43300.

With compounds 1, 9a, and 9f established as leads displaying minimal inherent toxicity, the effect of alkylation of the indole nitrogen on antibiofilm activity was interrogated. Synthesis of compounds 13a–n was adapted from a previous disclosure by Simon et al. (Scheme 1B).²² Acylation of a substituted indole at the 3 position proceeded as previously reported. Compounds 1 and 6a–b were then alkylated using dimethyl sulfate or the desired alkyl halide with tetrabutylammonium bromide as a phase transfer catalyst in a biphasic mixture of dichloromethane and 50% NaOH. Alkylation with isobutyl bromide failed under these conditions and required compounds to be refluxed in acetone with isobutyl bromide and potassium carbonate to yield the desired product. Compounds 10a–k and 11a–b were then transformed into enaminones 12a–m by stirring a mixture of DMA with pyrrolidine for 1 h at 80 °C followed by the addition of a solution of the appropriate n-alkylated acetyl indole (compounds 10a–k and 11a–b) dissolved in DMF and stirring the reaction at 110 °C for 3 h. Finally, cyclization of the enaminone with guanidine hydrochloride or commercially available substituted guanidine derivatives proceeded in 2-methoxyethanol at reflux with potassium carbonate for 16 h to yield compounds 13a–n. Methylation of the indolic nitrogen of the 6-bromo derivative, compound 13c, improved the IC₅₀ value to 59.6 ± 2.3 μM (Table 1) from 87.4 ± 4.0 μM (compound 1). Ethylation of the indole nitrogen, compound 13d, displayed no improvement compared to compound 13c, but the propyl derivative, compound 13a, displayed increased activity with an IC₅₀ of 34.3 ± 6.8 μM. The butyl derivative, compound 13j, displayed no antibiofilm activity with an IC₅₀ of >100 μM. Interestingly, alkylation of the indole nitrogen of the 5-bromo analogues followed a different activity trend than the 6-bromo analogues. Substitution with a methyl group, compound 13g, and n-pentyl group, compound 13h, abolished antibiofilm activity (IC₅₀s > 100 μM). Ethyl and n-propyl derivatives, compounds 13e and 13f, respectively, showed decreased activity compared to the 5-bromo derivative with a free indolic nitrogen with IC₅₀ values of 66.2 ± 7.6 and 69.3 ± 3.6 μM, respectively. The n-butyl derivative, compound 13b, displayed an IC₅₀ of 38.1 ± 2.8 μM but was more toxic than other analogues with an MIC of 50 μM, indicating that it may be acting via a toxic mechanism to prevent biofilm formation. Branching of the alkyl chains with isobutyl substitutions on both the 5-bromo and 6-bromo analogues, compounds 13k and 13l, respectively, abrogated antibiofilm activity with both N-isobutyl derivatives both displaying IC₅₀ values of >100 μM. Next, the 6-iodo analogue, compound 9f, was alkylated off the indolic nitrogen with a propyl group to yield compound 13m. Again, it was observed that the indolic alkyl substituent and halogen substitution of the indole did not correlate to each other as compound 13m displayed no antibiofilm activity. Finally, concurrent alkylation of the indolic nitrogen and the exocyclic 2-AP nitrogen, compound 13n, abolished all antibiofilm activity.

With a panel of 2-AP analogues in hand, we turned our interests toward replacement of the 2-AP ring with a 2-aminoimidazole (2-AI). Previously, 2-AIs have shown excellent antibiofilm activity against a wide variety of pathogenic bacteria, including MRSA.²⁵⁻²⁷ The synthesis of the 2-AI analogues proceeded with the acylation of a substituted indoles 5a and 5l with chloroacetyl chloride in toluene (Scheme 2). The indolic nitrogen of compounds 14a–b then were Boc protected to yield compounds 15a–b. Cyclization of the α-chloroketone with Boc guanidine and sodium iodide in DMF yielded compounds 16a–b. Subsequent Boc deprotection using TFA in DCM at 0 °C for 16 h delivered the 2-AI derivatives 17a–b. Interestingly, both compounds 17a–b were toxic to planktonic bacterial growth at 60 μM under the conditions of the biofilm inhibition assay and displayed no antibiofilm activity below this concentration. Both compounds did display moderate antimicrobial activity returning MICs of 25 μM (7.8 μg/mL) against our test MRSA strain.

Scheme 2. Synthetic Route to Compounds 17a–b.

Reagents and conditions: (a) (i) chloroacetylchloride, toluene, 60 °C, 2 h; (ii) MeOH, H₂O, rt, 1 h; (b) Boc-anhydride, 4-dimethylaminopyridine, THF, rt, 4 h; (c) Boc-guanidine, sodium iodide, DMF, rt, 48 h; (d) 30% trifluoroacetic acid, DCM, 0 °C to rt, 16 h; (e) MeOH/HCl.

Noting that various meridianin derivatives were capable of inhibiting MRSA biofilm formation, we next investigated whether they were capable of dispersing preformed MRSA biofilms. Interestingly, the alkylated 2-AP analogues displayed the greatest activity against preformed MRSA biofilms, with the methyl (compound 9e) and ethyl (compound 9d) derivatives displaying EC₅₀ values, the concentration at which a compound disperses 50% of a preformed biofilm, of 73.1 ± 2.4 and 75.8 ± 5.8 μM, respectively (Table 2). It does not appear that the ability to disperse preformed biofilms is related to increased toxicity because the more toxic compound 9b displayed only 26.2% dispersion at 80 μM. Interestingly, dimethylation of the exocyclic amine (compound 9c) and alkyl substituents on the indolic nitrogen (compounds 13a–b) did not impart the ability to disperse preformed biofilms with all displaying EC₅₀ values of greater than 160 μM. Compound 9a returned an EC₅₀ value of 138.4 ± 15.2 μM, demonstrating its ability to inhibit and disperse MRSA biofilms.

Table 2. EC₅₀ Values with and without Vancomycin for Active Antibiofilm Analogues^a.

compd	EC50 (μM)	EC50 (μM) + vancomycin
9a	138.4 ± 15.2	92.1 ± 17.5
9e	73.1 ± 2.4	71.4 ± 6.0
9d	75.8 ± 5.8	66.5 ± 7.5
17a	101.5 ± 0.5	>100
17b	101.0 ± 3.9	>100
9b	>80	>80
9c	>160	N/A
13a	>160	N/A
13b	>160	N/A

^aAll values displayed against MRSA 43300.

After confirming the ability of multiple analogues of meridianin D to disperse preformed MRSA biofilms, the most active compounds (9a,d,e) were tested for synergy with vancomycin at a concentration at which vancomycin does not affect preformed biofilms. This concentration was determined to be 19.0 μg/mL (Supporting Information) against MRSA 43300, which is 19 times its MIC. Accordingly, each compound was tested with 19 μg/mL of vancomycin to investigate whether the compounds would show a synergistic effect with vancomycin and disperse preformed biofilms at a lower concentration. Compound 9a displayed a 33% reduction in EC₅₀ value (Table 3) when combined with a nonactive concentration of vancomycin against preformed biofilms. Other compounds tested that were capable of dispersing preformed MRSA biofilms did not display a significant reduction in EC₅₀ value when combined with vancomycin.

Table 3. Colistin Potentiation by Select Compounds against Gram-Negative Bacteria, Colistin MIC (Fold Reductions)^a.

	colistin MIC	colistin + 1	colistin + 9d	colistin + 13f
E. coli ATCC 25922mcr-1	8	4 (2)	0.5 (16)	2 (4)
E. coli ATCC 25922parent	0.5	0.5 (0)	0.0625 (8)	0.03125 (16)
A. baumannii 17978mcr-1	16	0.5 (32)	0.125 (128)	4 (4)
A. baumannii 17978parent	1	0.5 (2)	0.0625 (16)	0.0625 (16)
A. baumannii 4106	1024	16 (64)	2 (512)	16 (64)
K. pneumoniae B9	512	16 (64)	1 (512)	1 (512)
A. baumannii 5075	1	0.25 (4)	0.0625 (16)	0.0078 (128)
K. pneumoniae ATCC 2146NDM-1	1	0.5 (2)	0.0625 (16)	0.0625 (16)

^aConcentrations are shown in μg/mL.

In a recent disclosure,¹⁹ synergy was found in concomitant treatment with colistin (polymyxin E) and compound 4 (Figure 1). This combination successfully disarmed colistin resistance in multiple strains carrying the mcr-1 plasmid-borne resistance gene. To our knowledge, there is no precedent for indole-containing compounds directly modulating polymyxin defense pathways in Gram-negative pathogens. Holistically, both compound 4 and meridianin D (compound 1) are small, indole-derived compounds with an additional nitrogenous heterocyclic appendage. Given their semblance, we postulated that compound 1 and its analogues potentially had the ability to modulate colistin resistance in Gram-negative bacteria.

To this end, Acinetobacter baumannii ATCC 17978^mcr-1 was chosen as a test strain.²⁸ As summarized in Table 3, we indeed found activity with multiple meridianin D analogues demonstrating synergy with colistin in a diverse panel of bacterial isolates comprising both colistin-resistant and colistin-sensitive A. baumannii, K. pneumoniae, and Escherichia coli. Of note, across all strains, all analogues with the exception of compounds 13a, 17a, and 17b had no inherent toxicity, with MICs of >200 μM. Accordingly, all analogues were dosed at 60 μM unless otherwise noted (Supporting Information).

Cross referencing activity in these 11 strains, limited activity is seen in compounds with varied indole halogenation, while alkylation of the indolic nitrogen increased activity in colistin-sensitive isolates. Alkylation of the exocyclic amine of the 2-AP delivered compounds with enhanced synergistic activity against colistin resistant strains while retaining synergy in colistin-sensitive strains, with compound 9d being a nearly universal modulator. Compound 9d outperforms all other analogues in both A. baumannii 17978^mcr-1 and E. coli 25922^mcr-1 with 128- and 16-fold reductions in colistin MIC, respectively.

A. baumannii 3941/4106 and K. pneumoniae C3/B9 are primary clinical isolates that contain genomically encoded colistin resistance. Such strains typically possess much higher colistin MICs than their mcr-1 counterparts. All four strains return colistin MICs of 512 μg/mL or greater, well exceeding the 4 μg/mL Clinical & Laboratory Standards Institute (CLSI) breakpoint. Compounds 9b and 9d recover a breakpoint MIC in A. baumannii 3941, while compound 9d delivers a 512-fold reduction in A. baumannii 4106, reducing the MIC to 1 μg/mL. Colistin sensitivity is also re-established in K. pneumoniae C3 by compounds 9b, 9d, and 9e, all producing 512-fold reductions, and this same activity is seen in K. pneumoniae B9 with compounds 9d, 9e, and 13f. No significant difference in colistin MIC value was observed when compounds 9b, 9d, 9n, and 9o were tested in the presence or absence of 0.01% triton X-100 against A. baumannii 4106 and K. pneumoniae B9.

A select cohort of analogues was capable of increasing colistin sensitivity in strains with no inherent resistance to the antibiotic. In the parent A. baumannii 17978 strain, 9d is equipotent with 9b, 13f, 13b, and 13k, all generating a 16-fold reduction in colistin MIC. A 16-fold reduction in parent strain E. coli 25922 is observed by 13b, 13f, and 13k, while 9d shows an eight-fold reduction. A total of six compounds (compounds 9b, 9d, 9e, 13a, 13b, and 13k) produced a colistin MIC of 0.0625 μg/mL (16-fold reduction) in A. baumannii ATCC 19606, while a more impressive 128-fold reduction (0.0078 μg/mL) was achieved with compound 13f in A. baumannii 5075. Compound 13f was equipotent with compound 9d in K. pneumoniae ATCC 2146^NDM-1, returning a final MIC of 0.0625 μg/mL (16-fold reduction).

In conclusion, after identifying the potential of the natural product meridianin D (compound 1) to control MRSA biofilm formation, a panel of analogues were synthesized in an effort to augment activity. Structural modification of the meridianin D core delivered molecules that modified both Gram-positive and Gram-negative bacterial defense mechanisms. In many cases, modification of key positions of the scaffold amended bacterial behavior in divergent manners. Of note, compound 9a inhibited and dispersed MRSA biofilms as a stand-alone treatment. Furthermore, the EC₅₀ of compound 9a was reduced by 33.4% in concomitant treatment with vancomycin dosed at levels that do not perturb preformed MRSA biofilms. Alkylation of the exocyclic amine of compound 9a with ethyl (9d) or methyl (9e) groups delivered analogues with increased ability to disperse MRSA biofilms; however, they do so with a tandem increase in toxicity and diminished synergy with vancomycin. Benzylated analogue 9b was the most potent MRSA biofilm inhibitor with an IC₅₀ of 9.62 ± 1.4 μM but did not disperse preformed MRSA biofilms. Interestingly, compound 9b also returned modest antimicrobial activity against a small panel of Gram-positive bacteria. Furthermore, we established that certain meridianin analogues increase the efficacy of colistin against Gram-negative bacteria. Compound 9d, bearing an ethyl substitution on the exocyclic nitrogen of the 2-AP ring, displayed the greatest range of colistin synergy across a panel of colistin-sensitive and colistin-resistant strains of Gram-negative bacteria, which included highly resistant primary clinical isolates of chromosomally encoded resistance, strains of E. coli and A. baumannii harboring the mcr-1 gene, and a strain of K. pneumoniae carrying the New Dehli metallo-beta-lactamase-1 (NDM-1) gene. Compounds 13a, 13b, 13f, and 13k, which bear an alkylation on the indolic nitrogen, displayed equivalent or greater potentiation of colistin activity against some strains but did not display the broad-spectrum potentiation of colistin activity of compound 9d. Mechanistic studies and additional structural modifications in an effort to increase activity of the analogues are currently underway.

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsmedchemlett.8b00161.

• Compound characterization for all novel compounds, ¹H and ¹³C NMR spectra, biofilm, colistin repotentiation, and full in vitro testing data (PDF)

Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

The authors thank the National Institutes of Health (GM055769) for funding.

The authors declare no competing financial interest.