Abstract

   [48]graphic file with name id4c00138_0007.jpg

   Tuberculosis (TB), caused by Mycobacterium tuberculosis (Mtb), is the
   leading cause of death worldwide by infectious disease. Treatment of
   Mtb infection requires a six-month course of multiple antibiotics, an
   extremely challenging regimen necessitated by Mtb’s ability to form
   drug-tolerant persister cells. Mtb persister formation is dependent on
   the trehalose catalytic shift, a stress-responsive metabolic remodeling
   mechanism in which the disaccharide trehalose is liberated from cell
   surface glycolipids and repurposed as an internal carbon source to meet
   energy and redox demands. Here, using a biofilm-persister model,
   metabolomics, and cryo-electron microscopy (EM), we found that
   azidodeoxy- and aminodeoxy-d-trehalose analogues block the Mtb
   trehalose catalytic shift through inhibition of trehalose synthase TreS
   (Rv0126), which catalyzes the isomerization of trehalose to maltose.
   Out of a focused eight-member compound panel constructed by
   chemoenzymatic synthesis, the natural product 2-trehalosamine exhibited
   the highest potency and significantly potentiated first- and
   second-line TB drugs in broth culture and macrophage infection assays.
   We also report the first structure of TreS bound to a substrate
   analogue inhibitor, obtained via cryo-EM, which revealed conformational
   changes likely essential for catalysis and inhibitor binding that can
   potentially be exploited for future therapeutic development. Our
   results demonstrate that inhibition of the trehalose catalytic shift is
   a viable strategy to target Mtb persisters and advance trehalose
   analogues as tools and potential adjunctive therapeutics for
   investigating and targeting mycobacterial persistence.
     __________________________________________________________________

   Mycobacterium tuberculosis (Mtb), which causes tuberculosis (TB), was
   responsible for 1.6 million deaths in 2021.^[49]1 TB endures as a
   leading cause of mortality in part due to the remarkable difficulty of
   treating Mtb infections. Conventional treatment of drug-susceptible Mtb
   infections requires the use of a multidrug cocktail administered for a
   minimum duration of six months, which increases the risk of deleterious
   side effects and patient nonadherence, and ultimately contributes to
   the emergence of multidrug-resistant (MDR) TB.^[50]2−[51]4 This lengthy
   and intensive drug treatment regimen is necessitated by—and its
   efficacy is limited by—the ability of Mtb to form persister populations
   of cells, which exhibit extraordinary tolerance to TB
   drugs.^[52]5,[53]6 Persisters form as a means for Mtb to respond and
   adapt to various stresses encountered during infection, thus promoting
   bacterial survival within the assaulting environment of the host. Given
   the urgent need to improve TB treatment, there has been increasing
   interest in understanding and targeting the molecular mechanisms that
   contribute to mycobacterial persistence.^[54]6−[55]8 Advancements in
   these areas may lead to adjunctive therapeutics that shorten TB
   treatment durations and help to curb acquired drug resistance.

   Using metabolomics in conjunction with an in vitro biofilm model of Mtb
   persistence, we recently discovered that Mtb persisters remodel
   trehalose metabolism to drive both transient drug tolerance and
   permanent drug resistance, an adaptive response mechanism termed the
   trehalose catalytic shift.^[56]10 Trehalose is a nonmammalian
   disaccharide consisting of two glucose residues linked by a
   1,1-α,α-glycosidic bond.^[57]11 Trehalose has long been known to play
   unique and critical roles in mycobacteria, namely, its involvement in
   the biosynthesis of essential immunomodulatory glycolipids, including
   trehalose monomycolate (TMM) and trehalose dimycolate (TDM), which are
   components of the mycobacterial outer membrane, or
   mycomembrane.^[58]12,[59]13 Actively replicating Mtb cells exhibit high
   glycolipid biosynthesis activity, whereas Mtb persister cells adapt to
   nutrient depletion by downregulating glycolipid synthesis, breaking
   down glycolipids to release free trehalose, which is recycled through
   central carbon metabolism (CCM) to support essential functions,
   including adenosine triphosphate (ATP) and nicotinamide adenine
   dinucleotide phosphate (NADPH) production ([60]Figure [61]1
   A).^[62]10,[63]14−[64]17 The trehalose catalytic shift depends on the
   Mtb trehalose synthase TreS (Rv0126),^[65]18 which catalyzes the
   isomerization of trehalose to maltose, thus serving as a link between
   trehalose metabolism and CCM ([66]Figure [67]1B). Although TreS is not
   strictly essential for Mtb viability under favorable growth conditions,
   it is required for the formation of biofilm persisters and can
   therefore, in principle, be conditionally targeted as a persistence
   factor. On the basis that our prior work had identified several
   monosubstituted trehalose analogues as selective inhibitors of biofilm
   formation in the nonpathogenic model organism Mycobacterium smegmatis
   (Msmeg),^[68]19 we then investigated the influence of
   6-azido-6-deoxy-d-trehalose (6-TreAz) on Mtb persister formation. Our
   initial studies revealed that 6-TreAz interrupted the trehalose
   catalytic shift through inhibition of TreS and that 6-TreAz
   significantly sensitized Mtb to treatment with bedaquiline (BDQ), a
   drug that was recently approved by the FDA to treat MDR-TB.^[69]10
   Thus, small-molecule TreS inhibitors, including trehalose analogues,
   are promising tools for interrogating the trehalose catalytic shift and
   represent lead compounds for the development of adjunctive therapeutics
   for TB.

Figure 1.

   Figure 1
   [70]Open in a new tab

   (A) Overview of the TreS-mediated trehalose (Tre) catalytic shift in
   Mtb. Under nutrient-replete conditions (left, green), actively
   replicating Mtb uses exogenously acquired nutrients to drive the
   synthesis of trehalose via OtsAB, and the subsequent synthesis of cell
   surface TMM and TDM via Pks13/CmrA and Ag85. In response to stresses
   such as nutrient-scarce conditions encountered during biofilm formation
   or antibiotic treatment (right, blue), Mtb degrades cell surface
   TMM/TDM via Tdmh and potentially other hydrolases, releasing free
   trehalose that is recycled via LpqY-SugABC and channeled into central
   carbon metabolism (CCM) via TreS, supporting ATP and NADPH synthesis.
   The present study investigates inhibition of the TreS-mediated
   trehalose catalytic shift. (B) TreS-catalyzed conversion of trehalose
   to maltose.

   Over half a century ago, the trehalosamine natural product
   2-amino-2-deoxy-d-trehalose (2-TreNH[2]) was isolated from Streptomyces
   and shown to potently inhibit Mtb growth.^[71]20,[72]21 However, the
   intriguing activity of 2-TreNH[2] on Mtb growth has not been further
   investigated, likely in part due to the difficulty of isolating or
   synthesizing this molecule.^[73]22,[74]23 Thus, the mechanism of action
   and therapeutic potential of 2-TreNH[2] and related compounds remains
   untested. In light of this history and the confluence of (i) our
   discovery of the trehalose catalytic shift as a persistence factor in
   Mtb, (ii) our finding that 6-TreAz inhibits the trehalose catalytic
   shift and enhances TB drug efficacy, and (iii) our lab’s ongoing
   development of novel chemoenzymatic methods to efficiently synthesize
   trehalose analogues,^[75]24,[76]25 here we synthesized a panel of
   azidodeoxy- and aminodeoxy-d-trehalose (TreAz and TreNH[2]) analogues
   and investigated them as potential inhibitors of the Mtb trehalose
   catalytic shift. Using our established biofilm model, metabolomics, and
   cryo-electron microscopy (EM), we found that a subset of the compounds
   impact Mtb growth and trehalose metabolism, with 2-TreNH[2] potently
   blocking the TreS-mediated trehalose catalytic shift, as well as
   potentiating front-line TB drugs both in liquid culture and in infected
   macrophages. Consistent with the cell-based results, the cryo-EM
   structure of TreS-D238A in complex with 6-TreAz shows that the
   inhibitor binds within the active site and induces a conformational
   change essential for conversion of trehalose to maltose, insights that
   may aid therapeutic development. Our data generate renewed interest in
   trehalosamines as TB therapeutic candidates and demonstrate that
   inhibition of the trehalose catalytic shift is a promising strategy to
   target mycobacterial persisters.

Results and Discussion

Design and Synthesis of Azido Trehalose (TreAz) and Trehalosamine (TreNH[2])
Analogues

   Prior studies from our group and others established that the synthetic
   compound 6-TreAz inhibits growth of Msmeg,^[77]19 Mtb,^[78]10,[79]26
   and Mycobacterium aurum.^[80]27 Importantly, we recently demonstrated
   that 6-TreAz is a selective inhibitor of biofilm-associated persisters
   in Msmeg and Mtb, and that 6-TreAz inhibits the TreS-mediated trehalose
   catalytic shift in Mtb.^[81]10,[82]19 In addition, the natural product
   2-TreNH[2] was shown in 1957 to inhibit Mtb growth,^[83]20,[84]21 and
   aside from synthetic studies,^[85]22,[86]28 to our knowledge it has not
   been further studied since, aside from synthetic studies. To
   investigate 6-TreAz and 2-TreNH[2] in more depth, identify related
   active compounds, and enable an initial structure–activity relationship
   study for this underexplored compound class, we designed a panel of 8
   regioisomeric TreAz and TreNH[2] analogues with systematic variation of
   the azidodeoxy or aminodeoxy group position ([87]Scheme [88]1
   A). In addition to 6-TreAz and 2-TreNH[2], the remaining target
   compounds have previously been synthesized and/or isolated from natural
   sources. In 2012, the TreAz series was chemically synthesized,
   including 2-, 3-, 4-, and 6-TreAz, and the analogues were tested for
   their ability to metabolically label TMM and TDM in mycobacteria to
   allow click chemistry-based imaging and other applications.^[89]29 Out
   of the four trehalosamine targets, 2-, 3-, and 4-TreNH[2] are naturally
   occurring aminoglycosides isolated from Streptomyces (2- and
   4-TreNH[2]) or Nocardiopsis (3-TreNH[2]), and 6-TreNH[2] was chemically
   synthesized, first by the Hanessian group.^[90]20,[91]30−[92]32
   However, aside from our prior study on 6-TreAz, the remaining 7
   compounds in the panel have not been tested for their impact on Mtb
   biofilm formation or the trehalose catalytic shift.

Scheme 1. (A) TreAz and TreNH[2] Analogues Investigated in This Study. (B, C)
TreT-Mediated Chemoenzymatic Synthesis of (B) 3- and 6-Position-Modified
Compounds and (C) 2-Position-Modified Compounds.

   [93]Scheme 1
   [94]Open in a new tab

   Comparison of chemoenzymatic synthesis step counts (blue) and chemical
   synthesis step counts (orange) is given for each target compound.

   To access the 8-member trehalose analogue panel, we used a combination
   of chemoenzymatic and chemical synthesis. Despite the relatively simple
   appearance of the TreAz and TreNH[2] target compounds, their
   preparation using conventional chemical synthesis approaches poses
   multiple challenges due to trehalose’s unique 1,1-α,α-glycosidic bond,
   eight similarly reactive hydroxyl groups, and C[2]-symmetric
   arrangement.^[95]33,[96]34 Previously reported chemical syntheses of
   several of the target compounds required between 4 and 12 steps and
   generally proceeded in <10% overall yield. Similarly, isolation of
   naturally occurring trehalosamines is lengthy and arduous. We
   previously developed a chemoenzymatic method for trehalose analogue
   synthesis, which utilizes the substrate-promiscuous trehalose synthase
   TreT from Thermoproteus tenax to convert glucose analogues and
   UDP-glucose into their respective trehalose analogues in one
   step.^[97]35,[98]36 Here, we applied TreT catalysis to generate 6 out
   of the 8 target compounds ([99]Scheme [100]1
   B,C). Commercially available 3- and 6-azido glucose analogues were
   reacted with UDP-glucose via TreT catalysis to generate 3- and 6-TreAz
   in one step, each in ≥90% yield. Then, Pd-catalyzed reduction converted
   3- and 6-TreAz to 3- and 6-TreNH[2] in 99 and 88% yield, respectively.
   Previously, we found that various 2-N-substituted glucoses were not
   tolerated by TreT, so we developed an alternative route whereby native
   glucose was successfully reacted with UDP-N-acetylglucosamine via TreT
   catalysis, generating 2-N-acetyltrehalosamine that could then be
   N-deacetylated with hydrazine to give 2-TreNH[2]; subjecting 2-TreNH[2]
   to diazo transfer reaction with triflic azide gave 2-TreAz.^[101]22
   Here, to access 2-TreAz and 2-TreNH[2], we used a combination of this
   chemoenzymatic procedure^[102]22 to support initial screening
   experiments and a reported chemical synthesis procedure^[103]29 to
   support experiments requiring more material. Because T. tenax TreT does
   not tolerate 4-position-modified glucose analogues, we chemically
   synthesized 4-TreAz and 4-TreNH[2] using a reported route.^[104]37
   Comparison of step counts for TreT chemoenzymatic synthesis and
   chemical synthesis, shown in [105]Scheme [106]1, reveals the improved
   efficiency of TreT catalysis.

TreAz and TreNH[2] Analogues Differentially Inhibit Planktonic and Biofilm
Growth of Mycobacteria via Action on Trehalose Metabolism

   We previously used in vitro mycobacterial biofilm culture, which is a
   model for drug-tolerant persisters, to discover the trehalose catalytic
   shift and identify 6-TreAz as a selective biofilm inhibitor that
   interferes with this pathway.^[107]10 To determine whether the
   synthesized trehalose analogues were selective biofilm inhibitors, we
   evaluated their effects on planktonic (i.e., free-living in liquid
   medium) and biofilm-associated mycobacterial growth. The compound panel
   was screened at 1 mM concentration for inhibition of Msmeg and Mtb
   grown under either planktonic or biofilm conditions and growth was
   monitored by optical density at 600 nm (OD[600]) or crystal violet
   staining, respectively ([108]Figure [109]2
   A). In the model organism Msmeg, 6-TreAz exhibited selective albeit
   partial biofilm inhibition at 1 mM, consistent with our reported
   results, while 2-TreAz and 2-TreNH[2] promisingly showed selective and
   complete biofilm inhibition at the same concentration ([110]Figure S1).
   None of the 3- or 4-position-modified analogues, nor 6-TreNH[2],
   exhibited inhibition of Msmeg in either growth mode at 1 mM. This
   finding demonstrates that subtle changes in modification type (azido or
   amino) and position on the disaccharide can influence activity.
   Dose–response curves in Msmeg revealed that the natural product
   2-TreNH[2] exhibited the highest potency, with selective biofilm
   inhibition occurring in the low-micromolar range ([111]Figures S2–S9).

Figure 2.

   [112]Figure 2
   [113]Open in a new tab

   6-TreAz and 2-TreNH[2] selectively inhibit biofilm formation in Mtb in
   an LpqY-SugABC-dependent manner. (A) Scheme for evaluating planktonic
   and biofilm growth in Mtb. Reproduced with permission from ref
   ([114]39). Copyright 2005 Elsevier. (B–E) Mtb H37Rv wild type (or
   mutant, if indicated) was cultured under planktonic or biofilm growth
   conditions in the presence of the indicated trehalose analogue, or left
   not treated (NT) as negative control, or treated with front-line TB
   drug isoniazid (INH) as positive control. Growth was measured using
   OD[600] reading for planktonic conditions and CV staining for biofilm
   conditions. In (B), all analogue concentrations are 1 mM. In (D),
   analogue concentrations are 2-TreAz, 1000 μM; 2-TreNH[2], 250 μM;
   3-TreNH[2], 1000 μM; 6-TreAz, 1000 μM. In (E), analogue concentrations
   are 2-TreAz, 1000 μM; 2-TreNH[2], 250 μM; 3-TreNH[2], 1000 μM; 6-TreAz,
   250 μM. Data are normalized relative to positive control at 100%. Error
   bars represent the standard deviation of three replicates and data are
   representative of three independent experiments. See [115]Figures
   S1–S16 for additional Mtb and Msmeg data.

   Inhibitory activities of TreAz and TreNH[2] analogues in the pathogen
   Mtb were largely consistent with those observed in Msmeg, with some
   notable differences ([116]Figure [117]2
   B,C; [118]Figures S10–S16). 6-TreAz selectively inhibited biofilm
   formation in Mtb, with modestly higher potency than in Msmeg,
   consistent with our prior work on this compound. 2-TreAz was less
   potent in Mtb than in Msmeg. 2-TreNH[2], the most potent compound in
   the panel against Mtb, fully inhibited both growth modes at 1 mM and
   dose–response analysis revealed selective, low-micromolar biofilm
   inhibition ([119]Figure [120]2C). Also consistent with the results in
   Msmeg, 3-TreAz, 4-TreAz, 4-TreNH[2], and 6-TreNH[2] lacked activity in
   Mtb. By contrast, 3-TreNH[2], which had no effect in Msmeg, completely
   blocked Mtb planktonic and biofilm growth at 1 mM but had lower potency
   than 6-TreAz and 2-TreNH[2] and was not selective for biofilm
   inhibition, suggesting an alternative mechanism of action. These data
   show that the activity of the compound panel was similar in Msmeg and
   Mtb and that modifications at the 2- and 6-positions were most likely
   to confer inhibitory activity. The natural product 2-TreNH[2] was
   identified as the most potent selective biofilm inhibitor in both Msmeg
   and Mtb.

   In a previous study, we found that the inhibitory activity of 6-TreAz
   in Msmeg was completely dependent on LpqY-SugABC,^[121]19 a plasma
   membrane-associated trehalose-specific transporter. The trehalose
   transporter is conserved in mycobacteria and required for virulence of
   Mtb.^[122]38 However, it is dispensable for growth and biofilm
   formation, and thus the genes encoding it can be knocked out.^[123]38
   To test whether the inhibitory action of TreAz and TreNH[2] analogues
   requires the trehalose transporter, we assessed their effects on
   planktonic and biofilm growth in Msmeg and Mtb wild type, ΔsugC
   knockout mutants (lack functional transporter), and ΔsugC::sugC
   complements (transporter restored). The activity of 6-TreAz, 2-TreAz,
   and 2-TreNH[2] against planktonic and biofilm growth in wild-type Mtb
   was reversed in the ΔsugC mutant and restored in the ΔsugC::sugC
   complement, demonstrating that LpqY-SugABC is required for inhibition
   by these compounds ([124]Figure [125]2
   D,E). Identical results were obtained for these compounds in Msmeg
   ([126]Figures S17 and S18). These findings confirm that the compounds
   interfere with trehalose metabolism and, furthermore, suggest that the
   compounds require active transport into the cell to act on an
   intracellular target. Of note, the inhibitory activity of 3-TreNH[2] in
   Mtb was independent of LpqY-SugABC, further suggesting an alternative
   mechanism of action for this analogue.

TreAz and TreNH[2] Analogues Inhibit the TreS-Mediated Trehalose Catalytic
Shift in Mycobacterium tuberculosis

   Our above data demonstrate that 2-TreNH[2] potently and selectively
   inhibits in vitro biofilm growth in both Msmeg and Mtb in a trehalose
   metabolism-dependent manner, suggesting that this compound may target
   Mtb persister formation through inhibition of the trehalose catalytic
   shift. Therefore, we performed additional mechanism-of-action studies
   in Mtb to determine whether 2-TreNH[2] exerts its activity through
   inhibition of TreS, the trehalose isomerase essential for the trehalose
   catalytic shift. First, we monitored the kinetics of planktonic growth
   and biofilm-persister formation in wild-type Mtb treated with
   2-TreNH[2], and the results were compared to Mtb ΔtreS, a
   TreS-deficient mutant. For comparison as a negative control, we also
   included 6-TreNH[2], which exhibited no inhibitory activity in Msmeg or
   Mtb. Neither treatment with 2- or 6-TreNH[2] nor treS deletion impacted
   Mtb planktonic growth, as expected ([127]Figure [128]3
   A). However, treatment with 2-TreNH[2], but not the control compound
   6-TreNH[2], selectively reduced biofilm-persister formation in
   wild-type Mtb to approximately the same level as Mtb ΔtreS ([129]Figure
   [130]3B). Thus, 2-TreNH[2] treatment phenocopies treS deletion with
   respect to Mtb biofilm-persister growth kinetics.

Figure 3.

   [131]Figure 3
   [132]Open in a new tab

   2-TreNH[2] treatment mimics the metabolic impact of treS deletion in
   Mtb. (A, B) Mtb H37Rv wild type was cultured under (A) planktonic or
   (B) biofilm growth conditions in the presence of 200 μM of the
   indicated trehalose analogue and growth was monitored over time. Mtb
   ΔtreS was included as a control. (C–J) Biofilm metabolomics analysis.
   Metabolomes from 28-day-old Mtb biofilm persisters after treatment with
   0, 2, 20, and 200 μM 2- or 6-TreNH[2] were collected and analyzed by
   liquid chromatography-mass spectrometry (LC-MS). ΔtreS and treatment
   with the same amount of 6-TreAz were used as positive controls. (C, D)
   Biofilm metabolomics pattern analysis by clustered heatmap analysis (C)
   and PCA (D) plots. See [133]Figure S18 for the chemical impact on Mtb
   metabolome in a concentration-dependent manner by PCA analysis. (E–J)
   Intracellular pool sizes of Mtb intermediates of trehalose metabolism
   (E, F), glycolysis (G, H), and pentose phosphate pathway (I, J) in Mtb
   wild type and ΔtreS biofilm persisters. All values are the average of
   independent biological triplicates ± standard error of the mean. **, P
   < 0.05; ns, not significant by Student t test.

   To investigate the mechanistic basis underlying 2-TreNH[2]-mediated
   biofilm-persister inhibition, we conducted metabolomic profiling of
   compound-treated wild-type Mtb biofilm, and compared with that of Mtb
   ΔtreS mutant. We collected the metabolomes from the 28-day-old
   biofilm-persister bacilli of 2-TreNH[2]-, 6-TreNH[2]-, or
   6-TreAz-treated (or untreated) wild-type Mtb or untreated Mtb ΔtreS,
   and quantified ∼250 known metabolites. Clustered heatmap analysis and
   principal component analysis (PCA) plots displayed a clear metabolic
   similarity between Mtb treated with 200 μM 2-TreNH[2] and ΔtreS, which
   was segregated from that of Mtb treated with 200 μM 6-TreNH[2] and
   untreated controls ([134]Figure [135]3
   C,D; [136]Figure S18). We found that the metabolic patterns of Mtb
   treated with increasing doses of 2-TreNH[2] gradually migrated toward
   that of ΔtreS in PCA results, indicating that Mtb metabolic damage due
   to 2-TreNH[2] was almost identical to that of ΔtreS. Given our prior
   finding that 6-TreAz inhibits the trehalose catalytic shift, we
   included this compound in our analysis as an additional control.
   Although Mtb metabolic patterns after 200 μM 6-TreAz treatment were
   quite similar to those of ΔtreS, the impact of 6-TreAz was slightly
   different than 2-TreNH[2], with the latter displaying the closest
   similarity to ΔtreS. Metabolic networks of Mtb after treatment with
   varying doses of 6-TreNH[2] showed no significant changes as compared
   to that of untreated controls. Using the metabolites that were altered
   in each condition, we conducted pathway enrichment analysis ([137]Table
   S1). As expected, a substantial number of pathways were damaged
   commonly in Mtb after chemical treatment with 2-TreNH[2] and treS
   genetic deficiency in ΔtreS under a biofilm-persister state.
   Intriguingly, 2-TreNH[2]-treated Mtb showed additional pathways that
   were remodeled, including pathways involved in arginine and proline
   metabolism, cysteine and methionine metabolism, carbapenem
   biosynthesis, and pentose and glucuronate interconversion. This
   suggested that 2-TreNH[2] may have additional targets. We also observed
   that Mtb in a biofilm-persister state remodeled several metabolic
   pathways after treatment with 6-TreNH[2] but still was able to form
   biofilm-persister bacilli at levels similar to untreated controls,
   suggesting that the affected metabolic pathways were not essential for
   persister formation ([138]Table S1).

   Next, targeted metabolomic analysis was conducted to monitor the
   abundance of trehalose metabolism and CCM intermediates. Similar to
   that of the ΔtreS biofilm-persister metabolome, 2-TreNH[2]-treated Mtb
   accumulated trehalose with reciprocal depletion of maltose, resulting
   in continued deficiency in carbon flux toward the biosynthesis of
   intermediates in upper glycolysis (e.g., glucose 6P) and pentose
   phosphate pathway (PPP) (e.g., pentose 5P and sedoheptulose 7P)
   ([139]Figure [140]3
   E–I). As expected, the metabolic patterns caused by treatment with
   2-TreNH[2] or treS genetic deficiency were not observed in Mtb after
   treatment with 6-TreNH[2]. Separately, we previously reported that
   substantial depletion of the lowest intermediate in glycolysis,
   phosphoenolpyruvate (PEP), also played an important role in provoking
   Mtb drug tolerance and persister formation.^[141]40 Intriguingly, ΔtreS
   failed to downregulate PEP abundance under a biofilm-persister state,
   which was also observed in Mtb after treatment with 2-TreNH[2] but not
   6-TreNH[2] ([142]Figure [143]3J). Overall, our metabolomic analysis
   demonstrated that the impaired Mtb biofilm-persister formation observed
   after treatment with 2-TreNH[2] was largely attributed to a defective
   TreS-mediated trehalose catalytic shift.

   To confirm that 2-TreNH[2] inhibits TreS-catalyzed isomerization of
   trehalose to maltose, we used our previously established in vitro TreS
   enzyme assay.^[144]10 We monitored the extent of the inhibitory effect
   of 2- or 6-TreNH[2] compounds against TreS activity. As expected,
   2-TreNH[2] showed 38–55% suppression of the maltose production activity
   of TreS at 100 μM, comparable to that of our previously reported TreS
   inhibitor 6-TreAz ([145]Figure S19). Intriguingly, despite its lack of
   activity in Msmeg and Mtb cells, 6-TreNH[2] also showed a weaker
   inhibitory effect, ranging from 10 to 50% on TreS activity. In
   addition, Mtb TreS did not convert 6-TreAz or 2-TreNH[2] to the
   corresponding maltose analogues, 6-azido or 2-amino maltose, indicating
   that neither compound served as a substrate of TreS-mediated trehalose
   isomerization. This result is consistent with a published report
   showing that Msmeg TreS does not isomerize 6-TreAz.^[146]41 Combined,
   our data suggest that 2-TreNH[2] is the best TreS-specific inhibitor
   among the compounds tested and harbors the most promising activity on
   Mtb biofilm persisters.

TreAz and TreNH[2] Analogues Sensitize Mycobacterium tuberculosis to Existing
Antitubercular Compounds and Are Active in a Macrophage Infection Model

   The trehalose catalytic shift is a metabolic strategy used by Mtb to
   survive environmental pressure, including antibiotic treatment.^[147]10
   Thus, we tested whether cotreatment with trehalosamine compounds
   increases the antimicrobial effects of known anti-TB antibiotics such
   as rifampicin (RIF), isoniazid (INH), and BDQ against Mtb growing
   planktonically. Consistent with the previous literature^[148]10 and our
   results monitoring biofilm-persister formation, 2-TreNH[2], but not
   6-TreNH[2], significantly enhanced the antimicrobial effects of RIF and
   BDQ ([149]Figure [150]4
   A,B). The IC[90] values of RIF and BDQ on wild-type Mtb were 0.84 ±
   0.12 and 0.18 ± 0.02 μg/mL, which were improved to 0.38 ± 0.08 and 0.09
   ± 0.01 μg/mL, respectively, as a result of cotreatment with 2-TreNH[2]
   ([151]Figure [152]4D). The improved IC[90] values represent a
   potentiation effect because Mtb planktonic growth was unaltered by
   treatment with 2-TreNH[2] alone at the same concentration ([153]Figure
   [154]3A). Interestingly, antimicrobial potentiation was not observed in
   cotreatment experiments with INH, presumably due to its strong and
   rapid bactericidal effects,^[155]42,[156]43 and in fact led to an
   unanticipated MIC increase ([157]Figure [158]4C,D). Thus, we measured
   the viable but nonculturable colonies after cotreatment of 2-TreNH[2]
   and INH or BDQ by monitoring the outgrowth kinetics in antibiotic-free
   7H9 medium after inoculating with each culture.^[159]44 Mtb cultures
   treated with INH or BDQ exited the lag phase at day 7, but ΔtreS and
   Mtb after cotreatment with 2-TreNH[2] showed a prolonged period of lag
   phase without evidence resuming the growth until day 20 ([160]Figure
   [161]4E,F). On the other hand, Mtb cultured with cotreatment with INH
   and the control compound 6-TreNH[2] showed regrowth kinetics similar to
   Mtb culture without any trehalose analogue. The results from Mtb
   outgrowth experiments suggest that 2-TreNH[2] also affects viable but
   nonculturable bacilli putatively formed after treatment with INH or
   BDQ.

Figure 4.

   [162]Figure 4
   [163]Open in a new tab

   TreAz and TreNH[2] analogues potentiate the activity of existing
   anti-TB drugs. (A–C) Dose–response growth of Mtb for (A) RIF, (B) BDQ,
   and (C) INH were measured with cotreatment with 200 μM trehalose
   analogues. The OD growth was monitored at 10 days of incubation and
   calculated by % growth relative to NT with no antibiotic. (D)
   Half-maximal inhibitory concentration (IC[50]) and 90% inhibitory
   concentration (IC[90]) shift resulting from treatments in (A–C) were
   determined by Prism software (Ver 9.5). (E, F) Effect of 200 μM
   trehalose analogues on Mtb outgrowth following treatment with 30× MIC
   of (E) INH and (F) BDQ for 5 days. (G, H) Effect of 200 μM trehalose
   analogues on colony-forming units (CFU)-based viability of Mtb within
   THP-1 macrophages following treatment with 10× MIC of (G) RIF and (H)
   BDQ. All values are the average of biological triplicates ± standard
   error of the mean.

   We next tested whether the antimicrobial potentiation resulting from
   cotreatment with conventional TB antibiotics and trehalose analogues
   during in vitro growth was also effective against intracellular Mtb
   bacilli. We used THP-1 human macrophages as a host and treated
   Mtb-infected THP-1 cells with 1X minimum inhibitory concentration (MIC)
   of RIF or BDQ in the presence or absence of trehalose analogue. RIF and
   BDQ showed a weak bactericidal effect on intracellular Mtb when
   administered alone. However, cotreatment with 2-TreNH[2] enhanced the
   activity of RIF from ∼10% to greater than 80%, nearly identical to the
   effect of treS deletion ([164]Figure [165]4
   G). Similar drug potentiation against intracellular Mtb was observed
   during cotreatment with BDQ and 2-TreNH[2] ([166]Figure [167]4H). In
   both the RIF and BDQ cotreatment experiments, the effect of 2-TreNH[2]
   was significantly better than that of 6-TreAz. Surprisingly,
   6-TreNH[2], which showed no impact on in vitro growth or
   antimycobacterial potentiation, showed a potentiation effect against
   intracellular Mtb at levels similar to that of 2-TreNH[2] ([168]Figure
   [169]4G,H). It is possible that 6-TreNH[2] has Mtb bactericidal targets
   only under intracellular conditions or host cell-specific targets to
   provoke the observed effects. Finally, none of the trehalose analogues
   shown to have inhibitory activity against Mtb in liquid culture or
   macrophages were cytotoxic to mammalian U-937 cells at concentrations
   up to 1 mM ([170]Figure S20), underscoring their potential as safe and
   selective inhibitors of Mtb persister formation within a host
   environment.

Cryo-Electron Microscopy Reveals That 6-TreAz Binds to the Active Site of Mtb
TreS and Induces a Conformational Change

   Our results demonstrate that small-molecule inhibitors of TreS
   sensitize Mtb to existing TB antibiotics and may have value as
   adjunctive therapeutics. To understand the mechanisms of TreS catalysis
   and inhibition, and to gain insight into specific interactions of
   inhibitors with TreS active site residues, we sought to determine the
   structure of TreS in complex with the inhibitors described herein. To
   date, no structures of TreS bound to native substrates or substrate
   analogues have been reported. We initially attempted cocrystallization
   of wild-type Mtb TreS with substrates or inhibitors. However,
   consistent with a previous report,^[171]45 our solved crystal
   structures, resolved to a resolution of 2.7 Å, lacked density
   corresponding to any of the compounds within the enzyme active site.
   Given that TreS preferentially crystallizes in the unliganded form, we
   turned to cryo-EM to elucidate the structures of enzyme–inhibitor
   complexes. In addition, we aimed to mitigate against the possibility of
   substrate or substrate analogues being reacted upon by the enzyme.
   Previously published results using Msmeg TreS unambiguously identified
   the aspartic acid residue corresponding to the D238 residue of Mtb TreS
   as the nucleophile that initiates the attack on the anomeric carbon of
   either trehalose or maltose.^[172]45−[173]47 Therefore, we constructed
   a catalytically inactive Mtb TreS-D238A variant, which was used to
   obtain a cryo-EM structure of Mtb TreS in complex with a trehalose
   analogue inhibitor.

   Cryo-EM experiments using TreS-D238A in the presence of 6-TreAz
   afforded the 3-dimensional reconstruction of the Mtb TreS
   tetramer/6-TreAz to a resolution of 3.6 Å (PDB ID: 8UQV) ([174]Figure
   S21). As the tetramer of the Mtb TreS crystal structure (PDB ID:
   [175]4LXF) was used as the preliminary model for fitting the initial
   maps, observed differences between the initial model and maps clearly
   illustrate a conformational change potentially induced by ligand
   binding. Following multiple rounds of manual and computational real
   space refinement, clear density corresponding to a disaccharide
   molecule was observed near the C-terminal end of the β-sheets of the
   barrel where the active site of a TIM-barrel fold protein is typically
   located ([176]Figure [177]5
   ).^[178]48 The observed density for 6-TreAz was stronger for the D
   chain of the tetramer compared to the other 3 chains. Therefore, for
   analysis purposes, the D chain was used for alignments and for defining
   the specific interactions within the active site. The 6-azido-glucosyl
   moiety of 6-TreAz fit within the cryo-EM map is modeled with multiple
   axial hydroxyls as that is the best fit ([179]Figure [180]5A). This
   higher-energy conformation may derive from steric hindrance between the
   C6 azido group and the enzyme active site residues, or it may reflect a
   conformational change in the substrate required to promote catalysis.
   As other glycosyl hydrolases generally bind glucosyl moieties and react
   upon them in the low-energy chair conformation,^[181]49,[182]50 it is
   reasonable to assume that the high-energy conformation of that portion
   of 6-TreAz is due to the addition of the azido moiety and is the direct
   cause of TreS inhibitory activity.

Figure 5.

   [183]Figure 5
   [184]Open in a new tab

   Cryo-EM structure of TreS/6-TreAz complex reveals a ligand-induced
   conformational change in TreS. (A) Density map for the 6-TreAz ligand.
   (B) Interactions formed between the ligand 6-TreAz and the catalytic
   amino acid residues at the active site of TreS-D238A. The D238A is
   shown as D238 and aligned to illustrate the likely interactions in
   wild-type TreS. (C) Conformational change observed from the alignment
   of the 6-TreAz-bound TreS-D238A to the published apo structure of Mtb
   TreS (PDB ID: [185]4LXF). Regions without significant changes in the
   alignment are colored in gray. (D) Changes of active site residues upon
   6-TreAz binding. (E) Structural changes of the L19 loop and the
   repositioning of L352.

   Based on published literature and the determined structures, the three
   conserved catalytic residues of TreS are D238 (nucleophile), E280
   (general acid/base), and D350 (substrate coordination).^[186]45−[187]47
   The side chain of the E280 residue is positioned 4.3 Å from the C1
   hydrogen of the pyranose possessing the azido group containing the
   azido group ([188]Figure [189]5
   B). When the D238A residue is modeled as aspartate to represent the
   wild-type form of the enzyme, this places the side chain 4 Å closer to
   the C1 carbon of the azide-containing pyranose moiety and able to
   undergo a nucleophilic attack on the true substrate.

   To understand the conformational changes that occur upon complex
   formation, the 6-TreAz-bound Mtb-TreS-D238A cryo-EM structure was
   aligned with the crystal structure of wild-type TreS protein incubated
   with 6-TreAz, treating that structure as a control since 6-TreAz is not
   observed in that crystal structure (PDB ID: 8UZH). Additionally, we
   determined the X-ray crystal structure of the ligand-free TreS-D238A
   variant. This structure is consistent with wild-type Mtb TreS, which
   indicates that the D238A mutation alone was insufficient to produce the
   observed structural changes in the cryo-EM structure of the
   TreS-D238A/6-TreAz complex. The 6-TreAz-bound complex and the
   ligand-free wild-type Mtb TreS structures were also superimposed with
   the published wild-type ligand-free TreS crystal structure (PDB ID:
   [190]4LXF). The structural alignment of the published ligand-free TreS
   structure with the solved crystal structure of wild-type TreS protein
   incubated with 6-TreAz shows no structural differences ([191]Figure
   S22). However, as shown in [192]Figure [193]5
   C, the 6-TreAz-bound cryo-EM structure shows significant movement of
   the L19 and L20 loops at the active site along with the H9 and H10
   helices and L8 loop vacating the space required to accommodate 6-TreAz
   binding. The structural alignment of the ligand-free and 6-TreAz
   complex also highlights the movement of the two catalytic residues of
   TreS upon binding 6-TreAz ([194]Figure [195]5D). The position of the
   D238A nucleophile is little changed. However, the E280 residue, which
   acts as the general acid, moves slightly away from the ligand binding
   site to better accommodate the ligand and afford the acid/base
   chemistry essential for the enzymatic reaction. Specifically, E280
   donates a proton to the C1 hydroxyl of the leaving group following
   nucleophilic attack by D238 and, following glycone rotation, subsequent
   proton abstraction from the C4 hydroxyl that functions as the
   nucleophile during the formation of the new α-1,4-glycosidic bond to
   produce maltose. Additionally, D350, which is essential for
   coordinating ligand binding, moves toward the substrate and forms a
   hydrogen-bonded interaction with each of the two rings of 6-TreAz. Like
   other enzymes catalyzing similar reactions on sugars, D350 may promote
   nucleophilic attack on the anomeric carbon by slightly modulating the
   electronics within the rings through hydrogen-bonded interactions with
   the C2 hydroxyl group. This is analogous to the role of D503 in the Mtb
   GlgE enzyme, D480 in the Streptomyces coelicolor GlgE1 enzyme that
   catalyzes the formation of 1,4-glycosidic bond in growing α-glucans,
   D295 of Leuconostoc mesenteroides sucrose phosphorylase that catalyzes
   sucrose phosphorolysis to produce α-d-glucose 1-phosphate and
   d-fructose, and other glycosyl hydrolase family 13
   members.^[196]51−[197]53 In each of the aforementioned systems, the
   conserved aspartate residue forms a bidentate interaction with the
   hydroxyls at positions 2 and 3 of the respective rings.

   Other structural differences in the ligand-free structure highlight the
   significant reorientation of active site loops, both to accommodate
   bound ligand and to prevent loss of the glucose intermediate during the
   required glycone rotation. The side chain of L352 harbored by the L19
   loop occupies the same volume of space in the ligand-free structure
   that is occupied by 6-TreAz in the TreS-D238A/6-TreAz complex
   structure. Specifically, the position of the L352 residue in the
   ligand-free structure is stabilized by hydrophobic interactions with
   Y101, V138, H141, I179, F202, and A239 ([198]Figure [199]5
   E). However, in the 6-TreAz-bound structure, the L19 loop containing
   L352 travels 10 Å from the active site with the new conformation
   stabilized via interactions between the L352 side chain and a different
   hydrophobic network formed by residues M311, P312, F315, L346, L354,
   M365, Y369, I381, and L388 ([200]Figure [201]5E). Additionally, the H9
   helix between the L19 and L20 loops and the L8 loop slightly shift
   toward the active site. This reorientation likely locks the compound at
   the active site and facilitates the catalytic reaction through
   preventing release of the glucose intermediate and promoting rotation
   and formation of the new 1,4-glycosidic linkage. Finally, to further
   highlight the unique active site structure in TreS stimulated by ligand
   binding, the structure was superimposed onto the TreS portion of the M.
   smegmatis TreS-Pep2 complex (PDB ID: [202]5JY7) to determine if the
   complex with Pep2 promotes a similar conformational change in
   TreS.^[203]54 Once again, the conformational differences in the 6-TreAz
   complex were clearly observed as TreS in TreS-Pep2 complex possesses
   the same conformation as the apo structure, which further strengthens
   the hypothesis that only substrates or substrate analogues stimulate
   this structural change.

   From a drug development perspective, it will be of interest to quantify
   the thermodynamic differences between the two TreS conformations to
   ascertain the need of stimulating this conformational change with novel
   TreS inhibitors, or if the ligand-free form is a druggable target. It
   may be possible to develop an allosteric inhibitor of TreS by targeting
   the ligand-free form, thereby preventing the conformational change
   required for substrate binding and catalysis. For example, targeting
   the hydrophobic pocket that accommodates L352 in the 6-TreAz-bound form
   of TreS could block the conformational shift necessary to promote
   catalysis.

Conclusions

   The discovery, characterization, and targeting of Mtb persistence
   factors is emerging as an attractive pathway toward adjunctive
   therapeutics that could be used alongside existing drugs to increase
   the speed and efficacy of TB treatment. The trehalose catalytic shift,
   which is required for Mtb persister formation in a biofilm model, was
   recently identified as a promising source of targets for
   persister-focused inhibitor development. Specifically, TreS-catalyzed
   isomerization of recycled trehalose into maltose is a key step in the
   trehalose catalytic shift that promotes Mtb survival during stress.
   Here, we took a multidisciplinary approach spanning synthetic
   chemistry, microbiology, metabolomics, and structural biology to
   develop substrate analogue inhibitors of TreS that selectively block
   Mtb biofilm-persister formation and potentiate TB drugs against Mtb
   growing in liquid medium or inside macrophages. Chemoenzymatic
   synthesis was instrumental in rapidly generating a panel of
   systematically altered TreAz and TreNH[2] analogues, which enabled
   initial SAR assessment on the focused 8-compound panel and
   identification of 6-TreAz and 2-TreNH[2] as lead compounds.
   Quantitative mass spectrometry-based metabolomic analysis of
   inhibitor-treated Mtb biofilm-persister cells identified TreS as a
   target of both compounds and showed that 2-TreNH[2] treatment most
   accurately mimics the metabolic damage of treS deletion. 2-TreNH[2] is
   a natural product whose antimycobacterial activity has been known for
   decades, but whose mechanism of action, TB drug potentiation, and lack
   of toxicity against mammalian cells were previously unknown and
   revealed here for the first time. Interestingly, in combination
   treatment experiments, different TB antibiotics were variably
   potentiated by 2-TreNH[2] during in vitro growth, and INH potency
   unexpectedly decreased. Although the precise reason for this
   variability is presently unknown, we nonetheless found in outgrowth
   kinetics and macrophage infection experiments that 2-TreNH[2] strongly
   potentiated multiple TB antibiotics, including INH in outgrowth
   experiments, nearly to the same level as deletion of treS. This is
   consistent with our prior work showing that Mtb engages the trehalose
   catalytic shift as an adaptive strategy to mitigate antibiotic
   treatment-induced reactive oxygen species (ROS) production, activate an
   alternate ATP biosynthetic pathway, and biosynthesize antioxidant
   chemicals, all of which are advantageous to survive antibiotic effects
   regardless of their modes-of-action.^[204]10 An important next step
   will be to investigate whether TreS inhibitors improve the efficacy of
   TB treatment in animal models of Mtb infection, which will require
   significantly larger compound quantities and may necessitate
   innovations to scale up the synthetic methods. We will also investigate
   the impact of TreS inhibitors on the frequency of the emergence of
   drug-resistant mutants through inhibition of persister formation. Given
   the differential activities of the trehalose analogues tested,
   extension of this research to related analogues (e.g., epimers,
   nonhydrolyzable analogues) to generate richer SAR data, as well as
   testing in mycobacterial pathogens besides Mtb, is warranted. Finally,
   this study reported the first three-dimensional (3D) structure of TreS
   in complex with a substrate or substrate analogue. Critically, because
   TreS crystallization is preferential for the unliganded form and
   wild-type enzyme presents potential issues with respect to analogue
   reactivity, we used cryo-EM to obtain the structure of a catalytically
   inactive TreS mutant (D238A) in complex with 6-TreAz. The
   TreS-D238A/6-TreAz structure showed that 6-TreAz binds to the enzyme
   active site, inducing a conformational change that is likely necessary
   for catalysis and could be exploited to develop new strategies for
   inhibiting TreS. These results also highlight the utility of cryo-EM as
   a key technique for obtaining high-resolution structures of TreS bound
   to other substrate analogues, including 2-TreNH[2]. Collectively, our
   study (i) validates TreS inhibition as a strategy to target Mtb
   persisters; (ii) identifies selective and cell-active substrate
   analogue inhibitors of TreS that are lead compounds for adjunctive
   therapeutic development; and (iii) provides TreS structural and
   mechanistic knowledge that can be leveraged in future TB drug
   development efforts. More broadly, our study serves as a model
   multidisciplinary approach to identifying inhibitors that target
   persistent infections, which are caused by diverse bacterial pathogens.

Materials and Methods

General Procedures for Compound Synthesis

   Reagents and solvents were procured from commercial sources without
   further purification, unless otherwise noted. TreT enzyme was expressed
   and purified as previously reported Meints et al.^[205]1
   Monosaccharides were obtained from Sigma (6-GlcAz), Synthose (3-GlcAz),
   Carbosynth (UDP-GlcNAc), and Abcam (UDP-glucose). Thin-layer
   chromatography (TLC) analysis was performed on glass-backed silica 60 Å
   plates (thickness 250 μm) from Silicycle and visualized either by
   charring with 5% H[2]SO[4] in ethanol or gentle warming with ninhydrin
   stain. Column chromatography was performed on flash-grade silica gel
   32–63 μm (230–400 mesh) from Silicycle. ^1H and ^13C NMR spectra were
   recorded at 500 and 126 MHz, respectively, on a Bruker Avance 500 NMR
   spectrometer.

Synthesis of 2-TreAz and 2-TreNH[2]

   2-TreAz and 2-TreNH[2] were synthesized using a combination of
   chemoenzymatic synthesis using the approach of Groenevelt et
   al.^[206]22 and chemical synthesis using the approach of Swarts et
   al.^[207]29^1H and ^13C NMR data ([208]Supporting Information) matched
   the literature data.

Synthesis of 4-TreAz and 2-TreNH[2]

   4-TreAz and 4-TreNH[2] were chemically synthesized using the approach
   of Bassily et al.^[209]37^1H and ^13C NMR data ([210]Supporting
   Information) matched the literature data.

Synthesis of 3-TreAz and 6-TreAz

   The syntheses of 3-TreAz and 6-TreAz were carried out using a reported
   chemoenzymatic synthesis method.^[211]35,[212]36 His-tagged TreT enzyme
   was expressed and purified from Escherichia coli as reported.^[213]36
   To a 15 mL conical tube were sequentially added
   3-azido-3-deoxy-d-glucose or 6-azido-6-deoxy-d-glucose (16.4 mg, 20 mM
   final concentration), UDP-glucose (97.6 mg, 40 mM final concentration),
   MgCl[2]·6H[2]O (16.2 mg, 20 mM final concentration), and 4 mL of
   His-tagged TreT enzyme (300 μg/mL) in Tris buffer. The contents of tube
   were mixed thoroughly by pipetting and the reaction was incubated at 70
   °C with shaking for 60 min, after which the reaction mixture was cooled
   on ice for 20 min. The reaction mixture was added to a pre-rinsed
   Amicon Ultra-15 centrifugal filter unit and centrifuged at 3900g for 20
   min. The upper chamber of the filter unit was washed twice with water
   (3 mL) and centrifuged at 3900g for 20 min. After discarding the upper
   chamber of the filter unit, mixed-bed ion-exchange resin (Bio-Rad
   Bio-Rex RG 501-X8, 3 g) was added and the mixture was stirred for 1 h.
   The supernatant was collected and the resin was washed twice with water
   (5 mL). The supernatants were combined and dried via rotary evaporation
   to give 3-TreAz (26.8 mg, 92% yield, white solid) or 6-TreAz (26.7 mg,
   92% yield, white solid). ^1H and ^13C NMR data ([214]Supporting
   Information) matched the literature data from Swarts et al.^[215]29

Synthesis of 3-TreNH[2] and 6-TreNH[2] Analogues

   TreAz analogues were reduced to the corresponding TreNH[2] analogues
   via Pd-catalyzed hydrogenation using the procedure of Rodriguez-Rivera
   et al.^[216]55 To an argon-flushed round-bottom flask containing a
   stirring solution of TreAz analogue (6-TreAz: 28.2 mg, 0.076 mmol;
   3-TreAz: 18.3 mg, 0.049 mmol) in water (6-TreAz: 5 mL; 3-TreAz: 5 mL)
   was added 10% wt. Pd/C (6-TreAz: 14 mg; 3-TreAz: 9 mg). A
   hydrogen-filled balloon was attached to the flask to replace the argon
   atmosphere. The reaction was stirred at room temperature overnight,
   after which the catalyst was filtered and the filtrate was dried via
   rotary evaporation to give the corresponding TreNH[2] analogue
   (3-TreNH[2]: 16.7 mg, 98% yield, white solid); for 6-TreNH[2], the
   sample was treated with glacial acetic acid (200 μL), filtered, and
   dried via rotary evaporation to give the acetate form (29.1 mg, 95%
   yield, white solid). ^1H and ^13C NMR data ([217]Supporting
   Information) matched the literature data from Lu et al.^[218]23

Bacterial Strains and Culture Conditions

   Msmeg WT, ΔsugC mutant^[219]38 (hygromycin B, 50 μg/mL), or sugC::sugC
   complement^[220]19 (hygromycin B, 50 μg/mL; apramycin, 10 μg/mL) was
   cultured at 37 °C in M63 medium (M63 salts minimal medium supplemented
   with 2% glucose, 0.5% casamino acids, 1 mM MgSO[4], and 0.7 mM CaCl[2])
   with or without Tween-80 as detergent and antibiotic if noted. Mtb
   H37Rv WT, ΔsugC mutant (hygromycin B, 50 μg/mL),^[221]38 ΔsugC::sugC
   complement (hygromycin B and kanamycin, 50 μg/mL each),^[222]38 and
   ΔtreS mutant (hygromycin B, 50 μg/mL)^[223]10 were cultured at 37 °C in
   Sauton’s medium^[224]56 (KH[2]PO[4] 0.5 g/L, MgSO[4]·7H[2]O 0.5 g/L,
   citric acid 2 g/L, ferric ammonium citrate 0.05 g/L, 1% ZnSO[4]
   solution 0.1 mL/L, l-asparagine 4 g/L, glycerol 6% v/v) supplemented
   with or without 0.4% tyloxapol as detergent and antibiotic if noted.
   Outgrowth experiments were conducted in Middlebrook 7H9 medium (2.5 g/L
   disodium phosphate, 1.0 g/L monopotassium phosphate, 0.5 g/L monosodium
   glutamate, 0.5 g/L ammonium sulfate, 0.1 g/L sodium citrate, 0.05 g/L
   magnesium sulfate, 0.04 g/L ferric ammonium citrate, 1.0 mg/L copper
   sulfate, 1.0 mg/L zinc sulfate, 1.0 mg/L pyridoxine HCl, 0.5 mg/L
   biotin, 0.5 mg/L calcium chloride) supplemented with 0.2% glycerol,
   0.2% dextrose, 0.04% tyloxapol, 0.5 g/L bovine serum albumin, and
   0.085% NaCl. All Mtb H37Rv strains were cultured in a biosafety level 3
   facility. Mtb mc^27000 auxotroph^[225]9 was cultured at 37 °C in
   Sauton’s medium supplemented with 100 μg/mL d-pantothenate and with or
   without 0.4% tyloxapol as detergent.

Planktonic and Biofilm Growth Assays

   Msmeg planktonic and biofilm growth assays were performed as previously
   reported by Wolber et al.^[226]19 For Mtb planktonic and biofilm growth
   assays, starter cultures of Mtb were grown in Sauton’s medium
   containing 0.04% tyloxapol detergent and antibiotic (if appropriate)
   until reaching exponential phase (OD[600] ∼ 0.8). To initiate
   planktonic growth assays, starter cultures were diluted to OD[600] 0.01
   in Sauton’s medium containing 0.04% tyloxapol. 100 μL of the diluted
   culture was added to a sterile polystyrene 96-well plate containing 100
   μL of 2-fold serially diluted trehalose analogues in Sauton’s medium to
   give a final compound concentration of 0–1000 μM, a final volume of 200
   μL, and a final OD[600] of 0.005. Liquid medium with no compound was
   used as a no treatment (NT) control and 5× MIC INH (5 μg/mL) was used
   as a positive control. The empty wells in the plate were filled with
   sterile water to help minimize evaporation. The plates were sealed with
   sterile breathable film and placed in a secondary container and
   incubated at 37 °C with shaking for 5 weeks. At regular intervals, the
   contents of plates were mixed by pipetting up and down and OD[600] was
   recorded using a Tecan Infinite M200 Pro microplate reader. At the end
   of the 5-week incubation period, biofilm growth assays were performed
   essentially as described by Lee et al.^[227]10 To initiate biofilm
   growth assays, starter cultures prepared as described above were
   diluted to OD[600] 0.01 in Sauton’s medium without tyloxapol. 100 μL of
   the diluted culture was added to a sterile polystyrene 96-well plate
   containing 100 μL of 2-fold serially diluted trehalose analogues in
   Sauton’s medium to give a final compound concentration of 0–1000 μM, a
   final volume of 200 μL, and a final OD[600] of 0.005. Liquid medium
   with no compound was used as a no treatment (NT) control and INH (5
   μg/mL) was used as a positive control. The empty wells in the plate
   were filled with sterile water to help minimize evaporation. Lids were
   placed on the plates, which were then triple-sealed with parafilm,
   placed in a secondary container, and incubated at 37 °C without shaking
   for 5 weeks. The biofilms were washed three times with
   phosphate-buffered saline (PBS) and stained using 1% aqueous CV
   solution. Stained biofilms were washed three times with PBS to remove
   excess unbound CV, then the biofilm-associated CV was extracted using
   95% ethanol. OD[595] of 20-fold ethanol-diluted CV extract was recorded
   using a Tecan Infinite M200 Pro microplate reader.

Metabolite Extraction and LC-MS Metabolomics

   For LC-MS metabolomics, at 28-day-old biofilm after media removal,
   biofilm culture was washed with ice-cold PBS three times and quenched
   by adding 1 mL of acetonitrile/methanol/H[2]O (40:40:20) precooled to
   −40 °C. Biofilm metabolites were extracted by mechanical lysis with 0.1
   mm zirconia beads in Precellys tissue homogenizer for 6 min (6000 rpm)
   twice under continuous cooling at or below 2 °C.^[228]10 Lysates were
   clarified by centrifugation and then filtered through a 0.22-μm spin-X
   column. The residual protein content of metabolite extracts was
   determined to normalize the samples to cell biomass. LC-MS
   differentiation and detection of extracted metabolites were performed
   using an Agilent Accurate Mass 6230 TOF coupled with an Agilent 1290
   Liquid Chromatography system. The metabolites were separated on a
   Cogent Diamond Hydride Type C column (gradient 3) (Microsolve
   Technologies) with solvents and configuration as reported
   previously.^[229]57 The mobile phase consisted of solution A (dd-H[2]O
   with 0.2% formic acid) and solution B (acetonitrile with 0.2% formic
   acid). An isocratic pump was used for continuous infusion of a
   reference mass solution to allow mass axis calibration. Detected ions
   were deemed to be metabolites on the basis of unique accurate
   mass-retention time identifiers for masses exhibiting the expected
   distribution of accompanying isotopologues. The abundance of extracted
   metabolites was analyzed using Profinder B06.00 software and Agilent
   Qualitative Analysis B.07.00 with a mass tolerance of <0.005 Da. The
   clustered heatmap, pathway enrichment assay, principal component
   analysis, and fold change analysis were performed using bioinformatics
   tools available in MetaboAnalyst v.5.0 ([230]www.metaboanalyst.ca).

Expression and Purification of Mtb TreS and In Vitro TreS Assay

   The treS gene (Rv0126) was amplified by PCR using the TreS-specific
   primers (forward, 5′- CAT ATG AAC GAG GCA GAA CAC AGC GTC – 3′;
   reverse, 5′ – AAG CTT TCA TAG GCG CCG CTC TCC C −3′) as previously
   reported.^[231]10 The amplified gene and pET28a were double digested
   with NdeI and HindIII and ligated to construct pET28a::TreS. After the
   sequence was confirmed, E. coli BL21 (DE3) pLysS was used as the
   expression strain. The E. coli BL21 (DE3) pLysS harboring pET28a::TreS
   was grown at 37 °C in LB medium containing 50 μg/mL of kanamycin and 25
   μg/mL of chloramphenicol to an OD[595] of 0.6–0.7. Expression of the
   treS gene was induced by isopropyl-β-d-thiogalactopyranoside (IPTG) to
   the cultures to a final concentration of 0.5 mM, and then cells were
   further grown for 16 h at 16 °C. The cells were harvested from 500 mL
   cultures and resuspended in 20 mL of lysis buffer (20 mM Tris-HCl [pH
   7.5] and 200 mM NaCl) containing protease inhibitor. The resuspended
   cells were disrupted using sonication, and cell-free crude extracts
   were obtained by centrifugation at 15,000 rpm for 30 min. The crude
   extracts were loaded into a column packed with 1 mL of an 80% (v/v)
   slurry of Ni-Sepharose resin. The resin was washed with 10 bed volumes
   of lysis buffer containing 5 mM imidazole and washed further with 10
   bed volumes of lysis buffer containing 75 mM imidazole. His[6]-tagged
   TreS was eluted from the resin with 5 bed volumes of lysis buffer
   containing 250 mM imidazole and 10% glycerol. Imidazole and NaCl were
   removed from purified TreS by means of a PD-10 desalting column
   equilibrated with 20 mM Tris-HCl (pH 7.5) containing 10% glycerol.

   In vitro TreS assays were conducted as previously reported.^[232]10
   TreS activity was measured using a 100 μL in vitro enzyme reaction
   containing 40 mM MOPS (pH 7.0), 10 mM trehalose, and 30 ng purified
   TreS enzyme in the presence or absence of 100 μM 2-TreNH[2],
   6-TreNH[2], or 6-TreAz. The reactions without TreS enzyme or trehalose
   substrate were also included. The TreS enzyme was incubated at 37 °C
   with the inhibitor for 10 min, after which the reaction was initiated
   by adding the substrate-buffer mixture. The mixture was incubated at 37
   °C for an additional 10 min, then the reaction was quenched by adding
   ice-cold acetonitrile containing 0.2% formic acid to yield a final 70%
   acetonitrile mixture. After centrifugation, reaction supernatants were
   analyzed by LC-MS to quantify maltose (or maltose analogue).

Antibiotic Potentiation Assays

   To monitor the antimycobacterial effects, Mtb cultures were
   growth-synchronized to late log-phase and back-diluted to an OD[595] of
   0.05 before plating in Middlebrook 7H9 medium. Plates were incubated
   standing at 37 °C with 5% CO[2]. OD[595] was evaluated using a plate
   reader at 10–15 days postplating and percent growth was calculated
   relative to the untreated control for each. IC[50] measurements were
   calculated using a nonlinear fit in GraphPad Prism. For all IC[50]
   curves, data represent the mean ± standard error for technical
   triplicates. Data are representative of at least two independent
   experiments.

   Outgrowth experiments were conducted as previously reported.^[233]44
   Mid log phase Mtb cultures were resuspended in 5 mL of Middlebrook 7H9
   medium with 50× MIC-equivalent BDQ or 20× MIC-equivalent INH and
   incubated at 37 °C for 5 days. The cultures were then diluted 21-fold
   (10 μL into 200 μL) into fresh 7H9 medium in a new 96-well plate
   (outgrowth plate). OD[595] was measured using a plate reader every 2–3
   days. All values are the average of biological triplicates ± standard
   error.

THP-1 Cell Culture and Intramacrophage Killing Assay

   THP-1 cells were grown RPMI-1640 with 10% v/v fetal bovine serum (FBS),
   1 mM sodium pyruvate, 2 mM l-glutamine, and PenStrep (100 U/mL
   penicillin and 100 μg/mL streptomycin) at 37 °C using an incubator with
   5% CO[2]. The cells were fed every 3–4 days by removing half of the
   culture medium and replacing with fresh medium. THP-1 cells were washed
   with fresh RPMI-1640 without PenStrep medium and adjusted 1 × 10^5
   cells/mL. The cells were treated with 20 nM phorbol myristate acetate
   (PMA) for 24 h before infection and used to infect a single-cell
   suspension of Mtb at a mean of infection (MOI) of 10 for 84 h. After 84
   h, the cells were extensively washed and the media was replaced with
   RPMI-1460 with 50 μg/mL gentamycin to remove extracellular bacteria
   overnight. The cells were and then treated with 1× MIC-equivalent RIF
   or BDQ in the presence of 100 μM of 2-TreNH[2], 6-TreNH[2], or 6-TreAz.
   To estimate intracellular Mtb growth, infected macrophages were lysed
   using 0.5% Triton X-100, and serial dilutions were plated on m7H10 and
   incubated at 37 °C. CFU were determined 21 days later.

Mammalian Cell Cytotoxicity Assay

   Cytotoxicity of 2-TreAz, 2-TreNH[2], 3-TreNH[2], 6-TreAz, and
   6-TreNH[2]was assessed in pro-monocytic U-937 cells using commercially
   available the lactate dehydrogenase (LDH) assay kit (Cayman Chemical).
   The assay was performed essentially as per kit instructions. Briefly, 2
   × 10^5 U-937 cells/mL in 100 μL RPMI medium with 10% FBS in a
   flat-bottom 96-well plate in triplicate were treated with 1 mM
   concentration of 2-TreAz, 2-TreNH[2], 3-TreNH[2], 6-TreAz, 6-TreNH[2],
   or trehalose. 10% Triton X-100 solution was used as a maximum LDH
   release control and cell culture grade water was used as the
   spontaneous LDH release control. The cells were incubated for 48 h at
   37 °C with 5% CO[2], after which the 96-well plate was centrifuged at
   400g for 5 min and 50 μL of supernatant was transferred to another
   96-well plate. To each well, 50 μL of the kit reaction mixture was
   added, and the plate was incubated at 37 °C with gentle shaking for 30
   min in a Tecan Infinite M200 Pro plate reader and absorbance was
   monitored at 490 nm. From the absorbance readings, percent cytotoxicity
   was calculated as per kit instructions.

Expression and Purification of Mtb TreS and Cryo-EM Structural Elucidation

   The production of recombinant Mtb wild-type TreS is as previously
   described but with slight modifications.^[234]45 The amino acid range
   from 11 to 587 with a cleavable N-terminal 6× histidine tag was
   expressed in E. coli. A synthetic gene, codon optimized for E. coli
   protein production encoding Mtb TreS-11-587-D238A (IDT-DNA
   technologies), was amplified by PCR and inserted at the PshA1 cut site
   of a modified pET32 plasmid (EMD biosciences) and used to transform E.
   coli T7 Express cells (New England BioLabs). A single colony was used
   to inoculate LB containing 100 μg/mL of carbenicillin and cultured at
   37 °C. Gene expression was induced at 16 °C with 1 mM IPTG for 24 h,
   and the cells were harvested by centrifugation at 4000 rpm for 30 min.
   Pelleted cells were resuspended in Buffer A containing 50 mM Tris pH
   7.5, 500 mM NaCl, 1% (v/v) glycerol, 0.3 mM (tris (2-carboxyethyl)
   phosphine) (TCEP), and 5 mM imidazole. Final concentrations of 1 μM
   lysozyme and 0.1 μM DNase I were added to the suspension of cells and
   the sample was sonicated (Sonicator 3000, Misonix). The lysate was
   centrifuged at 10,000 rpm for 40 min and the supernatant was loaded
   onto a His Trap 5 mL column equilibrated with Buffer A. After washing
   the column with 15 column volumes of Buffer A, bound TreS was eluted
   with Buffer B containing 150 mM imidazole and the other components of
   Buffer A. The protein was incubated with rhinovirus 3C protease
   overnight while dialyzing against buffer A. The tag-cleaved protein was
   applied to a His Trap 5 mL column equilibrated with Buffer A and the
   flowthrough was collected. The protein was subjected to size exclusion
   chromatography to remove any aggregates using 50 mM Tris pH 7.5 buffer,
   containing 300 mM NaCl and 0.3 mM TCEP.

   The sample was diluted to a concentration of 0.5 mg/mL and 3.2 μL of
   that same sample was blotted to cryo-EM grids (Quantifoil R 1.2/1.3 300
   Mesh, Cu) following glow discharge using a PELCO easiGlow glow
   discharge cleaning system and blotted for 3 s using the Vitrobot Mark
   IV System (Thermo Fisher). Cryo-EM data were collected at the Iowa
   State University cryo-EM facility using a Glacios microscope at 200 kV
   (Thermo Fisher) and a K3 direct electron detector (DED). Data were
   analyzed and maps were generated using CryoSPARC. A tetramer of the
   TreS wild-type apo structure available in PBD ID: [235]4LXF was used as
   the starting model for the initial fitting within the density map. The
   structure was further refined using Coot and Phenix. The final model
   was visualized using Pymol.

Acknowledgments