Activity of RX-04 Pyrrolocytosine Protein Synthesis Inhibitors against Multidrug-Resistant Gram-Negative Bacteria

Pyrrolocytosines RX-04A to -D are designed to bind to the bacterial 50S ribosomal subunit differently from currently used antibiotics. The four analogs had broad anti-Gram-negative activity: RX-04A—the most active analog—inhibited 94.7% of clinical Enterobacteriaceae, Acinetobacter baumannii, and Pseudomonas aeruginosa at 0.5 to 4 μg/ml, with no MICs of >8 μg/ml.

ABSTRACT

Pyrrolocytosines RX-04A to -D are designed to bind to the bacterial 50S ribosomal subunit differently from currently used antibiotics. The four analogs had broad anti-Gram-negative activity: RX-04A—the most active analog—inhibited 94.7% of clinical Enterobacteriaceae, Acinetobacter baumannii, and Pseudomonas aeruginosa at 0.5 to 4 μg/ml, with no MICs of >8 μg/ml. MICs for multidrug-resistant (MDR) carbapenemase producers were up to 2-fold higher than those for control strains; values were highest for one Serratia isolate with porin and efflux lesions. mcr-1 did not affect MICs.

TEXT

One approach in the search for new antibacterial agents is to model the target interactions of natural antibiotics that are unsuitable for pharmaceutical development, due to toxicity or instability, and to use this information to design synthetic molecules that achieve similar binding without the unfavorable traits of the original compounds.

Melinta Pharmaceuticals has applied this strategy to blasticidin S, a natural product of Streptomyces griseochromogenes that inhibits both eukaryotic and prokaryotic ribosomes but which has proved useful only as a fungicide, deployed to control rice blast disease in Japan (1). Modeling of the ribosomal interactions of blasticidin (2), TAB-1057A/B (3), and amecitin (4)—which have overlapping targets that are distinct from those of clinically used bacterial protein synthesis inhibitors—has led to several new antibacterial scaffolds, including pyrrolocytosines (5, 6). These are chemically unrelated to blasticidin, but mimic its principal interactions with the bacterial 50S subunit (6). In vitro antibacterial activity indicates that the pyrrolocytosines penetrate bacterial cells, and further development has sought to optimize this penetration for Gram-negative bacteria while reducing vulnerability to efflux (5). Chemical properties of the pyrrolocytosine derivatives, along with synthetic methods, are outlined in the relevant patents (7–9).

We evaluated four pyrrolocytosine derivatives, RX-04A to -D (Fig. 1), against a panel of 96 Gram-negative clinical isolates, biased to overrepresent carbapenemase producers, Enterobacteriaceae with mcr-1, and Pseudomonas aeruginosa with upregulated efflux. We additionally tested Escherichia coli HB10B and its transformant, carrying plasmid p594, which encodes expression of mcr-1 (10). The mcr-1 and carbapenemase genes were detected by PCR or sequencing (10, 11), while efflux levels in P. aeruginosa isolates were inferred by interpretive reading of antibiogram data, which predict mechanisms from phenotypes (12). MICs of the four RX-04 analogs and comparators (amikacin, cefepime, colistin, meropenem, and tigecycline) were determined by CLSI broth microdilution (13) using preprepared plates (Trek Diagnostic Systems, Thermo Fisher, Oakwood, OH). DNA from four Serratia isolates differing in susceptibility to the pyrrolocytosines was extracted using a QIAsymphony automated instrument. Sequencing libraries were prepared using the Nextera XT DNA library preparation kit and sequenced on the Illumina HiSeq 2500 system using the 2 × 100-bp paired-end mode. Genomes were assembled de novo with VelvetOptimiser 2.1.9 software (http://bioinformatics.net.au/software.velvetoptimiser.shtml) and then compared with each other to seek genetic modifications that were specific to the Serratia isolate with the highest pyrrolocytosine MICs, particularly in genes encoding porins, efflux pumps, and the rRNA targets of these antimicrobial agents.

FIG 1.

RX-04 pyrrolocytosine structures.

MICs by species, irrespective of resistance mechanism, are shown in Table 1, while Table 2 shows geometric mean MICs for major resistance types represented in the test panels. Nonsusceptibility rates to comparators for the Enterobacteriaceae isolates (n = 66) at CLSI breakpoints were as follows: amikacin, 14%; cefepime, 50%; colistin, 33% (2 μg/ml EUCAST breakpoint); meropenem, 47%; and tigecycline, 15% (1-μg/ml EUCAST breakpoint); those for the same agents against the A. baumannii isolates (n = 10) were as follows: amikacin, 40%; cefepime, 50%; colistin, 0%; meropenem, 50%; and tigecycline, 50%, respectively. Nonsusceptibility rates for the P. aeruginosa isolates (n = 20) were as follows: amikacin, 15%; cefepime, 45%; colistin, 25%; and meropenem, 45%.

TABLE 1.

Pyrrolocytosine MIC distributions by species, irrespective of resistance mechanism

Analog and speciesa	No. of isolates with MIC (μg/ml) of:
	0.25	0.5	1	2	4	8	16	>16
RX-04A
E. coli	1	8	14
S. enterica			11
K. pneumoniae		2	14	4
E. cloacae		1	5	2
Serratia spp.			1	2		1
P. aeruginosa		1	4	4	10	1
A. baumannii			3	4	1	2
All	1	12	52	16	11	4
RX-04B
E. coli	1	6	15	1
S. enterica			10	1
K. pneumoniae		1	14	5
E. cloacae			5	3
Serratia spp.			1		2		1
P. aeruginosa		1	3	4	7	2	2	1
A. baumannii			2	4	3	1
All	1	8	50	18	12	3	3	1
RX-04C
E. coli	1		12	10
S. enterica				11
K. pneumoniae		1	8	6	5
E. cloacae			1	6	1
Serratia spp.				1	2			1
P. aeruginosa		1		4	3	3	6	3
A. baumannii			3	1	2	4
All	1	2	24	39	13	7	6	4
RX-04D
E. coli		1	2	18	2
S. enterica				11
K. pneumoniae			2	11	5	2
E. cloacae			1		5	2
Serratia spp.				1	2			1
P. aeruginosa				4		6	7	3
A. baumannii				2	1	3	4
All		1	5	47	15	13	11	4

^aThe species included are Escherichia coli, Salmonella enterica, Klebsiella pneumoniae, Enterobacter cloacae, Serratia spp., Pseudomonas aeruginosa, and Acinetobacter baumannii.

TABLE 2.

Geometric mean MICs for different resistance groups

Resistance group by species (n)	Geometric mean MIC (μg/ml)
	RX-04A	RX-04B	RX-04C	RX-04D
E. coli
Wild type (5)	0.5	0.6	0.9	1.3
Carbapenemase (15)a	0.8	0.9	1.3	2.1
E. coli/Salmonella
mcr-1 (14)b	1.0	1.1	2.0	2.0
K. pneumoniae
Wild type (5)	1.0	1.0	1.0	2.0
Carbapenemase (15)a	1.1	1.2	2.0	2.8
E. cloacae
Wild type (4)	1.0	1.2	1.7	3.4
Carbapenemase (4)c	1.2	1.4	2.4	4.8
Serratia spp.
Wild type (2)	1, 2d	1, 4d	2, 4d	2, 4d
Carbapenemase (2)e	2, 8d	4, 16d	4, >16d	4, 16d
P. aeruginosa
Low efflux (5)	1.5	1.7	3.5	5.3
Normal efflux/wild type (5)	2.6	3.0	7.0	11.3
High efflux (5)	2.6	3.0	7.0	6.1
Carbapenemase (5)f	3.5	6.7	5.7	12.7
A. baumannii
Wild type (5)	1.7	1.7	2.0	4.6
OXA-23 positive (5)	3.0	3.5	5.3	12.1

^aFive isolates each with KPC, NDM, and OXA-48-like enzymes.

^bEleven S. enterica and 3 E. coli isolates.

^cTwo isolates with KPC enzymes and single strains with OXA-48 and NDM.

^dSingle isolates with SME and OXA-48-like enzymes.

^eSince only two isolates were tested, actual MICs are shown, not the mean.

^fTwo isolates with VIM, two with NDM carbapenemases, and one with an IMP enzyme.

Despite this heavy loading with isolates resistant to established agents, MIC distributions of RX-04A to -D were all unimodal and tightly clustered. MICs were lowest for RX-04A, where 94.7% of values for all species pooled lay between 0.5 and 4 μg/ml, with no values greater than 8 μg/ml. MICs were highest for analogs RX-04C and RX-04D, particularly for P. aeruginosa. Irrespective of the analog, the general pattern was for MICs to be lowest for E. coli, slightly higher for other Enterobacteriaceae, particularly Serratia spp., and highest for P. aeruginosa.

MICs for a single Serratia marcescens isolate, which also had OXA-48 carbapenemase, were raised markedly, at 8, 16, >16, and >16 μg/ml for molecules RX-04A, -B, -C, and -D, respectively, compared with 1 to 2, 1 to 4, 2 to 4, and 2 to 4 μg/ml, respectively, for the remaining three Serratia isolates tested. Comparison of the four sequenced genomes revealed the high-MIC Serratia isolate to have both (i) a premature stop codon (Tyr211) in omp2, which is an ompC/F homolog, and (ii) multiple unique changes (compared with all three low-MIC Serratia isolates) in the sdeCDE operon, encoding an RND pump system (14), specifically, Asn407Ser, Ser432Asn, Glu433Ala, Ala437Thr, Ala438Asn, Asn439Lys, Ala440Thr, Glu443Gln, and ArgR448Gly in sdeC, Glu111Asp and Thr363Met in sdeD, and Glu240Asp in sdeE. None of these changes was observed in the three low-MIC Serratia genomes. No lesions specific to the high-MIC isolate were found (i) in other recognized porin genes (omp1 and omp3), (ii) in porin regulatory genes (ompR and envZ), (iii) in efflux pump genes (smdAB, sdeXY, smfY, and ssmE), or (iv) in genes encoding the 16S or 23S rRNA targets of the RX-04A-D molecules. Inactivation of omp2 seems likely to reduce pyrrolocytosine uptake, and the sdeCDE lesions may increase efflux, explaining the phenotype of the high-MIC Serratia isolate. These uptake and efflux lesions also are congruent with an observed meropenem MIC of 32 μg/ml, which is unusually high for an Enterobacteriaceae strain with an OXA-48 β-lactamase.

Geometric mean MICs of the four analogs for carbapenemase-producing Enterobacteriaceae were slightly above those for the susceptible control strains, although the differentials never exceeded 1 doubling dilution (Table 2). These small rises again probably reflected widespread reductions in permeability or upregulations in efflux among the carbapenemase-producing Enterobacteriaceae. The MIC differential for carbapenemase-producing versus nonproducing A. baumannii was larger, exceeding 2-fold for analogs RX-04B to -D, although not for RX-04A; however, the numbers were small, and 3/5 OXA-23-producing isolates belonged to the same lineage (international clone II [15]), raising the possibility that the mean was skewed by overrepresentation of this lineage.

The effect of mcr-1 was of interest because the pyrrolocytosines are polybasic (Fig. 1), raising the hypothetical concern that MCR-1-mediated substitution of lipopolysaccharides (LPSs) with positively charged phosphoethanolamine (16) might impede their initial interaction with the cell surface, reducing uptake. MICs of the RX analogs for the mcr-1-positive isolates were around 1 doubling dilution above those for control strains. However, most (11/14) of these isolates were Salmonella enterica, being compared with E. coli controls, and the differential may reflect species rather than mechanism. Crucially, transformation of E. coli DH10B with the mcr-1-carrying plasmid p594 had no effect on MICs of RX-04A, -B, -C, and -D, which remained at 0.25, 0.5, 0.5, and 1 μg/ml, respectively, whereas the MIC of colistin was raised from 0.25 to 4 μg/ml. A caveat is that we do not know the extent of LPS modification achieved by p594-mediated carriage of mcr-1 nor the mode of expression, meaning that we cannot definitively exclude the possibility that induction by the pyrrolocytosines was weaker than by colistin. This seems unlikely, though: if LPS substitution with positively charged alcohols and sugars compromised the pyrrolocytosines, then generalized resistance would be expected in colistin-resistant genera such as Serratia, and this was not seen.

In the case of P. aeruginosa, geometric mean MICs of all analogs were ca. 1.5-fold higher for the isolates with “normal” versus low efflux, but did not rise further for those with elevated efflux-mediated resistance to β-lactams and fluoroquinolones (Table 2).

In conclusion, these data indicate that the four pyrrolocytosine molecules have broad activity against Enterobacteriaceae and nonfermenters, with RX-04A the most active analog. Near-full activity was retained against isolates with resistance mechanisms of current concern, including against carbapenemase producers, those with mcr-1-mediated colistin resistance, and (perhaps most surprisingly) P. aeruginosa with upregulated efflux. A caveat is that the strain panel was small, and we cannot exclude the possibility that resistance might arise from novel or unsuspected mechanisms only detectable with a larger panel. Notably, raised MICs were seen for one Serratia isolate with inactivated omp2 and upregulated sdeCDE efflux, suggesting that combinations of impermeability and upregulated efflux can compromise activity, at least against this species.

Given this spectrum, the new target, and demonstrable activity in experimental infections (17), the pyrrolocytosine class warrants further evaluation with a view to possible clinical development.