New Structure Activity Relationship Insight into the Role of the C‐3 Extension on Rifamycin Antimycobacterial Activity

Abstract

Herein, the antimycobacterial screening of a series of rifamycin analogues, modified at their C‐3 extension, is reported. Overall, these compounds display potent activity against a wild‐type Mtb strain assayed in three different growth media. Several promising C‐3 extensions are identified through this screen, with compounds featuring rigid tertiary alicyclic hydrazones displaying superior activity to amino compounds. In addition, a general correlative trend between logP and biological activity is observed. This study adds to the growing literature surrounding structure activity relationship pertaining the important C‐3 extension of rifamycin, which in addition to a poorly understood role in target engagement, has utility for modulating physicochemical properties, a key condition in antimycobacterial drug discovery.

Rifamycin analogues with variably functionalized C‐3 extensions provide new insight into the role of this region for MTb inhibitory activity. In addition to potentially optimizing target engagement, this region can also be used to modulate physicochemical properties, an important consideration for optimizing antimycobacterial activity. This study provides structural cues for future development of rifamycin congeners for Tb drug discovery.

1. Introduction

The archetypal semisynthetic rifamycin derivative, rifampicin (1) has formed the cornerstone of Mycobacterium tuberculosis (Mtb) treatment regimens for several decades.^{[ 1 ]} Its dramatic impact on Mtb therapeutic outcomes, coupled to the unique sterilizing activity of the rifamycins has seen several studies exploit this important pharmacophore to identify new rifamycin derivatives.^{[ 2} , ^{3 – 4 ]} Structure activity relationship (SAR) studies, which involve the systematic evaluation of the impact of structural modifications on biological activity,^{[ 5 ]} have to date culminated in the approval of three additional semisynthetic rifamycin's analogues including rifapentine (2) and rifabutin (3), whose structural alterations at C3 and C4, retain the rifamycin's underlying antimycobacterial potency whilst modulating pharmacokinetic properties.^{[ 6 ]} The availability of additional antimycobacterial rifamycin analogues has proven useful for adjusting treatment regimens to optimize dosing and mitigate adverse drug reactions.^{[ 7 ]} For example, rifampicin‐induced severe cutaneous adverse reactions have been observed in certain individuals coinfected with TB and HIV. Importantly, patients who develop this drug hypersensitivity seemingly tolerate rifabutin, allowing for rifamycin‐centered Mtb chemotherapy to continue.^{[ 8 ]} It is therefore important to continue growing the array of biologically active rifamycin analogues.

Mechanistically, rifamycins exert their antimicrobial effect by binding to bacterial RNA polymerase (RNAP) in the DNA/RNA channel, distal to the RNAP catalytic site, and distinct from the binding site of RNAP inhibitory Fidaxomicin.^{[ 9 ]} This interaction blocks elongation of the nascent RNA chain, thus halting RNA transcription and downstream protein translation.^{[ 10 , 11 ]} Rifamycins share a core polyketide‐derived ansa chain, bridging either a naphthoquinone or hydro naphthoquinone core.^{[ 12 ]} Modifications to the core scaffold are challenging and generally have negative consequences for biological activity. This phenomenon is most pronounced for the three hydroxyl groups found at C‐8, C‐21, and C‐23, which form key hydrogen bonds with the S447 (O‐8) and H442 (O‐21) and F430 (O‐23) amino acid residues found in its RNAP binding site (Figure  1 , residue numbering varies between organisms), where modification is not well tolerated.^{[ 11 , 13 , 14 ]} The critical importance of these residues is also reflected in the problematic S447L and H442Y, RNAP mutations, which in addition to the D432V/Y mutation, account for the majority rifampicin inducing resistance in Mtb.^{[ 15 ]}

Figure 1.

X‐ray cocrystal structure of rifampicin bound to M. Smegmatis RNAP (PDB 6CCV, resolution 3.05 Å). Oxygen containing residues found at C‐8 (S447), C‐21 (H442), and C‐23 (F430) on the rifamycin core form key hydrogen bonds with the highlighted phenylalanine, histidine, and serine residues. A. Simplified view of rifampicin binding, showing the C‐3 extension orientating into an open pocket (white surface), and despite influencing activity does not interact directly with any amino acid residues. B. Detailed description of rifampicin binding including additional interactions with E368 sigma and Q426.

Successful modifications to the rifamycin core have been reported at C‐11 and C‐25,^{[ 16 , 17 ]} and more recently Peek et al. reported the identification of several naturally occurring rifamycin congeners, the kanglemycins (e.g., Kang V1, 4), modified at positions C‐20 and C‐27.^{[ 18 ]} The kanglemycins, which lacked the characteristic C‐3 and C‐4 extensions found on compounds 1–3, were between 2 and 25 fold less active than rifampicin against wild‐type MTb, however, the unusual additions to the rifamycin scaffold which facilitated the formation of additional electrostatic interactions with RNAP ameliorated biological activity against rifampicin‐resistant Mtb featuring either a S447L or D432Y RNAP point mutation. It was notable that the degree of resistance mitigation was inconsistent among the kanglemycins.^{[ 18 ]}

X‐ray‐cocrystal data for rifamycin analogues in complex with either Escherichia. coli ^{[ 15 ]} suggest that the C‐3/C‐4 extensions orientate toward an open pocket of the RNAP active site, without making any direct electrostatic interactions with binding site amino acids. The x‐ray‐cocrystal of rifampicin in complex with Msm RNAP point to a similar pattern, with the exception of a cocrystallized SO₄ ²⁻ anion in close proximity to the rifampicin piperazine (Figure 1).^{[ 18 ]} Despite this, the majority of rifamycin modification studies, which have focused on synthetic extensions from the C‐3 and C‐4 region, have successfully seen improvements in antimicrobial performance.^{[ 15 , 19} , ²⁰ , ²¹ , ²² , ^{23 – 24 ]} This phenomenon was also observed upon modification to the kanglemycins, including partial mitigation of S447L and H442Y resistance.^{[ 7 ]} Furthermore, enzymatic modification of the C‐3 extension of rifampicin has evolved as a secondary resistance factor in Nocardia farcinica.^{[ 25 ]} It was earlier demonstrated that the pattern of substitution on the naphthalenoid core of rifamycins at the C‐3/C‐4 position and the nature of the substituent attached to the core (neutral or basic) played an important role in the observed antibacterial activity profile.^{[ 26 ]} Here, the zwitterionization process was found to be dependent on the basicity of C‐3 substituent and the environment (protic or aprotic solvents) and is an important additional SAR consideration.^{[ 26 , 27 ]} Therefore, understanding the influence of variation of C‐3/C‐4 extensions on rifamycin biological activity remains important in antimicrobial drug discovery. In addition to influencing target engagement, this region also offers a useful handle to modulate the physicochemical properties of rifamycin analogues. This includes zwitterionizability and lipophilicity, which in the context of antimycobacterial drug discovery, and the purported role of passive diffusion in accessing the mycobacterial intracellular environment, is particularly relevant.^{[ 28 ]}

A recent study reported the synthesis of a series of rifamycin congeners, modified either with an aliphatic amine (Figure  2 ) or hydrazone (Figure  3 ), extending from the C‐3 position via either an sp ³ or sp ² hybridized carbon atom attached at C‐3, respectively. At that time, the library was evaluated against various strains of Staphylococcus aureus, S. epidermidis, Enterococci faecalis, and E. coli. ^{[ 24 ]} In general, increased rigidity in the C‐3 arm was required for activity against Staphylococcus species. While activity against Enterococci was limited, more lipophilic compounds showed moderate activity against E. faecalis. ^{[ 24 ]} However, despite the important role rifamycins play in the fight against tuberculosis, these compounds were not assessed for their antimycobacterial activity. Accordingly, in this study, we report the rescreening of this cohort of rifamycin analogues (5–30)^{[ 24 ]} against drug‐sensitive and drug‐resistant Mtb culture with the aim of further developing our understanding of the SAR requirements of the important C‐3 extension.

Figure 2.

Aliphatic amine rifamycin congeners investigated in this study. Overall, the flexible amine‐containing compounds were less active than the hydrazone analogues.

Figure 3.

Aliphatic hydrazone rifamycin congeners investigated in this study. The rigid hydrazone combined with an alicyclic tertiary amine was beneficial for Mtb inhibitory activity.

2. Results and Discussion

Compounds 5–30 were subjected to a modified microplate Alamar blue assay (MABA)^{[ 29 ]} against the drug‐sensitive H37RvMA Mtb strain. The influence of growth media, on the biological activity of compounds in Mtb assay is a well‐known phenomenon. This is especially pronounced by the inclusion of the dispersal agent Tween 80, whose presence promotes the removal of layers of the Mtb cell envelope, thus increasing cellular permeability and with it mycobacterial sensitivity to certain classes of drugs.^{[ 30 , 31 ]} Furthermore, Tween 80 and to a lesser extent Tyloxapol modify the metabolome of Mtb growth.^{[ 32 , 33 ]} Accordingly, we conducted three separate biological assessments in three different Middlebrook 7H9 growth media, either in the presence of Tween 80 (7H9ADCGLUTW), or where Tween 80 was substituted with Tyloxapol (7H9ADCGLUTX or 7H9CASGLUTX). The entire library displayed good antimycobacterial with MIC₉₀ values below 3 μM in all three media and displayed generally superior antimycobacterial activity to isoniazid (Inh) and moxifloxacin (Mox), and activity comparable to rifampicin (Rif). In addition, several compounds displayed MIC₉₀ activity below the experimental cutoff of 0.01 μM. In these instances, biological activity was reported as 0.01 μM, and since all the replicates were reported at the same value, standard deviations were not determinable. Variation in biological activity of up to two orders of magnitude between growth media was observed for a number of compounds. Given this large variation, we opted to visualize these data on a logarithmic scale as pMIC₉₀ with a value of 8 representing the best activity in this screen (Figure  4 ). Here, we considered a pMIC₉₀ > 7 to be the threshold of noteworthy biological activity.

Figure 4.

pMIC₉₀ data of compounds 5–30 alongside isoniazid, moxifloxacin, and rifampicin against wild‐type M. tuberculosis (H37RvMA) in three different growth media. The presence of Tween 80 had an overall positive impact on biological activity. Greater variation was observed in the absence of Tween 80 and showed that hydrophobic hydrazones (23–26) are important structural features biological activity. Values are the mean ±SD of technical duplicates.

The assays conducted in 7H9ADCGLUTW, resulted in the majority of compounds exceeding the “noteworthy” activity threshold, with a few exceptions including compounds 8, 22, and 30. MIC₉₀ determinations conducted in the absence of Tween 80 showed a more varied activity landscape, including a comparative reduction in activity for rifampicin. The activity variation was particularly pronounced in compounds 6, 7, 10, 11, 13, 17, 18, and 19, whose MTb inhibitory activity fell by at least one order of magnitude, when assayed in Tween 80 free media. With the exception of compounds 5 and 16, none of the aliphatic amine analogues (5–21) displayed pMIC₉₀ values greater than 7. In general, compounds possessing more flexible secondary amines on the C3 substituent (e.g., 6–9) were moderately less active than those featuring ring‐constrained tertiary amines as either piperazines (e.g., 13–16) or piperidines (e.g., 20, 21). In Tween 80 free media, the presence of a piperazine substituted with either a heteroaromatic pyridine (15) or pyrimidine (16) was seemingly beneficial for activity, while larger benzylic substituents (17 and 19) had a negative effect on activity. The comparatively moderate activity of compound 12 which is the amino analogue of rifampicin suggested that an amine and the accompanying rotatable C—N bond represented an activity cliff for antimycobacterial activity. This result is consistent with the earlier studies where compound 12 was markedly less active (MICs > 0.25 µg mL^–1) than rifampicin (MICs 0.031 µg mL^–1)in tests with other Gram‐positive bacterial strains.^{[ 24 ]} This structural modification contributes to differences in direction, polarity, and strength of intramolecular H‐bonds for 12 and 1 (_lactamN–H^…N vs _lactamC=O…H–N⁺; Figure  5 ). This variation in the patterns of rifamycin intramolecular hydrogen bonding in rifamycins impacts the arrangement of the ansa bridge relative to the naphthalenoid core, as well as the conformation of the C‐3 extension, which together can impact RNAP engagement.

Figure 5.

Zwitterionic structures of rifampicin 1 (left) and derivative 12 (right), in protic solvents. These data indicate that basic flexible amines can substantially alter the conformation of the ansa bridge and C‐3 extension, thus likely impacting its interaction with RNA. calculated using the Scigress software package.^{[ 36 ]}

Compounds 23–26 and rifampicin, which feature a cyclic tertiary hydrazone displayed substantial antimycobacterial activity in all three growth media. While the phenyl hydrazone compound (28) showed some encouraging activity in all three media, in general, the presence of a secondary hydrazone (22, 27, 29, and 30) resulted in a substantial reduction in bioactivity, which was also a trend observed amongst the amino rifamycin's. The azepane containing compound 25 was the only compound to exceed the experimental threshold of the screen in all three media, displaying overall superior activity to rifampicin and was considered the most potent compound in this library, while the slightly smaller piperidine (24) and morpholine (26) analogues were also encouraging. As alluded to, Tween 80, has been known to increase the sensitivity of mycobacteria to certain classes of drugs. However, in this study, the direct impact that the growth media on the compound stability was not determined and cannot be entirely ruled out as contributing factor to the observed intermedia variability.

Lipophilicity is commonly cited as an important parameter for passive diffusion though the complex mycobacterial cell wall, and a conspiring factor which underpins the poor biological performance of many potential antimycobacterials. The variation in cell permeably is often a confounding factor in the interpretation of mycobacterial biological data. As mentioned, the data obtained in the Tween 80 media showed generally higher activity, while a more varied SAR was observed in Tween‐free media. Accordingly, we investigated the potential relationship between lipophilicity with biological activity obtained in the 7H9ADCGLUTX media. Plotting pMIC₉₀ vs experimentally determined logP showed a general correlative relationship between the two parameters (Figure  6 ). This is likely due to improved cell penetration and partially explains the potent activity displayed by compounds 23–26. However, what was more informative were compounds whose activity and logP values did not correlate which pointed toward C‐3 extensions that possibly enhanced (1) or disrupted (18, 19, 22, and 29) target engagement.

Figure 6.

Plot of pMIC₉₀ (7H9ADCGLUTX) versus logP shows a general correlative trend, between lipophilicity and biological activity. Compounds with high lipophilicity and poor activity likely point toward C‐3 extensions which are less favorable for target engagement.

Based on these data, we selected a small subset of compounds for assessment against rifampicin and fidaxomicin resistant Mtb (Table  1 ). While biological activity was maintained against the fidaxomicin‐resistant strain, this did not extend to the rifampicin resistant strain featuring the S447L RNAP mutant, where significant shifts in activity were observed, thus indicating that these C‐3 modifications were unable to overcome the S447L mutation. This result is understandable since the wild‐type S447 makes a crucial interaction with the naphthalenoid core. Given the lack of activity against the rifampicin resistant strain, we did not reassess these compounds in Tween 80 free media.

Table 1.

Biological evaluation of selected analogues against rifamycin and fidaxomicin resistant Mtb.

Compound name/number	MIC90 [µM] wild‐type	MIC90 [µM] Rifampicin resistanta)	MIC90 [µM] Fidaxomicin resistant
5	<0.01	>62.5	<0.01
23	<0.01	>62.5	<0.01
25	<0.01	62.5	<0.01
16	<0.01	62.5	<0.01
26	<0.01	62.5	<0.01
24	<0.01	62.5	<0.01
Moxifloxacin	0.19	0.19	0.19
Isoniazid	3.9	3.9	1.9
Fidaxomicin	0.039	0.078	>6.25b)
Rifampicin	0.002	>0.15b)	0.002

Assays conducted in 7H9ADCGLUTW media.

^a) S450L RNAP mutation.

^b) Rifampicin and fidaxomicin activity not measured beyond the indicated thresholds.

3. Conclusion

Herein, we report the biological evaluation of a series of previously synthesized hydrophobic rifamycin congeners for their Mtb inhibitory activity. Here, several compounds displayed potent biological activity against drug‐sensitive MTb. Several important insights were gleaned, including that activity is most greatly enhanced when the C‐3 extension comprises of a rigid hydrazone functionality attached to a tertiary alicyclic amine (Figure  7 ). In addition, we found a strong correlation between biological activity and lipophilicity, although several compounds did not follow this trend indicating that these C‐3 extensions were detrimental to activity beyond their ability to modulate physicochemical properties. With the exception of compounds 5 and 16, compounds featuring flexible amine containing C‐3 extensions, were generally less active, while the presence of aliphatic chains, (8) noncyclic (22) or large benzylic N‐substituents (17, 19, 29) had a negative effect on activity. In silico calculations suggest that this phenomenon is at least partially due to zwitterion induced alterations to ansa bridge conformation. By contrast, compounds whose C‐3 extension contained a more rigid alicyclic tertiary hydrazone, displayed potent antimycobacterial activity. Compounds 23–26 in particular, were potent antimycobacterials, with activity superior to rifampicin in Tween 80 free media. While activity was maintained against a fidaxomicin‐resistant strain, it did not extend to rifampicin resistance strains. Given the importance of the rifamycins as front‐line Mtb chemotherapeutics, expanded understanding of rifamycin SAR such as that demonstrated in this study is important in the continued fight against Mtb. However, the findings in this study also suggest that modification of the C‐3/C‐4 extension in isolation may not be sufficient for overcoming the challenge of Mtb resistance. Accordingly, the development of clinically relevant rifamycin analogues requires a multifaceted approach. This may include incorporating recent advances in Mtb chaperone inhibition, which partially mitigates rifampicin mutations.^{[ 34 ]} Furthermore, the structural features present in compounds 5, 16, 23–26, which had an overall positive impact on Mtb inhibitory activity, may prove useful in augmenting the antimycobacterial activity of rifamycin congeners which circumvent RNAP resistance such as the kanglemycins.

Figure 7.

Pictural summary of the key SAR finding from this study. Compounds whose C‐3 extension possessed a rigid hydrazone functionality attached to a tertiary cyclic aliphatic amine generally displayed the most encouraging activity. In these instances, activity was further enhanced by increased lipophilicity. By contrast, C‐3 extensions which were conformationally flexible and did not possess a tertiary alicyclic amine, tended to be less active.

4. Experimental Section

The synthesis, purification, characterization, and logP determination for all compounds can be found in following references.^{[ 21 , 24 ]} Compounds featuring in this study were all synthesized in the same batch as previously described and stored under deep freeze. Prior to analysis all compounds were assessed for purity, which ranged from 94–95%

4.1.

4.1.1.

Bioassay

10 mM stock samples of compounds 5–30 were prepared in 100% DMSO. Stocks were diluted twofold in the requisite growth media in 96‐well round bottom plates on the day of the experiment. A concentration range of 6.25 μM—0.012 μM was used for each sample. Isoniazid (Inh) and moxifloxacin (Mox) and rifampicin (Rif) were included as control at concentration ranges of 62.5–0.122, 6.25–0.012 , and 6.25–0.012 μM, respectively. A maximum inhibition control (Rif at 0.15 μM) and a minimum inhibition control (0.625% DMSO) were included on each test plate. A culture of M. tuberculosis H37RvMA,^{[ 35 ]} in 7H9ADCGLUTW, 7H9ADCGLUTX, and 7H9CASGLUTX was grown to an optical density (OD600) of 0.5–0.7, and a 500‐fold dilution was prepared in each of the three freshly prepared media. A volume of 50 μL of the diluted culture (≈1 X 104 bacilli) was added to each well of each test plate, for a final volume of 100 μL per well. Each compound was tested in duplicate. The assay plates were incubated at 37 °C with 5% CO₂ and humidification for 7 days. Cell pellets were visually scored and then alamar blue reagent added to each well of the assay plate and re‐incubated for 24 hr. The relative fluorescence units (RFU) (excitation 540 nm; emission 590 nm) of each well were measured using a SpecraMax i3x Plate reader on day 8 (Serial no. 36,370 3271, Molecular Devices Corporation 1311 Orleans Drive Sunnyvale, California). Data analysis was performed using GraphPad Prism‐10, Version 10.0.3. Briefly, raw RFU data were normalized to the minimum and maximum inhibition controls to generate a dose response curve (% inhibition), using the Levenberg–Marquardt damped least‐squares method, from which the MIC is calculated using the 4‐parameter curve fit protocol. The lowest concentration of drug that inhibits 90% of growth of the bacterial population is the MIC.

Assays involving rifampicin and fidaxomicin‐resistant strains followed the same protocol, with the exception that the biological assessment of compounds was conducted over a concentration range of 62.5–0.122 μM. rifampicin, fidaxomicin, moxifloxacin, and isoniazid were included as control drugs for the relevant strains at concentration ranges of 0.15–0.0002 , 0.625–0.0012 , 6.25–0.012, and 62.5–0.122 μM, respectively. A minimum growth control (Isoniazid at 20 μM) and a maximum growth control (0.625% DMSO) were included on each test plate.

Media Composition

7H9_ADC_GLU_TW: Middlebrook 7H9 media (Difco) supplemented with 0.2% Glucose, Middlebrook albumin‐dextrose‐catalase (ADC) enrichment (Difco), and 0.05% Tween 80.

7H9_ADC_GLU_TX: Middlebrook 7H9 media (Difco) supplemented with 0.2% Glucose, Middlebrook albumin‐dextrose‐catalase (ADC) enrichment (Difco), and 0.05% Tyloxapol.

7H9_CAS_GLU_TX: Middlebrook 7H9 media (Difco) supplemented with 0.4% Glucose, 0.03% Casitone (Gibco Bacto), and 0.05% Tyloxapol.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supplementary Material

Veale Clinton G. L., Smolarz Ewelina, Leśniewska Aleksandra, Pyta Krystian, Przybylski Piotr, ChemMedChem 2025, 0, e202500673. 10.1002/cmdc.202500673

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.