rpoB Mutations are Associated with Variable Levels of Rifampin and Rifabutin Resistance in Mycobacterium tuberculosis

Introduction

Tuberculosis (TB) continues to be a major global public health threat causing 6.4 million new cases and 1.6 million deaths in 2021.¹ The alarming increase in drug-resistant, especially rifampicin-resistant TB (RR-TB), represents the major challenge for TB control. RR-TB requires prolonged use of a combination of drugs that are less efficient and more toxic. In the last year, there was a worldwide estimate of 450,000 incident cases of RR-TB.¹

Rifampin (RIF) is one of the principal drugs in TB treatment and serves as a hallmark for the detection of multidrug-resistant TB (MDR-TB).² Rifabutin (RFB) is a semisynthetic derivative of rifamycin S, which together with RIF, belongs to the rifamycin family. These two drugs target the DNA-dependent RNA polymerase β-subunit, encoded by the rpoB gene. RFB is recommended for TB treatment in HIV-coinfected patients because it is less prone to drug–drug interactions than RIF in patients receiving antiretroviral therapy.^3,4 However, reports on the successful treatment of MDR-TB patients with RFB, even in patients with isolates susceptible in vitro to RFB, are very limited.^5,6 One possible reason is due to the fact that the clinical efficacy of RFB for the treatment of MDR-TB has not been well established because of general concerns about potential cross-resistance among the rifamycins.^7,8

Certain mutations in the RIF resistance-determining region (RRDR) in rpoB gene, located between codons 426 and 452 in M. tuberculosis, are associated with cross-resistance to both RIF and RFB.^2,9–11 The mutations, S450L, H445Y, and H445D are most frequently observed in RIF- and RFB-resistant strains.^11–13 Although cross-resistance to RIF and RFB is common, some RIF-resistant strains carrying certain rpoB mutations may remain susceptibility in vitro to RFB.^14,15 Thus, RFB has the potential to be clinically effective against RIF-resistant strains.

Some reports have shown RFB susceptibility in RIF-resistant strains with specific rpoB mutations.^16–18 These studies also pointed out the possible uses for RFB in these cases with rpoB polymorphisms and their association with phenotypic drug susceptibility data.^16,17 However, most studies focused on the analysis of mutations in the RRDR region of rpoB.^11,17,18 Furthermore, several rpoB mutations were also detected in rifamycin-susceptible phenotypes. Thus, a detailed investigation of the different rpoB mutations and their association with different rifamycin resistance levels may be helpful to clinicians treating RIF-resistant TB, even multi-drug resistant TB (MDR-TB). To this end, we systematically performed quantitative RIF and RFP resistance phenotyping with MIC measurements for 177 M. tuberculosis clinical isolates from China and compared the effect of mutation within the whole rpoB gene on MIC changes for these two drugs.

Materials and Methods

M. tuberculosis Isolates

Overall 177 M. tuberculosis clinical strains, isolated from 177 epidemiologically unrelated patients with pulmonary tuberculosis (126 males; age range, 16–80 years; mean ± standard deviation, 42.2 ± 16.6), were collected in nine provinces of China, including Fujian (15 isolates), Guangxi (21 isolates), Guizhou (19 isolates), Hunan (25 isolates), Gansu (14 isolates), Jilin (26 isolates), Inner Mongolia (23 isolates), Xinjiang (16 isolates) and Tibet (18 isolates). All isolates were cultured on Lowenstein-Jensen (L-J) medium and freshly subcultured before being used for MIC testing.

MIC Testing

MICs were determined by in vitro 96-well microplate-based assay as described previously.¹⁹ RIF concentrations prepared in Middlebrook 7H9 were 0.125, 0.25, 0.5, 1, 2, 4, 8, 16, 64, 128, and 256 µg/mL, and RFB concentrations tested were 0.125, 0.25, 0.5, 1, 2, 4, 8, and 16 µg/mL. The critical concentration was taken as 0.5 µg/mL for RIF and 0.5 µg/mL for RFB.^20,21

Quality Control for MIC Testing

M. tuberculosis (ATCC 27294) was used as a quality control and was tested with each batch of MIC testing. This quality control strain is susceptible to both rifamycin drugs with MICs ≤0.125 µg/mL²¹ in the present study.

DNA Extraction, Amplification, and Sequencing

All isolates on the L-J slants were collected and inactivated by heating at 95°C for 20 min. Supernatants containing genomic DNA were collected by centrifugation and stored at −20°C for further use.

The entire rpoB gene of each isolate was amplified using the primers and reaction conditions as previously described.¹⁹ All amplified products were purified, dried, and loaded onto an ABI 3730XL DNA Analyzer (Applied Biosystems, Foster City, CA). The sequences generated were compared with the H37Rv reference genome (GenBank accession number NC_000962) using BioEdit v7.05.3.

Data Analysis

The effects of rpoB mutations on rifamycin drugs were evaluated by the regression multivariate model and P value less than 0.05 was considered statistically significant. All statistical analyses were performed using SAS version 9.3 (SAS Institute, Cary, USA).

Results

MIC Results

Of the 177 tested isolates, 106 (59.9%) were resistant to RIF and RFB, 21 (11.9%) were RIF-resistant and RFB-susceptible, and 50 (28.2%) were susceptible to RIF and RFB. All the RIF-resistant isolates were categorized into the low-level group (MIC 1–64 µg/mL) and the high-level group (MIC ≥128 µg/mL). Among 127 RIF-resistant isolates, 38 (29.9%) and 89 (70.1%) had low and high levels of RIF MIC, respectively. Accordingly, MIC ranges for RFB resistance were categorized as high (MIC ≥8 µg/mL) and low (MIC 1–4 µg/mL) level groups, which contained 63 (59.4%) and isolates 43 (40.6%), respectively.

Mutations Within rpoB

DNA sequencing showed that 120 isolates harbored at least one non-synonymous mutation within the sequenced rpoB gene (Table 1), and 96 isolates (80.0%) had a single mutation, while 24 isolates (20.0%) had multiple mutations. It was notable that there were two isolates carrying both missense mutation and deletion mutation (Table 1). For all mutations, 38 genotype patterns distributed across 19 different sites were detected, including 28 polymorphisms in the RRDR and 10 polymorphisms outside the RRDR. The most frequent mutations were observed at codons 450, 445, and 435, which had mutated frequencies of 46.7% (56/120 isolates), 27.5% (33/120 isolates), and 17.5% (21/120 isolates), respectively. Other mutations were also observed at codons 45, 170, 400, 429, 430, 431, 432, 441, 446, 452, 460, 488, 491, 759, and 1056, which had a total mutated frequency of only 28.3% (34/120 isolates). Of these 34 isolates, 21 (61.8%, 21/34 isolates) were combined with additional mutations at 435, 445, or 450. It was notable that a single mutation V170F, located outside RRDR, occurred in one RIF and RFB-resistant isolate. Two novel deletions in rpoB, D435del (nucleotide 1303–1305 deletion) and F433-D435del (nucleotide 1296–1304 deletion) were also identified.

Table 1 Distribution of rpoB Mutations and MICs of RIF and RFB

Correlation of rpoB Mutations with Drug Resistance

Most of the rpoB mutations (single or double) occurred in RIF- and RFB-resistant isolates. Only one RIF-susceptible isolate (MIC = 0.5 µg/mL) and 19 RFB-susceptible isolates (MIC ≤ 0.5 µg/mL) bearing the rpoB mutation were detected (Table 1).

As 20.0% of isolates with rpoB mutations harbored more than one mutation, the correlation between mutations and drug resistance was estimated through multivariate regression (Table 2). In the multivariate analysis, resistance to both RIF and RFB was predominantly correlated with S450L, H445D, H445Y, and H445R (P<0.05), with the OR values of 226.69, 45.99, 37.39, and 15.89 for RIF, and 425.87, 86.16, 70.06 and 17.98 for RFB, respectively (Table 2). Notably, mutations D435V (OR = 37.39) and L452P (OR = 11.59) were associated with resistance to only RIF (P<0.05), but not RFB, while mutations Q432K (OR = 13.75), Q432L (OR = 13.75), and L430P (OR = 15.99) were associated with resistance to only RFB (P<0.05), but not RIF. Furthermore, other mutations, except for those locating in the RRDR, were not associated with drug resistance according to multivariate analyses. In addition, there were still 8 RIF-resistant isolates and 5 RFB-resistant isolates that harbored no mutation in the whole rpoB.

Table 2 Logistic Regression Multivariate Model Results Between rpoB Mutations and Drug Resistance

Association Between rpoB Mutations and High-Level MIC

Isolates with higher MICs were more likely to harbor the S450L, H445D, and H445Y mutations (Table 1). Among the isolates harboring these three mutations, the proportions of high-level MIC isolates were 90.7% (49/54 isolates), 100.0% (11/11 isolates), and 88.9% (8/9 isolates) for RIF, and 64.8 (35/54 isolates), 54.5% (6/11 isolates) and 77.8% (7/9 isolates), and for RFB, respectively. The multivariate model also showed that these three mutations were most significantly correlated with high-level MICs (P<0.01), with the OR values of 50.39, 75.53, and 40.80 for RIF, and 17.35, 11.12, and 31.43 for RFB, respectively (Table 3). In addition, the H445R mutation (OR = 26.09) was significantly associated with high-level RIF MIC (P<0.01), but not high-level RFB MIC. However, mutations Q432K (OR = 20.46) and Q432L (OR = 20.46) were significantly associated with high-level RFB MIC (P<0.05), but not high-level RIF MIC.

Table 3 Logistic Regression Multivariate Model Results Between rpoB Mutations and High-Level MIC

There were 29 high-level RIF MIC isolates with rpoB mutations belonging to the low-level RFB MIC group. However, the number of high-level RFB isolates with low-level RIF MIC was only six, among which four isolates had RIF MICs of 64 µg/mL, very close to the MIC definition of high-level RIF MIC (128 µg/mL). Interestingly, the V170F mutation was exclusively presented in two high-level RFB MIC but low-level RIF MIC isolates.

Discussion

Rapid and accurate determination of drug resistance in M. tuberculosis is essential for early initiation of appropriate anti-TB treatment. Although several molecular diagnostic techniques, such as GeneXpert MTB/RIF (Cepheid Inc., Sunnyvale, CA, USA) and GenoType MTBDRplus (HAIN Lifescience GmbH, Nehren, Germany), provide the rapid identification of M. tuberculosis drug resistance directly from specimens,^22–24 these molecular methods may miss isolates with mutations outside the target region, and cannot determine the level of resistance. Furthermore, previous reports indicated that some isolates with rpoB mutations are RIF-susceptible according to phenotypic drug susceptibility testing.^10,11,25 Similarly, RFB-susceptible isolates with rpoB mutations have also been reported,^{11,14,15,18,26} highlighting the importance of the potential inclusion of RFB in the treatment regimen. For these reasons, it is crucial to understand the association of rpoB mutations with different RIF and RFB resistance levels.

Although cross-resistance to RIF and RFB is common, we still observed discordance between RIF and RFB in 21 isolates. Previous studies showed a proportion of RIF-resistant isolates with RFB-susceptibility ranging from 13% to 28%.^11,17 In our study, the proportion of RIF-resistant/RFB-susceptible isolates was 16.5% (21/127 isolates). The replacement of RIF by RFB might positively affect the treatment outcome in the cases carrying these isolates.

It is reported that the RNA polymerase β-subunit encoded by rpoB is a target for rifamycin drugs, and some amino acid changes in this protein can lead to rifamycin resistance.^27–30 Our study showed that 93.7% (119/127 isolates) of RIF-resistant isolates and 95.3% (101/106 isolates) of RFB-resistant isolates harbored at least one mutation in rpoB. Overall, 19 distinct codons in rpoB had mutations contributing to amino acid replacements, of which 11 codons are located in the RRDR and 8 are located outside the RRDR. Interestingly, almost all mutations outside the RRDR were accompanied by mutations in the RRDR. These results show the potential for using DNA sequencing of RRDR to rapidly predict RIF and RFB resistance.^2,10 However, a single mutation V170F outside the RRDR was also observed in one resistant isolate, similar to some reports.^20,31 This highlights the complexity of RIF and RFB resistance and emphasizes the requirement to investigate beyond the RRDR, especially when patient treatment effects are not as expected.

Our study also showed the mutations associated with RIF and RFB. Statistical analysis indicated that, of the 38 different rpoB polymorphisms, S450L, H445D, H445Y, and H445R were significantly correlated with RIF and RFB resistance. Miotto et al³² and Whitfield et al¹⁷ also classified these four mutations as RIF-resistant and RFB-resistant. Mutations D435V and L452P were significantly associated with only RIF resistance, but not RFB resistance, consistent with the prior data.^17,32 Cases carrying the isolates with these two mutations may be RFB-susceptible and benefit from the RFB-containing treatment regimen. In addition, our results indicated that Q432K, Q432L and L430P were significant with only RFB-resistance. Although mutation L430P was classified as RFB-susceptible by Whitfield et al,¹⁷ this mutation was observed in four RFB-resistant isolates. Of them, three isolates were accompanied by another rpoB mutation (Table 1). Interestingly, one isolate harboring double mutations (L430P/H445N) classified as RFB-susceptible by Whitfield,¹⁷ had a MIC of ≥16 µg/mL. Thus, the actual effects of these double rpoB mutations on the RFB resistance deserve further investigation.

We also found that mutations 450L, H445D, and H445Y were strongly associated with high-level RIF/RFB MIC. This observation is similar to those of previous reports.^11–13,33 Moreover, mutation H445R was well correlated with high-level RIF MIC, while mutations Q432K and Q432L were associated with high-level RFB MIC. Our results, therefore, add important data on the association between mutations and rifamycin-resistant levels in TB. Notably, V170F mutation occurred exclusively in two high-level RFB MIC but low-level RIF MIC isolates. The function of this mutation on the interaction between RpoB and RFB is unknown and requires further exploration.

This study used the WHO current recommended clinical breakpoints (0.5 µg/mL) for susceptibility testing of RIF.²⁰ However, we still identified one RIF-susceptible isolate (MIC = 0.5 µg/mL) with the D435Y mutation, which has also been reported in some RIF-susceptible isolates.^14,15,31 It was notable that 19 RFB-susceptible isolates also harbored rpoB mutations, with the use of CLSI recommended clinical breakpoints (0.5 µg/mL) for susceptibility testing of RFB.²¹ Of them, 9 had MICs of 0.5 µg/mL. Although most mutations presenting in these isolates were classified as RFB-susceptible, some reports also suggested that the critical testing concentration, currently recommended at 0.5 µg/mL, may be too high.^15,16 Moreover, some RIF/RFB resistant isolates lacking rpoB mutations were identified, indicating that resistance in these isolates may be attributed to other mechanisms, such as enhanced efflux pump activity or lowered cell wall permeability.^11,34

This study also suffers some limitations. In total, large amounts of clinical isolates (n = 177) were recruited in this study. However, the number of isolates carrying mutations outside RRDR, such as T400I, G759S, and Q1056K, was very small, leading to insufficient data for further statistical analysis. Also, we only analyzed mutations in the whole rpoB gene, and changes in other genes or other mechanisms such as efflux pumps could not be identified in the present study.

In summary, we determined comprehensive profiles of rpoB mutations and their associations with RIF and RFB resistance levels in a variety of M. tuberculosis clinical isolates. Revealing the SNP-specific differential rifamycin phenotypic resistance will have the potential for using RFB to optimize the treatment of some RIF-resistant patients. However, clinical investigations are required to verify whether RFB can be applied effectively in these patients to promote outcomes.