Reprogramming of proline and glutamine metabolismcontributes to the proliferative and metabolic responsesregulated by oncogenic transcription factor c-MYCWei Liua,1, Anne Leb, Chad Hancocka, Andrew N. Lanec,d, Chi V. Dange, Teresa W.-M. Fanf,d,1, and James M. Phanga,1

aMetabolism and Cancer Susceptibility Section, Basic Research Laboratory, National Cancer Institute, Frederick, MD 21702; bDepartment of Pathology, TheJohns Hopkins University School of Medicine, Baltimore, MD 21205; Departments of cMedicine and fChemistry, Center for Regulatory EnvironmentalAnalytical Metabolomics and dJames Graham Brown Cancer Center, University of Louisville, KY 40202; and eAbramson Cancer Center, Abramson FamilyCancer Research Institute, University of Pennsylvania, Philadelphia, PA 19104

Edited by Matt Vander Heiden, Massachusetts Institute of Technology, Cambridge, MA, and accepted by the Editorial Board April 23, 2012 (received forreview February 24, 2012)

In addition to glycolysis, the oncogenic transcription factor c-MYC(MYC) stimulates glutamine catabolism to fuel growth and pro-liferation of cancer cells through up-regulating glutaminase (GLS).Glutamine is converted to glutamate by GLS, entering the tri-carboxylic acid cycle as an important energy source. Less well-recognized, glutamate can also be converted to proline throughΔ1-pyrroline-5-carboxylate (P5C) and vice versa. This study sug-gests that some MYC-induced cellular effects are due to MYC reg-ulation of proline metabolism. Proline oxidase, also known asproline dehydrogenase (POX/PRODH), the first enzyme in prolinecatabolism, is a mitochondrial tumor suppressor that inhibits pro-liferation and induces apoptosis. MiR-23b* mediates POX/PRODHdown-regulation in human kidney tumors. MiR-23b* is processedfrom the same transcript as miR-23b; the latter inhibits the trans-lation of GLS. Using MYC-inducible human Burkitt lymphomamodel P493 and PC3 human prostate cancer cells, we showed thatMYC suppressed POX/PRODH expression primarily through up-regulating miR-23b*. The growth inhibition in the absence ofMYC was partially reversed by POX/PRODH knockdown, indicatingthe importance of suppression of POX/PRODH in MYC-mediatedcellular effects. Interestingly, MYC not only inhibited POX/PRODH,but also markedly increased the enzymes of proline biosynthesisfrom glutamine, including P5C synthase and P5C reductase 1. MYC-induced proline biosynthesis from glutamine was directly con-firmed using 13C,15N-glutamine as a tracer. The metabolic link be-tween glutamine and proline afforded by MYC emphasizes thecomplexity of tumor metabolism. Further studies of the relation-ship between glutamine and proline metabolism should providea deeper understanding of tumor metabolism while enabling thedevelopment of novel therapeutic strategies.

amino acids | redox signaling | reactive oxygen species | miRNA |metabolic tumor suppressor

Growing tumors alter their metabolic profiles to meet thebioenergetic and biosynthetic demands of increased cell

growth and proliferation (1–3). Many oncogenes and tumorsuppressors have been linked to tumor metabolic regulation(4, 5). Proline oxidase, also known as proline dehydrogenase(POX/PRODH), as a mitochondrial inner membrane enzyme isinvolved in the first step of proline catabolism and has beenidentified as one of a few mitochondrial tumor suppressors (6–10). The gene encoding POX/PRODH also known as PRODHwas initially identified as a p53-induced gene in a screening study(8). Here, we will refer to this enzyme as POX/PRODH and tothe gene as PRODH. Intensive investigation has led to the rec-ognition of important functions of proline metabolism in humantumors. Earlier work in our laboratory has demonstrated theimportant roles of POX/PRODH in the apoptosis induced bycytotoxic agents and by peroxisome proliferator activated re-ceptor γ (PPARγ) and its ligands (6, 10). Hyperexpression ofPOX/PRODH in cancer cells is sufficient to initiate apoptosis

because of its ability to generate reactive oxygen species (ROS)(6, 7). In addition to inducing apoptotic cell death, POX/PRODH negatively regulates the growth of various cancer cells,causes cell cycle arrest at the G2-M checkpoint, and inhibitstumor formation in a mouse xenograft model (7, 9). Sub-sequently, the absence or reduction of POX/PRODH was ob-served in a variety of human tumor tissues compared with theirnormal tissue counterparts (9, 11), and miR-23b* was found tobe one of the mechanisms mediating POX/PRODH down-reg-ulation in renal tumors (11).The MYC oncogene is frequently dysregulated in human can-

cers. It encodes a transcription factor, c-MYC (herein termedMYC), which links altered cellular metabolism to carcinogenesis.In addition to its known function in regulating glucosemetabolism(12), MYC recently has been documented to induce the expres-sion of mitochondrial glutaminase (GLS) to stimulate glutaminecatabolism (13, 14). The importance of glutamine catabolism incancer cell metabolism was reemphasized by these findings. Inaddition to its use for synthesis of proteins, nucleotides, and lipids(15, 16), glutamine is converted to glutamate by GLS. Glutamatenot only is an essential component of glutathione, but also animportant energy source via anaplerotic input into the tri-carboxylic acid (TCA) cycle after conversion to α-ketoglutarate(α-KG). Less well-recognized, glutamate can also be converted toproline through Δ1-pyrroline-5-carboxylate (P5C) catalyzed se-quentially by P5C synthase (P5CS) and P5C reductase (PYCR)(Fig. 1A). Conversely, proline can be converted to glutamatethrough proline catabolism sequentially catalyzed by POX/PRODH and P5C dehydrogenase (P5CDH). The proline-derivedglutamate can be further converted to α-KG, or used to generateglutamine through glutamine synthetase (GS). Thus, we won-dered whether oncogenic MYC could affect proline metabolismand, specifically, modulate proline catabolism by the tumor sup-pressor POX/PRODH. In this study, we investigated the effectsof MYC on proline catabolism and proline biosynthesis fromglutamine, especially its effect on the expression of POX/PRODHand its regulatory mechanisms. The contribution of POX/PRODHto MYC-mediated tumor cell behavior was further established.

Author contributions: W.L., A.L., C.H., A.N.L., C.V.D., T.W.-M.F., and J.M.P. designed re-search; W.L., A.L., C.H., A.N.L., and T.W.-M.F. performed research; W.L., C.H., A.N.L., C.V.D.,T.W.-M.F., and J.M.P. analyzed data; andW.L., A.N.L., T.W.-M.F., and J.M.P. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission. M.V.H. is a guest editor invited by the EditorialBoard.

Freely available online through the PNAS open access option.1To whom correspondence may be addressed. E-mail: phangj@mail.nih.gov, liuwei7997@gmail.com, or twmfan@gmail.com.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1203244109/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1203244109 PNAS | June 5, 2012 | vol. 109 | no. 23 | 8983–8988

CELL

BIOLO

GY

ResultsMYC Robustly Suppresses the Expression of POX/PRODH Protein. Toinvestigate whether MYC plays a role in the regulation of POX/PRODH expression in human cancer, we used human Burkittlymphoma model P493 cells that bear a tetracycline-repressibleMYC construct. In the absence of tetracycline, ectopic MYC isinduced in a tumorigenic state that resembles human Burkittlymphoma. We analyzed the changes of POX/PRODH proteinand mRNA expression in response to MYC. As shown in Fig. 1B,POX/PRODH protein increased in a time-dependent fashion andreached approximately 3.8-fold and 7.5- fold when P493 cells weretreated with tetracycline for 24 h and 120 h, respectively. WhenMYC expression was recovered by tetracycline removal, the POX/PRODH protein diminished markedly (Fig. 1D). However, POX/PRODHmRNAdid not show any obvious increase until 72 h aftertetracycline treatment (∼1.7-fold), and even at 120 h, it increasedonly ∼4.7-fold (Fig. 1 C and E). Thus, the changes of POX/PRODH mRNA levels were delayed and plateaued at a lowerlevel compared with changes in POX/PRODH protein levels.Because overexpression of MYC was also found in human

prostate cancer, we tested whether MYC had the same effects onthe expression of POX/PRODH in the PC3 human prostatecancer cells. MYC knockdown in PC3 cells by short interferingRNA (siRNA) resulted in the inhibition of POX/PRODH ex-pression with a pattern similar to the P493 cells (Fig. S1A). WhenMYC was knocked down 85%, POX/PRODH protein expressionincreased ∼4.2-fold, whereas POX/PRODH mRNA levels onlyincreased ∼1.6-fold (Fig. S1 A and B).

POX/PRODH Suppression Is Essential for MYC-Mediated Cancer CellProliferation and Survival. As mentioned above, POX/PRODHhas been identified as a tumor suppressor, whose expression issuppressed in a variety of tumors; it induces apoptosis throughthe generation of ROS and inhibits tumor growth in a xenograftmouse model. Whether the suppression of POX/PRODH by

MYC is functionally linked to MYC-induced proliferation andcell survival became an interesting question. We knocked downthe expression of POX/PRODH by siRNA in P493 cells withMYC suppressed by tetracycline. Western blot confirmed theknockdown of POX/PRODH (Fig. 2A). In Fig. 2B, at differenttime points of tetracycline treatment, POX/PRODH siRNAconsistently reduced the production of ROS, although the sup-pression of MYC by tetracycline decreased the accumulation ofROS at day 4 (-Tet + siNeg vs. +Tet + siNeg as shown in Fig.2B), which may reflect the different effects of various MYCregulated genes on ROS production at various stages (17–19).Correspondingly, the apoptosis assay by flow cytometry showedthat knockdown of POX/PRODH by siRNA decreased thepercentage of apoptotic and dead cells occurring with MYCsuppression (Fig. 2C). In contrast, the number of living cellssuggested that POX/PRODH siRNA could significantly rescue30∼40% of the diminished growth rates, which were 46% and82% at day 2 and day 4, respectively, resulting from MYC sup-pression by tetracycline (Fig. 2D). These results indicated thatPOX/PRODH suppression participates in MYC-mediated can-cer cell proliferation and survival.To extend the above conclusion from the tetracycline-con-

trolled cells to cancer cells overexpressing MYC, we performedthe same assays in PC3 prostate cancer cells. We first confirmedthe decreased expression of MYC and POX/PRODH by theirrespective siRNAs using Western blots (Fig. S2A). As shown inFig. S2B, knockdown of MYC by siRNA first resulted in thedecrease of ROS production at 3 d followed by a small increaseat 6 d. When POX/PRODH siRNA was used to reduce theincreased expression of POX/PRODH resulting from MYCsiRNA, the production of ROS was decreased at both timepoints. The reduction of MYC by MYC siRNA reduced cellgrowth 53.0% and 68.6% at 3 and 6 d, respectively, which couldbe recovered 19.5% and 71.6% by POX/PRODH siRNA (Fig.S2C). Apoptosis assay showed that the percentage of apoptotic

0 24 48 72 96 120

0.0

0.2

0.4

0.6

0.8

POX/PRODH

MYC

Time (h)

+T

Pro

tie

n e

xp

re

ss

ion

(R

ela

tiv

e o

ptic

al d

en

sit

y)

0

1

2

3

4

72 72 48 72 120 120 48 72 Time (h)

-T +T Wash -T +T Wash

p<0.05

p<0.001

p<0.05

p<0.05p<0.05

PO

X/P

RO

DH

mR

NA

re

lativ

e e

xp

re

ss

ion

0 24 48 72 96 120

0

1

2

3

4

5

6

7

*

** **

Time (h)

+T

PO

X/P

RO

DH

mR

NA

re

lativ

e e

xp

re

ss

ion

B

C

D

E

GAPDH

POX/PRODHMYC

0 24 48 72 96 120 Time (h)

+T

A

POX/PRODH

MYC

GAPDH

-T -T +T Wash -T +T Wash 72 72 72 48 72 120 120 48 72 Time (h)

0.0

0.2

0.4

0.6

0.8

1.0

POX/PRODH

MYC

72 72 48 72 120 120 48 72 Time (h)

-T +T Wash -T +T Wash

Pro

te

in e

xp

re

ss

ion

(R

ela

tiv

e o

ptic

al d

en

sit

y)

Fig. 1. MYC robustly suppresses the expres-sion of POX/PRODH protein. (A) Scheme ofproline and glutamine interconversion. GLS,glutaminase; GS, glutamine synthetase; GSA,glutamic-γ-semialdehyde; α-KG, α-ketogluta-rate; P5C, Δ1-pyyroline-5-carboxylate; P5CDH,P5C dehydrogenase; P5CS, P5C synthase;POX/PRODH, proline oxidase/dehydrogenase;PYCR1, P5C reductase 1; TCA cycle, tricar-boxylic acid cycle. (B and D Upper) Westernblots for POX/PRODH and MYC in P493 cellswith or without tetracycline (T) treatment,using GAPDH as a loading control. (B and DLower) Densitometry analysis shows the banddensity ratios of MYC and POX/PRODH toloading control. Data shown represent one ofthree independent experiments. (C and E)POX/PRODH mRNA levels were measured byreal-time RT-PCR using GAPDH as an internalcontrol. The relative folds were calculated tothe group without tetracycline treatment. Theresults shown are mean ± SEM, n = 3. P valueswere obtained by one-way analysis of vari-ance. *P < 0.05 and **P < 0.001 comparedwith 0 h control.

8984 | www.pnas.org/cgi/doi/10.1073/pnas.1203244109 Liu et al.

cells increased threefold with MYC siRNA, which was decreased∼40% by POX/PRODH knockdown (Fig. S2D).The mechanism about ROS production by POX/PRODH

through complex III of the electron transport chain, and thecontribution of ROS generated by POX/PRODH to reducedgrowth resulting from MYC suppression are described in SIResults and Discussion and shown in Fig. S3.

MYC Indirectly Suppresses POX/PRODH Expression at the TranscriptionalLevel. As described above, MYC decreased POX/PRODHmRNA expression although it was not comparable with its in-hibition of POX/PRODH protein levels. To further confirm itsregulation of POX/PRODH transcription, we tested the effect ofMYC on PRODH promoter activity in PC3 prostate cancer cellsby transfecting the PRODH promoter/luciferase reporter con-struct containing PRODH promoter region (10). As shown inFig. 3A, knockdown of MYC resulted in the increase of PRODHpromoter activity, indicating MYC regulates POX/PRODH atthe transcriptional level. However, it is unknown whether MYCacts as a transcription factor by binding directly to the PRODHpromoter area. Analysis of PRODH promoter nucleotide se-quence revealed one canonical MYC binding site 5′-CACGTG-3′ (E-box) and one noncanonical binding site (5′-ACGGTG-3′)at −2808 to −2813 bp and −637 to −642 bp of the PRODH pro-moter region, respectively. To investigate whether MYC binds

directly to the PRODH gene, we performed the chromatin im-munoprecipitation (ChIP) assay in P493 cells, using cyclin-dependent kinase 4 (CDK4), which has been reported to beup-regulated by MYC (20), as a positive control. None of thePRODH promoter regions containing either canonical or non-canonical MYC binding sites showed significant PCR amplifi-cation (Fig. 3B), demonstrating that MYC does not directlyinteract with the PRODH gene, and the decreased POX/PRODH mRNA expression may be mediated through othertranscription factors. This result is consistent with the delayedchanges of POX/PRODH mRNA levels regulated by MYC asshowed in Fig. 1.

MYC Suppresses POX/PRODH Protein Expression Primarily ThroughIncreasing miR-23b*. The above data showed that POX/PRODHprotein, but not mRNA, robustly responds to the alteration ofMYC levels in P493 and PC3 cells. Thus, we sought regulatoryeffects of MYC on POX/PRODH at the posttranscriptionallevel. Earlier study showed that MYC suppresses the expressionof miR-23b in P493 and PC3 cells (14). Our previously publishedstudy indicated that miR-23b* mediates the loss of POX/PRODHprotein in renal cancers (11). Because miR-23b and miR-23b*originate from the same transcript, we first determined the effects

-T+siN

eg

-T+siPOX

+T+s

iNeg

+T+s

iPOX0

5

10

15

20

25 p<0.01p<0.001

% A

nnex

in V

+ c

ells

-T+siNeg -T+siPOX

+T+siNeg +T+siPOX

2ds 4ds0.0

0.5

1.0

1.5

2.0+T+siNeg

p<0.05p<0.01

-T+siNeg

+T+siPOX

p<0.05-T+siPOX

RO

S(R

elat

ive

Fluo

resc

ence

)

0d 2ds 4ds0

2

4

6

+T+siPOX+T+siNeg

p<0.001p<0.05

-T+siNeg-T+siPOX

Rel

ativ

e ce

ll nu

mbe

r

A B

C

D

POXMYCGAPDH

Fig. 2. POX/PRODH suppression is necessary for MYC-mediated cancer cellproliferation and survival in P493 cells. P493 cells were firstly transfectedwith siRNA against POX/PRODH (designated as siPOX) or negative controlsiRNA (siNeg) for 24 h, then treated with tetracycline (Tet) for 2 or 4 d. (A)The knockdown of POX/PRODH was confirmed by Western blot. (B) ROSproduction was performed by dichlorofluorescein (DCF) assay. (C) After 4 dof tetracycline treatment, apoptosis in the cells was monitored by Annexin V-FITC and propidium iodide (PI) staining. The results are representative oftwo separate experiments in triplicate. The number in the bottom rightcorner shows the percentage of Annexin V-positive cells, expressed as mean ±SEM. (D) The relative living cell number was determined by trypan blueexclusion assay. The results are shown as mean ± SEM (n = 3). All P valueswere obtained by the Student t test.

72 h 96 h0.0

0.5

1.0

1.5

2.0

siNeg + PRODH-LucsiMyc + PRODH-Luc

p<0.001 p<0.01

Rela

tive

luci

fera

seac

tivity

A

B

PRODH promoter-1040 to -812

(229bp)

Positive control:CDK4 (110bp)

PRODH promoter-2858 to -2783

(76bp)

-T +T

Fig. 3. MYC indirectly suppresses POX/PRODH expression at the transcrip-tional level. (A) PC3 cells were cotransfected with luciferase reporter PRODH-Luc containing PRODH promoter and siMYC or siNeg, using pRL-null renillaluciferase reporter as an internal control for normalizing transfection effi-ciency. PRODH promoter luciferase activity was determined by using thedual luciferase assay kit. Data shown are mean ± SEM (n = 3). P values wereobtained by the Student t test. (B) Chromatin immunoprecipitation assay ofthe PRODH promoter in P493 cells treated with or without tetracycline.Soluble chromatin was immunoprecipitated by using anti-MYC antibody orno antibody as control. A portion of the sonicated chromatin was used asDNA input control. A known target gene of MYC, cyclin-dependent kinase 4(CDK4), was used as a positive control. Immunoprecipitates were analyzed byPCR with specific primers for the PRODH and CDK4 gene areas containingcanonical or noncanonical MYC binding sequence (E-box).

Liu et al. PNAS | June 5, 2012 | vol. 109 | no. 23 | 8985

CELL

BIOLO

GY

of MYC on miR-23b* and then tested whether the suppressionof POX/PRODH by MYC was related to its effects on miR-23b*in P493 cells. In Fig. 4 A and B, real-time PCR assays showedthat miR-23b* levels decreased with diminished MYC expres-sion and then increased on MYC reinduction in a mannercompatible with the POX/PRODH protein levels observed inFig. 1 B and D. PC3 prostate cancer cells also showed the samerelationship between MYC and the levels of miR-23b*, i.e., MYCknockdown by siRNA resulted in the decrease of miR-23b* ex-pression (Fig. S4A). These results were distinct from the reporteddecrease of miR-23b by MYC. The nonparallel expression of thesibling miRNAs regulated by MYC implied that MYC may dif-ferentially affect the processing of miRNA, including their sta-bilization and degradation. SI Results and Discussion and Fig. S5provide the preliminary data that showed that MYC regulatedthe differential expression of miR-23b* and miR-23b partiallythrough up-regulating Agonaute 2 protein (Ago 2), a key playerin miRNAs stability and degradation.To assess whether the suppression of POX/PRODH by MYC

relates to the increase of miR-23b*, we transfected P493 cellswith antagomirs against miR-23b* to inhibit the high expressionof miR-23b* induced by MYC overexpression in the absence oftetracycline. As shown in Fig. 4C, POX/PRODH protein levelincreased 1.5-fold after miR-23b* was inhibited by antagomirs.Additionally, we transfected P493 cells with mimic miR-23b*under MYC inhibition by tetracycline. Ectopic miR-23b* ex-pression at high levels was confirmed by real-time PCR after thetransfection. As expected, mimic miR-23b* resulted in a markeddecrease in POX/PRODH protein (Fig. 4D). However, the de-crease of POX/PRODH still was not comparable with thatwithout tetracycline treatment, indicating that MYC could sup-press POX/PRODH expression through pathways other than

miRNA, such as the indirect regulation at the transcriptionallevel as shown above.We further confirmed the suppression of POX/PRODH by

MYC through miR-23b* in PC3 cells by transfecting the lucif-erase reporter with the POX/PRODH mRNA 3′ UTR sequencecontaining the miR-23b* binding site, designated as pPOX 3′UTR (11). pMIR-Report, the original reporter without thePOX/PRODH mRNA 3′ UTR was used as a control. NegativesiRNAs (siNeg) or MYC siRNAs (siMYC) were cotransfectedinto PC3 cells. As shown in Fig. S4B, MYC knockdown by siR-NAs increased luciferase activity of pPOX 3′ UTR significantly,indicating the decrease of miR-23b* by siMYC. As expected,without MYC knockdown, the luciferase activity of pPOX 3′UTR was much lower than that of the original pMIR-Reportbecause of high levels of miR-23b* binding to POX/PRODHmRNA 3′ UTR, thereby suppressing luciferase expression.

MYC Markedly Increases the Biosynthesis of Proline from Glutamine.MYC has been shown to enhance GLS expression and glutaminecatabolism (13, 14). Because glutamine and proline are in-terconvertible (Fig. 1A), we investigated the MYC regulation ofglutamine and proline metabolism by studying the enzymes ofglutamine-proline metabolic pathway in P493 cells. As shown inFig. 5A, MYC robustly increased the expression of GLS, P5CS,and PYCR1 in the pathway from glutamine to proline and de-creased the expression of POX/PRODH, P5CDH, and GS in thepathway from proline to glutamine. PC3 human prostate cancercells displayed the same correlation between MYC and gluta-mine and proline metabolism. The reduction of MYC by MYCsiRNA resulted in the increase of POX/PRODH, P5CDH, andGS and decrease of GLS, P5CS, and PYCR1 (Fig. 5B). We thenmeasured the intracellular levels of proline in P493 MYC-Onand MYC-Off cells. As expected, MYC dramatically increasedthe intracellular levels of proline (Fig. 5C).To confirm directly the production of proline from glutamine

induced by MYC, we traced the conversion of glutamine toproline by gas chromatography–mass spectrometry (GC-MS),Fourier transform-ion cyclotron resonance MS (FT-ICR-MS),and NMR using [U-13C,15N]-glutamine (Gln) as a tracer in P493cells. Glutamine was labeled at all five carbon and both nitrogenatoms, which were incorporated into newly synthesized proline.Fig. 5D shows the expected labeling patterns from [U-13C,15N]-Gln, including the intermediates of the TCA cycle and proline(Pro) via glutamate (Glu). As described in detail (21), MYCinduced glutamine oxidation via the TCA cycle such that all ofthe intermediates of the TCA cycle derived from glutamine wereincreased by MYC, including α-KG, succinate, fumarate, malate,and citrate. To confirm the previous results, the GC-MS analysisof 13C,15N-Gln contribution to citrate in P493 cells with MYC-On and MYC-Off is shown in Fig. 5E. The production of 13C5-citrate (m+5) from Gln tracer can be explained by the reductivecarboxylation of α-KG, which entails a reversal of the citrate toα-KG reaction catalyzed by aconitase and isocitrate de-hydrogenase (IDH) as recently reported in other cells (22–25).However, this process is unlikely to be the major pathway in P493cells under normoxia, as discussed (21). This finding is alsoconsistent with the recent reports that showed the glutamine-dependent reductive carboxylation is most prevalent either underhypoxia or in cells with a TCA cycle deficiency (23–25).Fig. 5F shows the GC-MS analysis of 13C,15N-Gln contribution

to proline synthesis with or without tetracycline treatment. m+1to m+6 represent incremental increases of neutron massesresulting from one to five 13C and zero or one 15N incorporationinto proline. Thus, the m+6 labeled species of proline (13C5,

15N-Pro) is the direct product from 13C5,

15N2-Gln catabolism,whereas the rest of the labeled species can be derived from theparent glutamine tracer via metabolic scrambling through theTCA cycle, transamination (TA), and/or glutamate dehydrogenase(GDH) activity (Fig. 5D). The increased expression of MYC withtetracycline withdrawal markedly and consistently increased thelevels of m+1 to m+6 isotopologues of proline, although the

0 24 48 72 96 1200.0

0.5

1.0

1.5

** **** ** **

Time (h)+T

miR

-23b

* rel

ativ

eex

pres

sion

0.0

0.5

1.0

1.5

72 72 48 72 Time (h) -T +T Wash

p<0.001

p<0.001p<0.001

miR

-23b

* rel

ativ

eex

pres

sion

-T+Neg RNA -T+Anti-23b*0.0

0.5

1.0

1.5

2.0

POX/

PRO

DH

pro

tein

(Rel

ativ

e op

tical

den

sity

) GAPDH

POX/PRODH

MYC

-T

-T +T

+T+Mimic 23b*

+T+Neg RNA0.0

0.2

0.4

0.6

0.8

POX/

PRO

DH

prot

ein

(Rel

ativ

e op

tical

den

sity

) GAPDH

POX/PRODH

MYC

- + + + T- - + - Mimic 23b*- - - + Neg RNA

A

C

B

D

Fig. 4. MYC suppresses POX/PRODH protein expression primarily throughincreasing miR-23b*. (A and B) The expression of miR-23b* was monitoredby real-time RT-PCR in P493 cells treated with or without tetracycline (T). U6was used as an internal control. Results were determined in triplicate andrepeated in two independent experiments. Data are shown as mean ± SEM.P values were obtained by one-way analysis of variance. **P < 0.001 com-pared with 0 h control in A. (C) P493 cells were transfected with miR-23b*antagomir (Anti-23b*) to inhibit the expression of miR-23b*, and scrambledRNA was used as negative control (Neg RNA). (D) P493 cells were transfectedwith mimic miR-23b* to enhance the expression of miR-23b* when tetra-cycline was added to the medium to inhibit MYC expression. (C and D Upper)POX/PRODH protein was detected by Western blot. (C and D Lower) Thenormalization of POX/PRODH protein relative to loading control bydensitometry analysis. Data shown represent one of three independentexperiments.

8986 | www.pnas.org/cgi/doi/10.1073/pnas.1203244109 Liu et al.

ratios of concentrations for the various labeled isotopologues ofproline were not constant. The nonconstant ratios of prolineisotopologues between MYC On and MYC Off are likely due tothe following reasons. As stated above, the various labeled spe-cies of proline except for m+6 can be derived from the variouscarbon and nitrogen scrambling processes through the TCA cy-cle, transamination, and/or GDH activity, as shown in Fig. 5Dafter three turns. Scrambling of carbon detected in proline viathe TCA cycle is consistent with the production of labeled citratewith different scrambled labeling patterns (e.g., m+2 and m+4isotopologues in Fig. 5E). The m+2 and m+4 isotopologues ofcitrate are the respective products of three and one turn of theKrebs cycling (Fig. 5D), whereas the m+5 (13C5-citrate) and m+6(13C6-citrate) can be contributed from pyruvate carboxylationand TCA cycle-independent ATP citrate lyase plus malic enzymereaction sequence as shown in Le et al. (21). MYC suppressionmay have differential effects on these processes. Moreover, theabundant m+5 isotopologue of Pro implies the replacement of15N by 14N as all five carbons should be labeled because of thehigh abundance of its precursor 13C5-Glu (21). The FT-ICR-MSanalysis in Fig. S6 showed the fractional distribution of Glu,aspartate (Asp) and proline isotopologues. Consistent with thatof proline, 13C and 15N isotopologue distributions of Glu andAsp were differentially modulated by MYC expression.In addition, the NMR spectral analysis in Fig. S7 showed the

extensive 13C enrichment in proline from glutamine induced byMYC, which further confirmed the GC-MS data. These findingsare consistent with the marked increases in the enzymes medi-ating proline synthesis from glutamate. Thus, MYC not onlystimulates the conversion of glutamine to glutamate, but alsomarkedly enhances subsequent conversion of glutamate to pro-line via proline biosynthetic enzymes.

DiscussionAs the only proteinogenic secondary amino acid, proline is me-tabolized by its own family of enzymes that respond to variousstresses and participate in redox regulation and metabolic sig-naling. Recent studies defining the regulation of this systemsuggest that proline is a “stress substrate” and proline metabo-lism may be a potential antitumor target. POX/PRODH, the firstenzyme in proline catabolism, is induced by genotoxic (p53) (8),inflammatory (PPARγ and its ligands) (10), and nutrient stress(glucose deprivation) (26). Proline catabolism catalyzed by POX/PRODH generates electrons to produce ROS and initiatesa variety of downstream effects, including blockage of the cellcycle and initiation of apoptosis. In this sense, POX/PRODHfunctions as a metabolic tumor suppressor, which is supported byits low expression or loss in tumors and the inhibition of tumorformation in a mouse xenograft tumor model by its ectopic ex-pression. In this work, we showed that oncogenic transcriptionfactor MYC inhibits POX/PRODH expression and, thereby,inhibits its tumor suppressor function. When MYC is suppressed,the increase of POX/PRODH induces ROS generation and ap-optosis, leading to decreased cell proliferation and growth.These results suggest that MYC-induced suppression of POX/PRODH contributes to MYC-mediated changes of cell behaviorincluding proliferation and metabolic reprogramming that, inturn, contributes to tumorigenesis and tumor progression.Transformed cells from different origins typically up-regulate

both glucose and glutamine consumption as sources of metabolicenergy and as precursors for biosynthesis of macromolecules (13,27). The MYC oncogene, which plays a critical role in manyhuman cancers, is considered a master regulator of cell metab-olism and proliferation. It not only promotes glucose uptake andinduces aerobic glycolysis, but also enhances glutamine uptakeand stimulates glutamine catabolism. As mentioned above, glu-tamine metabolism is linked to biosynthesis of protein, nucleotide

POX/PRODH

PYCR1

P5CS

GLS

MYC

GAPDH

Tet- + - + - +

P5CDH

1 d 3ds 4ds

GS

A B

D

POX/PRODH

PYCR1

P5CS

GLS

MYC

GAPDH

P5CDH

GS

MYC ON MYC OFF

0.0

0.5

1.0

1.5

p=0.001

Intra

ce

llu

lar p

ro

lin

e le

ve

l

(R

ela

tiv

e t

o M

yc

-O

n)

C

E

F

Fig. 5. MYC markedly increases the bio-synthesis of proline from glutamine (A) Westernblots of the enzymes in proline and glutaminecatabolism pathway in P493 cells treated withtetracycline for different lengths of time. (B)PC3 cells were transfected with siRNA againstMYC (siMYC) or negative control siRNA (siNeg).The enzymes in the proline and glutaminemetabolic pathways were analyzed by Westernblots. Experiments were replicated with similarresults. (C) Intracellular proline levels in MYC-On and MYC-Off cells. (D) The expected labelingpatterns from [U-13C,15N]-glutamine (Gln) andpreexisting unlabeled Gln to different iso-topologues of proline (Pro) and the inter-mediates of the TCA cycle via glutamate (Glu).[13C5,

15N2]-Gln can be catabolized via gluta-minase (GLS) to produce 13C5,

15N-glutamate(Glu) (m+6), which can be converted directly to13C5,

15N-Pro (m+6) via the pathway depicted inFig. 1A. Alternatively, 13C5,

15N-Glu can be con-verted to α-ketoglutarate (α-KG) and metabo-lized via the TCA cycle. The reverse reaction ofα-KG to Glu enables 15N reincorporation intoGlu (e.g., production of 15N1-Glu) via trans-aminases (TA) and/or Glu dehydrogenase (GDH)activity. The carbon tracings shown illustratethe 13C fate after three turns of the TCA cycle.●, 12C or 14N; red circle, 13C in the first turn;green circle, 13C in the second turn; pink circle,13C in the third turn; blue circle and N, 15N; AA,amino acids; single and double-headed solidarrows: single irreversible and reversible reactions, respectively; dashed arrows, multistep reactions; AA, amino acids; CS, citrate synthase; PDH, pyruvatedehydrogenase. (E and F) The GC-MS analysis of [13C,15N]-Gln contribution to citrate and proline synthesis in P493 cells with MYC On and Off. All GC-MS datawere corrected for natural abundance isotopic contribution and normalized to cell pellet wet weight. Each value is an average of duplicate samples. Theentire experiment was repeated three times.

Liu et al. PNAS | June 5, 2012 | vol. 109 | no. 23 | 8987

CELL

BIOLO

GY

and lipids, redox homeostasis, and energy metabolism. However,the report from Wise et al. suggests that little of the glutamineuptake stimulated by MYC is used for macromolecular synthesis(13). MYC-induced glutamine catabolism is the reprogrammingof mitochondrial metabolism to sustain cellular viability andTCA cycle anaplerosis (13). The most recent findings reportedby Le et al. (21) and Wang et al. (28) have emphasized themetabolic reprogramming controlled by MYC in transformedcells and activated T cells. The latter showed that MYC-drivenglutamine catabolism couples with multiple biosynthetic path-ways, especially ornithine and polyamine biosynthesis (28).Interestingly, our current studies show that MYC markedlyincreased glutamine-derived proline biosynthesis. However,how does this biosynthetic pathway fit into the MYC-drivenmetabolic reprogramming?The physiological relationship between glutamine and proline

metabolism was observed during the 1970s by Windmueller andSpaeth (29). In rat small intestine, a rapidly proliferating tissue,glutamine was an important substrate, with a utilization ratenearly two-thirds of that for glucose. Interestingly, proline wasa quantitatively important product from glutamine. Similarly, incultured L-M cells, Stoner and Merchant showed a net increaseof free proline accompanying the utilization of glutamine (30). Inthe tumor microenvironment, a metabolic commensalism betweenareas under the varying influence of MYC and HIF-1 or betweentumor and stromal cells has been proposed (3, 31–33). Themetabolic advantage afforded by the increased conversion ofglutamine to proline remains unclear, but the conversion of onenonessential amino acid to another for protein synthesis seemsunlikely. The previously described metabolic interlock betweenproline synthesis and pentose phosphate pathway offers an

attractive model for understanding the up-regulation of theproline synthetic pathway by MYC (34–36).In summary, we showed the suppression of POX/PRODH by

MYC, and demonstrated the effect of this reprogramming on cellproliferation and survival, which makes POX/PRODH a poten-tial target for cancer therapeutics. The regulatory mechanism forthe decreased expression of tumor suppressor POX/PRODH inhuman cancer is in part mediated posttranscriptionally by MYCvia miRNA. The metabolic link between glutamine and prolineafforded by MYC emphasizes the complexity of tumor metabo-lism, an area deserving additional studies.

Materials and MethodsCells and Cell Culture. MYC-inducible human Burkitt lymphoma model P493cells, PC3 human prostate cancer cells were maintained in RPMI medium 1640with 10% (vol/vol) FBS.

Additional Methods. Detailed descriptions of methods for real-time RT-PCR analysis, small RNAs transfection, Western blot, luciferase assay, mea-surement of ROS, apoptosis assay, chromatin immunoprecipitation assay,measurement of intracellular proline levels, measurement of glutamate andGSH, and GC-MS, FT-ICR-MS and NMR studies of [13C,15N]-Gln contribution toproline biosynthesis are available in supporting information.

ACKNOWLEDGMENTS. We thank Dr. Ziqiang Zhu for insightful commentsand Dr. Ziqiang Zhu, Julie Tan, and Radhika Burra for excellent technicalassistance. This research was supported by the Intramural Research Programof the National Institutes of Health (NIH); the National Cancer Institute;the Center for Cancer Research; the National Science Foundation, theExperimental Program to Stimulate Competitive Research (EPSCoR) GrantEPS-0447479; NIH Grants P20RR018733 from the National Center for Re-search Resources 1R01CA118434-01A2 (to T.W.-M.F.), 3R01CA118434-02S1(to T.W.-M.F.), and R21CA133688 (to A.N.L.); and the Brown Foundation.

1. Kroemer G, Pouyssegur J (2008) Tumor cell metabolism: Cancer’s Achilles’ heel. CancerCell 13:472–482.

2. Vander Heiden MG, Cantley LC, Thompson CB (2009) Understanding the Warburgeffect: The metabolic requirements of cell proliferation. Science 324:1029–1033.

3. Dang CV (2010) Rethinking the Warburg effect with Myc micromanaging glutaminemetabolism. Cancer Res 70:859–862.

4. Suzuki S, et al. (2010) Phosphate-activated glutaminase (GLS2), a p53-inducible reg-ulator of glutamine metabolism and reactive oxygen species. Proc Natl Acad Sci USA107:7461–7466.

5. Dang CV (1999) c-Myc target genes involved in cell growth, apoptosis, and metabo-lism. Mol Cell Biol 19:1–11.

6. Donald SP, et al. (2001) Proline oxidase, encoded by p53-induced gene-6, catalyzes thegeneration of proline-dependent reactive oxygen species. Cancer Res 61:1810–1815.

7. Liu Y, Borchert GL, Surazynski A, Hu CA, Phang JM (2006) Proline oxidase activatesboth intrinsic and extrinsic pathways for apoptosis: The role of ROS/superoxides, NFATand MEK/ERK signaling. Oncogene 25:5640–5647.

8. Polyak K, Xia Y, Zweier JL, Kinzler KW, Vogelstein B (1997) A model for p53-inducedapoptosis. Nature 389:300–305.

9. Liu Y, et al. (2009) Proline oxidase functions as a mitochondrial tumor suppressor inhuman cancers. Cancer Res 69:6414–6422.

10. Pandhare J, Cooper SK, Phang JM (2006) Proline oxidase, a proapoptotic gene, is in-duced by troglitazone: Evidence for both peroxisome proliferator-activated receptorgamma-dependent and -independent mechanisms. J Biol Chem 281:2044–2052.

11. Liu W, et al. (2010) miR-23b targets proline oxidase, a novel tumor suppressor proteinin renal cancer. Oncogene 29:4914–4924.

12. Eilers M, Eisenman RN (2008) Myc’s broad reach. Genes Dev 22:2755–2766.13. Wise DR, et al. (2008) Myc regulates a transcriptional program that stimulates mito-

chondrial glutaminolysis and leads to glutamine addiction. Proc Natl Acad Sci USA105:18782–18787.

14. Gao P, et al. (2009) c-Myc suppression of miR-23a/b enhances mitochondrial gluta-minase expression and glutamine metabolism. Nature 458:762–765.

15. DeBerardinis RJ, et al. (2007) Beyond aerobic glycolysis: Transformed cells can engagein glutamine metabolism that exceeds the requirement for protein and nucleotidesynthesis. Proc Natl Acad Sci USA 104:19345–19350.

16. Tong X, Zhao F, Thompson CB (2009) The molecular determinants of de novo nu-cleotide biosynthesis in cancer cells. Curr Opin Genet Dev 19:32–37.

17. Vafa O, et al. (2002) c-Myc can induce DNA damage, increase reactive oxygen species,and mitigate p53 function: A mechanism for oncogene-induced genetic instability.Mol Cell 9:1031–1044.

18. DeNicola GM, et al. (2011) Oncogene-induced Nrf2 transcription promotes ROS de-toxification and tumorigenesis. Nature 475:106–109.

19. Wonsey DR, Zeller KI, Dang CV (2002) The c-Myc target gene PRDX3 is required formitochondrial homeostasis and neoplastic transformation. Proc Natl Acad Sci USA 99:6649–6654.

20. Zeller KI, et al. (2006) Global mapping of c-Myc binding sites and target gene net-works in human B cells. Proc Natl Acad Sci USA 103:17834–17839.

21. Le A, et al. (2012) Glucose-independent glutamine metabolism via TCA cycling forproliferation and survival in B cells. Cell Metab 15:110–121.

22. Des Rosiers C, et al. (1995) Isotopomer analysis of citric acid cycle and gluconeogenesisin rat liver. Reversibility of isocitrate dehydrogenase and involvement of ATP-citratelyase in gluconeogenesis. J Biol Chem 270:10027–10036.

23. Mullen AR, et al. (2012) Reductive carboxylation supports growth in tumour cells withdefective mitochondria. Nature 481:385–388.

24. Metallo CM, et al. (2012) Reductive glutamine metabolism by IDH1 mediates lipo-genesis under hypoxia. Nature 481:380–384.

25. Wise DR, et al. (2011) Hypoxia promotes isocitrate dehydrogenase-dependent car-boxylation of α-ketoglutarate to citrate to support cell growth and viability. Proc NatlAcad Sci USA 108:19611–19616.

26. Pandhare J, Donald SP, Cooper SK, Phang JM (2009) Regulation and function ofproline oxidase under nutrient stress. J Cell Biochem 107:759–768.

27. Fan TW, Tan JL, McKinney MM, Lane AN (2011) Stable isotope resolved metabolomicsof lung cancer in a SCID mouse model. Metabolomics 7:257–269.

28. Wang R, et al. (2011) The transcription factor Myc controls metabolic reprogrammingupon T lymphocyte activation. Immunity 35:871–882.

29. Windmueller HG, Spaeth AE (1974) Uptake and metabolism of plasma glutamine bythe small intestine. J Biol Chem 249:5070–5079.

30. Stoner GD, Merchant DJ (1972) Amino acid utilization by L-M strain mouse cells ina chemically defined medium. In Vitro 7:330–343.

31. Zhang W, Huang P (2011) Cancer-stromal interactions: Role in cell survival, metabo-lism and drug sensitivity. Cancer Biol Ther 11:150–156.

32. Lisanti MP, et al. (2010) Understanding the “lethal” drivers of tumor-stroma co-evolution: Emerging role(s) for hypoxia, oxidative stress and autophagy/mitophagy inthe tumor micro-environment. Cancer Biol Ther 10:537–542.

33. Dang CV, Kim JW, Gao P, Yustein J (2008) The interplay between MYC and HIF incancer. Nat Rev Cancer 8:51–56.

34. Phang JM, et al. (1982) Stimulation of the hexosemonophosphate-pentose pathwayby pyrroline-5-carboxylate in cultured cells. J Cell Physiol 110:255–261.

35. Yeh GC, et al. (1984) The effect of pyrroline-5-carboxylic acid on nucleotide metab-olism in erythrocytes from normal and glucose-6-phosphate dehydrogenase-deficientsubjects. J Biol Chem 259:5454–5458.

36. Phang JM (1985) The regulatory functions of proline and pyrroline-5-carboxylic acid.Curr Top Cell Regul 25:91–132.

8988 | www.pnas.org/cgi/doi/10.1073/pnas.1203244109 Liu et al.