Folate Pathway Disruption Leads to Critical Disruption of Methionine Derivatives in Mycobacterium...

Please cite this article in press as: Nixon et al., Folate Pathway Disruption Leads to Critical Disruption of Methionine Derivatives in Mycobacteriumtuberculosis, Chemistry & Biology (2014), http://dx.doi.org/10.1016/j.chembiol.2014.04.009

Chemistry & Biology

Article

Folate Pathway Disruption Leadsto Critical Disruption of Methionine Derivativesin Mycobacterium tuberculosisMolly R. Nixon,1,2 Kurt W. Saionz,3 Mi-Sun Koo,5 Michael J. Szymonifka,5 Hunmin Jung,3 Justin P. Roberts,3

Madhumita Nandakumar,4 Anuradha Kumar,2 Reiling Liao,2 Tige Rustad,2 James C. Sacchettini,3 Kyu Y. Rhee,4

Joel S. Freundlich,5 and David R. Sherman1,2,*1Interdisciplinary Program in Pathobiology, Department of Global Health, University of Washington, Seattle, WA 98195, USA2Seattle Biomedical Research Institute, Seattle, WA 98109, USA3Department of Biochemistry and Biophysics, Texas A&M University, College Station, TX 77843, USA4Departments of Microbiology and Immunology, Weill Medical College of Cornell University, New York, NY 10065, USA5Department of Pharmacology and Physiology and Medicine, Center for Emerging and Reemerging Pathogens, Rutgers University–NewJersey Medical School, Newark, NJ 07103, USA

*Correspondence: [email protected]

http://dx.doi.org/10.1016/j.chembiol.2014.04.009

SUMMARY

In this study, we identified antifolates with potent,targeted activity against whole-cell Mycobacteriumtuberculosis (MTB). Liquid chromatography-massspectrometry analysis of antifolate-treated culturesrevealed metabolic disruption, including decreasedpools of methionine and S-adenosylmethionine.Transcriptomic analysis highlighted altered regula-tion of genes involved in the biosynthesis and utiliza-tion of these two compounds. Supplementation withamino acids or S-adenosylmethionine was sufficientto rescue cultures from antifolate treatment. Insteadof the ‘‘thymineless death’’ that characterizes folatepathway inhibition in a wide variety of organisms,these data suggest thatMTB is vulnerable to a criticaldisruption of the reactions centered around S-ad-enosylmethionione, the activated methyl cycle.

INTRODUCTION

Mycobacterium tuberculosis (MTB), the causative agent of

tuberculosis (TB), is arguably one of the most successful of all

human pathogens. Within recorded history, it is likely the leading

cause of death in Europe and the United States (Bloom and

Murray, 1992). Although the advent of modern diagnostics, anti-

bacterials, and hygiene have dramatically reduced the death rate

from TB in developed nations, MTB still caused 1.3 million

deaths worldwide in 2012 (Zumla et al., 2013)—more than from

any other single bacterial pathogen (Nathan et al., 2008).

Currently, ‘‘short course’’ therapy for TB consists of four

drugs—isoniazid, rifampicin, ethambutol, and pyrazinamide—

that are administered for 2 months, followed by 4 months of

isoniazid and rifampicin (Plorde, 2004). Although highly effective

if taken to completion, this regimen is lengthy, poorly tolerated,

and incompatible with many other medications, including some

antiretrovirals prescribed for the treatment of HIV. Unfortunately,

these issues have frequently led to low rates of completion for

Chemistry & Biolo

the full course of treatment (Gough and Kaufman, 2011). The

recent emergence of multidrug-resistant, extensively drug-

resistant, and totally drug-resistant tuberculosis strains further

challenges the utility of the current front-line regimen (Babu

and Laxminarayan, 2012; Gandhi et al., 2006; Velayati et al.,

2009), which has not changed since the inclusion of rifampicin

in 1965 (Williams and Duncan, 2007).

New drugs to treat TB are desperately needed. Ideally, any

new TB drugs would be active following oral administration,

exhibit minimal side effects, be compatible with anti-retroviral

treatments, have activity against strains resistant to other TB

drugs, rapidly clear latent infections, and have pharmacokinetic

and pharmacodynamic properties amenable to relatively infre-

quent dosing.

Antifolates, which interrupt the production of reduced folate

cofactors, are one class of antibacterials with potential to

meet this need (Figure 1). Reduced folate cofactors are used

as one-carbon donors in a wide variety of essential cellular

processes, including the biosynthesis of methionine, serine,

and glycine; purine production; the conversion of deoxyuridine

monophosphate (dUMP) to thymidine monophosphate (dTMP);

and the formation of formyl-methionyl tRNAs in bacteria

(Kompis et al., 2005). Verified drug targets within the folate

pathway include dihydropteroate synthase (DHPS) in bacteria

and dihydrofolate reductase (DHFR) in bacteria, parasites,

and mammals (Hirsch, 1942; Kompis et al., 2005; Potts et al.,

2005).

As early as 1940, promin, a pro-drug of the DHPS inhibitor

dapsone (DDS), was shown to clear MTB infection in guinea

pigs (Barr, 2011). A current second-line therapy for TB, para-

amino salicylic acid (PAS), is also a pro-drug substrate of

DHPS that subsequently poisons as-yet unidentified down-

stream reactions (Chakraborty et al., 2013). Another DHPS inhib-

itor, sulfamethoxazole (SMX), is active against MTB in vitro, with

90% growth inhibition at 8 mg/ml (Wallace et al., 1986). The MTB

DHFR enzyme, encoded by the gene dfrA, is exquisitely sensitive

to the inhibitor methotrexate, but whole-cell MTB is resistant to

killing by this drug (Kumar et al., 2012). WR99210, another inhib-

itor of DHFR, was shown to have bactericidal activity against

whole-cell MTB at 1 mM (Gerum et al., 2002). Additionally, it

gy 21, 1–12, July 17, 2014 ª2014 Elsevier Ltd All rights reserved 1

mailto:[email protected]


Figure 1. The Folate Pathway and Relevant Inhibitors

(A) The folate pathway in Mycobacterium tuberculosis. Compounds in red are known inhibitors of DHPS and DHFR, respectively.

(B) Test compounds.

Chemistry & Biology

Folate Pathway Disruption in M. tuberculosis


has recently been demonstrated that a combination of the DHFR

inhibitor trimethoprim and sulfamethoxazole maintains its acti-

vity in culture against some MDR-TB strains (Vilcheze and Ja-

cobs, 2012). However, there are currently no DHFR inhibitors

used clinically for the treatment of TB.

Structural studies on MTB DHFR have revealed a general fold

similar to the human ortholog, as well as features within the

2 Chemistry & Biology 21, 1–12, July 17, 2014 ª2014 Elsevier Ltd All

active site and potential conformational changes that are spe-

cific to MTB (Dias et al., 2014; Li et al., 2000). The possibility

of exploiting unique active site binding, combined with the

demonstrated susceptibility of whole-cell MTB to inhibition by

WR99210, inspired further work to develop antifolates with

activity against MTB. We recently reported a high-throughput

screen for inhibitors of the MTB DHFR enzyme (Kumar et al.,

rights reserved

Table 1. Activity of Test Compounds against MTB DHFR Enzyme

(IC50) and against Whole-Cell MTB H37Rv or Strains with Altered

DHFR Expression—MIC99

Test

Compound

Mtb DHFR

IC50 (mM)

H37Rv

(mM)

H37Rv::

pMRN1 (mM)

H37Rv:dfrA-

TetON (mM)

WR99210 0.014 1.5–3 30 0.75

Trimetrexate 0.017 6.25 >100 not tested

Trimethoprim 19 >100 not tested not tested

Piritrexim 0.022 >12.5 >100 not tested

Methotrexate 0.0068 >100 >100 25

JSF-1183 0.030 0.6 50 0.08

JSF-1187 0.050 0.2–0.4 >100 0.13

JSF-2104 0.024 33 >100 not tested

JSF-2105 0.172 33 >100 not tested

In H37Rv::pMRN1, dfrA expression levels are approximately 8 times wild

type, whereas in the absence of tetracycline, dfrA expression levels in the

H37Rv:dfrA-TetON are approximately 0.3 times wild type.

Chemistry & Biology



2012). Here we report the synthesis and characterization of

methotrexate derivatives (Figure 1) with activity against MTB.

In addition, we describe the molecular consequences associ-

ated with folate pathway disruption in MTB. Unlike the ‘‘thymine-

less death’’ associated with folate pathway inhibition in many

other organisms, we find that MTB undergoes specific perturba-

tion in reactions that produce and utilize S-adenosylmethionine,

collectively known as the activated methyl cycle (Halliday et al.,

2010; Parveen and Cornell, 2011).

RESULTS

MTB Strains with Altered dfrA ExpressionTo differentiate between compounds targeting DHFR and those

aimed at other targets, we created strains with altered expres-

sion of dfrA. H37Rv::pMRN1 has two copies of dfrA, resulting

in overexpression of the cognate mRNA. The second copy was

introduced via the integrating plasmid pMRN1, which contains

the MTB dfrA sequence driven by the high-level constitutive

MOP promoter. Transformants were identified using kanamycin

selection and nearly 8-fold dfrA overexpression was confirmed

via quantitative RT-PCR (qRT-PCR) (data not shown). We also

created H37Rv:dfrA-TetON, in which the native dfrA promoter

was swapped for an artificial promoter responsive to tetracy-

cline. In the absence of tetracycline, expression of dfrA is

approximately 30% of wild-type levels (Kumar et al., 2012). An

anti-DHFR antibody was not available to determine the degree

of altered expression at the protein level, but the results below

strongly suggest that DHFR levels are in fact altered as expected

in these strains.

WR99210 Targets MTB DHFRPreviously published studies have demonstrated not only that

the triazine antimalarial WR99210 binds to and inhibits the

MTB DHFR enzyme, but also that it inhibits the growth of

MTB at approximately 1 mM (Gerum et al., 2002; Li et al.,

2000). However, the unusually low rate of resistant mutants

(<1.2 3 10�10) suggested that WR99210 might have multiple-

Chemistry & Biolo

or off-target effects in MTB (Gerum et al., 2002). To confirm

that the observed activity of WR99210 resulted from inhibition

of DHFR, minimum inhibitory concentrations (MICs) were

compared between wild-type H37Rv and the dfrA over- and

underexpressor strains. The MICs increased in the overexpres-

sor and decreased in the underexpressor, showing that sensi-

tivity to WR99210 is materially linked with intracellular levels

of DHFR (Table 1).

Test Compounds Exhibit Activity against H37RvTo evaluate the susceptibility of MTB to antifolates, select com-

pounds were assayed for their ability to inhibit growth of the

wild-type MTB strain H37Rv (Table 1). These compounds

included commercially available antifolates, as well as metho-

trexate analogs that to the best of our knowledge have not pre-

viously been tested against MTB. Of the commercially available

antifolates tested, piritrexim and trimetrexate both showed ac-

tivity targeting the folate pathway, as evidenced by potent inhi-

bition of the MTB DHFR enzyme in vitro and the shift in MIC in

the DHFR overexpressor strain (Table 1). As expected, metho-

trexate and trimethoprim showed no activity against whole-

cell MTB. Although trimethoprim’s lack of whole-cell activity

may be attributed to its relatively poor inhibition of DHFR

(IC50 = 19 mM), methotrexate is a potent DHFR inhibitor

(IC50 = 6.8 nM). Cognizant of previous efforts outside of the

anti-infectives realm (Rosowsky 1973), we also synthesized

and tested analogs of methotrexate. We harnessed published

chemical methods (Dawson et al., 1987; Piper and Montgom-

ery, 1977; Rosowsky, 1973) to prepare the dimethyl (JSF-

1187) and diethyl (JSF-1183) esters of methotrexate and

demonstrated that they inhibit DHFR with IC50 values of 50

and 30 nM, respectively. Unlike methotrexate, however, the

esters also exhibited significant potency against whole-cell

MTB (MIC = 600 nM for JSF-1183 and 200–400 nM for JSF-

1187). Critically, the MIC values increased approximately two

orders of magnitude when dfrA was overexpressed and

decreased approximately 10-fold when dfrA expression was

reduced (Table 1). To investigate the active species within cells,

we first measured intracellular compound levels by mass

spectrometry and found a mixture of parent compound, each

mono-ester, and some methotrexate (data not shown). We

then synthesized and tested both mono-ethyl esters (JSF-

2104 and JSF-2105). Each of these compounds had weak but

measurable activity on MTB, whereas neither showed improved

activity on the DHFR enzyme (Table 1).

WR99210 and JSF-1187 Are Effective against ClinicalMTB StrainsIsolated in 1905, H37Rv is a laboratory strain of MTB that retains

full virulence in animal models (Cole et al., 1998); however, there

are likely significant differences between currently circulating

strains of MTB and H37Rv (Ioerger et al., 2010). To test whether

clinical strains might be similarly susceptible to growth inhibition

by antifolates, a panel of five circulating MTB strains was treated

with JSF-1187 and WR99210. All the strains tested were known

to be pan-sensitive to current antitubercular drugs. Although

there was some divergence in the observed MICs, all strains

were susceptible to both WR99210 and JSF-1187 at levels

similar to H37Rv (Table S1 available online).


Figure 2. DHFR and DHPS Inhibitors Behave Synergistically in Cul-

ture against MTB

Each point represents a combination of drug concentrations that result in an

MIC level of inhibition. All combinations shown were synergistic. The results

shown are representative of at least three replicate experiments.

(A) JSF-1187 and dapsone. FIC = 0.47.

(B) JSF-1187 and sulfamethoxazole. FIC = 0.54.

(C) WR99210 and dapsone. FIC = 0.45.

(D) WR99210 and sulfamethoxazole. FIC = 0.49.

Chemistry & Biology



DHFR and DHPS Inhibitors Are Synergistic in MTBIn other systems, it is well established that the combination of

DHFR and DHPS inhibitors can result in synergy that reduces

the rate at which resistance develops while decreasing the total

amount of drug that must be used, potentially minimizing the risk

of unwanted side effects (Bushby, 1975). To assesswhether syn-

ergy also occurs in MTB, the DHFR inhibitors JSF-1187 and

WR99210 were each combined with the DHPS inhibitors SMX

(Wallace et al., 1986) and DDS (Karlson, 1963).

A traditional checkerboard titration format (Elion et al., 1954)

was used to determine a fractional inhibitory concentration

(FIC) for each set of compounds, where FIC is defined as the

sum of the [concentration of each agent in combination]/[con-

centration of each agent individually] for both agents at a given

level of inhibition. An FIC of 1 denotes an additive effect, whereas

an FIC greater than 1 represents an antagonistic relationship and

an FIC less than 1 represents a synergistic relationship between

compounds. For all four compound combinations tested, the FIC

was approximately 0.5, signifying synergy (Figure 2). JSF-1187

and WR99210 were also tested for synergy with the antimyco-

bacterials isoniazid, streptomycin, kanamycin, and rifampicin;

additive relationships were observed for all such combinations

(data not shown).

Exposure toAntifolates Leads toDownstreamMetabolicDisruptionTo probe the metabolic effects of these compounds in MTB,

metabolite levels were quantified following treatment with


WR99210 using LC-MS-based methods (de Carvalho et al.,

2010b). Cells were grown on nitrocellulose filters and exposed

to drug at 0.8, 20, and 100 mM (approximately 0.253, 73, and

333 MIC) for 0, 4, and 24 hr. Cells were then rapidly quenched

in a cold acetonitrile/methanol mixture and lysed. Ion profiles

were generated and queried for 168 chromatographically

resolved m/z species corresponding to known metabolites, of

which 92 were detected (Table S2). For 80% of these peaks,

there was little or no change in metabolite abundance following

treatment. As expected, abundance of the m/z species corre-

sponding to WR99210 increased with both time and concentra-

tion (data not shown). Increased WR99210 concentrations and

longer exposure times were associated with increased metabo-

lite dysregulation.

Twelve chromatographically resolved m/z species produced

peaks whose relative abundance increased more than 2-fold

in duplicate experiments, whereas six produced peaks whose

relative abundance decreased more than 2-fold in both experi-

ments (Table S2). Where available, we used coelution with

chemical standards to confirm select metabolite identities and

the method of standard addition to calculate metabolite con-

centrations. The largest increases were observed for 5-aminoi-

midazole-4-carboxamide ribonucleotide (AICAR) and dUMP

(�40- and 30-fold, respectively), and the largest decrease was

for S-adenosylmethionine (SAM, at least 20-fold, to the limit of

detection; Figure 3A). Other metabolites (and/or their corre-

sponding masses) that changed in concentration include

homocysteine, S-ribosylhomocysteine, S-methylthioadenosine,

lysine, serine, methionine, and uracil. The kinetics of metabolite

disruption were also noteworthy (Figure 3B). Homocysteine

levels increased within 4 hr, suggesting that reduced folate

pools were already depleted. Methionine concentrations were

halved at 4 hr but showed no change after that. In contrast,

SAM levels were barely affected after 4 hr and then dropped

substantially by 24 hr. At no time were significant changes in

metabolite concentration observed for thymine, adenine, gua-

nine, adenosine, or glycine.

In aggregate, exposure to WR99210 produced a striking

pattern of metabolic disruption, with an emphasis on species

associated with the active methyl cycle, reactions that produce

or utilize S-adenosylmethionine. To assess which features of

this pattern were shared by exposure to other antifolates, metab-

olomics experiments were also conducted with JSF-1183 and

JSF-1187. MTB cells were grown on nitrocellulose filters, treated

with 25 mM (�333 MIC) of each compound for 24 hr, and then

processed for metabolomic analysis as above. Treatment with

JSF-1183 or JSF-1187 each produced disruptions in core folate

metabolites remarkably similar to those seen following exposure

to WR99210. Key metabolites affected by WR99210 all trended

in the same directions following treatment with JSF-1183 or JSF-

1187. In particular, homocysteine and AICAR were highly

induced, whereas SAM levels were highly depleted by all three

drugs. The patterns were not absolutely identical; dUMP was

more powerfully induced by WR99210, whereas methionine

levels were decreased to a greater extent following treatment

with JSF-1183 and JSF-1187. Such modest differences in

metabolite levels following 24 hr of exposure to drugs are not

surprising, and the overall pattern is similar enough to define a

metabolic signature of antifolate inhibition in MTB.

rights reserved

Figure 3. Treatment of MTB with WR99210 Leads to Concentration-Dependent and Time-Dependent Metabolic Disruptions

(A)WR99210 disruptsmetabolite pools in a concentration-dependent fashion. Cells were treated for 24 hr withWR99210 at 0.8, 20, and 100 mM (�0.253, 73, and

333 MIC), and the data represent the fold change relative to time zero.

(B) WR99210 disrupts metabolite pools in a time-dependent fashion. Conversion of homocysteine to methionine depends on the availability of reduced folate

cofactors (indicated by red bars). Metabolites were identified by LC-MS (see Experimental Procedures), and putative metabolite identities were based on ac-

curate mass (m/z) and chromatographic retention time matching. In most cases, metabolite identities were confirmed using chemical standards demonstrating

coelution with experimental samples, and levels are expressed as the ratio of metabolite concentration compared to untreated cells. For AICAR, S-ribosylho-

mocysteine and S-methylthioadenosine, levels are expressed as the ratio of signal intensity compared to untreated cells. The results are representative of two

independent experiments, with three technical replicates per experiment. S-Adenosylhomocysteine could not be detected in these experiments.

Chemistry & Biology



Exposure to Antifolates Alters Expression of Geneswithin the Activated Methyl CycleMicroarray analysis was used to track mRNA levels following

DHFR inhibition. Cultures were grown to early log phase

Chemistry & Biolo

(OD600 = 0.1) and then exposed to WR99210, JSF-1183, or

JSF-1187, each at�30 times the MIC, the highest concentration

used for metabolomic analysis. 10 ml aliquots were taken before

drug treatment and after 24 hr of exposure. The RNAwas purified


Chemistry & Biology



from these samples and labeled with fluorescent probes, and

then transcript abundance was quantified by hybridization to

custom Nimblegen tiling arrays with 135,000 probes covering

the MTB H37Rv genome (Minch et al., 2012). Gene expression

changes were considered significant if they produced a moder-

ated t test p < 0.01 and a 2-fold change after Benjamini Hoch-

berg multiple testing correction.

After 24 hr of treatment, a large number of genes were differ-

entially regulated in response to each of the drugs, leading to up-

regulation of 300–500 genes and downregulation of 300–900

genes. The response to the three drugswas strikingly consistent;

245 genes were upregulated, and 312 were downregulated in all

three treatments. The full data set is available online (NCBI GEO

accession number GSE55979).

Expression changes were dominated by broad downregula-

tion of basic cellular processes. Almost 70%of the genes encod-

ing ribosomal proteins were downregulated, along with ATP

synthase-related genes and the genes involved in cell division

(ftsEQWXY). Also downregulated were nearly all of the NADH-

dehydrogenase complex I genes and genes of the mce1 and

mce4 loci. thyX, which encodes the flavin-dependent alternate

thymidylate synthase, was also modestly downregulated.

Biosynthetic pathways for several amino acids were downregu-

lated, including those involved in arginine, asparagine, and tryp-

tophan biosynthesis. The set of downregulated genes also

included seven methyltransferases known to be dependent on

S-adenosylmethionine. These genes produce proteins used in

a variety of metabolic pathways, including mycolic acid biosyn-

thesis, RNA methylation, and production of phenolic glycolipids

and pthiocerol dimycocerosates.

Against this backdrop of broad downregulation, the genes

required for production of methionine and SAM were markedly

upregulated (Figure 5). The genesmetA andmetC were upregu-

lated �15-fold, making them 2 of the 10 most strongly upregu-

lated genes. Transcript levels of the methionine synthase gene

metH and the SAM synthase gene metK increased �3-fold and

nearly 2-fold, respectively. In contrast, genes associated with

SAM utilization, including speE, mtn, and Rv2952, were among

those repressed by antifolate treatment.

Supplementation with Serine/Methionine/Glycine, orwith S-Adenosylmethionine, Rescues MTB fromAntifolate TreatmentSupplementation assays have long been used as tools to under-

stand the effects of drug treatment and pathway inhibition in a

variety of organisms (Daniel et al., 1947). To probe the effects

of folate pathway inhibition in MTB, we supplemented MTB

cultures exposed to MIC levels of WR99210 or JSF-1187

with a variety of metabolites. Nucleosides, folic acid, leucovorin

(5-formyl-tetrahydrofolate), and dTMP all failed to alter cell

survival at concentrations up to 1 mM (data not shown). How-

ever, MTB lacks the annotated enzymes required to assimilate

exogenous folate species (Fivian-Hughes et al., 2012) and is un-

likely to efficiently transport phosphorylated nucleotide species

across its thick and hydrophobic cell envelope.

The folate pathway is directly involved in the metabolism of

three amino acids (Figure 1). The serine hydroxymethyltrans-

ferases GlyA1 and GlyA2 use serine as the methyl donor to pro-

duce 5,10-methylene-tetrahydrofolate from tetrahydrofolate.


The reverse reaction transfers the methyl group from 5,10-meth-

ylenetetrahydrofolate to glycine, regenerating tetrahydrofolate.

Another reduced folate cofactor, 5-methyltetrahydrofolate, is

the methyl donor for the production of methionine from homo-

cysteine. When cultures treated with MIC-level concentrations

of WR99210 were grown in media supplemented with combined

serine/methionine/glycine, cell growth was restored to approxi-

mately 75% of untreated cells (Figure 6A). A similar rescue

from inhibition by DDS, SMX, WR99210, and JSF-1187 was

observed when cells were grown in media containing all 20

L-amino acids (data not shown). However, when cultures were

supplemented with a full amino acid mixture lacking serine,

methionine, and glycine, there was no rescue from antifolate

treatment (Figure 6B). There was also no rescue observed

when treated cultures were grown on media containing D-amino

acids (Figure 6D). As expected, supplementation with amino

acids did not affect the outcome of treatment with isoniazid, an

inhibitor of MTB cell wall biosynthesis (data not shown).

To test the hypothesis that antifolate treatment leads to a critical

depletion of methionine derivatives, we also supplemented

WR99210-treated cultures with S-adenosylmethionine (Fig-

ure 6C). Supplementation with 80 mM SAM restored growth to

approximately 50% that of untreated cultures, and growth was

completely restored at 320 mM SAM. However, supplementation

withmethioninealonewasunable to restoregrowthofcells treated

with either WR99210 or JSF-1187 (data not shown). Altogether,

these results suggest that disruption of the activatedmethyl cycle,

rather than perturbation of folate-dependent amino acid levels, is

the critical consequence of folate pathway disruption.

DISCUSSION

In an effort to identify antifolates with antitubercular properties,

we explored methotrexate as a scaffold. Although methotrexate

is inactive against whole-cell MTB, themethotrexate dialkyl ester

analogs JSF-1187 and JSF-1183 exhibited MIC values between

200 and 600 nM against H37Rv (Table 1). JSF-1187 was also

tested and found active against a panel of MTB clinical isolates

(Table S1). Although the mechanistic rationale for this increase

in activity remains unresolved, a plausible explanation is that

JSF-1183 and JSF-1187 benefit from improved target access.

The alkylations afford more lipophilic molecules that may pene-

trate the hydrophobic mycobacterial cell wall better than metho-

trexate, which should form a dianion under assay conditions.

These substitutions may also interfere with polyglutamylation

of the molecule (Kwon et al., 2008), although polyglutamylation

of methotrexate has traditionally been associated with increased

compound retention within the cell and increased affinity for

DHFR (Liani et al., 2003). Additionally, the mono- and di-esters

do not display improved inhibition of DHFR but show a 4- to

25-fold decrease in DHFR enzyme inhibition in vitro (Table 1).

Similar trends were observed in inhibition studies using DHFR

from rabbit, mouse, and Trypanosoma cruzi (Rosowsky et al.,

1978). A plausible scenario is that the methotrexate esters are

sufficiently lipophilic and stable to penetrate the cell wall and

then inhibit DHFR through a combination of the dialkyl ester,

one or both mono-esters, and methotrexate.

Part of the rationale for studying antifolates in MTB is the

wealth of information available about these molecules in other

rights reserved

Figure 4. The Signature of Antifolate Inhibi-

tion in MTB

(A) Map of the core folate metabolome, with key

metabolites that collectively comprise the signa-

ture of folate pathway disruption circled in red.

DHF, dihydrofolate; THF, tetrahydrofolate.

(B) The signature of folate pathway disruption.

MTB cells were treated with JSF-1183 or JSF-

1187 at 100 mM for 24 hr, and results for core folate

pathway metabolites were compared with those

shown above for WR99210. Metabolite signal in-

tensities were normalized to internal control me-

tabolites. Results shown are on a log2 scale and

are representative of two independent experi-

ments, with three technical replicates per experi-

ment. Metabolite identities were confirmed using

chemical standards demonstrating coelution with

experimental samples, with the exception of

AICAR.

Chemistry & Biology



systems. For example, we show that WR99210 and JSF-1187

display synergy with the DHPS inhibitors dapsone and sulfa-

methoxazole (Figure 2), consistent with data from Plasmodium

and other bacteria (Bushby, 1975). However, we also show

that the downstream consequences of folate inhibition in MTB

differ substantially from the model developed primarily in Plas-

modium and Escherichia coli. In particular, our data are not

consistent with the ‘‘thymineless death’’ scenario that is the hall-

mark of antifolate treatment in other organisms (Ahmad et al.,

1998; Hartman, 1993; Huennekens, 1996).

Antifolates restrict cellular metabolism at several discrete

steps (Figure 4). Thymineless death describes a process in which

antifolate-generated thymine starvation leads to increased rates

of erroneous uridine incorporation into DNA and a subsequent

accumulation of fatal single- and double-stranded breaks

(Ahmad et al., 1998; Sangurdekar et al., 2011). This idea was

originally proposed after experiments in which bacteria treated

with antifolates were rescued by thymine supplementation

(Amyes and Smith, 1974). Recent analyses support this concept,

showing a rapid and significant drop in dTMP/thymidine/thymine

levels in E. coli and Plasmodium falciparum following antifolate

treatment (Kwon et al., 2010; Yeo et al., 1997), although antifo-

lates produce other metabolic disruptions as well. For example,

in E. coli treated with antifolates, the glycine pool is depleted 10-

fold within 5 min, and methionine levels decrease 85% within

30 min, whereas most amino acid levels gradually increase

(Kwon et al., 2010). In fact, when amino acid levels are low,

E. coli is rescued from the bactericidal effects of thymineless

death by a RelA-mediated stringent response (Kwon et al.,

2010). In MTB, however, the picture is quite different. We see

little evidence to support the thymineless death scenario, and

instead, our metabolomics, transcriptomics, and supplementa-

tion experiments are all consistent with a disruption in the acti-

vated methyl cycle.

Following antifolate treatment, our LC-MS experiments re-

vealed limited disruption of MTB metabolite pools (Figures 3

and 4). Even after 24 hr of treatment with WR99210 at �333

MIC, the putative thymine and glycine pools, as well as 80% of

all compounds assayed, were unperturbed. Many of the metab-

olites that do change are linked to the activated methyl cycle.

Comparing metabolite profiles following treatment with three

Chemistry & Biolo

antifolates revealed striking similarities, including substantial

increases in AICAR and homocysteine and a drop in levels of

SAM and methionine (Figure 4). These changes are accompa-

nied by consistent shifts in expression of the relevant genes (Fig-

ure 5). Exposure to the DHPS substrate and TB drug PAS yields a

very comparable metabolic profile (Chakraborty et al., 2013),

suggesting that this pattern reflects a signature of folate pathway

inhibition in MTB.

Supplementation experiments also produced a unique

response in MTB. It is not surprising that supplementation with

thymidine and dTMP failed to support bacterial growth after

treatment with WR99210 and JSF-1187. MTB lacks an obvious

homolog to thymidine kinase and thus cannot convert exoge-

nous thymidine into the dTMP needed for DNA incorporation

(Fivian-Hughes et al., 2012). Additionally, the relatively imperme-

able mycobacterial cell wall is likely to limit the movement of

dTMP into the cell, even when levels in the growth media are

high (Nguyen and Thompson, 2006). However, addition of SAM

or a combination of serine, methionine, and glycine reversed

the inhibitory effects of antifolates in MTB (Figure 6). In contrast,

amino acid supplementation enhances antifolate killing in E. coli

(Kwon et al., 2010).

Altogether, our experiments suggest a model where methio-

nine pools are depleted by antifolate treatment, whereas SAM

pools are protected out to 4 hr of treatment. At some point, how-

ever, methionine levels stabilize, and other components of the

activated methyl cycle are allowed to deplete. Continued flux

through the activated methyl cycle is indicated by the drop in

SAM and the increase in levels of putative S-ribosylhomocys-

teine (one of the end products of SAM demethylation) between

4 and 24 hr (Figure 3). We were unable to detect S-adenosylho-

mocysteine, a feedback inhibitor of many SAM-dependent

methyltransferases (Reddy et al., 2008), although it seems likely

that it accumulates as well. In multiple cell lines, perturbation of

the ratio of SAM and SAH, typically maintained at 10:1, leads to

growth arrest (Reddy et al., 2008). Whether such a mechanism

exists in MTB and is induced following antifolate treatment re-

mains to be seen.

Disruption of the activated methyl cycle should have multiple

consequences. SAM is an integral donor of reactive methyl

groups essential for cellular processes such asDNAmethylation,


A

B

Figure 5. Treatment with Antifolates Leads

to Altered Expression of Genes of the Acti-

vated Methyl Cycle

Cells were treated with WR99210, JSF-1183, or

JSF-1187 at �333 MIC for 24 hr. RNA was ex-

tracted from treated cells, and labeled probes

were hybridized to custom Nimblegen arrays.

(A) Map of the activated methyl cycle, with relevant

MTB genes and compounds (in boxes).

(B) Fold change in expression of activated methyl

cycle genes following antifolate treatment. Results

shown are the averages of three independent

biological replicates.

Chemistry & Biology



polyamine biosynthesis, and biotin production (Parveen and

Cornell, 2011). In MTB, the SAM-dependent methyltransferase

Hma catalyzes the formation of both keto- andmethoxy-mycolic

acids (Boissier et al., 2006); another SAM-dependent methyl-

transferase, Rv2952, is responsible for the unique phthiocerol

dimycocerosates and phenolic glycolipids found in the Beijing

family of MTB strains (Huet et al., 2009). SAM is also necessary

for proper methylation of ribosomal RNAs in MTB (Kumar et al.,

2011; Rahman et al., 2010). Depletion of SAM pathway compo-

nents likely disrupts these essential cellular reactions, as well as

others that are not yet appreciated.

8 Chemistry & Biology 21, 1–12, July 17, 2014 ª2014 Elsevier Ltd All rights reserved

Our current experiments cannot defini-

tively exclude the possibility that starva-

tion for dTMP plays a role in the death of

MTB following antifolate treatment.

Consistent with the thymineless death

scenario, the MTB genes induced by

exposure to antifolates include several

associated with DNA repair (NCBI GEO

accession number GSE55979). We also

observe a 30-fold increase in dUMP after

24 hr of treatment withWR99210 at�333

MIC. Unfortunately, we were unable to

measure dTMP or thymidine at any time

point, although thymine levels remained

unchanged. It should be noted, however,

that MTB is unusual in possessing two

thymidylate synthase enzymes (Figure 1).

ThyA, a canonical thymidylate synthase,

uses 5,10-methylene-tetrahydrofolate as

both methyl donor and reductant. ThyX,

a flavin-dependent thymidylate synthase,

uses 5,10-methylene-tetrahydrofolate as

methyl donor and FADH2 as reductant

(Fivian-Hughes et al., 2012). This unique

pairing may skew the response to folate

pathway inhibition away from thymidylate

starvation and toward the depletion of

methionine derivatives.

Antifolates are an attractive class of

inhibitors, whose potential for treating

TB has yet to be realized. Increased un-

derstanding of the effects of folate

pathway inhibition should allow rational

design of better inhibitors and combinations. Although the folate

pathway has been the subject of study for over 50 years, it is

becoming clear that the established paradigms may not entirely

apply to MTB, providing us with unique challenges to confront

and unique opportunities to exploit.

SIGNIFICANCE

New agents to combat tuberculosis (TB) are badly needed,

yet one of the world’s most utilized drug targets—folate

pathway inhibition—is not a part of front line TB

Figure 6. Supplementation with Amino

Acids or S-Adenosylmethionine Rescues

MTB from Treatment with WR99210

MTB constitutively expressing GFP as a reporter

for growth was treated with 780 nM WR99210

(white bars) or 1% DMSO (no drug control; gray

bars) and supplemented with serine/methionine/

glycine (A), amino acids (minus serine/methionine/

glycine) (B), S-adenosylmethionine (C), and

D-amino acids (D). The results shown are the

means of two technical replicates ± SD. All results

were confirmed by optical density (data not

shown). RFU, relative fluorescence units.

(A and B) Graphs show representative data from

three to five experiments.

(C and D) Graphs show representative data from

two experiments.

Chemistry & Biology



chemotherapy, and widely used antifolates methotrexate

and trimethoprim are inactive on M. tuberculosis (MTB) in

culture. In this study, however, we identified antifolate

variants of methotrexate with good activity versus MTB di-

hydrofolate reductase in vitro and potent activity against

whole-cell MTB. Metabolomic analysis of antifolate-treated

cultures revealed unique metabolic disruptions, including

decreased pools of methionine and S-adenosylmethionine.

Gene expression analysis highlighted altered regulation of

genes involved in the biosynthesis and utilization of these

two compounds. Supplementation with amino acids or

S-adenosylmethionine was sufficient to rescue cultures

from antifolate treatment. These data contrast substantially

with the well-studied effects of folate pathway inhibition in

other organisms. Instead of the ‘‘thymineless death’’ that

characterizes antifolate addition in E. coli, Plasmodium,

and other systems, these data suggest that MTB is vulner-

able to a critical disruption of the reactions centered around

S-adenosylmethionione, the activated methyl cycle. As a

result, folate pathway inhibition in MTB may present novel

opportunities and challenges for drug development.

EXPERIMENTAL PROCEDURES

Bacterial Strains and Growth Conditions

Unless otherwise indicated, liquid cultures of mycobacteria were grown in 7H9

media supplementedwith 0.05%Tween 80 and albumin, dextrose, and catalase

(ADC;BectonDickinson)oronMiddlebrook7H10platessupplementedwitholeic

acid, ADC at 37�C. For metabolomic analysis cultures were grown on Middle-

brook 7H10 plates supplemented with 5% bovine serum albumin, 2% dextrose,

and 0.85% NaCl. Strains were grown to an optical density (OD600) of approxi-

mately 1 and stored at�80�C in 15% glycerol. If needed, kanamycinwas added

at 30 or 50 mg/ml for mycobacteria and E. coli, respectively. H37Rv (American

Type Culture Collection 25618) was used as the Mycobacterium tuberculosis

Chemistry & Biology 21, 1–12, July 17, 20

wild-type strain, whereas E. coli DH5-a (Invitrogen)

was used for cloning. H37Rv::pHigh22, bearing gfp

driven by a strong 16S rRNA promoter, was used

for all assays using a fluorescent read-out.

Construction of MTB Strains with Altered

dfrA Expression

For the constitutively overexpressing construct

pMRN1, PCR-amplified MTB dfrA sequence was

cloned into the shuttle vector pJEB402 downstream of the mycobacterial

constitutive promoter MOP (George et al., 1995). This vector integrates into

the MTB chromosome at the phage L5 attachment site (Lee and Hatfull,

1993). DNA was isolated from E. coli transformants and introduced into MTB

via electroporation as previously described (Park et al., 2003; Wards and

Collins, 1996). Kanamycin was used to select for transformants, and qRT-

PCR was used to quantify the level of dfrA overexpression. Construction of

the strain H37Rv:dfrA-TetON used for underexpression in the absence of

tetracycline was described previously (Kumar et al., 2012).

Expression and Purification of MTB-DHFR

Competent E. coli BL21 (DE3) cells (Novagen) were transformed with recom-

binant plasmid containing dfrA gene and grown at 37�C to OD600 of 0.7 in

LB medium containing 70 mg/ml of kanamycin. The cells were cooled to

20�C and equilibrated for an hour, and protein expression was induced by

the addition of 0.8 mM IPTG. After an overnight induction, the cells were

spun down at 4,000 rpm. The pellet was resuspended in the buffer (20mM trie-

thanolamine [TEA], 30 mM imidazole, 50 mMKCl [pH 7.8]) containing protease

inhibitor cocktail (Novagen), 20mg of egg white lysozyme (Sigma), and 5 of mg

bovine pancreas DNase I. The resuspended cells were disrupted by cell lysis

machine and were spun down at 16,000 rpm for 50 min to remove the cell

debris. The supernatant, filtered through 0.22 mm, was applied to a 15 ml

column of high performance HisTrap column (GE Healthcare) pre-equilibrated

by the loading buffer (20mMTEA, 30mM imidazole, 300mMKCl [pH 7.8]). The

cell-loaded column was washed with extensive amounts of the loading buffer

and then eluted with a linear gradient (350 ml) from 0% to 100% of elution

buffer (20 mM TEA, 500 mM imidazole, 300 mM KCl [pH 7.8]). The pure

DHFR-containing fractions were pooled and concentrated down to 4 ml and

treated with 30 ml of thrombin (Novagen) at 16�C for 3 days. The cleaved

protein was passed through a 5 ml HisTrap HP column (GE Healthcare) pre-

equilibrated with the loading buffer. The unbound protein was collected,

dialyzed against 4 l of dialysis buffer (25 mM potassium phosphate, 50 mM

KCl, 0.1 mM EDTA, 5% glycerol [pH 7.2]) for 6 hr, and concentrated down to

15 mg/ml.

In Vitro Assay for MTB-DHFR

The enzyme assay for Mtb DHFR was performed in 100 mM HEPES, 50 mM

KCl (pH 7.0) at 25�C. The absorbance decrease at 340 nm, representing the

14 ª2014 Elsevier Ltd All rights reserved 9

Chemistry & Biology



consumption of NAPDH, was monitored with a Cary 50 spectrophotometer.

The inhibitors at various concentrations (20 nM of DHFR and 40 mMof NADPH)

were added in a 1 ml cuvette, and the reaction was initiated by the addition of

35 mM of dihydrofolate. The calculated IC50 values for trimethoprim and meth-

otrexate are consistent with those previously reported in the literature (White

et al., 2004).

IC50 Determination

20 nM of DHFR was incubated with 40 mM of cofactor NADPH and seven or

eight different concentrations of the inhibitor (where applicable) for 1 min.

The reaction was initiated by the addition of 35 mM dihydrofolate. The reaction

progress was measured for 2 min, and the linear region was used for

measuring the initial velocity parameters. Percent inhibition values from

different concentration points were acquired and analyzed by a curve-fitting

program supported by Collaborative Drug Discovery.

Preparation of Test Compounds

WR99210 was kindly provided by Jacobus Pharmaceuticals. Synthesized

compounds are designated as JSF-####. Trimetrexate, trimethoprim, metho-

trexate, piritrexim, sulfamethoxazole, and dapsone were purchased from

commercial sources. All compounds were dissolved in DMSO. All drug ali-

quots were stored at �20�C until immediately before use.

Assessment of Antitubercular Activity

MTB cultures were grown to mid-log phase and then diluted to a calculated

starting OD600 of 0.001 and aliquoted into 96-well plates. Test compounds in

DMSO were added at 1% of total volume. A 1:100 dilution was used to deter-

mine the MIC cutoff for each assay. Plates were incubated for 7 days at 37�C.The BacTiter-Glo Microbial Cell Viability Kit (Promega) was used to measure

intracellular ATP levels as a proxy for cell survival and growth. Cultures were

incubated with reconstituted, room temperature BacTiter-Glo Reagent for

10–20 min prior to luminescence quantification by a FLUOStar Omega plate

reader. MICs were confirmed with an MTB H37Rv strain that constitutively ex-

presses GFP. Growth was monitored by quantitating fluorescence (excitation:

485 nm; emission: 520 nm) using the FLUOStar Omega plate reader.

Metabolomic Sample Preparation

For metabolomic analysis samples were prepared according to previously

published procedures (de Carvalho et al., 2010a). In short, MTB was grown

to mid-log phase and then diluted to an OD600 of 0.1. A 1 ml culture was

then inoculated onto 22 mm 0.2 mm nitrocellulose filters (Millipore) using

vacuum filtration, placed on 7H10+ADN plates, and incubated at 37�C. On

day 5 postinoculation, theMTB-laden filters were transferred to plates contain-

ing appropriate concentrations of drug. At appropriate times, cell metabolism

was quenched by freezing MTB-laden filters in �40�C acetonitrile:metha-

nol:H2O (40:40:20). Cells were lysed by bead beating, with cooling on ice

between steps, and the extracts purified by centrifugation and 0.22 mm filtra-

tion. Each technical replicate combined the biomass from eight nitrocellulose

filters, and an experiment included three technical replicates for every time

point. Each experiment was performed twice.

Liquid Chromatography-Mass Spectrometry

A Cogent Diamond Hydride Type C column (Gradient 3) was used to separate

metabolites as described previously (Pesek et al., 2009). An Agilent Accurate

Mass 6220 TOFwas coupled to an Agilent 1200 liquid chromatography system

for analysis. This configuration achieves mass errors of approximately 5 ppm,

mass resolution ranging from 10,000 to 25,000 (over m/z 121–955), and a 5

log10 dynamic range. Metabolite identities were queried using a mass

tolerance of <0.005 Da. Signal intensities were normalized within experiments

using acetyldiaminopimelate, acetylaminoadipate, indolepyruvate, aminooxo-

pimelate, and triose phosphate, 5 of the 74metabolites whose abundance was

unaffected by WR99210. Putative metabolite identities were based on accu-

rate mass (m/z) and chromatographic retention time matching. Metabolite

identities were verified by coelution with a chemical standard and metabolite

concentrations were calculated using the method of standard addition, where

available.

10 Chemistry & Biology 21, 1–12, July 17, 2014 ª2014 Elsevier Ltd A

RNA Preparation

RNA was isolated as described previously (Rustad et al., 2008). Briefly, pellets

were condensed from 40 ml of culture (OD600 0.1) and then resuspended in

TRIzol. The cells were lysed by bead beating, with cooling on ice between

steps. The cell lysates were then centrifuged, and the supernatant was trans-

ferred to a tube with Heavy Phase Lock Gel (Eppendorf North America) and

chloroform, inverted for 2 min, and then centrifuged at max speed for 5 min.

The aqueous phase was precipitated with isopranol and high salt solution

(0.8 M sodium citrate, 1.2 M NaCl). RNA was purified using an RNeasy kit

(Qiagen).

Microarray Analysis

RNA was converted to CyDye-labeled cDNA probes as described previously

(Rustad et al., 2008). For all experiments described here, 2 mg of total RNA

was used to generate probes. Sets of fluorescent probes were then hybridized

as previously described (Minch et al., 2012). Briefly, the arrays were custom

NimbleGen tiling arrays consisting of 135,000 probes spaced at�100 bp inter-

vals around the MTB H37Rv genome (NCBI GEO accession number

GPL14896). Three biological replicate experiments were used for each drug

at each time point. Arrays were scanned, and spots were quantified using a

Genepix 4000B scanner with GenePix 6.0 software. These data were exported

to NimbleScan for mask alignment and ArrayStar for robust multichip average

(Bolstad et al., 2003) normalization and statistical analysis (NCBI GEO acces-

sion number GSE55979). Altered gene expression was considered significant

if it produced amoderated t test p < 0.01 and a log2 changeR 1 after Benjamini

Hochberg multiple testing correction.

Supplementation Assays

Supplemented stockmedia wasmade by adding filter sterilizedmetabolites to

7H9 for a final concentration of 2 mM for each individual metabolite. MTB

cultures constitutively expressing GFP as a reporter for growth were grown

to mid-log phase in 7H9 and then diluted to a calculated starting OD600 of

0.001 in supplemented media and dispensed into 96-well plates. Test com-

pounds in DMSO were added at 1% of total volume. Cells diluted 1:100 at

day 0, without drug treatment, were used to determine the MIC cutoff for

each assay. Plates were incubated for 7 days at 37�C, with growth monitored

by fluorescence as described above. All results were confirmed by optical

density readings.

SUPPLEMENTAL INFORMATION

Supplemental Information includes two tables and can be found with this

article online at http://dx.doi.org/10.1016/j.chembiol.2014.04.009.

ACKNOWLEDGMENTS

D.R.S. acknowledges support from the Paul G. Allen Family Foundation and

NIH (U19 AI106761, and training grant T32AI055396 in bacterial pathogenesis

awarded to M.R.N.). J.S.F. acknowledges funding from UMDNJ–NJMS and

the Foundation of UMDNJ. K.Y.R. acknowledges funding from a Burroughs

Wellcome Career Award in the Biomedical Sciences and the Bill and Melinda

Gates TB Drug Accelerator Program (OPP1024050). J.C.S. acknowledges

support from the Welch Foundation (grant A-0015). We thank Jacobus Phar-

maceuticals for the gift of WR99210 compound, Sebastien Gagnieux for ac-

cess to clinical strains of MTB, Kyle Minch for help with figures, and members

of the Sherman Lab for technical assistance and valuable advice.

Received: November 19, 2012

Revised: April 11, 2014

Accepted: April 23, 2014

Published: June 19, 2014

REFERENCES

Ahmad, S.I., Kirk, S.H., and Eisenstark, A. (1998). Thymine metabolism and

thymineless death in prokaryotes and eukaryotes. Annu. Rev. Microbiol. 52,

591–625.

ll rights reserved


Chemistry & Biology



Amyes, S.G., and Smith, J.T. (1974). Trimethoprim action and its analogy with

thymine starvation. Antimicrob. Agents Chemother. 5, 169–178.

Babu, G.R., and Laxminarayan, R. (2012). The unsurprising story of MDR-TB

resistance in India. Tuberculosis (Edinb.) 92, 301–306.

Barr, J. (2011). A short history of dapsone, or an alternative model of drug

development. Pharm. Hist. 53, 123–147.

Bloom, B.R., andMurray, C.J. (1992). Tuberculosis: Commentary on a reemer-

gent killer. Science 257, 1055–1064.

Boissier, F., Bardou, F., Guillet, V., Uttenweiler-Joseph, S., Daffe, M.,

Quemard, A., and Mourey, L. (2006). Further insight into S-adenosylmethio-

nine-dependent methyltransferases: Structural characterization of Hma, an

enzyme essential for the biosynthesis of oxygenated mycolic acids in

Mycobacterium tuberculosis. J. Biol. Chem. 281, 4434–4445.

Bolstad, B.M., Irizarry, R.A., Astrand, M., and Speed, T.P. (2003). A compari-

son of normalization methods for high density oligonucleotide array data

based on variance and bias. Bioinformatics 19, 185–193.

Bushby, S.R. (1975). Synergy of trimethoprim-sulfamethoxazole. Can. Med.

Assoc. J. 112, 63–66.

Chakraborty, S., Gruber, T., Barry, C.E., 3rd, Boshoff, H.I., and Rhee, K.Y.

(2013). Para-aminosalicylic acid acts as an alternative substrate of folatemeta-

bolism in Mycobacterium tuberculosis. Science 339, 88–91.

Cole, S.T., Brosch, R., Parkhill, J., Garnier, T., Churcher, C., Harris, D., Gordon,

S.V., Eiglmeier, K., Gas, S., Barry, C.E., 3rd., et al. (1998). Deciphering the

biology ofMycobacterium tuberculosis from the complete genome sequence.

Nature 393, 537–544.

Daniel, L.J., Norris, L.C., Scott, M.L., and Heuser, G.F. (1947). Growth inhibi-

tion of bacteria by synthetic pterins: Studies with Streptococcus faecalis,

Lactobacillus casei, and Lactobacillus arabinosus. J. Biol. Chem. 169,

689–697.

Dawson, M.I., O’Krongly, D., Hobbs, P.D., Barrueco, J.R., and Sirotnak, F.M.

(1987). Synthesis of the 7-hydroxy metabolites of methotrexate and 10-ethyl-

10-deazaaminopterin. J. Pharm. Sci. 76, 635–638.

de Carvalho, L.P., Fischer, S.M., Marrero, J., Nathan, C., Ehrt, S., and Rhee,

K.Y. (2010a). Metabolomics of Mycobacterium tuberculosis reveals compart-

mentalized co-catabolism of carbon substrates. Chem. Biol. 17, 1122–1131.

de Carvalho, L.P., Zhao, H., Dickinson, C.E., Arango, N.M., Lima, C.D.,

Fischer, S.M., Ouerfelli, O., Nathan, C., and Rhee, K.Y. (2010b). Activity-based

metabolomic profiling of enzymatic function: Identification of Rv1248c as a

mycobacterial 2-hydroxy-3-oxoadipate synthase. Chem. Biol. 17, 323–332.

Dias, M.V., Tyrakis, P., Domingues, R.R., Paes Leme, A.F., and Blundell, T.L.

(2014). Mycobacterium tuberculosis dihydrofolate reductase reveals two

conformational states and a possible low affinity mechanism to antifolate

drugs. Structure 22, 94–103.

Elion, G.B., Singer, S., and Hitchings, G.H. (1954). Antagonists of nucleic acid

derivatives: VIII. Synergism in combinations of biochemically related antime-

tabolites. J. Biol. Chem. 208, 477–488.

Fivian-Hughes, A.S., Houghton, J., and Davis, E.O. (2012). Mycobacterium

tuberculosis thymidylate synthase gene thyX is essential and potentially

bifunctional, while thyA deletion confers resistance to p-aminosalicylic acid.

Microbiology 158, 308–318.

Gandhi, N.R., Moll, A., Sturm, A.W., Pawinski, R., Govender, T., Lalloo, U.,

Zeller, K., Andrews, J., and Friedland, G. (2006). Extensively drug-resistant

tuberculosis as a cause of death in patients co-infected with tuberculosis

and HIV in a rural area of South Africa. Lancet 368, 1575–1580.

George, K.M., Yuan, Y., Sherman, D.R., and Barry, C.E., 3rd. (1995). The

biosynthesis of cyclopropanated mycolic acids in Mycobacterium tubercu-

losis. Identification and functional analysis of CMAS-2. J. Biol. Chem. 270,

27292–27298.

Gerum, A.B., Ulmer, J.E., Jacobus, D.P., Jensen, N.P., Sherman, D.R., and

Sibley, C.H. (2002). Novel Saccharomyces cerevisiae screen identifies

WR99210 analogues that inhibit Mycobacterium tuberculosis dihydrofolate

reductase. Antimicrob. Agents Chemother. 46, 3362–3369.

Gough, A., and Kaufman, G. (2011). Pulmonary tuberculosis: Clinical features

and patient management. Nurs. Stand. 25, 48–56.

Chemistry & Biolog

Halliday, N.M., Hardie, K.R., Williams, P., Winzer, K., and Barrett, D.A. (2010).

Quantitative liquid chromatography-tandem mass spectrometry profiling of

activated methyl cycle metabolites involved in LuxS-dependent quorum

sensing in Escherichia coli. Anal. Biochem. 403, 20–29.

Hartman, P.G. (1993). Molecular aspects and mechanism of action of dihydro-

folate reductase inhibitors. J. Chemother. 5, 369–376.

Hirsch, J. (1942). The ‘‘Sulfanilamide Effect’’ of Substances Devoid of Sulfo

Groups. Science 96, 139–140.

Huennekens, F.M. (1996). In search of dihydrofolate reductase. Protein Sci. 5,

1201–1208.

Huet, G., Constant, P., Malaga, W., Laneelle, M.A., Kremer, K., van Soolingen,

D., Daffe, M., and Guilhot, C. (2009). A lipid profile typifies the Beijing strains of

Mycobacterium tuberculosis: Identification of a mutation responsible for a

modification of the structures of phthiocerol dimycocerosates and phenolic

glycolipids. J. Biol. Chem. 284, 27101–27113.

Ioerger, T.R., Feng, Y., Ganesula, K., Chen, X., Dobos, K.M., Fortune, S.,

Jacobs, W.R., Jr., Mizrahi, V., Parish, T., Rubin, E., et al. (2010). Variation

among genome sequences of H37Rv strains of Mycobacterium tuberculosis

from multiple laboratories. J. Bacteriol. 192, 3645–3653.

Karlson, A.G. (1963). The in vitro activity of 4,40-diaminodiphenyl sulfone

against various acid-fast microorganisms. Int. J. Lepr. 31, 183–187.

Kompis, I.M., Islam, K., and Then, R.L. (2005). DNA and RNA synthesis:

Antifolates. Chem. Rev. 105, 593–620.

Kumar, A., Saigal, K., Malhotra, K., Sinha, K.M., and Taneja, B. (2011).

Structural and functional characterization of Rv2966c protein reveals an

RsmD-like methyltransferase from Mycobacterium tuberculosis and the role

of its N-terminal domain in target recognition. J. Biol. Chem. 286, 19652–

19661.

Kumar, A., Zhang, M., Zhu, L., Liao, R.P., Mutai, C., Hafsat, S., Sherman, D.R.,

and Wang, M.W. (2012). High-throughput screening and sensitized bacteria

identify an M. tuberculosis dihydrofolate reductase inhibitor with whole cell

activity. PLoS One 7, e39961.

Kwon, Y.K., Lu, W., Melamud, E., Khanam, N., Bognar, A., and Rabinowitz,

J.D. (2008). A domino effect in antifolate drug action in Escherichia coli. Nat.

Chem. Biol. 4, 602–608.

Kwon, Y.K., Higgins, M.B., and Rabinowitz, J.D. (2010). Antifolate-induced

depletion of intracellular glycine and purines inhibits thymineless death in E.

coli. ACS Chem. Biol. 5, 787–795.

Lee, M.H., and Hatfull, G.F. (1993). Mycobacteriophage L5 integrase-medi-

ated site-specific integration in vitro. J. Bacteriol. 175, 6836–6841.

Li, R., Sirawaraporn, R., Chitnumsub, P., Sirawaraporn, W., Wooden, J.,

Athappilly, F., Turley, S., and Hol, W.G. (2000). Three-dimensional structure

of M. tuberculosis dihydrofolate reductase reveals opportunities for the design

of novel tuberculosis drugs. J. Mol. Biol. 295, 307–323.

Liani, E., Rothem, L., Bunni, M.A., Smith, C.A., Jansen, G., and Assaraf, Y.G.

(2003). Loss of folylpoly-gamma-glutamate synthetase activity is a dominant

mechanism of resistance to polyglutamylation-dependent novel antifolates

in multiple human leukemia sublines. Int. J. Cancer 103, 587–599.

Minch, K., Rustad, T., and Sherman, D.R. (2012).Mycobacterium tuberculosis

growth following aerobic expression of the DosR regulon. PLoS One 7,

e35935.

Nathan, C., Gold, B., Lin, G., Stegman, M., de Carvalho, L.P., Vandal, O.,

Venugopal, A., and Bryk, R. (2008). A philosophy of anti-infectives as a guide

in the search for new drugs for tuberculosis. Tuberculosis (Edinb.) 88 (Suppl 1 ),

S25–S33.

Nguyen, L., and Thompson, C.J. (2006). Foundations of antibiotic resistance in

bacterial physiology: The mycobacterial paradigm. Trends Microbiol. 14,

304–312.

Park, H.D., Guinn, K.M., Harrell, M.I., Liao, R., Voskuil, M.I., Tompa, M.,

Schoolnik, G.K., and Sherman, D.R. (2003). Rv3133c/dosR is a transcription

factor that mediates the hypoxic response of Mycobacterium tuberculosis.

Mol. Microbiol. 48, 833–843.

y 21, 1–12, July 17, 2014 ª2014 Elsevier Ltd All rights reserved 11

Chemistry & Biology



Parveen, N., and Cornell, K.A. (2011). Methylthioadenosine/S-adenosylhomo-

cysteine nucleosidase, a critical enzyme for bacterial metabolism. Mol.

Microbiol. 79, 7–20.

Pesek, J.J., Matyska, M.T., Hearn, M.T., and Boysen, R.I. (2009). Aqueous

normal-phase retention of nucleotides on silica hydride columns.

J. Chromatogr. A 1216, 1140–1146.

Piper, J.R., andMontgomery, J.A. (1977). Preparation of 6-(bromomethyl)-2,4-

pteridinediamine hydrobromide and its use in improved syntheses of metho-

trexate and related compounds. J. Org. Chem. 42, 208–211.

Plorde, J. (2004). Sherris Medical Microbiology. (New York: McGraw-Hill).

Potts, S.J., Edwards, D.J., and Hoffman, R. (2005). Challenges of target/com-

pound data integration from disease to chemistry: A case study of dihydrofo-

late reductase inhibitors. Curr. Drug Discov. Technol. 2, 75–87.

Rahman, A., Srivastava, S.S., Sneh, A., Ahmed, N., and Krishnasastry, M.V.

(2010). Molecular characterization of tlyA gene product, Rv1694 of

Mycobacterium tuberculosis: A non-conventional hemolysin and a ribosomal

RNA methyl transferase. BMC Biochem. 11, 35.

Reddy, M.C., Kuppan, G., Shetty, N.D., Owen, J.L., Ioerger, T.R., and

Sacchettini, J.C. (2008). Crystal structures of Mycobacterium tuberculosis

S-adenosyl-L-homocysteine hydrolase in ternary complex with substrate

and inhibitors. Protein Sci. 17, 2134–2144.

Rosowsky, A. (1973). Methotrexate analogs. 2. A facile method of preparation

of lipophilic derivatives of methotrexate and 30,50-dichloromethotrexate by

direct esterification. J. Med. Chem. 16, 1190–1193.

Rosowsky, A., Beardsley, G.P., Ensminger, W.D., Lazarus, H., and Yu, C.S.

(1978). Methotrexate analogues. 11. Unambiguous chemical synthesis and

in vitro biological evaluation of alpha- and gamma-monoesters as potential

prodrugs. J. Med. Chem. 21, 380–386.

Rustad, T.R., Harrell, M.I., Liao, R., and Sherman, D.R. (2008). The enduring

hypoxic response of Mycobacterium tuberculosis. PLoS One 3, e1502.

12 Chemistry & Biology 21, 1–12, July 17, 2014 ª2014 Elsevier Ltd A

Sangurdekar, D.P., Zhang, Z., and Khodursky, A.B. (2011). The association of

DNA damage response and nucleotide level modulation with the antibacterial

mechanism of the anti-folate drug trimethoprim. BMC Genomics 12, 583.

Velayati, A.A., Farnia, P., Masjedi, M.R., Ibrahim, T.A., Tabarsi, P., Haroun,

R.Z., Kuan, H.O., Ghanavi, J., Farnia, P., and Varahram, M. (2009). Totally

drug-resistant tuberculosis strains: evidence of adaptation at the cellular level.

Eur. Respir. J. 34, 1202–1203.

Vilcheze, C., and Jacobs, W.R., Jr. (2012). The combination of sulfamethoxa-

zole, trimethoprim, and isoniazid or rifampin is bactericidal and prevents the

emergence of drug resistance in Mycobacterium tuberculosis. Antimicrob.

Agents Chemother. 56, 5142–5148.

Wallace, R.J., Jr., Nash, D.R., Steele, L.C., and Steingrube, V. (1986).

Susceptibility testing of slowly growing mycobacteria by a microdilution MIC

method with 7H9 broth. J. Clin. Microbiol. 24, 976–981.

Wards, B.J., and Collins, D.M. (1996). Electroporation at elevated tempera-

tures substantially improves transformation efficiency of slow-growing myco-

bacteria. FEMS Microbiol. Lett. 145, 101–105.

White, E.L., Ross, L.J., Cunningham, A., and Escuyer, V. (2004). Cloning,

expression, and characterization ofMycobacterium tuberculosis dihydrofolate

reductase. FEMS Microbiol. Lett. 232, 101–105.

Williams, K.J., and Duncan, K. (2007). Current strategies for identifying and

validating targets for new treatment-shortening drugs for TB. Curr. Mol.

Med. 7, 297–307.

Yeo, A.E., Seymour, K.K., Rieckmann, K.H., and Christopherson, R.I. (1997).

Effects of dual combinations of antifolates with atovaquone or dapsone on

nucleotide levels in Plasmodium falciparum. Biochem. Pharmacol. 53,

943–950.

Zumla, A., George, A., Sharma, V., and Herbert, N.; Baroness Masham of Ilton

(2013). WHO’s 2013 global report on tuberculosis: Successes, threats, and

opportunities. Lancet 382, 1765–1767.

ll rights reserved

Date post:	27-Jan-2017
Category:	Documents
Upload:	davidr
View:	216 times
Download:	0 times

Folate Pathway Disruption Leads to Critical Disruption of Methionine Derivatives in Mycobacterium...

Documents