Cancer Models in Caenorhabditis elegans · 2020. 6. 17. · C. elegans is a free-living soil nema-tode that is approximately 1 mm in length. As the ﬁrst metazoan to have had its

a SPECIAL ISSUE REVIEWS-A PEER REVIEWED FORUM

Cancer Models in Caenorhabditis elegansNatalia V. Kirienko, Kumaran Mani, and David S. Fay*

Although now dogma, the idea that nonvertebrate organisms such as yeast, worms, and flies could inform,and in some cases even revolutionize, our understanding of oncogenesis in humans was not immediatelyobvious. Aided by the conservative nature of evolution and the persistence of a cohort of devotedresearchers, the role of model organisms as a key tool in solving the cancer problem has, however, becomewidely accepted. In this review, we focus on the nematode Caenorhabditis elegans and its diverse andsometimes surprising contributions to our understanding of the tumorigenic process. Specifically, we dis-cuss findings in the worm that address a well-defined set of processes known to be deregulated in cancercells including cell cycle progression, growth factor signaling, terminal differentiation, apoptosis, themaintenance of genome stability, and developmental mechanisms relevant to invasion and metastasis.Developmental Dynamics 239:1413–1448, 2010. VC 2010 Wiley-Liss, Inc.

Key words: C. elegans; cancer

Accepted 9 January 2010

Caenorhabditis

elegans AS A MODEL

ORGANISM FOR STUDYING

CANCER

Extensive research into the underlyingbasis of cancer has led to the generalconsensus that alterations in a definedset of biological activities are requiredfor cells to attain a fullymalignant state(Thompson et al., 1989; Renan, 1993;Kinzler and Vogelstein, 1996; Hanahanand Weinberg, 2000; Hahn and Wein-berg, 2002a,b). Key traits associatedwith oncogenic transformation includecell cycle deregulation, independencefrom growth factor signaling, immortal-ization, metastasis and invasion, avoid-ance of apoptosis and immune surveil-lance, genomic instability, and theinduction of angiogenesis. Although vir-tually all malignant cancers acquiremost or all of these defined characteris-tics during the course of their progres-

sion, the particular spectrum of muta-tions leading to these standardizedtraits can vary dramatically among dif-ferent types of cancers. This fact aloneis responsible for increasing the com-plexity of cancer biology by orders ofmagnitude. Importantly, the collectiveinsights required to attain a compre-hensive picture of the oncogenic processwill necessitate the use ofmany comple-mentary approaches. It is in this con-text that model organisms can continuetomake theirmark.

C. elegans is a free-living soil nema-tode that is approximately 1 mm inlength. As the first metazoan to havehad its genome sequenced, C. eleganshas benefited greatly from both classi-cal genetic and modern functionalgenomic approaches. Several character-istics of the worm make it highly ame-nable for cancer research. C. eleganshas a completely characterized andessentially invariant somatic cell line-

age, thereby facilitating the analysis ofphenotypes that disrupt normal prolif-eration and patterning (Sulston andHorvitz, 1977; Sulston et al., 1983). Inaddition, C. elegans is transparentthroughout all stages of development,allowing for the direct visualization ofall cells, including their divisions andmovements. Both forward and reversegenetics are well established andstraightforward, allowing for the com-prehensive analysis of genetic path-ways and protein functions. Finally,and quite critically, many human genesand pathways involved in cancer arehighly conserved in C. elegans. Indeed,these regulatory networks are ofteneasier to parse in C. elegans, as thegene families involved contain fewermembers, thus reducing the likelihoodfor genetic redundancy. Examples ofthis include the pRb andp53 tumor sup-pressors, which are estimated to bedirectly or indirectly compromised in

Dev

elop

men

tal D

ynam

ics

University of Wyoming, College of Agriculture, Department of Molecular Biology, Laramie, WyomingGrant sponsor: NIH; Grant number: GM066868; Grant sponsor: INBRE; Grant number: P20RR016474.*Correspondence to: David S. Fay, University of Wyoming, College of Agriculture, Department of Molecular Biology,Department 3944, 1000 E. University Avenue, Laramie, WY 82071. E-mail: [email protected]

DOI 10.1002/dvdy.22247Published online 19 February 2010 in Wiley InterScience (www.interscience.wiley.com).

DEVELOPMENTAL DYNAMICS 239:1413–1448, 2010

VC 2010 Wiley-Liss, Inc.

nearly all human cancers (Hunter andPines, 1994; Sherr, 2000, 2004; Nevins,2001; Sherr and McCormick, 2002;Yamasaki, 2003; Bindra and Glazer,2006; Knudsen and Knudsen, 2006).The pRb and p53 gene families containthree members each in mammals,whereas C. elegans contains only a sin-gle member of each family, LIN-35/pRband CEP-1/p53, respectively (Lu andHorvitz, 1998; Mulligan and Jacks,1998; Chen, 1999; Kaelin, 1999; Derryet al., 2001).

Perhaps surprisingly, of the acquiredmalignant characteristics listed above,the large majority can be addressedquite directly by studies in C. elegans.In this review, we will attempt to touchupon most of these areas, although therelative depth of coverage will varysomewhat among subjects. Our goal isto provide a comprehensive descriptionof the contributions of worm research tocancer biology and to serve as a startingpoint for readers to delve more deeplyinto the primary literature and more-specialized reviews that address thesesubject areas.

CELL CYCLE

DEREGULATION AND

CANCER

Mostmammalian cells receive a varietyof antiproliferative signals from theextracellular environment. These canlead to either transient arrest or perma-nent withdrawal from the cell cycle andterminal differentiation (Zetterberget al., 1995; Sun et al., 2007; Nishikawaet al., 2008). Thus, for tumorigenesis tooccur, cancer cells must develop theability to evade the influence of nega-tive growth signals to remain in a statethat is permissible for continuous pro-liferation. Most typically, this is accom-plished by acquiringmutations in genesthat regulate cell cycle entry and pro-gression. Importantly, many of the corecomponents of the cell cycle networkare conserved from yeast to mammals.In the established model, cell cycle pro-gression is driven by the physical asso-ciation of cyclin proteins with small ser-ine/threonine kinases known as cyclin-dependent kinases (CDKs; Bird, 2003).This interaction confers activity on theCDKs and allows them to phosphoryl-ate several downstream targets (Bird,2003). One of the most critical CDK

substrates involved in the G1-to-Sphase progression is pRb (Blomen andBoonstra, 2007). When in a relativelyunder-phosphorylated state, pRb blocksthe ability of ‘‘activating’’ E2F-familytranscription factors to promote theexpression of genes required for S phaseentry. Simultaneously, this hypophos-phorylated form of pRb (or its familymembers p107 and p130; collectivelytermed the pocket proteins), also associ-ates with distinct set of ‘‘inhibitory’’E2F-family members to directly medi-ate the transcriptional repression ofE2F targets (Schafer, 1998). CDK-de-pendent phosphorylation of pRb allevi-ates both inhibitory activities by induc-ing the dissociation of hyperpho-sphorylated pRb from E2Fs, therebyallowing S phase to proceed (StevauxandDyson, 2002).

As the regulation of cell cycle controlis critical for proper growth and devel-opment, CDKs are subjected toextremely tight regulatory control.Some of the mechanisms involvedinclude the controlled expression anddegradation of cyclins (Bird, 2003), theexpression of cyclin-dependent kinaseinhibitors (CKIs), which block cyclin-CDK activity (Sherr and Roberts,1999), targeted degradation of cell cycleregulatory proteins by the proteasome(van Leuken et al., 2008; Kitagawaet al., 2009), and the phosphorylationand dephosphorylation of key sites onCDKs by other regulatory protein ki-nases (Kellogg, 2003; Lolli and John-son, 2005; Queralt and Uhlmann, 2008;Lavecchia et al., 2009).

Importantly, most human cell cyclegenes have orthologs in C. elegans, andtheir functions have generally beenshown to be well conserved (van denHeuvel, 2005a,b). Notably,mutations inmany of the genes involved in control-ling the expression, activity, and degra-dation of the cyclin-CDK complexeslead to hyperproliferation of postem-bryonic cell lineages (Table 1). Twonoteworthy characteristics are sharedby these genes. First, virtually all ofthem are involved specifically in G1/Sprogression. This is also consistent witha relative lack of mutations that havebeen reported to lead to embryonichyperproliferation, as embryonic cycleslargely bypass G1 phase (Edgar andMcGhee, 1988). Second, all ofthe known hyperproliferation-inducingmutations affect genes that directly or

indirectly regulate the abundance or ac-tivity of G1/S cyclin-CDK complexes,indicating that the G1/S transition isthe most critical step in controlling celldivision in C. elegans. This is similar tothe case in human cancers, where alter-ations in cyclin D, CDK4, and p16INK4A

are detected more frequently thanmutations in genes that regulate othercell cycle events (Sherr, 1996). Althoughevidence to date has not directly shownthat overexpression of G1/S cyclin-CDKcomplexes can induce hyperprolifera-tion inC. elegans, their loss does lead toa reduction in cell numbers (Park andKrause, 1999; Fay and Han, 2000;Boxemand van denHeuvel, 2001).

CKIs REGULATE CELL

CYCLE QUIESCENCE

Two families of mammalian CKIs pro-mote terminal differentiation and cellcycle exit (Matsuoka et al., 1995; Parkeret al., 1995; Casaccia-Bonnefil et al.,1997). One, the INK4 family, which iscomposed of p16INK4a, p15INK4b,p18INK4c, and p19INK4d, does not appearto be conserved inC. elegans. The other,the human Cip/Kip family, consisting ofp21Cip1, p27Kip1, and p57Kip2 (Sherr andRoberts, 1999; Besson et al., 2008), hasa pair of orthologs in C. elegans, CKI-1and CKI-2, which share equal similar-ity to p21Cip1 and p27Kip1 (van denHeu-vel, 2005a). Homozygous null p27Kip1

mice exhibit several striking pheno-types, including gigantism, multiorgancell hyperproliferation, and increasedrates of cancer (Fero et al., 1996;Nakayama et al., 1996; Fero et al.,1998). In addition, p57Kip2 has beenimplicated in tumorigenesis in humans(Hatada et al., 1996a,b; Matsuokaet al., 1996; Thompson et al., 1996).In C. elegans, homozygous cki-

1(gk132) deletion mutants exhibit em-bryonic lethality, whereas heterozy-gotes display weak hyperproliferationof vulval precursor cells (VPCs), a set oflarval blast cells from which the vulvais derived (Saito et al., 2004; Buck et al.,2009). Strikingly, RNAi of cki-1 resultsin excessive proliferation of other post-embryonic tissues, including in thehypodermis, vulva, and somatic gonad(Hong et al., 1998). In contrast, homozy-gous cki-2(ok2105) mutants, as well ascki-2(RNAi), display only weak hyper-proliferation in VPCs (Buck et al.,2009), although double RNAi of cki-1

Dev

elop

men

tal D

ynam

ics

1414 KIRIENKO ET AL.

and cki-2 leads to intestinal cell hyper-proliferation, indicating that thesegenes are partially redundant in theirfunctions. Furthermore, overexpres-sion of either CKI-1 or CKI-2 triggerscell cycle arrest (Hong et al., 1998;Fukuyama et al., 2003), indicating thatCKI-1 and CKI-2 are sufficient toinduce cell cycle quiescence.

Notably, the excess vulval cells thatare generated in cki-1(RNAi) animalsresult from the precocious division ofnormally quiescent VPCs during earlylarval development, indicating that de-velopmental timing is also misregu-lated in cki-1 mutants (Hong et al.,1998). Although it is generally believedthat cell cycle control must be inti-

mately coordinated with developmen-tal-stage progression, direct linksbetween these two processes have gen-erally been lacking. Thus, a uniquefacet to the analysis of the C. elegansCKI orthologs has been the identifica-tion of proteins that regulate the timingof CKI-1 expression (Fig. 1). Two suchregulators, LIN-14 and LIN-29, posi-tively regulate expression of cki-1 inVPC and hypodermal lineages, respec-tively (Hong et al., 1998). LIN-14 andLIN-29 belong to the heterochronicgroup of proteins, which are involved inthe global regulation of developmentaltiming during postembryonic develop-ment (Moss, 2007). Notably, mutationsin lin-14 and lin-29 also cause supernu-

merary cell divisions, as cell divisionpatterns that are characteristic of earlydevelopmental events are typically reit-eratedmultiple times (Ambros andHor-vitz, 1984). Thus, in the absence of LIN-14 or LIN-29 activity, CKI-1 expressionis reduced, resulting in the prematuretermination of G0 quiescence and ec-topic divisions within several postem-bryonic lineages. Using a clever geneticscreen, Saito and colleagues identifiedtwo additional cki-1 transcriptional reg-ulators, lin-1/Ets and lin-31/FoxB1(Clayton et al., 2008). Ets family tran-scription factors have important rolesin proliferation and differentiation dur-ing development and have been sug-gested as targets for cancer therapies

TABLE 1. C. elegans Hyperplasia-Associated Genes

Common name Worm gene Mammalian ortholog Affected tissue(s)

cdc-14 C17G10.4 Cdc14 Hypodermis1, VPCs1

cdc-25.1(gf) K06A5.7 Cdc25A Intestine2,3

cdk-8 F39H11.3 Cdk8 VPCs4

cki-1 T05A6.1 Cip/Kip family Hypodermis5, VPCs5, DTC5,Germ cells5, Intestine6,Multiple6,7,8,a, Multiple7,8,b

cki-2 T05A6.2 Cip/Kip family VPCs6, Multiple6,7,8,c,Multiple7,8,d

cpb-1 C40H1.1 CBP/p300 Embryonic cells9

cul-1 D2045.6 Cul1 All postembryonic blast cells10

dpl-1 T23G7.1 DP Intestine8,e

efl-1 Y102A5C.18 E2F4, E2F5 Intestine8,e

fzr-1 ZK1307.6 Cdh1/Hct1/FZR Intestine11, Multiple11,f

let-7 C05G5.6 Let7 Hypodermis12

lin-1 C37F5.1 Ets VPCs4

lin-4 F59G1.6 miR-125 Hypodermis13, Intestine13,Sex myoblasts13, VPCs13

lin-14 T25C12.1 None known Hypodermis14, Intestine14,Sex myoblasts14, VPCs14

lin-23 K10B2.1 bTRCP All postembryonic blast cells15

lin-29 W03C9.4 None known Hypodermal seam cells14

lin-31 K10G6.1 FoxB1 VPCs4

lin-35 C32F10.2 Retinoblastoma, p107, p130 Intestine16,17,Multiple7,8,e,Multiple11,g

lin-36 F44B9.6 None known Intestine8,e

mdt-1.1/sop-3 Y71F9B.10 None Known VPCs4

mdt-12/dpy-22 F47A4.2 TRAP230 VPCs4

mdt-13/let-19 K08F8.6 TRAP240 VPCs4

mdt-23/sur-2 F39B2.4 MED23 VPCs4

skr-1 F46A9.5 p19Skp1 All postembryonic blast cells18

skr-2 F46A9.4 p19Skp1 All postembryonic blast cells18

ain combination with cki-2; bin combination with cki-2 and lin-35; cin combination with cki-1; din combination with cki-1 and lin-35; ein combination with cki-1/2; fin combination with lin-35; gin combination with fzr-1; 1(Saito et al., 2004); 2(Clucas et al.,2002); 3(Kostic and Roy, 2002); 4(Clayton et al., 2008); 5(Hong et al., 1998); 6(Buck et al., 2009); 7(Boxem and van den Heuvel,2001); 8(Boxem and van den Heuvel, 2002); 9(Shi and Mello, 1998); 10(Kipreos et al., 1996); 11(Fay et al., 2002); 12(Reinhart et al.,2000); 13(Chalfie et al., 1981); 14(Ambros and Horvitz, 1984); 15(Kipreos et al., 2000); 16(Grishok and Sharp, 2005); 17(Ouellet andRoy, 2007); 18(Nayak et al., 2002).

Dev

elop

men

tal D

ynam

ics

CANCER MODELS IN C. elegans 1415

(Hahne et al., 2008). Although LIN-29and LIN-14 do not have obvious ortho-logs in humans and little is knownabout the role of mammalian FoxB1,the regulation of cki-1 by these tran-scription factors suggests that furtherinvestigation is justified.

Mammalian CKI proteins are alsoposttranslationally regulated. For ex-ample, phosphorylation of CKI familymembers by Akt alters their subcellularlocalization and affinity for particulartargets, as well as triggering their deg-radation (Montagnoli et al., 1999; Fujitaet al., 2002; Liang et al., 2002). It is notcurrently knownwhether theC. elegansAkt co-orthologs AKT-1/2 directly phos-phorylate CKI-1/2. AKT-1/2 does, how-ever, inhibit DAF-16/FOXO activity byphosphorylation, and DAF-16/FOXO isrequired for the transcription of cki-1during the L1 larval stage (Ogg et al.,1997; Henderson and Johnson, 2001;Baugh and Sternberg, 2006). Therefore,in nutrient-deprived daf-16/FOXO mu-tants, CKI-1 levels are insufficient tomediate the normal pattern of starva-tion-induced cell division arrest (Baughand Sternberg, 2006). Notably, thismechanism provides a direct linkbetween food availability, sensed by theinsulin signaling pathway, and cell cyclecontrol.

Phosphorylational control of CKI-1 inC. elegans is also suggested by the obser-vation that the CDC-14/Cdc14 proteinphosphatase is required for the stabiliza-tion of CKI-1 expression (Saito et al.,2004). Inhibition of CDC-14 activityleads to the hyperproliferation of VPCs,intestinal nuclei, and hypodermal cells(Saito et al., 2004). Furthermore, overex-pression of CDC-14 in wild-type leads toa reduction in the normal number of in-testinal cells in aCKI-1–dependentman-ner. These results suggest that in C. ele-gans, CDC-14 is critical for promotingcellular quiescence through the mainte-nance of CKI-1 stability (Fig. 1). It is alsoworth noting that, although Saccharo-myces cerevisae cdc14 was originallyimplicated in controlling transit throughmitosis, a recent report suggests thatCDC14 may also function in G1/S regu-lation in other organisms (Dulev et al.,2009).

PHOSPHORYLATION REGU-

LATES CDK ACTIVITY

As mentioned above, CDK phosphoryl-ation is a well-conserved means for reg-

ulating cell cycle progression in manysystems (Fattaey and Booher, 1997;Schmitt and Nebreda, 2002; Kellogg,2003; Han et al., 2005; Burrows et al.,2006). Inhibitory phosphorylation ofCDKs is carried out by members of theWee1/Myt1 family of protein kinases.Acting in opposition to Wee1/Myt1 arethe CDC25 family of protein phospha-tases, which remove inhibitory phos-phate groups from CDKs, thereby pro-moting cell division (Boutros et al.,2007). In C. elegans, the best-studiedmember of the Wee1 family is wee-1.3,which is actively transcribed through-out development. Putative null allelesof wee-1.3 lead to embryonic lethality,suggesting a general role in cell cycleregulation (Lamitina and L’Hernault,2002). One well-established target ofWEE-1.3 is theC. elegansCDK-1/cyclinB complex, thus directly implicatingWEE-1.3 in the control of the G2/Mtransition (Burrows et al., 2006).

Although no hyperproliferation de-fects have been reported for wee-1.3loss-of-function mutants (or for otherC.elegans Wee1 homologs), postem-bryonic intestinal hyperplasia has beenobserved in strains containing gain-of-function (gf)mutations in cdc-25.1 (Clu-cas et al., 2002; Kostic and Roy, 2002).Of interest, although cdc-25.1(gf)mutants do not exhibit hyperprolifera-tion outside of the intestinal lineage,loss of function of CDC-25.1 activityleads to pleiotropic defects including awidespread failure in cell proliferation.Thus, although CDC-25.1 appears toplay an important role in cell cycle pro-gression in multiple tissues, the intes-tine may be particularly sensitive toperturbations in this pathway. Addi-tionally, cdc-25.1(gf) alleles result inreduced rates of protein turnover byaltering internal phosphodegronsequences that are targeted by the LIN-23/bTRCP ubiquitin-ligase complex(Hebeisen andRoy, 2008).

Ubiquitin-Mediated Proteolysis

and Cell Cycle Control

Protein degradation through ubiquitin-mediated proteolysis is known to playan important role in regulating theexpression of cyclins, CKIs, and manyother cell cycle proteins in mammals(Hochstrasser, 1995). Initial insightinto the role of ubiquitin-mediated pro-teolysis in C. elegans cell cycle control

beganwith the characterization of cul-1mutants, which display extensivehyperplasia of the hypodermis, vulva,uterus, and sex muscles (Kipreos et al.,1996). CUL-1 has served as the found-ing member of the C. elegans cullin pro-tein family, which is conserved fromyeast to humans. A second gene, lin-23/bTRCP, was identified through muta-tions that conferred similar hyperprolif-eration defects to those of cul-1 (Kipreoset al., 2000). Both lin-23 and cul-1mutants show normal embryonic devel-opment but undergo hyperplasia begin-ning in larval stages. However, in con-trast to cki-1 and cdc-14 mutants,which initiate cell divisions preco-ciously, excess cells in lin-23 and cul-1mutants occur not as a result of prema-ture cell cycle entry but instead becauseof an inability to exit the cell cycle inresponse to normal developmental cues(Kipreos et al., 1996, 2000; Hong et al.,1998; Saito et al., 2004).In mammals, bTRCP and Cul1, to-

gether with p19Skp1 and the RING Boxprotein Rbx1, function as an E3 ubiqui-tin-ligase complex (termed SCF). Mam-malian SCF complexes catalyze theubiquitylation, and subsequent de-struction, of several proteins that arecritical for proper cell cycle regulation,including CDC25A; p27Kip1; cyclins A,B, D, and E; Emi1; and Wee1 (Yu et al.,1998b; Carrano et al., 1999;Dealy et al.,1999; Sutterluty et al., 1999; Tsvetkovet al., 1999; Nakayama et al., 2000;Busino et al., 2003; Guardavaccaroet al., 2003; Margottin-Goguet et al.,2003; Watanabe et al., 2004). Likewise,C. elegans SCF complex componentshave been implicated in the degrada-tion of CDC-25.1 and CYE-1/cyclin E(Fay et al., 2002; Hebeisen and Roy,2008). Consistent with this, mutationsin two worm homologs of p19Skp1, skr-1and skr-2, lead to the hyperproliferationof multiple postembryonic lineages(Nayak et al., 2002). Although C. ele-gans RBX-1 and RBX-2 have roles inmultiple aspects of mitotic and meioticprogression, no hyperproliferation phe-notypes have been detected in mutants(Sasagawa et al., 2003). In addition tocell cycle control, both the worm andmammalian SCF complexes regulateadditional targets that may be relevantto tumorigenesis, including the Wntpathway member and cell adhesioncomponent b-catenin (Fuchs et al.,1999; Hart et al., 1999; Latres et al.,

Dev

elop

men

tal D

ynam

ics


1999; Winston et al., 1999; Dreier et al.,2005). Furthermore, SCF-related E3complexes target additional cell cycleregulators. For example, CKI-1 is a tar-get for ubiquitylation by an E3 ubiqui-tin-ligase complex involving CUL-4/Cul4 and DDB-1/DDB1 in C. elegans(Kim et al., 2007), which is consistentwith a similar reported activity in Dro-sophila and humans (Bondar et al.,2006;Higa et al., 2006; Li et al., 2006).

pRb IS A

PHYLOGENETICALLY CON-

SERVED CELL CYCLE

REGULATOR

One of the most widespread eventsleading to oncogenesis in humans is theloss of G1/S regulation resulting fromthe inactivation of pRb (Weinberg,1995). Loss of pRb activity can occurthrough either direct mutations withinthe RB1 locus or by alterations to otherloci that result in the functional inacti-vation of pRb. Mutations in this latterclass include loss of function of theINK4 family of CKIs or the amplifica-tion and overexpression of Cyclin D andCdk4 family members (Sherr, 1996;Nevins, 2001). C. elegans, with its sin-gle pocket protein family member, LIN-35, provides a streamlined system fordeciphering pRb family functionswithin the context of an intact develop-ing organism. In addition, the C. ele-gans genome encodes two bona fideE2F family members (EFL-1 and EFL-2) and one E2F dimerization partner,DPL-1/DP; mammals contain eightE2Fs and two DP family proteins (LaThangue, 1994; Yamasaki, 1998; Ceoland Horvitz, 2001; Page et al., 2001;DeGregori and Johnson, 2006; van denHeuvel and Dyson, 2008). This mini-malistic, but complete, pathway config-uration lowers the potential for intra-gene family functional redundancy. Forexample, in mammals, pRb, p107, andp130 have significant functional over-lap, thereby complicating the analysisof both the collective and individualactivities of these paralogs (Cobriniket al., 1996; Lee et al., 1996; Mulliganand Jacks, 1998; Robanus-Maandaget al., 1998; Sage et al., 2000; Dannen-berg et al., 2004).

Despite the fact that many cell cycle-related genes are misregulated follow-ing loss of LIN-35 activity, lin-35 null

single mutants are viable and displayonly minimal cell proliferation defects(Lu and Horvitz, 1998; Kirienko andFay, 2007; Ouellet and Roy, 2007). Forthis reason, the great majority of cellcycle functions ascribed to LIN-35 haveresulted from the analysis of compoundmutants. For example, whereas bothlin-35 and fzr-1/Cdh1 single mutantsexhibit only weak hyperproliferation ofthe intestine, lin-35; fzr-1 doublemutants display pronounced hyperpla-sia in multiple lineages (Fay et al.,2002). This synthetic phenotype isthought to result from an inability tonegatively regulate G1 cyclins througheither transcriptional repression bymeans of LIN-35–EFL-1 or at the levelof protein stability through FZR-1,which functions as a specificity compo-nent of the anaphase-promoting-com-plex (APC) E3 ubiquitin-ligase (Schwabet al., 1997; Sigrist and Lehner, 1997;Visintin et al., 1997; Fay et al., 2002).Likewise, inhibition of lin-35 activityenhances hyperproliferation in cki-1/cki-2 (RNAi), cdc-14, and lin-31mutants (Boxem and van den Heuvel,2001; Saito et al., 2004). Hyperprolifer-ation in these latter examples is likelydue to the cumulative effects of inacti-vating multiple pathways that inhibitthe G1/S transition. Loss of lin-35 canalso suppress developmental defectsin cyd-1/cyclin D and cdk-4/Cdk4mutants, which is consistent with itsrole as a downstream target of thesefactors in mammals (Boxem and vandenHeuvel, 2001).

It is worth noting that LIN-35/Rbalso carries out several functions thatare distinct from its role in cell cyclecontrol. These include promoting germ-line apoptosis (Grote and Conradt,2006; Reddien et al., 2007; Schertel andConradt, 2007), inhibiting ectopicgrowth factor signaling (Lu and Hor-vitz, 1998), and preventing the expres-sion of germline traits in somatic cells(Wang et al., 2005; Kirienko and Fay,2007). In addition, LIN-35 redundantlycontrols organ morphogenesis (Fayet al., 2003; Bender et al., 2007; Maniand Fay, 2009), cell fate specification(Bender et al., 2004a), maintenance oforgan homeostasis (Kirienko et al.,2008), and other developmental proc-esses essential for fertility and viability(Cui et al., 2004; Cardoso et al., 2005;Chesney et al., 2006; Ceron et al., 2007).LIN-35 also plays a role in controlling

ribosome biogenesis (Voutev et al.,2006), promoting the expression oftransgenes (Hsieh et al., 1999), and in-hibiting the RNAi response in somatictissues (Wang et al., 2005; Lehner et al.,2006). Notably, many of these reportedfunctions have correlates with some ofthe proposed tumor-suppressing func-tions of mammalian pocket proteins.Thus, the role of pRb family membersin repressing malignant transforma-tion may extend well beyond cell cycleregulation (Fig. 2).

NOVEL CELL CYCLE

REGULATORS

Although the roles and activities ofmany hyperplasia-associated genes inC. elegans and mammals are welldefined and can be traced to specific cellcycle functions, critical mechanisticdetails for others are still lacking. Forexample, themammalianCdk8 protein,a component of the conserved Mediatorsubcomplex (Akoulitchev et al., 2000),functions generically in target generepression (Kobor and Greenblatt,2002; Knuesel et al., 2009a,b). CDK-8,along with several other members ofthe Mediator complex, including MDT-13, is required in C. elegans for themaintenance of cell cycle quiescence inVPCs (Clayton et al., 2008). It is unclearwhy loss of the Mediator complex com-ponents leads to ectopic divisions, asexpression of CKI-1 and G1 cyclins arenot altered inmdt-13mutants (Claytonet al., 2008). Nevertheless, these find-ings suggest that additional means ofcell cycle control remain to be identifiedthrough further studies inC. elegans.

GROWTH SIGNAL SELF-

SUFFICIENCY

As discussed above, the core machineryand molecular mechanisms controllingcell cycle progression are highlyconserved between mammals andC. elegans, making direct comparisonsstraightforward. The signals that trig-ger cell proliferation, on the other hand,are quite different. In mammals, mito-genic factors (MFs; also referred to asgrowth factors) are generally requiredto initiate exit from the quiescent state.Most often, these signals are conveyedfrom the exterior of the cell through re-ceptor tyrosine kinases (RTKs), leadingto the activation of intracellular

Dev

elop

men

tal D

ynam

ics


signaling pathways and changes ingene expression (Hanahan and Wein-berg, 2000). As these pathways arerequired for the promotion of cell divi-sion in mammals, it is not surprisingthat alterations in growth factor signal-ing components are commonly observedin tumors. These include mutationsthat lead to the autonomous productionof MFs by cancerous cells (Andraeet al., 2008; Palumbo et al., 2008), over-expression or constitutive activation ofgrowth factor receptors or their down-stream signaling components (Kalekoet al., 1990; Cappellen et al., 1999; Onoand Kuwano, 2006; Palumbo et al.,

2008; Acevedo et al., 2009; Werner andBruchim, 2009; Zahorowska et al.,2009), and the expression of variantextracellular matrix-associated pro-teins that influence signaling strength(Wilkins-Port and Higgins, 2007;Dydensborg et al., 2009).

C. elegans, as previously mentioned,has an invariant somatic cell lineage(Sulston and Horvitz, 1977; Sulstonet al., 1983). In addition, growth inadult worms occurs by cell expansionrather than cell division, a processlargely dependent upon endoreduplica-tion events in the expanding cells(Flemming et al., 2000; Lozano et al.,2006). Furthermore, cell proliferation

per se is not dependent on growth factorsignaling. Rather, these pathways areused instead to regulate developmentalevents such as the induction of specificcell fates and the coordination of cellmigration events. In addition, growthfactor signaling pathways are used inC. elegans to respond to environmentalchallenges, such as starvation, stress,and infection by pathogens (Johnsonet al., 2002; Kim et al., 2002, 2004; Gar-sin et al., 2003; Troemel et al., 2006;Hu,2007). Despite its apparent lack ofinvolvement in cell proliferation con-trol, there are nevertheless compellingreasons to study growth factor signal-ing in C. elegans, as the components of

Fig. 3.

Fig. 2. Pocket protein functions are phyloge-netically conserved. Comparison of the reportedfunctions for mammalian pocket proteins in tu-mor suppression and the known functions of thesole C. elegans pocket protein ortholog, LIN-35/pRb. Identically colored lines correspond torelated or analogous functions.

Fig. 1. Regulation of CKI-1. Model of factorscontrolling the expression and activity of CKI-1. LIN-1, LIN-14, LIN-29, LIN-31, and DAF-16promote transcription of cki-1, whereas CDC-14 is thought to stabilize CKI-1 proteinthrough removal of an inhibitory phosphate.The antagonist of the CDC-14 protein phos-phatase, a presumed kinase, is unknown in C.elegans. Yellow shading indicates the exis-tence of human orthologs. Dashed arrow indi-cates translation of cki-1 mRNA into protein.

Dev

elop

men

tal D

ynam

ics


these pathways are often conserved,even if their specific biological functionsare not. More generally, C. elegans hasprovided seminal insights into the regu-lation of many canonical signalingpathways connected to tumorigenesisincluding Delta-Notch, Wnt, insulingrowth factor (IGF), and epidermalgrowth factor (EGF; Joneson and Bar-Sagi, 1997; Polakis, 1999; Waltzer andBienz, 1999; Brown, 2001; Lustig andBehrens, 2003; Weng et al., 2004; Dunnet al., 2005; Dhillon et al., 2007; Boloset al., 2007, 2009; Khavari and Rinn,2007; Leicht et al., 2007; Samani et al.,2007; Frasca et al., 2008; Pollak, 2008;Thurston andKitajewski, 2008;Wernerand Bruchim, 2009). For illustrativepurposes, we will focus here on the con-tributions ofC. elegans research towardelucidating conserved components ofEGF signaling, asmany of these factorshave subsequently been demonstratedto play a role in human tumorigenesis.

The EGFR-Ras-MAPK pathway isthought to be mutated in approxi-mately one-fifth of human cancers(Montagut and Settleman, 2009).

Notably, mutations that lead toenhanced activity of the C. elegansRas ortholog, LET-60, occur at identi-cal residues to those found in consti-tutively activated Ras from humantumors (Ferguson and Horvitz, 1985;Bos, 1989; Beitel et al., 1990). Asdescribed below, several distinctstrategies have been used to identifynovel components of this pathway inC. elegans, although most are basedon forward-genetic approaches. Themost well-studied phenotypes associ-ated with C. elegans EGF signalingare those that affect the specificationof vulval tissue (Sundaram, 2006).The development of the vulva isinduced in response to signaling fromthe somatic gonad anchor cell, whichsecretes an EGF-like signal, LIN-3,which binds to the LET-23/EGFR re-ceptor on the surface of VPCs (Hilland Sternberg, 1992). This signaltriggers the activation of a canonicalRas-MAPK cascade that results in thespecification of three vulval cell pro-genitors, which then undergo stereo-typical divisions and morphogenetic

events to form the mature organ.In addition, coordinated signalingthrough the Delta-Notch and Wntpathways is also required to furtherrefine vulval fates and patterningevents (Sternberg, 2005). Loss of func-tion in EGFR-Ras-MAPK signalingleads to a reduction in or absence ofvulval tissue in the adult, such thatgravid hermaphrodites cannot lay eggs[the Vulvaless (Vul) or Egg laying de-fective (Egl) phenotype]. Conversely,gain of function in EGFR-Ras-MAPKactivity leads to an over-production ofvulval tissue, causing a multivulval(Muv) phenotype. Importantly, bothphenotypes are straightforward toscore and neither result in inviability.The complete EGFR-Ras-MAPK

pathway involved in vulval cell induc-tion is depicted in Figure 3. Although asignificant number of these pathwaycomponents were first identified usinggenetic screens in C. elegans, the func-tions of these proteins are generallyconserved across diverse phyla. Manyof the core components were identifiedin simple screens for mutations thatcause either Vul or Muv phenotypes. Inthe case of loss-of-function mutations,the Vul and Muv phenotypes suggestedthat the affected genes encode positiveand negative regulators of the signalingpathway, respectively. More than20 genes were identified using thisapproach including LIN-3/EGF, LET-23/EGFR, LET-60/Ras, and the down-stream effectors LIN-1/Ets and LIN-31/FoxB1 (Horvitz and Sulston, 1980; Fer-guson and Horvitz, 1985). In addition,several pathway modulators were iden-tified in secondary screens for muta-tions that suppress or enhance loss-of-function alleles of let-23/EGFR, sli-1/c-Cbl, lip-1/MKP, lin-2/CASK, and lin-10/APBA2. These include SLI-1/c-Cblitself, UNC-101/AP-1m1-1, ARK-1/Ack,DEP-1/Dep-1/Scc, and GAP-1/Gap1(Lee et al., 1994; Jongeward et al., 1995;Sternberg et al., 1995; Yoon et al., 1995;Hajnal et al., 1997; Hopper et al., 2000;Berset et al., 2005).SLI-1 provides a particularly inter-

esting example, as its mammaliancounterpart, c-Cbl, had previously beenimplicated as a proto-oncogene,although its molecular functions wereunknown (Langdon et al., 1989). Muta-tions in sli-1were discovered in screensfor mutations that would suppress theVul phenotype of a let-23 partial loss-of-

Fig. 3. The EGFR-Ras-MAPK pathway in C. elegans vulval development. Activation of theEGFR-Ras-MAPK pathway in C. elegans vulval development is initiated when LIN-3/EGF ligandsecreted by the gonadal anchor cell (AC) binds to the LET-23/EGFR receptor, causing it todimerize and undergo autophosphorylation (Aroian et al., 1990; Hill and Sternberg, 1992). Thisevent allows the SEM-5/Grb2 adaptor to bind phosphorylated LET-23 and recruit the SOS-1/Sos1 guanine nucleotide exchange factor (GEF; Clark et al., 1992a; Chang et al., 2000). SOS-1promotes the exchange of GDP for GTP on the LET-60/Ras GTPase, which in turn activates aprotein kinase cascade that includes LIN-45/Raf, MEK-2/MEK, and MPK-1/MAPK/ERK (Hanet al., 1993; Sternberg et al., 1993; Lackner et al., 1994; Wu and Han, 1994; Kornfeld et al.,1995a; Wu et al., 1995; Chong et al., 2003). Targets of MPK-1 phosphorylation include the com-plex of LIN-1/Ets and LIN-31/FoxB1, which dissociates after phosphorylation, thereby allowingactivated LIN-31 to promote the acquisition of vulval cell fates (Tan et al., 1998). Additional posi-tive regulators of vulval fates include SUR-2/Med23 and LIN-39/Hox-B5, a member of thehomeobox family of proteins (Clark et al., 1993; Singh and Han, 1995). Known negative regula-tors include members of the SynMuv group of proteins and CBP-1/p300 (Eastburn and Han,2005; Fay and Yochem, 2007). Pathway modulators include LIN-2/CASK, LIN-7/Lin-7C, and LIN-10/APBA2, which mediate the subcellular localization of LET-23 and UNC-101/AP-1m1-1 andAPM-1/AP-1m1-2, which promote LET-23 endocytotic recycling (Lee et al., 1994; Sternberget al., 1995; Kaech et al., 1998; Shim et al., 2000). SLI-1/c-Cbl negatively regulates the pathwayby targeting activated LET-23 for internalization and degradation (Rubin et al., 2005; Swamina-than and Tsygankov, 2006). The roles of ARK-1/Ack and DEP-1/Dep1/Scc in inhibiting LET-23signaling are not yet understood. RHO-1/RhoA and its GEF, ECT-2/Ect2, positively regulate thepathway downstream of SEM-5 (Canevascini et al., 2005). The role of the PTP-2/PTPN11 proteintyrosine phosphatase is not well understood, but it may act at several nodes upstream of LET-60 signaling (Gutch et al., 1998; Chang et al., 2000). LET-60 GTP hydrolysis is stimulated byGAP-1/Gap1 (Hajnal et al., 1997). The putative scaffold proteins KSR-1/Ksr1 and SUR-8/Sur8are required for robust signaling downstream of LET-60 (Kornfeld et al., 1995b; Sundaram andHan, 1995; Sieburth et al., 1998; Stewart et al., 1999). SUR-5/Sur5/AACS, an aminoacyl CoAsynthase, inhibits the pathway at the level of LET-60 through an unknown mechanism (Gu et al.,1998). MPK-1/MAPK/ERK is dephosphorylated by the MAPK phosphatase (MKP) family proteinLIP-1/MKP, which renders it inactive (Berset et al., 2005). In addition, KSR-1 activity is regulatedin part by zinc ion concentrations that are controlled through CDF-1/Znt1 and SUR-7/CDF, bythe SUR-6/PP2A protein serine/threonine phosphatase, and by the PAR-1/MARK1 kinase (Jaku-bowski and Kornfeld, 1999; Sieburth et al., 1999; Bruinsma et al., 2002; Yoder et al., 2004). Yel-low fill indicates proteins that were first implicated in EGFR-Ras-MAPK signaling through studiesin C. elegans.

Dev

elop

men

tal D

ynam

ics


function allele (Jongeward et al., 1995).This finding implicated SLI-1/c-Cbl as anegative regulator of LET-23/EGFRsig-naling, thus suggesting that it may nor-mally function as a tumor suppressor.Both SLI-1 and c-Cbl possess RING fin-ger and SH3 domains, and c-Cbl wassubsequently shown to directly bindGrb2, the human ortholog of the SEM-5adapter (Odai et al., 1995; Yoon et al.,1995). Furthermore, the presence of aRING finger domain suggested a rolefor SLI-1/c-Cbl in ubiquitin-mediatedproteolysis. Consistent with this, c-Cblwas shown to ubiquitylate EGFR invitro (Levkowitz et al., 1998). Moreover,ubiquitylation of EGFR by c-Cbl occursfollowing ligand binding and receptoractivation (Haglund et al., 2003; Moses-son et al., 2003; Rubin et al., 2005;Swaminathan and Tsygankov, 2006).One model to emerge is that Grb2,when bound to phosphorylated EGFR,recruits c-Cbl for the purpose of degrad-ing the receptor and terminating RTKsignaling in a classic negative-feedbackloop (Swaminathan and Tsygankov,2006). Thus, in the absence of c-Cbl,EGFR signaling goes unchecked, thusleading to hyperproliferation.

A second series of suppressor screenssought to identify negative regulatorsof EGFR-Ras-MAPK signaling basedon the ability of mutations to suppressthe Muv phenotype. A screen using alin-15 Muv background led to the iden-tification of sem-5/Grb2, sos-1/let-341/Sos1, let-60/Ras, and lin-45/Raf as coredeterminants of vulval cell fate acquisi-tion (Han et al., 1990; Aroian andStern-berg, 1991; Clark et al., 1992a,b; Hanet al., 1993; Hsu et al., 2002), whereasalleles of let-23/EGFR, lin-45/Raf, mek-2/MEK, and mpk-1/ERK were identi-fied using gain-of-function let-60/Ras al-leles (Lackner et al., 1994;Wu andHan,1994; Kornfeld et al., 1995a; Wu et al.,1995; Hsu et al., 2002). In addition, alarge number of noncore pathway regu-lators, including CDF-1/Znt1, KSR-1/Ksr1, SUR-2/MDT-23/Sur2, SUR-6/PPP2R2A, SUR-7/CDF, and SOC-2/SUR-8/Sur8, were also identified assuppressors of let-60(gf) Muv, whereasSUR-5/AACS was identified as a sup-pressor of a dominant-negative let-60Vul allele (Kornfeld et al., 1995b; Singhand Han, 1995; Sundaram and Han,1995; Gu et al., 1998; Sieburth et al.,1998; Jakubowski and Kornfeld, 1999;Sieburth et al., 1999; Bruinsma et al.,

2002; Yoder et al., 2004). Finally, a fewimportant pathway members, includ-ing MEK-2/MEK, and the MAPK phos-phatase, LIP-1, along with PTP-2/PTPN11 and APM-1/AP-1m2, wereidentified by homology searches forEGFR-Ras-MAPK pathway membersknown in either C. elegans or othermodel organisms (Wu et al., 1995;Gutch et al., 1998; Shim et al., 2000;Berset et al., 2005).

The identification and analysis ofKSR-1/Ksr1 exemplifies the ability ofgenetic screens in C. elegans and othermodel systems to provide novel insightsinto the regulation of mammaliangrowth factor signaling pathways(Kornfeld et al., 1995b; Sundaram andHan, 1995; Therrien et al., 1995). Nota-bly, Drosophila Ksr/SR3-1 was identi-fied at the same time as the C. elegansortholog, using an analogous screen tostudy the role of Ras signaling in eye de-velopment (Therrien et al., 1995). Fur-thermore, both reports implicatedKSR-1/Ksr in the positive regulation ofEGFR-Ras-MAPK signaling. Follow-upstudies in mammalian cell lines alsoindicated that Ksr1 binds to multiplecomponents of Ras-MAPK pathwaysincludingMEK2, ERK, andRAF, a find-ing that has been partially corroboratedin C. elegans (Michaud et al., 1997;Xing et al., 1997; Denouel-Galy et al.,1998; Joneson et al., 1998; Yu et al.,1998; Stewart et al., 1999). However,overexpression of mammalian Ksr1was initially reported to stabilize MEKin an inactive form and thus inhibitRas-mediated transformation in cellculture, suggesting a negative regula-tory role inEGFR-Ras-MAPK signaling(Denouel-Galy et al., 1998). Construc-tion of a mouse Ksr1 knockout recon-ciled this discrepancy, showing thatmammalian Ksr1 is required specifi-cally for Ras-mediated transformation,which is consistent with findings inworms and flies (Lozano et al., 2003).The most likely explanation for theincongruity in these reports is that, as ascaffolding protein, Ksr1 expressionmust be tightly regulated as fluctua-tions in either direction could poten-tially lead to the assembly of nonfunc-tional protein complexes and reducedsignaling.

Notably, the functions of the Raspathway-associated oncogenes CASK,Grb2, Sur2, Sur8, and c-Cbl were firstunderstood in C. elegans (Baines, 1996;

Hoskins et al., 1996; Kaech et al., 1998;Wang et al., 2002b). In addition, severaldownstream effectors of the C. elegansRas-MAPK pathway may also be rele-vant to cancer progression. At least twoknown targets of this pathway, lin-1/Ets and lin-31/FoxB1, are involved inmaintaining cell cycle quiescence byregulating cki-1/CKI (Clayton et al.,2008). Similarly in mammals, Ets tran-scription factors are targets of Ras-MAPK signaling and regulate p21Cip1

(Zhang et al., 2003).Interestingly, a direct link has been

established between the C. elegans Ras-MAPK pathway and the LIN-35/pRb tu-mor suppressor through the study ofsynthetic multivuvlal (SynMuv) mu-tants. Two primary classes of SynMuvgenes have been described, termed Aand B (Ferguson and Horvitz, 1985; Fayand Yochem, 2007; van den Heuvel andDyson, 2008).Whereas in general, singlemutations in either A or B class genesfail to display vulval induction defects,when combined, class A–B doublemutants show a highly penetrant Muvphenotype. Molecular cloning of theclass B genes identified LIN-35, alongwith several conserved transcriptionalregulators including EFL-1/E2F, DPL-1/Dp, and HDA-1/histone deacetylase (Luand Horvitz, 1998; Ceol and Horvitz,2001). These and other class Bmemberscomprise an overlapping set of con-served transcriptional repressor com-plexes (e.g., NURD, and DRM), whichtogether with the SynMuv class A pro-teins, inhibit the ectopic expression oflin-3 in the hypodermis (Cui et al.,2006a). Thus, in class A–B doublemutants, exogenous LIN-3 is producedfrom the hypodermis, leading to thehyper-induction of vulval cell fates andthe Muv phenotype (Cui et al., 2006a;Andersen et al., 2008). This linkbetweenLIN-35 and growth factor expression inC. elegans correlates with findings inmammals where pRb controls theexpression of growth factors includingVEGF, thus suggesting a novel mode bywhich pocket proteins may exert theirtumor suppressive effects (Gabelliniet al., 2006; Chien et al., 2007). It is alsoworth noting that additional factorsidentified through studies of SynMuvmutants (e.g., LIN-9), have been subse-quently been linked to cancer in humans(Gagrica et al., 2004).In mammals, many additional

growth factors act through Ras-MAPK

Dev

elop

men

tal D

ynam

ics


signaling including FGF, VEGF, IGF,and PDGF. Correspondingly in C. ele-gans, two FGF ligand orthologs (EGL-17and LET-756) and their receptor, EGL-15/FGFR, activate Ras-MAPK signalingto control sex myoblast migration, mus-cle development, neural migration, andosmotic balance (Bulow et al., 2004;Huang and Stern, 2004; Fleming et al.,2005; Dixon et al., 2006). This has led tothe suggestion that C. elegans can pro-vide a valuable model for noncanonicalroles for FGF signaling (Polanska et al.,2009). AsC. elegans lacks a vascularizedcirculatory system, studies of vascularendothelial growth factor (VEGF) andits receptors (VEGFR1–3), which are re-sponsible for the angiogenic growth thatis necessary for sustained tumorigene-sis, seem likely to be out of place. Yet sur-prisingly, orthologs of these proteins arepresent in the C. elegans genome (Plow-man et al., 1999; Popovici et al., 1999;Rikke et al., 2000; Popovici et al., 2002;Tarsitano et al., 2006).Moreover, a puta-tive PDGF/VEGF ortholog, PVF-1,behaves similarly to mammalian VEGFin its ability to dimerize, undergo secre-tion, and bind at least two mammalianVEGFreceptors inheterologous systems(Tarsitano et al., 2006). Furthermore,PVF-1 can activate angiogenesis whenexpressed in the chicken chorioallantoicmembrane and in cultured human um-bilical vein endothelial cells (HUVEC;Tarsitano et al., 2006). The four C. ele-gans VEGFR homologs, VER-1–4, areexpressed predominantly in neuronsand their associated cells, although theirfunctions are currently unclear (Popo-vici et al., 2002). This expression pat-tern, however, correlates with the obser-vation that some mammalian angio-genic factors also have a role in neuro-genesis, suggesting that some functionsof theC. elegansVEGF pathwaymay bephylogenetically conserved (Galvanet al., 2006).

It is also important to note that somedifferences have been observed betweenC. elegans and mammalian growth fac-tor signaling pathways. For example,the insulin growth factor receptor(IGFR) in mammals signals through acanonical Ras-MAPK pathway. In con-trast, the C. elegans DAF-2/IGFR path-way functions independently of Ras-MAPK (Sundaram, 2006). Anotherexample is the noncanonical Wnt sig-naling pathway originally identified instudies of C. elegans embryogenesis. In

this case, however, the noncanonicalWnt pathway was subsequently shownto be conserved across phyla (Han,1997; Kuhl et al., 2000; Peifer and Pola-kis, 2000; Korswagen, 2002; Kuhl,2002; Veeman et al., 2003). Thus, stud-ies of growth factor signaling pathwaysinC. elegans have been extremely fruit-ful in identifying both conserved corecomponents and accessory regulators ofthese pathways and linking their func-tions to oncogenesis.

THE GERMLINE: TUMORS,

IMMORTALITY, GENOME

INTEGRITY, AND SURVIVAL

As described in the preceding sections,mutations in several genes associatedwith cell cycle control can induce so-matic cell hyperproliferation defects inC. elegans. Such ectopic divisions, how-ever, rarely produce anything resem-bling a classic tumor. Rather, supernu-merary somatic cells undergo qui-escence, terminally differentiate, andintegrate more or less appropriatelyinto their respective organs. By con-trast,mutations leading to excess germ-line proliferation can result in bona fidetumors. Although generally confined tothe gonad, germline tumor cells aremitotically active and fail to differenti-ate into gametes. As described below,germline tumors can arise throughboth germline intrinsic (cell autono-mous) and somatically based (nonau-tonomous) mechanisms. In addition, asthe only tissue to harbor a population ofstem cells throughout adulthood, the C.elegans germline provides an opportu-nity to study mechanisms governingreplicative immortality and themainte-nance of pluri-potency, both of whichare relevant to traits acquired by cancercells. Finally, the germline is the onlytissue capable of undergoing apoptosisin C. elegans adults, both in response todamage cues and as part of a normal de-velopmental process. Thus, the germ-line ofC. elegans provides a particularlypowerful system in which to study sev-eral biological phenomena pertinent tomalignant transformation.

A BRIEF REVIEW OF

GERMLINE PROLIFERA-

TION CONTROL

At hatching,C. elegans hermaphroditescontain two primordial germ cells that

give rise to independently regulated an-terior and posterior germ cell popula-tions in the adult (for detailed reviews,seeHansen and Schedl, 2006;Hubbard,2007; Kimble and Crittenden, 2007;Byrd and Kimble, 2009). Proliferationof germ cells during larval and adultstages is controlled by the somaticgonad, most notably the distal tip cells(DTCs), which are essential for main-taining a mitotic stem cell niche at thedistal terminus of each gonad arm(Kimble and White, 1981; Fig. 4). TheDTCs promote mitotic proliferationthrough a conserved Delta-Notch sig-naling pathway. Specifically, DTCsexpress a Delta/Serrate-like transmem-brane ligand, LAG-2 (collectivelyreferred to as DSLs; Henderson et al.,1994; Tax et al., 1994), which binds tothe GLP-1/Notch receptor locatedwithin the plasma membrane of distalgerm cells (Crittenden et al., 1994;Hen-derson et al., 1994; Tax et al., 1994).Activation of GLP-1 by LAG-2 isthought to result in cleavage of theGLP-1 intracellular domain, whichthen translocates to the nucleus whereit interacts with several co-factors toinduce the expression of target genesthat promote mitosis and inhibit meio-sis (Christensen et al., 1996; Doyleet al., 2000; Petcherski and Kimble,2000; Kovall, 2008).Although a complete knowledge of

GLP-1 downstream effectors is cur-rently lacking, one established target isthe Pumilio and FBF (PUF) familymember fbf-2 (Lamont et al., 2004).FBF-2 and its close paralog, FBF-1 (to-gether, termed the FBFs), are bothexpressed in the distal mitotic region ofthe germline where GLP-1 is active(Zhang et al., 1997; Lamont et al.,2004). Although mutations that affectindividual fbfs show minimal defects,germline proliferation is largely abol-ished after the L4 larval stage in fbf-1fbf-2 double mutants, indicating thatthe FBFs are redundantly required formaintaining the germline stem cellniche in adults (Crittenden et al., 2002).The FBFs function by binding tosequence elements in the 30UTRs of tar-get mRNAs to inhibit their translation(Zhang et al., 1997; Bernstein et al.,2005). Targets for repression by theFBFs include two additional transla-tional regulators, GLD-1 and GLD-3(Crittenden et al., 2002; Eckmannet al., 2004; Fig. 5). GLD-3, a Bicaudal-

Dev

elop

men

tal D

ynam

ics


C family member (Eckmann et al.,2002, 2004), physically associates withthe cytoplasmic poly(A) polymerase(PAP), GLD-2 (Wang et al., 2002a). TheGLD-2–GLD-3 complex has been sug-gested to stimulate gene expression bylengthening the poly(A) tails of targetmRNAs, resulting in increased stabilityand translation. Conversely, GLD-1, amaxi-KH/STAR domain RNA-bindingprotein, acts primarily as a transla-tional repressor (Jones and Schedl,1995; Lee and Schedl, 2001; Crittendenet al., 2002; Hansen and Schedl, 2006).Also operating in the GLD-1 pathway isNOS-3 and GLD-4, which promoteGLD-1 accumulation redundantly withGLD-2–GLD-3 (Fig. 5; Hansen et al.,2004b; Schmid et al., 2009). NOS-3, ahomolog of theDrosophilaNanos RNA-binding protein, promotes GLD-1expression at the level of translationand can also physically interact withthe FBFs (Kraemer et al., 1999;Hansenet al., 2004b). GLD-4, a cytoplasmicPAP, acts with GLS-1 to promote poly-adenylation and increased stability ofgld-1mRNA (Schmid et al., 2009).

As a result of FBF translationalrepression, expression of GLD-1 andGLD-3 is inhibited in the distal mitoticniche (Crittenden et al., 2002; Eckmannet al., 2004). GLD-1 and GLD-3 arepresent at higher levels in cells that lieproximal to the niche, consistent withtheir role in promoting meiosis (Joneset al., 1996; Eckmann et al., 2004).Moreover, loss of GLP-1 or FBF activityleads to ectopic expression of GLD-1 inthe distal portion of the gonad and pre-mature entry of niche cells into meiosis(Crittenden et al., 2002; Hansen et al.,2004b). One known target for repres-sion by GLD-1 is GLP-1(Marin andEvans, 2003), and both GLD-1 andGLD-3 inhibit expression of the FBFs,thus indicating the presence of negativefeedback loops (Eckmann et al., 2002;Kimble and Crittenden, 2007; Fig. 5).Many other factors, including a host ofadditional RNA-binding proteins, havealso been implicated in controlling themitosis-to-meiosis transition (Hansenand Schedl, 2006; Kimble and Critten-den, 2007), but in most cases they willnot be discussed here.

It is worth underscoring that theoverall picture to have emerged fromstudies on C. elegans germline regula-tion is one of striking complexity. Thisis due in part to the presence of numer-

ous feedback loops, the utilization ofcertain pathway components at multi-ple nodes in the regulatory network,and the shifting roles of various factorsat discrete stages of development. Fur-thermore, many genes controlling themitosis-to-meiosis switch are alsorequired for progression through meio-sis and for promoting sex-specific germcell fates. Moreover, germline prolifera-tion is sensitive to environmental condi-tions. For example, in response to star-vation, germ cells in L1 larvae undergoa reversible cell cycle arrest. This arrestis dependent onDAF-18, a PTEN tumorsuppressor ortholog, and MDF-1, aMAD family protein required for G2/Mcheckpoint arrest (Fukuyama et al.,2006; Watanabe et al., 2008). Thus,complex intrinsic mechanisms, as wellas external cues, are critical for main-taining proper levels of germ cell prolif-eration and for controlling the balancebetweenmitosis andmeiosis.

GERMLINE TUMORS:

INTRINSIC MECHANISMS

Many of the genes controlling germlineproliferation and differentiation werefirst identified in genetic screens formutations that cause sterility. Furthercharacterization revealed a wide arrayof phenotypes including the inability ofcertain mutants to establish or main-tain a stem cell niche. Other mutationsled to an expansion of the niche andinterfered with the ability of germ cellsto initiate or complete meiosis. Asdescribed below, mutations in this lat-ter class can, under certain conditions,promote the formation of germlinetumors. In such cases, gonads becometightly packed with mitotic nuclei thatexhibit cellular configurations reminis-cent of mammalian tumors. In someinstances, excess proliferation leads tosevere swelling of the proximal gonadand the release of its contents, whichcan contribute to premature death(Francis et al., 1995a).

As indicated by the regulatory net-work depicted in Figure 5, gain-of-func-tion mutations in glp-1 or loss-of-func-tion mutations in other pathwaycomponents, such as the gld genes, canshift the balance toward mitotic prolif-eration. In the case of glp-1(gf) muta-tions, sequence alterations result inconstitutive activation of the receptor,which for strong alleles (e.g., glp-

1(oz112gf)) leads to a complete absenceof germline meiosis and the formationof a contiguous germline tumor [Fig. 4;(Berry et al., 1997)]. For weaker glp-1(gf) alleles (e.g., glp-1(ar202gf)), meio-sis and gamete differentiation occur ina relatively normal manner; however,such gonads also contain an ectopicmass of proliferating cells at the proxi-mal terminus (the Pro phenotype; Fig.4; Pepper et al., 2003a). In the case ofglp-1(ar202gf) mutants, tumors arisefrom a population of proximal mitoticgerm cells that fail to differentiate,because of the establishment of an ec-topic niche (also see below). Deviationsfrom the normal mitotic-to-meiotic spa-tiotemporal progression also occur forloss-of-function mutations in gld-1 andpuf-8; however, in these cases, tumorsarise from germ cells that initiate butfail to complete gametogenesis. Forexample, germ cells in gld-1 nullmutants (heretofore referred to as gld-1mutants) initiate meiosis but return tomitosis before completing meiotic pro-phase (Francis et al., 1995a,b). Like-wise, mutations in the PUF familymember puf-8 result in primary sper-matocytes that fail to correctly executereductional meiotic divisions andinstead de-differentiate into mitoticcells that form a proximal mass (Subra-maniamand Seydoux, 2003).Analyses of compound mutants have

revealed that components of the NOS-3–GLD-1 and GLD-2–GLD-3 pathwaysfunction redundantly with respect topromoting meiotic entry and progres-sion (Fig. 5). For example, tumorgrowth is significantly more pro-nounced in double mutants of gld-1 andeither gld-2 or gld-3 than in gld-1 singlemutants (Kadyk et al., 1997; Eckmannet al., 2004; Hansen et al., 2004a).Moreover, unlike gld-1 single mutants,germ cells in double mutants generallyfail to initiate meiosis and remain con-stitutively mitotic. As would beexpected, other double-mutant combi-nations that lead to the inhibition ofboth pathways (e.g., gld-3; nos-3 andgld-2 gld-4) also display ‘‘synthetic’’tumorous phenotypes (Hansen et al.,2004a,b; Schmid et al., 2009). Con-versely, synthetic tumor growth is notobserved in gld-1; nos-3 or gld-2; gld-3double mutants, as these combinationsdisrupt only the activities of the individ-ual pathway branches (Eckmann et al.,2004;Hansen et al., 2004b).

Dev

elop

men

tal D

ynam

ics


Consistent with NOS-3 and theGLDs functioning downstream of GLP-1, synthetic germline tumors are notsuppressed by loss of glp-1 activity(Kadyk and Kimble, 1998; Eckmannet al., 2004; Hansen et al., 2004b). Of in-terest, unlike the germlines in strongglp(gf) mutants, those in compoundmutants (e.g., gld-1 gld-2) can still har-bor small numbers of meiotic cells(Hansen et al., 2004a). This finding sug-gests that a third pathway, operatingdownstream of GLP-1, also contributesto the promotion of meiosis (Fig. 5). Wenote that the regulatory networkdescribed above and in Figure 5 is con-siderably more complex than we havedepicted. For example, contrary toexpectations, loss of fbf function cansuppress synthetic tumor formation insome compound mutants as well asenhance tumor growth in other geneticbackgrounds (Eckmann et al., 2004).This latter finding suggests that theFBFs, acting together or individually,may differentially influence outcomesat distinct nodes within the network orat different stages of development.

Several additional germline-intrinsicregulators of the mitosis-to-meiosisswitch have also been identified; how-ever, their precise connection to theGLP-1 pathway has yet to be estab-lished (Fig. 5). A loss-of-function muta-tion in teg-4 was isolated as anenhancer of germline over-proliferationin strains that were heterozygous forthe strong glp-1(oz112gf) allele (Man-tina et al., 2009). Furthermore, a reduc-tion in teg-4 activity leads to synthetictumors in gld-1, gld-2, and gld-3 mu-tant backgrounds, indicating that TEG-4 does not function specifically in eitherthe NOS-3–GLD-1 or GLD-2–GLD-3branches of the pathway. teg-4 encodesa homolog of human SAP130, a con-served component of the SF3b pre-mRNA splicing complex (Mantinaet al., 2009). Thus, TEG-4 may influ-ence germline proliferation and differ-entiation indirectly, possibly throughthe expression of one or more down-stream targets of GLP-1 signaling. Sim-ilarly, mutations in the mog genes(mog-1–6) produce synthetic germlinetumors when combined with mutationsin the gld genes (gld-1–3; Belfiore et al.,2004; Hansen and Schedl, 2006). Themolecular identities of theMOGs impli-cate a role in mRNA processing or me-tabolism, and genetic epistasis experi-

ments suggest that the MOGs, as wellas TEG-4, may function downstream ofor in parallel to GLP-1 (Puoti and Kim-ble, 1999, 2000; Belfiore et al., 2004;Hansen and Schedl, 2006; Mantinaet al., 2009).

Roles for epigenetic regulators ofgermline proliferation are indicated byfindings that loss-of-functionmutationsin him-17, which encodes a novel chro-matin-associated protein, and mett-10,an evolutionarily conserved putativehistone methyltransferase, enhancegermline over-proliferation in weakglp-1(gf) mutants (Reddy and Ville-neuve, 2004; Bessler et al., 2007; Dor-sett et al., 2009). In addition, certain al-leles of mett-10 cause a partiallypenetrant tumorous phenotype on theirown that can be suppressed by glp-1(lf)mutations (Dorsett et al., 2009). Of in-terest, mett-10 mRNA levels andMETT-10 protein accumulation innuclei are positively regulated by thedynein motor protein light and heavychain subunits DLC-1 and DCH-1,which in the case of nuclear import,involves direct binding between DLC-1and METT-10 (Dorsett and Schedl,2009). Consistent with this, weak inhi-bition of dlc-1 or dhc-1 strongly enhan-ces the tumorous phenotype of glp-1(gf)and mett-10(lf) mutants. Genetic andphenotypic data indicate thatMETT-10normally promotes meiotic entry andmay function either upstream or in par-allel to GLP-1 (Dorsett et al., 2009).Additional data suggest that DLC-1–DCH-1 may also promote meiotic entrythrough an unknown partner of METT-10 (Dorsett and Schedl, 2009). We alsonote that unlike the situation in gld-1mutants, germline tumors associatedwith mutations inmett-10, him-17, andteg-4 are not due to the de-differentia-tion of meiotic cells but result from pro-longed maintenance of the mitotic state(Bessler et al., 2007; Dorsett andSchedl, 2009; Dorsett et al., 2009; Man-tina et al., 2009).

Additional insight into the control ofgerm cell proliferation has come fromthe identification of mutations thatreduce the occurrence or severity ofgermline tumors. Loss-of-functionmutations in either ego-1 or atx-2 leadto partial suppression of synthetictumorous growth in gld-1 gld-2 doublemutants (Maine et al., 2004; Voughtet al., 2005). ego-1 encodes an RNA-directed RNA polymerase that is also

required in the RNAi response pathway(Smardon et al., 2000; Vought et al.,2005). atx-2 encodes an ortholog ofmammalian ataxin-2 and has beenshown to regulate mRNA metabolismand translation (Kiehl et al., 2000;Ciosk et al., 2004). Based on geneticandmolecular evidence, neither EGO-1nor ATX-2 is likely to function withinthe GLP-1 pathway, but they may pro-mote germline mitosis through inde-pendent mechanisms (Maine et al.,2004; Vought et al., 2005). Simultane-ous loss of puf-8 andmex-3, the latter ofwhich encodes aKH-domainRNA-bind-ing protein (Draper et al., 1996), inhib-its tumor formation in both gld-1 gld-2double mutants and in strong glp-1gain-of-function mutants (e.g., glp-1(oz112gf); Ariz et al., 2009). Thus,PUF-8, like the FBFs, can either pro-mote or inhibit tumor formation,depending on the genetic backgroundand developmental context (Subrama-niam and Seydoux, 2003; Ariz et al.,2009). Unlike the combined loss ofmex-3 and puf-8, inhibition of atx-2 does notsuppress glp-1(oz112gf) tumor growth(Maine et al., 2004), suggesting that tu-mor suppression in these backgroundsmay occur by distinct mechanisms.Suppression of tumors in gld-1 gld-2mutants also occurs with mutations inthe conserved DNA-replication check-point gene clk-2/TEL2, indicating thatcorrect cell cycle execution is also arequisite for germline tumor growth(Moser et al., 2009).Intriguingly, mutations that increase

longevity can suppress tumor growthand premature death in gld-1 mutants(Pinkston et al., 2006; Moser et al.,2009). In the case of daf-2/IGFR, a com-ponent of the C. elegans insulin growthfactor pathway, the mechanism of tu-mor suppression is attributable to bothdecreased germ cell proliferation andincreased apoptosis. Mutations in daf-2promote germline apoptosis in gld-1mutants through a mechanism thatrequires the functions of DAF-16/FOXO; CEP-1, a protein with homologyto p53 family members; and severalDNA damage–induced checkpointgenes (Pinkston et al., 2006). In addi-tion, loss of daf-2 activity reduces thenumber of actively dividing germ cellsin gld-1 mutants. Similarly, mutationsthat extend life span by caloric restric-tion (eat-2) and metabolic repression(clk-1) reduce gld-1 germline mitotic

Dev

elop

men

tal D

ynam

ics


Fig. 5. The regulatory network that controls germ cellproliferation. Increased activity or ectopic expression ofproteins in green boxes promotes mitotic proliferation. Inthe case of glp-1(gf)mutants, this results in germline tumorformation. Decreased activity of proteins in red boxes, ei-ther aloneor in combination, also induces hyperproliferationand tumor formation. For simplicity, only a subset of knowngermline regulators is shown. In some cases genesdepicted to act in parallel to GLP-1, such as TEG-4 andMETT-10, may integrate their functions more directly withinthe GLP-1 pathway. Note the high degree of regulatorycross-talk between the core factors (gray lines). Dashedlines and questionmarks indicate predicted regulatory con-nections andmissing components, respectively.

Fig. 4.

Fig. 6.

Dev

elop

men

tal D

ynam

ics


proliferation but, unlike daf-2 muta-tions, have no effect on gld-1 germlineapoptosis (Pinkston et al., 2006). Curi-ously,mutations in daf-2, eat-2, and clk-2 do not adversely affect germline pro-liferation in wild-type. This is in con-trast to mutations in ego-1, atx-2, andmex-3; puf-8, which suppress germ cellproliferation in both tumorous andwild-type animals (Qiao et al., 1995;Maine et al., 2004; Ariz et al., 2009).More recently, several putative DAF-16transcriptional targets have been iden-tified that mediate life-span extensionand tumor growth suppression in gld-1;daf-2 mutants (Pinkston-Gosse andKenyon, 2007). This analysis suggeststhat both positively and negativelyregulated DAF-16 targets contribute todaf-2-mediated tumor suppression andthat individual targets have specificroles with respect to regulating eitherproliferation or apoptosis. Given theestablished role of the DAF-2–DAF-16pathway in stress response and dietaryrestriction (Murakami et al., 2000; Ken-yon, 2005), these studies provide anadditional link between environmentalinput and germline regulation. Thisconnection is further supported by thefinding that inhibition of the C. elegansprohibitin orthologs, phb-1/2, also sup-presses tumor growth in gld-1 mutants

(Artal-Sanz and Tavernarakis, 2009).Loss of phb-1/2 can extend lifespanunder conditions of dietary restrictionthrough the control of fat mobilizationand energy metabolism. Notably, over-expression of human prohibitins isassociated with several cancers andprohibitins have been suggested as apotential target for cancer therapy(Mishra et al., 2005).

GERMLINE TUMORS:

SOMATIC INFLUENCES

As previously discussed, the somaticDTC is required for the establishmentof a germline stem cell niche (KimbleandWhite, 1981). Moreover, the DTC issufficient for niche induction, as evi-denced by experimental manipulationsormutations that lead to the generationof multiple DTCs within a single gonadarm (Kimble and White, 1981; Kipreoset al., 2000; Fay et al., 2002; Kidd et al.,2005; Lam et al., 2006). Furthermore,mutations or manipulations thatimpede the stereotypical pattern and/ortiming of DTC migration during larvaldevelopment result in abnormal distal-to-proximal patterns of germline prolif-eration (e.g., Kimble and White, 1981;Blelloch et al., 1999; Nishiwaki, 1999;Tamai and Nishiwaki, 2007). In partic-

ular, failure of theDTC tomigrate a suf-ficient distance from the proximal ter-minus leads to prolonged exposure ofproximal germ cells to LAG-2 signalingand can result in tumorigenesis (alsosee below).The DTC is not, however, the only so-

matic cell that influences germ cell pro-liferation and differentiation. Somaticcells of the proximal gonad can also pro-mote germ cell proliferation and, undercertain circumstances, are integral tothe genesis of germline tumors. Mostprominent among these are the gonadalsheath cells, which form a thin layersurrounding most of the germline(Hirsh et al., 1976; Kimble and Hirsh,1979; Fig. 6). Each gonad arm in theadult harbors five pairs of sheath cells(Sh1–5), the three most proximal ofwhich (Sh3–5) form a myoepithelialsheet that aids in ovulation (Ward andCarrel, 1979). Laser ablation of sheathcell precursors during larval develop-ment leads to pleiotropic germlinedefects, including a three- to five-folddecrease in the extent of germ cell pro-liferation, thus demonstrating that so-matic gonad cells other than the DTCpromote germline mitosis (McCarteret al., 1997).More recently, Sh1, the dis-tal-most sheath cell pair, has beenshown to be most critical for this activ-ity (Killian andHubbard, 2005).Initial evidence of a role for the proxi-

mal somatic gonad in germline tumori-genesis came from cell ablation studiesin wild-type and in strains carrying alin-12 null mutation, which leads to theformation of proximal tumors (Seydouxet al., 1990). Like glp-1, lin-12 encodes aNotch receptor ortholog that can beactivated by LAG-2 (Yochem et al.,1988; Austin andKimble, 1989; Yochemand Greenwald, 1989; Henderson et al.,1994; Tax et al., 1994). Rather thanbeing expressed in the germline, how-ever, LIN-12 functions in the somaticgonad to control several binary cell-fatedecisions (Greenwald, 2005). The abla-tion studies by Seydoux and colleaguessupport a model whereby aberrantLAG-2 signal, emanating from the so-matic gonad anchor cell (AC), ensueswhen LIN-12 expression is abolished insomatic cells that lie adjacent to the AC.Thus, rather than binding to LIN-12,its normal target in the proximal so-matic gonad, AC-derived LAG-2 acti-vates GLP-1 on neighboring proximalgerm cells, thereby promoting

Fig. 4. Hermaphrodite wild-type and tumorous gonads. In wild-type gonads, mitosis is re-stricted to the distal portion of the germline. In this region, GLP-1/Notch is activated by LAG-2/DSL by means of the somatic distal tip cell (DTC). As divisions occur and mitotic cells move outof range of the LAG-2 signal, GLP-1 becomes inactive and germ cells exit mitosis and entermeiosis (transition zone). Moving further proximally, germ cells transit several stages of meioticprophase, eventually forming male and female gametes that are used for fertilization. Tumorousgonads contain proliferating mitotic cells at ectopic locations. In the case of strong gain-of-func-tion alleles such as glp-1 (oz112gf), meiosis is completely abolished and mitotic cells aredetected throughout the gonad arm, resulting in a contiguous tumor. Loss-of-function mutationsin other genes (e.g., pro-1 and gld-1) lead to more-limited tumor formation in the central-to-proximal gonad region.

Fig. 6. A delay in meiotic entry, in combination with somatic gonad signaling, promotes proxi-mal tumor formation. Blue and red indicate mitotic and meiotic germ cell regions, respectively.The green crescent-shaped cell on the left represents the distal tip cell (DTC). Yellow-to-brownflattened oval cells indicate somatic gonad sheath cells; darker shading indicates more proximalsheath cells. The proximal sheath cells depicted in mid-L4 and adult gonads are the Sh3–5pairs. Rounder proximal cells in early L3 and mid-L3 larvae have not yet completed divisions.Green arrows indicate DSL ligand signaling from the DTC and sheath cells. For simplicity, centralportions of mid-L4 and adult gonads are not shown; the missing region is demarcated by paral-lel lines. Note the delay in the timing of initial meiosis in Pro-defective gonads at the mid-L3stage. In the case of glp-1(gf) mutants (e.g., [glp-1(ar202)]), this is due to increased GLP-1 activ-ity that prevents proximal germ cells from exiting mitosis even after moving a sufficient distancefrom the DTC. Because pro-1 mutants have reduced mitotic proliferation during early larval de-velopment, gonads fail to extend at the normal rate. This results in the prolonged exposure ofproximal germ cells to the DTC pro-mitotic signal. Similarly, hlh-12 mutants, which are defectiveat DTC migration, also result in continuous ligand exposure. At later developmental stages,these nondifferentiated proximal germ cells are susceptible to stimulation by DSL ligandsexpressed from the sheath, leading to sustained mitosis and the formation of tumors. This figureis modeled after Killian and Hubbard, 2005.

Dev

elop

men

tal D

ynam

ics


sustained mitotic proliferation and tu-morigenesis.

More recently, several genes affectingribosome biogenesis have been identi-fied in screens for mutants that developproximal tumors. Among these arepro1–3, which encode orthologs of yeastproteins that regulate ribosome activityat the level of rRNA processing and nu-cleolar transport (Killian and Hubbard,2004; Voutev et al., 2006). Furthermore,PRO-1 is specifically required withinthe sheath cell lineage to inhibit the for-mation of germline tumors (Killian andHubbard, 2004). These results indicatethat gross disruption of normal meta-bolic processes within sheath cells canlead to nonautonomous defects in theproximal germline. This conclusion isfurther supported by the finding thatablation of Sh2–Sh5 can fully revertgermline tumorigenesis in pro-1mutants (Killian and Hubbard, 2005)and that partial suppression of pro-1tumors can also be achieved throughsecondary mutations that increase pre-rRNA levels (Voutev et al., 2006). Of in-terest, one of the mutations found tosuppress pro-1 tumors is lin-35/Rb,which functions as a negative regulatorof POL I rRNA transcription (Hannanet al., 2000; Pelletier et al., 2000;Voutevet al., 2006). In other mutant back-grounds, loss of LIN-35 activity can,however, strongly enhance proximal tu-mor formation, thus underscoring theinfluence of genetic context on pheno-typic outcome (Fay et al., 2002; Benderet al., 2007).

Given the wide spectrum of germlineand somatic genes that have beenimplicated in proximal tumor forma-tion, one might expect there to be littlecommonality in the underlying mecha-nisms of tumorigenesis. This is not,however, the case: through a series ofdetailed studies, Hubbard and col-leagues have successfully uncovered aunifying theme to account for the largemajority of proximal tumor phenotypes.Key among their observations is thatmutations in either pro-1, a somatic fac-tor, or glp-1(ar202gf), a germline intrin-sic factor, result in similar temporaldelays in the onset of initial meiosis(i.e., the time at which germ cells firstexit themitotic cycle and begin to differ-entiate; Pepper et al., 2003b; KillianandHubbard, 2004). This delay leads toa discordance between the developmen-tal ages of the proximal germline and

the surrounding somatic gonad (Sh2–Sh5; Fig. 6). According to the model,these developmentally retarded germcells are then exposed to signals ema-nating from the maturing proximalsheath, thereby maintaining them in anondifferentiated mitotic state. Thismodel is supported by results showingthat ablation of Sh2–Sh5 can variablysuppress proximal tumor formationin glp-1(ar202gf), puf-8, and pro-1mutants but not in gld-1 mutants,which are not associated with a delay inthe onset of meiosis (Killian and Hub-bard, 2005). Also consistent with thismodel, the induction of synthetic proxi-mal tumors in various compoundmutants of lin-35 (e.g., lin-35; spr-1, lin-12(RNAi)) correlates tightly with late-onset meiosis in these germlines(Bender et al., 2007).

Most recently, Hubbard and col-leagues have demonstrated a direct rolefor DSL family ligands in the inductionof proximal tumors by the sheath(McGovern et al., 2009). Evidenceincludes the finding that RNAi ormuta-tions affecting the DSL-encoding genesapx-1, arg-2, and dsl-5 can variablysuppress tumor formation in pro-1 andglp-1(ar202gf) mutants, as well asin hlh-12 mutants, which exhibitimpaired DTC migration (Tamai andNishiwaki, 2007; McGovern et al.,2009). This is consistent with previousfindings showing that proximal prolif-eration in pro-1 mutants requires func-tional GLP-1 (Killian and Hubbard,2004). Moreover, simultaneous repres-sion of both apx-1 and arg-1 completelysuppressed tumor formation in pro-1mutants, indicating that the combinedactivities of apx-1 and arg-1 can largelyaccount for themitosis-promoting activ-ity of the proximal sheath (McGovernet al., 2009). In contrast, inhibition ofDSL ligands did not suppress tumor for-mation in gld-1 mutants, consistentwith previous sheath cell ablationresults. Finally, both apx-1 and arg-1are expressed within the proximalsheath lineage. Of interest, disruptionof DSL-Notch signaling in adults thatharbor proximal tumors led to the dif-ferentiation of tumor cells, indicatingthat continuous signaling is requiredfor tumormaintenance.

To frame their findings in broaderterms, the authors invoke the conceptof a ‘‘latent niche’’ (McGovern et al.,2009). The latent niche describes a cel-

lular microenvironment that does notnormally serve as a niche but that isnevertheless capable of sustaining theproliferation of renewal-competent cellswith which it comes in contact. In thecase of germline tumors, proximalsheath cells and their progenitors pro-vide a latent niche for nondifferentiatedgerm cells through DSL-Notch signal-ing. The authors speculate that the pro-ductive seeding of metastatic cancers inhumans may occur through the exploi-tation of latent niches. Intriguingly, tu-mor promotion by the latent nichewould not require additional mutationsto arisewithin either the niche or tumorcells, nor would it require the inductionof potential niche cells by the primarytumor. As this proposed mechanismmay allow for a greater understandingof tumor metastasis, as well as thepotential for directed therapeutic strat-egies, follow-up work in other systemsis clearlywarranted.

STEMNESS AND THE

MAINTENANCE OF A

NONDIFFERENTIATED

STATE

Parallels have long been drawnbetween the characteristics of tumorcells and those of normal stem cells,including their mutual ability to evadeterminal differentiation. Althoughexceptions have been reported, conven-tional wisdom holds that malignantcells are at most partially differenti-ated, either because tumor initiationmust necessarily precede terminal dif-ferentiation, or because the transforma-tion process itself can lead to tumor cellde-differentiation (Daley, 2008). More-over, patient prognosis, as predicted byhistological features, often correlateswith the degree of tumor cell differen-tiation, such that less-differentiatedcancers are typically associated with agreater health risk. Several studieshave also suggested that a sub-popula-tion of tumor cells, termed cancer stemcells (CSCs), comprise the critical self-renewing population of cells within aheterogeneous tumor, although thistheory remains controversial (Jordan,2004; Gupta et al., 2009; Rosen and Jor-dan, 2009).The protection of germline totipotency

(and other germ cell characteristics) in

Dev

elop

men

tal D

ynam

ics


C. elegans effectively begins before thefirst cell division with the asymmetricdistribution of germline-specific factorssuch as P granules, macromolecularparticles composed of heterogeneousproteins, and RNAs (Updike andStrome, 2009). Extensive analysis hasuncovered several distinct mechanismsthat are used at various times to pro-mote survival and prevent future germcells from undergoing programs of so-matic differentiation (for reviews seeSeydoux and Schedl, 2001; Shin andMello, 2003; Strome, 2005; Updike andStrome, 2009). These include the repres-sion of transcriptional elongation ingerm cell progenitors by PIE-1, the ini-tiation of epigenetically controlled pat-terns of gene expression in early germcells by Polycomb group–relatedproteins, and the actions of the Pgranules, which are likely to function atthe level of RNAmetabolism andmRNAtranslation.

Although some of the mechanismsinitiated during embryonic develop-ment are also relevant to germline func-tions at later stages, less is knownabout the specific maintenance of germcell totipotency in adults. Intriguingly,adult germ cells in mex-3 gld-1 doublemutants—and to a lesser extent gld-1single mutants—undergo differentia-tion into a diverse array of cell types, aphenotype that closely resembleshuman teratomas (Ciosk et al., 2006).That this trans-differentiation processrequires the initiation of meiosis is sup-ported by the finding that germ cell dif-ferentiation in mex-3 gld-1 mutants isblocked by the glp-1(oz112gf) mutation.Based on computational predictions,germline teratoma formation is likelyto involve the translational de-repres-sion of many individual or sharedmRNA targets of MEX-3 and GLD-1(Pagano et al., 2009). Several confirmedtargets provide insight into the generalcategories of factors that are likely to beinvolved. These include PAL-1, a home-odomain-containing transcription fac-tor that normally promotes muscle cellfates in the embryo and which is trans-lationally repressed by bothMEX-3 andGLD-1 (Mori et al., 1988; Draper et al.,1996; Hunter and Kenyon, 1996; Ciosket al., 2006). Inhibition of pal-1 largelyblocks muscle cell differentiation inmex-3; gld-1 germlines, indicating thatpal-1 is one of the critical targets forrepression byMEX-1 and GLD-1 in ter-

atomas. In addition, cye-1/cyclin E,which is required for progressionthrough G1/S, is a target for transla-tional repression by GLD-1 (Fay andHan, 2000; Biedermann et al., 2009).Furthermore, inhibition of cye-1 byRNAi prevents germ cell trans-differen-tiation in gld-1mutants, demonstratingthat re-entry into the mitotic cell cycleis also a critical step in teratoma forma-tion (Biedermann et al., 2009). Finally,the identification of glp-1 as a target forrepression by both GLD-1 and MEX-3suggests that the deregulation of germ-line intrinsic factors may also be impor-tant in germ cell trans-differentiation(Marin and Evans, 2003; Pagano et al.,2009).

STEMNESS AND

IMMORTALITY

Both stem cells and cancer cells possessthe capacity for long-term, if not limit-less, proliferation. As the sole self-renewing stem cell population in C. ele-gans adults, the germline presents anopportunity for uncovering the mecha-nisms that inhibit cellular senescenceand promote continuous growth and di-vision. One approach to the study of cellimmortality in C. elegans has been toisolate mutations that give rise to ‘‘mor-tal’’ germlines (The Mrt phenotype;reviewed by Smelick and Ahmed,2005). Such mutants become progres-sively sterile over multiple generations(4–20), most likely in response to theaccumulation of structural damage tochromosomes (Ahmed and Hodgkin,2000; Smelick and Ahmed, 2005). Nota-bly, estimates suggest that >100 inde-pendent loci may be required for themaintenance of germline immortalityin C. elegans, and the full range ofmechanisms awaits further exploration(Smelick andAhmed, 2005).

mrt-2, the first mortal germline mu-tant to be cloned, encodes a checkpointprotein that is homologous to S. pomberad1 (Dean et al., 1998; Ahmed andHodgkin, 2000). MRT-2 functions in theresponse to double-stranded DNAbreaks and is also required for the faith-ful replication of telomeres. mrt-2mutants exhibit progressive telomereshortening and chromosomal fusionsthat are characteristic of telomerase-de-fective mutants in other organisms.Furthermore, loss of function of hus-1and hpr-17, which encode proteins that

associate with MRT-2 family membersto promote assembly of the 9-1-1 check-point complex, also lead to telomere ero-sion and progressive sterility, as domutations that directly or indirectlyinterfere with telomerase activity (e.g.,trt-1 and mrt-1; Hofmann et al., 2002;Meier et al., 2006; Boerckel et al., 2007;Meier et al., 2009). These findings dove-tail with studies of mammalian cancercells, as well as with those of immortal-ized cells in culture, which are knownto select for mutations that activatetelomerase. Further analysis of mrtmutants may therefore identify novelfactors required for telomere mainte-nance by cancer cells.As mentioned in the preceding sec-

tion, chromatin-level transcriptionalregulation is critical for the preserva-tion of germline competence. Impli-cated factors include MES-2/E[Z],MES-6/ESC, and MES-3, which to-gether form a complex related to thePolycomb repressor of Drosophila (Hol-deman et al., 1998; Korf et al., 1998; Xuet al., 2001a,b; Bender et al., 2004b).This complex blocks inappropriate geneexpression in the germline by inhibitingMES-4, a histoneH3methyltransferasethat promotes the active chromatinstate (Kelly and Fire, 1998; Fong et al.,2002; Bender et al., 2006). Recently,another chromatin modifier, the SPR-5/LSD1 histone demethylase, has beenshown to be required for germlineimmortality (Katz et al., 2009). Similartomrt-1 andmrt-2mutants, loss of spr-5 leads to progressive sterility within�20–30 generations. Unlike mrt mu-tants, spr-5-deficient strains do not,however, exhibit telomere shorteningor other associated genomic aberra-tions, such as end-to-end chromosomalfusion and aneuploidy. Furthermore,loss of SPR-5 function correlateswith the progressive accumulation ofH3K4me2 methylation in germ cells to-gether with an increase in the expres-sion of genes associated with spermato-genesis. Of interest, fertility can berestored to late-generation spr-5mutants following a single outcross,indicating that the slow-onset epige-netic changes are nevertheless rapidlyreversible. Notably, epigenetic changesare associated with many cancer typesand are thought to play important rolesin cancer initiation and progression(Daley, 2008; Albert and Helin, 2009;Sharma et al., 2010).

Dev

elop

men

tal D

ynam

ics


Of interest, several studies haveimplicated genomic stability as arequirement for germline immortality(Chin and Villeneuve, 2001; Degtyar-eva et al., 2002; Tijsterman et al.,2002b). One example is MRE-11, a con-served protein that is required for bothDNA double-stranded break repair andmeiotic crossover (Chin and Villeneuve,2001). mre-11 homozygous mutantsbecome sterile after just several genera-tions, despite that fact that meioticrecombination is not required for long-term viability (Zalevsky et al., 1999;Kelly et al., 2000). These results sug-gest thatmaintenance of genomic integ-rity is essential for long-term reproduc-tive capacity. Another example of thisphenomenon is the Bergerac strain, anaturally occurring isolate ofC. elegansthat contains several hundred copies ofthe Tc1 transposon (Emmons et al.,1983; Anderson, 1995; Bessereau,2006). Like the previously discussedmrt mutants, the Bergerac strainexhibits progressive sterility when self-fertilized for multiple generations. Incontrast, the Bristol isolate (N2)—thestandard C. elegans laboratory strain—contains few copies of Tc1, and theseare silenced in the germline (Eide andAnderson, 1985; Collins et al., 1987).These observations point to an appa-rent paradox, as immortality andgenomic instability are dual character-istics of tumor cells. Thus, cancer cellsmaybe required to developmechanismsthat maintain immortality while hav-ing highly plastic genomes.

REPRESSION OF GERMLINE

TRAITS IN THE SOMA

In the preceding sections, we addressedmechanisms required for the promotionof stem cell characteristics in the C. ele-gans germline. An equally relevant lineof inquiry, particularly within the con-text of oncogenesis, is to understand themeans by w

Documents

Cancer Models in Caenorhabditis elegans · 2020. 6. 17. · C. elegans is a free-living soil nema-tode that is approximately 1 mm in length. As the ﬁrst metazoan to have had its