Báo cáo khoa học: Functional analysis of pyrimidine biosynthesis enzymes using the anticancer drug 5-fluorouracil in Caenorhabditis elegans docx

Knockdown of other pyrimidine biosynthesis enzyme homologs, such as uridine monophosphate kinase and uridine monophosphate synthetase, also resulted in 5-FU resistance.. Abbreviations 5d

using the anticancer drug 5-fluorouracil in

Caenorhabditis elegans

Seongseop Kim1,*, Dae-Hun Park2,*, Tai Hoon Kim1, Moogak Hwang1and Jaegal Shim1

1 Cancer Experimental Resources Branch, National Cancer Center, Gyeonggi-do, Korea

2 College of Pharmacy, Kangwon National University, Gangwon-do, Korea

Introduction

Enzymes responsible for pyrimidine biosynthesis play

critical roles in cellular metabolism, because they

pro-vide the pyrimidine nucleosides that are key

compo-nents of many biomolecules, such as RNA and DNA

Pyrimidine metabolism disorders can cause diseases

such as orotic aciduria, which results from uridine

monophosphate synthetase (UMPS) deficiency [1]

There are two routes for synthesizing pyrimidines:

de novo and salvage pathways Many genes encoding pyrimidine salvage pathway enzymes are genetic fac-tors influencing pyrimidine antagonist-based cancer chemotherapy [2]

5-Fluorouracil (5-FU) is a major pyrimidine anta-gonist that has been used for more than 40 years in

Keywords

5-fluorouracil; C elegans; UMPK; UMPS;

uridine phosphorylase

Correspondence

J Shim, Cancer Experimental Resources

Branch, National Cancer Center, 809 Madu

1-dong, Goyang-si, Gyeonggi-do, 411-769,

Korea

Fax: +82 31 920 2002

Tel: +82 31 920 2262

E-mail: jaegal@ncc.re.kr

*These authors contributed equally to this

work

(Received 19 May 2009, revised 22 June

2009, accepted 24 June 2009)

doi:10.1111/j.1742-4658.2009.07168.x

Pyrimidine biosynthesis enzymes function in many cellular processes and are closely associated with pyrimidine antagonists used in cancer chemo-therapy These enzymes are well characterized from bacteria to mammals, but not in a simple metazoan To study the pyrimidine biosynthesis path-way in Caenorhabditis elegans, we screened for mutants exhibiting resis-tance to the anticancer drug 5-fluorouracil (5-FU) In several strains, mutations were identified in ZK783.2, the worm homolog of human uridine phosphorylase (UP) UP is a member of the pyrimidine biosynthesis family

of enzymes and is a key regulator of uridine homeostasis C elegans UP homologous protein (UPP-1) exhibited both uridine and thymidine phos-phorylase activity in vitro Knockdown of other pyrimidine biosynthesis enzyme homologs, such as uridine monophosphate kinase and uridine monophosphate synthetase, also resulted in 5-FU resistance Uridine monophosphate kinase and uridine monophosphate synthetase proteins are redundant, and show different, tissue-specific expression patterns in C ele-gans Whereas pyrimidine biosynthesis pathways are highly conserved between worms and humans, no human thymidine phosphorylase homolog has been identified in C elegans UPP-1 functions as a key regulator of the pyrimidine salvage pathway in C elegans, as mutation of upp-1 results in strong 5-FU resistance

Abbreviations

5dFUR, 5¢-deoxy-5-fluorouridine; 5-FU, 5-fluorouracil; DPD, dihydropyrimidine dehydrogenase; dRib1P, 2-deoxy-a- D -ribose 1-phosphate; GFP, green fluorescent protein; MBP, maltose-binding protein; OMPDC, orotate monophosphate decarboxylase; OPRT, orotate phosphoribosyl transferase; PRPP, phosphoribosyl pyrophosphate; RNAi, RNA interference; SEM, standard error of the mean; SNP, single-nucleotide polymorphism; TK, thymidine kinase; TP, thymidine phosphorylase; TS, thymidylate synthase; UMPK, uridine monophosphate kinase; UMPS, uridine monophosphate synthetase; UP, uridine phosphorylase.

cancer chemotherapies 5-FU and other pyrimidine

antagonists, such as capecitabine and tegafur, have

been used to treat various cancers, including

colo-rectal, stomach, ovarian, head and neck cancers In

particular, 5-FU is a primary therapy for colorectal

cancer [2] Like other pyrimidine antagonists, 5-FU is

a prodrug that is converted to the active form via

the pyrimidine biosynthesis pathway [3] Therefore,

the function of this drug is closely associated with

the activity of pyrimidine synthesis enzymes,

includ-ing dihydropyrimidine dehydrogenase (DPD),

thymi-dylate synthase (TS), uridine phosphorylase (UP),

thymidine phosphorylase (TP), uridine monophosphate

kinase (UMPK), and orotate phosphoribosyl

trans-ferase (OPRT) Expression levels of these enzymes

in cancer cells are linked to 5-FU sensitivity and

resistance [2–6]

Uridine is a pyrimidine nucleoside that is essential

for the synthesis of RNA and biomembranes and is

involved in the regulation and function of the

cardio-circulatory, reproductive, nervous and respiratory

sys-tems [7] Furthermore, it modulates the cytotoxic

effects of fluoropyrimidines in both normal and

neo-plastic tissues [8] The concentration of uridine in

plasma and tissues is tightly regulated by cellular

transport mechanisms and by UP activity [7] UP

cata-lyzes the reversible phosphorolysis of uridine, yielding

uracil and Rib1P, and is an important enzyme in the

pyrimidine salvage pathway Human UP and TP each

exhibit both uridine and thymidine phosphorylase

activities Pyrimidine phosphorylases differ in activity

and substrate specificity, and play different roles in

flu-oropyrimidine sensitivity [9] UP is the major

phos-phorylase that regulates uridine homeostasis, but TP

also acts on uridine as a substrate to a certain extent

At least two UPs and one TP are present in humans

The first human UP was cloned in 1995 [10], and this

was followed by the cloning and expression analysis of

UPP-2 [11] UP expression is controlled

transcription-ally by oncogenes, tumor suppressor genes, and

cyto-kines [12] UP activity is typically upregulated in

various tumor tissues, conferring a therapeutic

advan-tage for 5-FU in cancer patients [13] Beyond

tran-scriptional regulation, UP activity is modulated by

specific inhibitors, such as 5-phenylthioacyclouridine

[14,15] During oncogenesis, ectopic expression of UP

is reported to support anchorage-independent cell

growth [16] Thus, UP is a possible prognostic factor

for several cancers, including breast cancer and oral

squamous cell carcinoma [13,17]

As a core enzyme of the pyrimidine salvage

path-way, UP is conserved across kingdoms, and many

studies on UP have been carried out in Escherichia

coli However, given that UP function is important for both normal physiology and cancer therapy, animal models are increasingly being used to study this enzyme Disruption of UP activity in mouse embryonic stem cells leads to increased 5-FU concentrations in plasma and reduced incorporation of 5-FU into nucleic acids [18] Moreover, UP) ⁄ ) mice exhibit increased uridine concentrations in the plasma, lung, gut, liver and kidney as compared with wild-type mice [19] The inhibition of TP activity also results in ele-vated pyrimidine levels in plasma and axonal swelling

in the brains of mice [20]

Previously, we demonstrated that 5-FU induces germ cell death and inhibits development in Caenor-habditis elegans [21] We also observed that C elegans DPD and TS expression levels are associated with 5-FU function [22] Here, we describe the results obtained from a 5-FU-resistant mutant screen in

C elegans, from which we identified upp-1 mutations from several 5-FU-resistant mutants In addition, we characterized C elegans UMPK and UMPS⁄ OPRT homologs using RNA interference (RNAi) and 5-FU Uncovering the mechanism of 5-FU resistance and characterization of pyrimidine biosynthesis enzymes in C elegans will help to further our under-standing of pyrimidine biosynthesis enzymes by fill-ing in a missfill-ing link between bacteria and higher organisms

Results

UPP-1 mediates 5-FU functions in C elegans

We performed a genetic screen for 5-FU-resistant mutant C elegans strains by scoring for larval growth

in the presence of 5-FU One of these mutants (jg1) was mapped to identify the mutated gene by using single nucleotide polymorphisms (SNPs) The jg1 mutation mapped near SNP uCE3-1087, which is in the )0.07 region of chromosome III (Fig 1G) and could be rescued with a single cosmid, ZK783 It could also be rescued by expression of a single ORF (ZK783.2) encoding UPP-1 Transgenic worms expressing these rescue constructs exhibited 5-FU sensitivity (Fig 1H)

Although the upp-1 mutant grew slowly on the 5-FU plate, it advanced to late larval and adult stages (Fig 1C), unlike wild-type worms, which arrested at L1 or L2 (Fig 1B,E) Larvae were evaluated 60 h after egg transfer from normal plates to 5-FU plates, during which time wild-type worms grown on control plates exhibit the vulval invagination typical of L4 larva (Fig 1A,D) Because early embryogenesis is more

sensitive to 5-FU than late larval development, we

decreased the 5-FU concentration (5 nm) to compare

the hatching ratios of wild-type and upp-1 mutant

worms Wild-type worms on the 5-FU plate exhibited

a low hatching ratio (10% of total eggs), whereas the

upp-1 mutant exhibited a high hatching ratio (over

90%) (Fig S1) Observation of upp-1 mutants under a

dissection microscope and a high-resolution differential

interference contrast microscope revealed that, with

the exception of 5-FU resistance, they did not differ

from wild-type worms in either morphology or

behav-ior However, the lifespan of upp-1 mutant worms was

reduced by about 30% as compared with wild-type

worms (Fig S2)

ZK783.2 encodes a protein with amino acid

homol-ogy to human UP (UPP-1 and UPP-2) (46% identical;

Fig S3) In order to verify the functional conservation

between human and worm UPP-1, we expressed

human UPP-1 under the control of the worm upp-1

promoter Human UPP-1 was able to rescue the

upp-1(jg1) mutant (Fig 1H) Sequencing of six upp-1

Fig 2 UPP-1 is highly conserved from C elegans to humans (A) The ZK783.2 ORF encodes a homolog of human UP Six upp-1 mutants were sequenced, and their mutations are indicated by an asterisk on the ZK783.2 genomic diagram Asterisks indicate the location of each mutation, and Q203* indicates that the gluta-mine 203 was changed to a stop codon (B, C) The UPP-1::GFP fusion proteins were expressed in several tissues, including the hypodermis, pharynx, spermatheca, and gonad Scale bars: 100 lm (B) and 10 lm (C).

Fig 1 The upp-1 mutant is highly resistant to 5-FU The upp-1 (jg1) mutant grows well on the 5-FU plate as compared with wild-type

C elegans (A–C) Although the growth of upp-1 animals on the 5-FU plate is slower than that of wild-type animals on normal plates, this mutant survives up to stage L4 and adulthood (D–F) The arrowhead (A) and asterisk (D) indicate vulval invagination, which is a character-istic of the L4 stage The arrow indicates turning of the gonad, which occurs at the early L4 larval stage (F) Wild-type worms growing

on the 5-FU plates are arrested at the L2 stage (B, E) Growth tests were done on plates containing 800 n M 5-FU (A–F) (A–C) Dissect-ing microscope images (D–F) Nomarski images from a high-resolution differential interference contrast microscope (A, D) Wild-type worms on control plates (B, E) Wild-type worms on 5-FU plates (C, F) upp-1 (jg1) mutant worms on 5-FU plates All pictures were taken 60 h after egg transfer Scale bars: 100 lm (A–C) and 10 lm (D–F) (G) The SNP mapping method using the Hawaiian strain CB4856 was used for upp-1 (jg1) mutant cloning The upp-1 mutation mapped near the SNP uCE3-1087 in the )0.07 region of LG III (H) Fifteen cosmids in that region were obtained from the Sanger Center, and the phenotype of the mutant was rescued with a single ZK783 cosmid The upp-1 mutant was also rescued with a genomic PCR product including a single ORF (ZK783.2) The upp-1 mutant worm carrying only the control pRF4 (rol-6gf) plasmid [upp-1; Ex (pRF4)] was the same as the nontransgenic upp-1 mutant Transgenic worms carrying the ZK783 cosmid [upp-1; Ex (ZK783; pRF4)] or ZK783.2 PCR product [upp-1; Ex (ZK783.2; pRF4)] were sensitive to 5-FU The upp-1 mutant worm expressing human UPP-1 under the control of the worm upp-1 promoter [upp-1; Ex (hUPP1; pRF4)] also exhibited 5-FU sensitivity Coinjection of the marker pRF4 was used to identify transgenic worms Error bars represent standard error of the mean (SEM) *P < 0.001 as compared with control Ex (pRF4) worms on the 5-FU plate, determined by unpaired Student’s t-test.

**P > 0.1.

mutants (jg1, jg2, jg3, jg7, jg8, and jg11) revealed

mis-sense mutations in all but jg3, which had a nonmis-sense

mutation at glutamine 203 (Fig 2A)

As upp-1 mutants exhibited strong 5-FU resistance,

and as there is only one UPP gene in the C elegans

genome, we hypothesized that the expression of UPP-1

was ubiquitous Transgenic worms expressing

UPP-1::green fluorescent protein (GFP) showed bright GFP

signal in the hypodermis, pharynx, and spermatheca

(Fig 2C) UPP-1::GFP was also expressed in the

gonad (Fig 2D), consistent with our observation that

germ cell death normally induced by 5-FU is

sup-pressed in upp-1 mutants [21]

UPP-1 has both UP and TP activities

As the upp-1 mutant is resistant to 5-FU, and no TP

homolog has been identified in C elegans, we

hypoth-esized that C elegans UPP-1 functions as both a UP

and a TP Thus, a single mutation in upp-1 may

con-fer strong 5-FU resistance Indeed, C elegans UPP-1

exhibited both UP and TP activity in vitro, with TP

activity being greater than UP activity (Fig 3A) We

also tested the activities of several mutant UPP-1

proteins (T128I, Y209F, and Q203*), and compared

larval growth of these mutant strains on 5-FU plates

All three UPP-1 mutant proteins exhibited very low

levels of enzyme activity ( 20% of that of wild-type

UPP-1) in vitro (Fig 3A) The growth rates of these

upp-1 mutants were also similar to each other

(Fig 3B)

Next, we treated wild-type and upp-1 mutant worms

with 5¢-deoxy-5-fluorouridine (5dFUR) to further

examine the function of UPP-1 in the pyrimidine

bio-synthesis pathway 5dFUR is converted to 5-FU by

UP and TP [9] The effects of 5dFUR in the upp-1

mutant are questionable, as no TP homolog has been

discovered in the C elegans genome Both 5dFUR and

5-FU, however, exhibited similar effects on wild-type

worms, and growth of the upp-1 mutant on the

5dFUR plate resembled that on the 5-FU plate

(Fig 3C)

The pyrimidine biosynthesis pathway is well

conserved between C elegans and humans

5-FU is converted to FdUMP by several sequential

steps of the pyrimidine salvage pathway, which

involves several metabolic enzymes UP mediates the

first step of 5-FU conversion We searched for

homo-logs of the human pyrimidine biosynthesis enzymes,

including uridine kinase, UMPK, and UMPS, in

C elegans to test their roles in 5-FU function and

were able to identify homologs for most of them in

C elegans

To explore the functional relationships of these pyrimidine biosynthesis enzymes using 5-FU, we used

Fig 3 C elegans UPP-1 exhibits TP and UP activities (A) UPP-1 converted both Rib1P and dRib1P to uridine and thymidine in the presence of uracil and thymine, respectively The enzymatic activi-ties of three mutant UPP-1 proteins were very low as compared with that of wild-type UPP-1 in vitro The TP activity of UPP-1 is two times higher than the UP activity No differences in enzymatic activity were observed between the three UPP-1 mutant proteins Both *P and **P (as compared with wild-type UPP-1 activity) are

< 0.001 (B) The ratios of L4 and adult worms compared with the total for the three upp-1 mutants on 5-FU plates are shown No dif-ferences in growth were observed among three the upp-1 alleles.

*P-values (as compared with wild-type worms on 5-FU plates) cal-culated by unpaired Student’s t-test were < 0.001 (C) The upp-1 (jg1) mutant also grew well on 5dFUR plates 5dFUR, a precursor

of 5-FU, is converted by UP and TP activities Growth test results are reported as percentages of L4 and adult animals out of total progeny (y-axis) *P-values (as compared with wild-type worms on 5-FU or 5dFUR plates) calculated by unpaired Student’s t-test were

< 0.001 Error bars represent SEM.

RNAi to knock down these genes by the bacterial

feeding method Some genes, such as T23G5.1 (rnr-1)

and C03C10.3 (rnr-2), which are homologs of the

genes encoding ribonucleotide reductase a and b

subunits, respectively, exhibited a lethal phenotype, so

we could not test the growth of these worms on 5-FU

plates Knockdown of T07C4.1 (UMPS) or C29F7.3

(UMPK) resulted in 5-FU resistance, whereas RNAi

for other genes had no effect on 5-FU sensitivity

(Fig 4) Our results indicate that the pyrimidine

biosynthesis and 5-FU functional pathways are well

conserved between humans and C elegans

Two UMPK homologs were expressed in

different tissues

Using homology searches and RNAi, we determined

that C29F7.3 and T07C4.1 were associated with 5-FU

function (Fig 4) Interestingly, C29F7.3 shows amino

acid similarity to human uridine kinase, UMPK, and

UDPK Uridine kinases are downstream of UP in the

pyrimidine salvage pathway Homology searches

revealed that several uridine kinases exist in the C

ele-gansgenome Three of these were selected on the basis

of length and sequence homology, and their expression

patterns and enzymatic activities were characterized

Deletion (tm2740) or knockdown of B0001.4 did not

result in altered 5-FU sensitivity C29F7.3 and

F40F8.1 share 82% identity (Fig S4), but knockdown

of C29F7.3 results in 5-FU resistance, whereas knock-down of F40F8.1 does not (Fig 4) In order to evalu-ate how these proteins function differently in the 5-FU pathway, we studied their expression patterns using GFP reporter fusion constructs (Fig 5A) C29F7.3::GFP is expressed in the hypodermis, intes-tine, and pharynx, whereas F40F8.1::GFP signal is observed mostly in neurons and the pharynx These distinct expression patterns may account for the differ-ences between these two proteins in the 5-FU treat-ment and RNAi experitreat-ments The activities of these enzymes in the hypodermis and intestine may be important for mediating 5-FU effects

As C29F7.3 and F40F8.1 share considerable sequence identity, we wished to rule out the possibility

of RNAi cross-effects C29F7.3 RNAi was very effec-tive, with most GFP disappearing in the hypodermis and intestine, but neuronal expression of C29F7.3::GFP appeared (data not shown) In general,

C elegans neurons are resistant to RNAi [23] It was difficult to dissect the effects of F40F8.1 RNAi, because F40F8.1::GFP was expressed strongly in neu-rons and the pharynx, but weakly in the intestine F40F8.1 RNAi resulted in a slightly decreased GFP signal We made a transgenic worm expressing F40F8.1::GFP under the control of the C29F7.3 pro-moter to further evaluate the F40F8.1 RNAi efficiency and specificity Ectopically expressed F40F8.1::GFP was diminished by F40F8.1 RNAi, but not by C29F7.3 RNAi (data not shown), indicating that knockdown of these two genes is very specific

To more precisely understand the functions of these uridine kinase homologs, we analyzed their enzyme activity in vitro The proteins purified from the bacte-rial induction system and several substrates were incu-bated together, and products were detected by HPLC Both C29F7.3 and F40F8.1 exhibited only UMPK activity, but B0001.4 showed no uridine kinase activity (Fig 5B) C29F7.3 and F40F8.1 exhibited similar UDP peaks when UMP was added as a substrate Both C29F7.3 and F40F8.1 may function downstream

of UPP-1, but show different responses in mediating 5-FU function, probably because of their different expression patterns

T07C4.1 and R12E2.11 proteins have OPRT function

Both de novo synthesis and salvage pathways are used

to synthesize UMP The salvage pathway includes

UP⁄ uridine kinase and OPRT, and the de novo path-way includes enzymes such as orotate monophosphate

Fig 4 The pyrimidine biosynthesis pathway is conserved from

humans to C elegans Some enzymes of the pyrimidine

biosynthe-sis pathway are also involved in 5-FU rebiosynthe-sistance Growth tests

were performed on 5-FU plates following RNAi for worm homologs

of various human genes F25H2.5 is a putative homolog of uridine

diphosphate kinase, and Y43C5A.5 is TK R12E2.11 and T07C4.1

are OPRT domain proteins F19B6.1, B0001.4, C29F7.3 and

F40F8.1 are putative uridine kinase or UMPK homologs ZK783.2

(upp-1) RNAi was used as a positive control Depletion of T07C4.1

and C29F7.3 by RNAi resulted in 5-FU resistance Error bars

represent SEM, and *P-values (as compared with wild-type worms

on 5-FU plates, calculated by unpaired Student’s t-test) were

< 0.001.

decarboxylase (OMPDC) T07C4.1 and R12E2.11 both

have an OPRT domain, but their functional roles are

unclear Bacteria and fungi have separate genes for

OPRT and OMPDC, but animals and plants have a

single UMPS protein comprising both OPRT and

OM-PDC [24] Interestingly, worms have both UMPS

(OPRT plus OMPDC) and OPRT forms (Fig 6A)

Sequence alignments indicate that T07C4.1 protein is

very similar to human UMPS, and R12E2.11 is very

similar to the OPRT domain of human UMPS and

T07C4.1 (Fig S5) The T07C4.1::GFP fusion construct

is expressed in neurons and intestinal cells, but

R12E2.11::GFP is expressed in the body wall muscle,

spermatheca, intestine, and vulval muscle (Fig 6B)

Finally, we verified the OPRT and OMPDC

activi-ties of T07C4.1 and R12E2.11 proteins in vitro Both

proteins can synthesize UMP from uracil and

phos-phoribosyl pyrophosphate, but only T07C4.1 exhibited OMPDC activity (Fig 6C) The single UMPS of higher organisms is more efficient than the separate OPRT and OMPDC system for UMP synthesis [24] T07C4.1 shows stronger OPRT activity than R12E2.11, and mixing the two proteins has an additive effect on UMP synthesis Interestingly, R12E2.11 itself has no OMPDC activity, but mixing R12E2.11 and T07C4.1 results in a higher UMP peak than observed with T07C4.1 protein alone This suggests that R12E2.11 and T07C4.1 may cooperate to synthesize UMP in C elegans intestinal cells

Discussion

The upp-1 mutant is highly resistant to 5-FU, even when compared with other 5-FU resistant mutant and

Fig 5 Characterization of UMPKs in C ele-gans (A) The expression patterns of C29F7.3::GFP and F40F8.1::GFP are shown C29F7.3::GFP expression is robust in the pharynx, hypodermal cells, and intestine, whereas F40F8.1::GFP is expressed strongly in neurons and the pharynx, but weakly in the intestine Scale bars: 100 lm (B) In vitro enzymatic assays of three uridine kinase homologs using analytical HPLC No proteins exhibited uridine kinase activity when uridine was used as a substrate, but both C29F7.3 and F40F8.1 produced UDP when UMP was used as a substrate Arrows indicate UDP peaks Detection times are shown on the x-axes, and UV absorbance at 260 nm on the y-axes.

transgenic worms Expression levels of DPD and TS

are closely related to the 5-FU response and sensitivity

in human cancers [25,26] Transgenic worms

overex-pressing DPD and TS, however, showed only small

increases in survival ratios on 5-FU plates [22] In

addition, 5-FU-induced cell death is dependent on p53

[27], but the C elegans cep-1⁄ p53 mutant exhibited

only minimal improvement in germ cell death as

com-pared with the upp-1 mutant [21] Thus, UPP-1 is a

key player mediating 5-FU functions in C elegans

TP and UP participate in both uridine and

thymi-dine synthesis, and humans possess at least two

different UPs and one TP This complex redundancy makes the relationship between UP⁄ TP and 5-FU sen-sitivity in humans difficult to decipher In contrast,

C elegans has only one UP, which functions as both

UP and TP (Fig 3A) Human UPP1 can rescue the

C elegans upp-1 mutant phenotype (Fig 1H), suggest-ing that UP function is evolutionarily conserved and mediates 5-FU function in vivo in humans Addition-ally, the upp-1 mutant showed similar resistance to 5dFUR and 5-FU (Fig 3C) These results also indicate that the single upp-1 gene in C elegans plays a key role

in the pyrimidine salvage pathway

Fig 6 Characteristics of UMPS and OPRT

homologs in C elegans (A) Domains of

UMP synthesis enzymes Bacteria and fungi

have separate OPRT and OMPDC proteins,

but higher animals have a single protein

with both OPRT and OMPDC functions.

C elegans has a long UMPS homolog and a

short OPRT homolog (B) The expression

patterns of T07C4.1::GFP and

R12E2.11::GFP are shown T07C4.1::GFP

expression is robust in neurons and the

intestine, whereas R12E2.11::GFP is

expressed strongly in the body wall muscle,

spermatheca, and vulval muscle Scale bars:

100 lm (C) Enzymatic activities of T07C4.1

and R12E2.11 proteins in vitro OPRT (left)

and OMPDC (right) activities were

mea-sured by adding phosphoribosyl

pyropho-sphate (PRPP) with uracil (Ura) and orotate

(Oro), respectively, as a substrate Both

T07C4.1 and R12E2.11 have OPRT activity,

but only T07C4.1 has OMPDC activity, as

expected from the protein domain

struc-tures R12E2.11 itself has no OMPDC

activ-ity, but it promotes the OMPDC activity of

T07C4.1 Detection times are shown on the

x-axes, and UV absorbance at 260 nm on

the y-axes.

RNAi for other pyrimidine biosynthesis pathway

enzymes revealed that the depletion of only three genes

resulted in 5-FU resistance One explanation for the

observed results is that knockdown of a single gene is

not sufficient to abolish pyrimidine biosynthesis, owing

to the existence of redundant genes or pathways

Inter-estingly, two UMPK genes, C29F7.3 and F40F8.1, are

very similar in amino acid sequence, and their protein

products show similar abilities to synthesize UDP from

UMP (Fig 5), but their knockdown produces different

5-FU responses (Fig 4) This difference is probably

due to the distinct expression patterns of these genes in

the intestine and hypodermis, which appears to be

important for 5-FU function

Another gene mediating 5-FU function is that

encod-ing T07C4.1, which contains OPRT and OMPDC

domains As the OPRT activity of T07C4.1 is

impor-tant for mediating 5-FU function and the salvage

path-way of pyrimidine biosynthesis, the differences in 5-FU

metabolism between T07C4.1 and another OPRT

pro-tein, R12E2.11, is puzzling Both proteins have OPRT

enzymatic activity (Fig 6C), but knockdown of only

the gene encoding T07C4.1 resulted in 5-FU resistance

The expression patterns of these proteins differ, but not

enough to explain the RNAi results It has been

reported that UMPS, which is a single protein with

both OPRT and OMPDC domains, is more stable and

has higher activity than separate OPRT and OMPDC

proteins [28,29] R12E2.11 exhibited lower activity than

T07C4.1 in vitro, and it is possible that this difference is

amplified in vivo The higher OPRT activity and strong

intestinal expression of T07C4.1 may account for the

difference In addition, knockdown of R12E2.11

pro-moted T07C4.1::GFP expression (data not shown),

indicating a complex relationship between these two

proteins in vivo

As UP⁄ uridine kinase and OPRT both synthesize

UMP from uracil in the pyrimidine salvage pathway,

the strong 5-FU resistance resulting from single gene

knockdown for UPP-1 or UMPS was unexpected

However, UPP-1 is strongly expressed in the

hypo-dermis, and T07C4.1 is mainly expressed in the

intes-tine Knockdown of C29F7.3, on which both UP⁄

uridine kinase and OPRT converge, resulted in high

5-FU resistance As C29F7.3 is expressed in the

hypo-dermis and intestine, both UP⁄ UMPK in the

hypo-dermis and OPRT⁄ UMPK in the intestine are essential

for mediating 5-FU function in C elegans It is also

possible that UPP-1 and OPRT cooperate to mediate

5-FU function and UMP synthesis, because

knock-down of T07C4.1 in upp-1 mutants resulted in similar

5-FU resistance as knockdown of either T07C4.1 or

UPP-1 (data not shown)

On the basis of these results, we propose a model of pyrimidine biosynthesis and 5-FU conversion in humans and in C elegans (Fig 7) In a human cancer model, the conversion of 5-FU to FdUMP is mediated

by three independent pathways involving UPP, OPRT, and TPP In contrast, the C elegans genome does not include a TP homolog, and downstream signaling via tyrosine kinase (TK) does not appear to be associated with 5-FU function, given the lack of 5-FU resistance following knockdown of the worm TK candidate gene, Y43C5A.5 (Fig 4) Our data do not explain all of the similarities and differences in the pyrimidine salvage pathway and 5-FU function between humans and

C elegans, but it is clear that UP and OPRT activity mediated by UMPK is a major 5-FU conversion path-way in C elegans

Although C elegans has been used as a model sys-tem in pharmacogenetics and chemical genetics, it has only recently begun to be used to study anticancer

Fig 7 Comparison of the human and C elegans 5-FU conversion and pyrimidine biosynthesis pathways Two pathways allow the conversion of 5-FU to FdUMP in humans The C elegans genome has homologs for the enzymes in these pathways Humans show redundancy of pyrimidine synthesis enzymes, including two UPs and one TP, but C elegans has only one uridine and thymidine phosphorylase, UPP-1 The C elegans uridine kinase has not been identified yet, but C29F7.3 and F40F8.1 proteins were identified by their UMPK activities Two OPRT homologs, T07C4.1 and R12E2.11, mediate conversion of uracil to UMP in C elegans.

drugs, such as farnesyl transferase inhibitors [30].

Here, we evaluate how the C elegans upp-1 mutant

interacts with the anticancer drug 5-FU As C elegans

is a simple metazoan, interpreting the relationships

between anticancer drugs and gene function may be

less complex than in higher organisms At the same

time, the example of a single, dual-function protein in

humans that takes on the roles that two separate

enzymes play in worms underscores the challenges and

discoveries that await us in C elegans Our findings

support a close relationship between pyrimidine

salvage enzymes in 5-FU function and resistance in

both C elegans and humans

Experimental procedures

C elegans strains and culture

The Bristol strain N2 was used as a wild-type strain The

Hawaiian strain CB4856 was used as a reference strain for

mapping mutant genes by SNPs [31] The B0001.4 deletion

mutant (tm2740) was a gift from S Mitami (Tokyo

Women’s Medical University, Japan) Animals were

cul-tured as described by Brenner [32]

Chemicals

Rib1P, 2-deoxy-a-d-ribose 1-phosphate (dRib1P), uridine,

UMP, UDP, UTP, ATP, 2-deoxyuridine, 5-FU, 5dFUR,

orotidine 5¢-phosphate and PPRP were purchased from

Sigma-Aldrich Chemicals (St Louis, MO, USA) [6-14C]5-FU

(specific activity 52 mCiÆmmol)1) was purchased from

Moravek Biochemicals, Inc (Brea, CA, USA)

5-FU sensitivity and mutant phenotype analysis

Analysis of 5-FU sensitivity was performed on plates

con-taining 5-FU (800 nm) Synchronized embryos were

trans-ferred to 5-FU plates After 60 or 72 h, the numbers of L4

larvae⁄ adult worms and total worms were counted, and the

ratios of L4 larvae⁄ adult worms to total worms were

calcu-lated 5-FU plates were kept in the dark during experiments

to avoid fluorine degradation by light 5dFUR sensitivity

was tested using the same method For upp-1 (jg1) mutant

rescue experiments, total L4 and adult animals from

trans-genic worms carrying additional genes, such as the ZK732

cosmid, were counted, and the ratio of roller worms

con-taining coinjected pRF4 plasmid to nonroller worms was

calculated

Sequence alignment

Amino acid sequences of human and C elegans proteins

were aligned using macvector (MacVector Inc., Cary, NC,

USA) The GenBank accession number of human UPP1 is AAH07348, and that of UMPS is CAG33068 The amino acid sequences of UPP-1 (ZK783.2), C29F7.3, F40F8.1, T07C4.1 and R12E2.11 are from wormbase (http:// www.wormbase.org)

Plasmid construction and protein purification

Rescue experiments were performed using PCR-amplified upp-1genomic DNA and the ZK783 cosmid obtained from the Sanger Institute (Cambridge, UK) The following primers were used to amplify upp-1 genomic DNA: 5¢-AGC ATC TGC AGC AAC CAC C-3¢ and 5¢-TGG ATC CGA TCC CGG TCT GCT TGC G-3¢ To construct the upp-1::gfp fusion construct, the GFP expression vector pPD95.77 (obtained from A Fire, Stanford University, CA, USA) was used The following primers were used to amplify upp-1 geno-mic DNA (4793 bp): 5¢-TTT CTG CAG GAG AGT TGT ACC TAA AGG CGC G-3¢ and 5¢-TTT GGT ACC ATC CCG GTC TGC TTG CGA ATG-3¢ Amplified PCR frag-ments were digested with PstI and KpnI, and used as insert DNA

To generate the human UPP1 rescue construct, the

C elegans upp-1promoter region was fused with the human UPP1cDNA in pPD95.77 To amplify the C elegans upp-1 promoter region (3105 bp), the following primers were used: 5¢-TTC TGC AGG TGA TGC CTT TGA GCA CT

T AGC-3¢ and 5¢-TTT TCT AGA CTT GAT GGA TCT GAA AAA ATT CC-3¢ Amplified PCR fragments were digested with PstI and XbaI, and ligated into pPD95.77 Human UPP1 cDNA (995 bp) was then amplified, using a human cDNA library (Clontech Laboratories, Inc., Moun-tain View, CA, USA) as a template and the following prim-ers: 5¢-TTT CCC GGG CAC TGC AGA CGT CTG TCC G-3¢ and 5¢-TTT GGT ACC CAG GCC TTG CTC AGT TTC TTC-3¢ PCR products were digested with SmaI and KpnI, and ligated to the amplified C elegans upp-1 pro-moter The same vector and methods were used to make C29F7.3::GFP, F40F8.1::GFP, T07C4.1::GFP, and R12E2.11::GFP The primers and restriction enzyme sites used were as follows: 5¢-TTT AAG CTT CTT TAT CAG TAG TTT TGA GGC CG-3¢ (HindIII) and 5¢-AAT CTG CAG TTT TTG GTT GGC AGC CGC GAA TAC-3¢ (PstI) for C29F7.3::GFP, 5¢-TTG TCG ACC AGT CTT CAA AAT AGC GCA GG-3¢ (SalI) and 5¢-TTT TCT AGA TTT TTT GTT GGC AGC GTC G-3¢ (XbaI) for F40F8.1::GFP, 5¢-AAT GGG CTG CAG AAG AAA AGG GTG GC-3¢ (PstI) and 5¢-TGG ATC CAA TGC TAT CGT CGC TTC TCG-3¢ (BamHI) for T07C4.1::GFP, 5¢-TTT CTG CAG TTG TCC TTG ATA TCT C-3¢ (PstI) and 5¢-AAT CTA GAA GCA GAT GAG CAA TAA TCT G-3¢ (XbaI) for R12E2.11::GFP

To construct the plasmid expressing the maltose-binding protein (MBP)::UPP-1 fusion protein, full-length upp-1 cDNA (888 bp) was cloned from first-strand worm cDNA

by PCR and inserted in-frame, downstream of the MBP

sequence in the E coli expression vector pMAL-c2X (New

England Biolabs, Ipswich, MA, USA) PCR was performed

using the following primers: 5¢-TGG ATC CAT GAA

CGGACT TGT CAA GAA CGG-3¢ and 5¢-TTT AAG

CTT TTA GAT CCC GGT CTG CTT GC-3¢ The

ampli-fied PCR fragments were digested with BamHI and

HindIII, and were ligated into pMAL-c2X The same vector

and methods were used to make MBP::C29F7.3,

MBP::F40F8.1, MBP::B0001.4, MBP::T07C4.1 and

MBP::R12E2.11 constructs The primers and restriction

enzyme sites used were as follows: 5¢-TTA GAT CTA TGT

ACA ACG TCG TCT TTG TTC-3¢ (BglII) and 5¢-AAG

GTA CCC TAT TTT TGG TTG GCA GCC G-3¢ (KpnI)

for C29F7.3, 5¢-AAG GAT CCA TGC ACA ACG TGG

TTT TTG TTC-3¢ (BamHI) and 5¢-AAG GTA CCT TAT

TTT TTG TTG GCA GCG TC-3¢ (KpnI) for F40F8.1,

5¢-AAG GAT CCA TGA AAA ACA CTC TGA AAT

TGC-3¢ (BamHI) and 5¢-AAG GTA CCT TAA TGT GGA

CGG GAG AAT GG-3¢ (KpnI) for B0001.4, 5¢-AAG GAT

CCA TGC ACA ACG TGG TTT TTG TTC-3¢ (BamHI)

and 5¢-TTT TCT AGA TCA AAT GCT ATC GTC GCT

TCT CG-3¢ (XbaI) for T07C4.1, and 5¢-TTT GAA TTC

ATG ACC GCC GCC ACC G-3¢ (EcoRI) and 5¢-AAG

GTA CCT TAA TGT GGA CGG GAG AAT GG-3¢

(KpnI) for R12E2.11

Microinjection and RNAi

All transgenic strains were generated by microinjection to

achieve germline transformation For rescue experiments,

the ZK783 cosmid carrying the upp-1 (ZK783.2) PCR

prod-uct and the constrprod-uct containing the C elegans upp-1

pro-moter fused with human UPP1 cDNA were injected

(75 lgÆmL)1) along with the marker pRF4 (75 lgÆmL)1)

into upp-1 (jg1) mutants Control transgenic worms were

injected with pRF4 plasmid DNA (100 lgÆmL)1) only To

generate transgenic worms that express UPP-1::GFP, the

upp-1::gfp fusion construct was injected (75 lgÆmL)1) into

adult N2 animals along with the pRF4 plasmid

(75 lgÆmL)1) The C29F7.3::GFP, F40F8.1::GFP,

B0001.4::GFP, T07C4.1::GFP and R12E2.11::GFP

plas-mids were injected using the same method and at the same

DNA concentration

RNAi by bacterial feeding was performed as previously

described [33] Briefly, synchronized L4 larvae were

transferred onto plates containing 1 mm isopropyl

thio-b-d-galactoside and the HT115-RNAi bacterial clone The

next day, adult worms were transferred to new RNAi

plates Embryos from RNAi plates were transferred to

both a control plate without 5-FU and an experimental

plate containing 5-FU (800 nm) After 60 or 72 h,

L4⁄ adult animals and total worms were counted, and the

ratio of L4 and adult animals to total worms was

calculated The empty vector L4440 was used as a

control

In vitro enzymatic assays

To induce the MBP–UPP1 fusion protein, 0.3 lm IPTG was added to the culture Induced fusion proteins were purified using amylose resin (New England Biolabs), according to the manufacturer’s protocol, and the UPP-1 proteins were cleaved and eluted by factor Xa digestion (New England Biolabs) Modified methods described by Kouni et al [34] were used, and the activity assay mixture (35 lL) consisted of 1 lg of purified UPP-1 fusion protein,

10 mm Tris⁄ HCl buffer (pH 7.4), 0.8 mm EDTA, 2.5 mm Rib1P or dRib1P, 5 mm MgCl2, and 192 lm [6-14C]5-FU The reaction was incubated at 37C for 1 h After incuba-tion, samples were boiled for 3 min to stop the enzymatic reaction, and then chilled on ice Compounds were sepa-rated by TLC All assay mixtures were spotted onto PEI cellulose sheets with 4 lL of nonradioactive tracer (100 lg

of 5-FU and 100 lg of uridine for the UP assay mixture;

100 lg of 5-FU and 100 lg of deoxyuridine for the TP assay mixture) After development with distilled water, spots were excised using 254 nm UV light The activity was counted after addition of 4 mL of scintillation fluid Evaluation of the enzymatic activity of C29F7.3, F40F8.1, B0001.4, T07C4.1 and R12E2.11 was performed as described

by Li et al [35] and Krungkrai et al [36], with a few modifi-cations All reaction mixtures contained 10 lg of recombi-nant proteins in a total reaction volume of 100 lL The reaction mixture was incubated for 12 h at room tempera-ture, and then boiled at 100C for 3 min to stop the reaction The 10· reaction buffer mixture contained 500 mm Tris ⁄ HCl buffer (pH 7.4), 100 mm MgCl2, 2.5 nm dithiothreitol, and

10 mm EDTA Analytical HPLC using the method described

by Di Pierro et al [37], with some modifications, was carried out on a Waters 2695 Separation Module The separation system consisted of a Prevail C-18 column (250· 4.6 mm,

5 lm particle size) (Alltech Associates, Inc., Deerfield, IL, USA) and a mobile phase developed with buffer A (10 mm

KH2PO4, and 8 mm tetrabutyl ammonium hydrogen sulfate

as the ion-pairing reagent, pH 7.0) and buffer B (100 mm

KH2PO4, 10 mm tetrabutyl ammonium hydrogen sulfate, 30% MeOH, pH 5.3) The gradient was formed as follows:

6 min with 100% buffer A; 1 min with 75% buffer A; 7 min with 58% buffer A; 2 min with 45% buffer A; 16 min with 20% buffer A; and 10 min with 100% buffer B The flow rate was 1.0 mLÆmin)1, and absorbance was monitored at

260 nm with a 2996 Photodiode Array Detector (Waters Corporation, Milford, MA, USA)

Microscopy and photography

Images of worms were captured using an AxioCam HRc digital camera attached to a Zeiss Axio Imager M1 micro-scope (Zeiss Corporation, Jena, Germany) axiovision Release 4.6 software (Zeiss) was used for image acquisition and processing