Though this may seem to be a tidy expla-nation for the initiation of cyst growth, aconnection between calcium mobilizationand the JAK2–STAT1–p21 pathway is notimmediately obvious.

Other functions?A further complication is that a number ofstudies have localized PC1 and PC2 to lat-eral membrane junctional complexes, inparticular to desmosomes, thus implicat-ing the polycystins in cell–cell communi-cation12. Furthermore, the bulk of PC2 isfound in cytosolic membranes, probablythose of the endoplasmic reticulum, andthere is direct evidence that PC2 acts as acalcium-activated calcium-release chan-nel requiring activation of the InsP3 recep-tor13. The work by Nauli et al.2, incontrast, argues that ciliary PC2 acts as a

calcium-entry channel, triggering calciumrelease from intracellular stores in a ryan-odine receptor–dependent, but InsP3receptor–independent process. Thus,there are knotty questions about wherethe polycystins function in the cell andwhat they are doing.

Integrating calcium signalingOne possible explanation for the aboveobservations is that the polycystins do dif-ferent things in different settings. Forexample, evidence is accumulating thatPC1 is a G protein–coupled receptor12,which can activate heterotrimeric G pro-teins and release Gβγ subunits that canthen inhibit calcium-channel activity14,among other activities14. Yet, the study byNauli et al.2 shows that the polycystin-dependent calcium increase does not

require G-protein signaling. Thus, it ispossible that the ciliary polycystins pro-vide a calcium signal in response tomechanosensory stimulation, but thatinterpretation of this signal requires inputfrom other polycystin-activated pathwaysand, ultimately, the integration of multi-ple signals into an appropriate cellularresponse. The transformation of a renalepithelial cell into a hyperproliferating(yet somewhat differentiated) cyst epithe-lial cell will probably involve changes ingene expression that may reflect aberrantcalcium mobilization. If so, there will cer-tainly be other affected pathways. Defin-ing these pathways and their gene targetswill be the next challenge. �

1. Igarashi, P. & Somlo, S. J. Am. Soc. Nephrol. 13,2384–2398 (2002).

2. Nauli, S.M. et al. Nat. Genet. 33 (2003); advanceonline publication, 6 January 2003 (doi:10.1038/ng1076).

3. Murcia, N.S. et al. Development 127, 2347–2355(2000).

4. Yoder, B.K. et al. Am. J. Physiol. Renal Physiol. 282,F541–552 (2002).

5. Hou, X. et al. J. Clin. Invest. 109, 533–540 (2002).6. Barr, M.M. & Sternberg, P.W. Nature 401, 386–389

(1999).7. Qin, H., Rosenbaum, J.L. & Barr, M.M. Curr. Biol. 11,

457–461 (2001).8. Haycraft, C.J. et al. Development 128, 1493–1505

(2001).9. Pazour, G.J. et al. Curr. Biol. 12, R378–380 (2002).10. Yoder, B.K., Hou, X. & Guay-Woodford, L.M. J. Am.

Soc. Nephrol. 13, 2508–2516 (2002).11. Bhunia, A.K. et al. Cell 109, 157–168 (2002).12. Scheffers, M.S. et al. Hum. Mol. Genet. 11, 59–67

(2002).13. Koulen, P. et al. Nat. Cell Biol. 4, 191–197 (2002).14. Delmas, P. et al. J. Biol. Chem. 277, 11276–11283

(2002).15. Schrier, R.W. & Gottschalk, C.W. (eds.) Diseases of

the Kidney 6th edn. vol. I. 529 (Little Brown andCo., Boston, 1997).

Fig. 2 Scanning electron micro-graph of the inside of a collect-ing-duct cyst from a humanautosomal dominant polycystickidney. This region of the cystwall shows a single interca-lated cell in the center of thefigure surrounded by principalcells, each with one (or several)primary cilia (white spikes).Although the cilia on thesecells appear normal, they arepresumably defective becauseof the mutation in PKD1 orPKD2. Reprinted with permis-sion from ref. 15.

news & views

114 nature genetics • volume 33 • february 2003

Surprise endingJohn M. Sedivy1, Dorothy E. Shippen2 & Eugene V. Shakirov2

1Department of Molecular Biology, Cell Biology and Biochemistry, Brown University, Providence, Rhode Island 02912, USA.e-mail: John_Sedivy@brown.edu. 2Department of Biochemistry and Biophysics Texas A&M University College Station, Texas 77843, USA.

e-mail: dshippen@tamu.edu

As the caps of chromosomes, telomeres are essential for genome integrity and stability. A highly accurate method for measuringthe length of a single human telomere has been developed and reveals previously unrecognized variation in telomere length.

Telomeres are found at the ends of alllinear chromosomes and comprise along stretch of simple repeat DNA (seefigure) bound by specialized proteins.The ensuing structure protects chromo-somes from end-to-end fusion, thusmaintaining genome stability. Telomerescan be elongated independently of the

rest of the genome by a unique poly-merase, telomerase, that utilizes aninternal RNA template. This nifty mech-anism prevents the shortening of chro-mosome ends that would otherwiseresult from incomplete replication oflagging strands by conventional DNApolymerases.

It is easy to imagine how subvertingtelomere maintenance could lead to all sortsof genomic mischief. So, it was somewhatsurprising to discover that in humans (andmany other long-lived vertebrate species),telomerase activity is present in the germline and some stem cells, but turned off inthe vast majority of somatic tissues. The

©20

03 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://w

ww

.nat

ure

.co

m/n

atu

reg

enet

ics

news & views

nature genetics • volume 33 • february 2003 115

absence of telomerase means that telomeresbegin to erode during development andcontinue to diminish throughout life. Giventhat cells ultimately cannot survive withouttelomeres, the absence of telomeraseimposes a replicative limit determined inlarge part by initial telomere length.

In recent years, telomeres have been rec-ognized as having important roles in thebiology of cancer and aging. As such, mea-suring telomere lengths has become animportant issue. Although much provoca-tive data has already been generated, theabsence of a method for accurately measur-ing telomere length has limited the conclu-sions that can be drawn at this stage. A newmethod reported by Duncan Baird and col-leagues1 on page 203 may provide the nec-essary resolving power to work out the finedetails of telomere dynamics.

Sizing up with STELAPrevious methods have been developedbased on either hybridization or PCR(Table 1). However, PCR, which is inher-ently more sensitive and accurate, hasbeen hampered by the lack of appropriateprimer binding sites. In vertebrates, thesubtelomeric regions where one primerwould optimally be directed (see figure)mainly comprise large sequence blocks,often tandemly repeated, that are dis-persed to varying extents between differ-ent chromosomes. Thus, while tagging thechromosome end with a unique primer isnot that difficult, anchoring the referencepoint inside the chromosome is muchmore problematic.

The new approach described by Bairdet al.1, termed STELA (single telomerelength analysis), combines knowledge ofunique sequences in the XpYp subtelo-meric region with an elegant and efficientmethod of tagging the telomere end with aunique-sequence primer site to measuretelomere lengths accurately on a singlehuman chromosome arm. This methodhas numerous advantages over previousapproaches: it can be performed on cellsthat have reached replicative senescence;discrete bands that, in all likelihood rep-resent individual chromosome mole-cules can be resolved; and, importantly,allele-specific internal primers allow thetwo homologous chromosomes to bedistinguished.

Variable endsUsing STELA, Baird et al.1 found thattelomere lengths were consistently shorter(∼ 1.3 kbp) than those obtained using classi-cal terminal restriction fragment (TRF)analysis. This is consistent with the knownunderestimation of short telomeres by TRF.Unexpectedly, single-allele amplification oftelomeres revealed in some cases differences> 6 kbp between homologous chromo-somes. Thus, at senescence cells containedone very short (0–2 kbp) and one long (4–8kbp) telomere. Analysis of inheritance in ahuman pedigree indicated that the telomerelength at senescence segregated in a non-mendelian fashion, suggesting that randomsampling from a range of telomeric lengthsoccurs in gametes. Interestingly, expressionof telomerase in presenescent fibroblasts

homogenized the interallelic difference, butthis process took many (>20) generations.

Notably, although the mean length atsenescence of the shorter telomere was 1.2kbp, some individual chromosomes con-sistently displayed extremely short telo-meres (∼ 45 bp), suggesting that essentiallycomplete telomere loss may occur on asubset of telomeres. Could ultra-shorttelomeres be a common factor in senes-cent cells? Although such catastrophic lossappears to afflict only a few percent oftotal XpYp telomeres in a population ofsenescent cells, if the remainder of humantelomeres behave in as heterogeneous andstochastic a fashion, one wonders whethereach cell at senescence might not containat least one essentially naked chromosomeend. Recent studies of mice lacking telo-merase indicate that critically shortenedtelomeres intitiate genome instability byserving as the preferred substrates forend-to-end fusion2. Whether denudedtelomeres are the elusive ‘clock’ that trig-gers senescence remains to be determined.

In many experiments reported by Bairdet al.1, occasional bands were observedthat were significantly larger or smallerthan the bulk population. Such apparentlystochastic events could be caused by repli-cation slippage, telomere recombinationor oxidative damage. Thus, it may be fea-sible to use STELA to investigate poorlyunderstood aspects of telomere dynamics,such as oxidative damage in resting (non-dividing) cells. Undoubtedly, the tech-nique can be applied to investigate moreprecisely the extent of telomere erosion

Table 1 • Methods available for analyzing telomere lengths

TRF3 Q-FISH4 Flow-FISH5 Telomere PCR6 STELA1

Number of starting cells large intermediate intermediate small smallDetect small changes in telomere length √ intermediate √ √Applicable to non-dividing cells √ √ √ √Detect individual telomeres √ √ √Signal includes 3′ overhang √ √ √ √Single-chromosome average √ √ √Genome-wide average √ √ √Non-telomere background signal √General ease of use √ √ √ √

TRF, terminal restriction fragment analysis; Q-FISH, quantitative fluorescent in situ hybridization; STELA, single telomere length analysis.

chromosome

subtelomeric region repetitive elements

imperfect telomeric (TTAGGG) repeats

telomeric (TTAGGG) repeats

G-rich overhang

region detected by hybridization methods (1–3 in Table 1)

possible locations for chromosome-specific primers

TdT tail (PCR method 4)

end-specific primer (PCR method 5)

Telomere structure. The telomere repeat regions are estimated to be 10–20 kbp in length in humans, and end with a 3′ single-stranded region (G-rich overhang) of 200to 600 nucleotides. In addition to being complexed with proteins, the G-rich overhang may fold back, displacing a portion of the telomere duplex to form a t-loop7.

©20

03 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://w

ww

.nat

ure

.co

m/n

atu

reg

enet

ics

The X-linked form of dyskeratosis con-genita (DC) is caused by mutations inDKC1, encoding dyskerin1, formerlyknown as NAP57 (ref. 2), or Cbf5 inyeast3. Dyskerin is an essential struc-tural component of small nucleolarRNA-protein complexes (snoRNPs) andof the mammalian telomeraseRNP4,5. Although DC was pro-posed to be caused by a telo-merase deficiency, a newmouse model published in Sci-ence by Davide Ruggero andcolleagues6 sheds a differentlight on its origins.

Ruggero et al.6 developed amouse model of DC that hasreduced expression of Dkc1.

These hypomorphic Dkc1m mice repro-duce the phenotype of human DC to aremarkable degree. The Dkc1m mice havesevere anemia, lymphopenia, hypocellu-larity of the bone marrow, and reducedlevels of erythroid and lymphoid colony-forming units, telltale signs of bone mar-

row failure, the predominant cause ofdeath in people with DC. The hallmark ofDC, dyskeratosis of the skin, is present inDkc1m mice, along with abnormalities inthe lungs and kidneys. Additionally, halfof the Dkc1m mice develop tumors, mir-roring the increased risk of malignancy in

Tipping the scales. The Dkc1m mousetips the scales of the proposed molecu-lar mechanism of DC from telomeremaintenance to protein synthesis, rep-resented by the telomerase RNP andthe box H/ACA snoRNPs, respectively.purple, box H/ACA snoRNA; light blue,box H/ACA–specific core proteins; red,dyskerin; dark green, TR; light green,telomerase-specific proteins.

Dissecting dyskeratosisU. Thomas Meier

Department of Anatomy & Structural Biology, Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, New York 10461, USA.e-mail: meier@aecom.yu.edu

Dyskeratosis congenita is a rare but fatal syndrome characterized by bone marrow failure. A new mouse model informs the ongo-ing debate on its molecular pathogenesis.

news & views

116 nature genetics • volume 33 • february 2003

dyskeratosiscongenita

telomere maintenanceprotein synthesis

telomerase RNP

box H/ACA snoRNPs

5´ ACA 3A ´

5´

ACA 3A ´

Dkc1mdyskerindyskerindyskerindyskerin

dyskerindyskerindyskerindyskerin

with age in living organisms. Further-more, the sensitivity and accuracy ofSTELA will likely find clinical applica-tions. For example, the ability to use verysmall tissue samples should allow theanalysis of premalignant lesions for telo-mere loss and subsequent chromosomeinstability.

A means to other endsHow far are we from applying this methodto other chromosomes? The most signifi-cant barrier is finding chromosomeend–specific sequence tags in the sub-

telomeric regions. This requires sequenc-ing the ends of human chromosomes—adaunting task that is now being systemati-cally attempted. Current data indicate thatSTELA will be feasible for 7q, 12q, 16p and16q. Although the technique may ulti-mately prove impractical for all telomeres,data from even one or two additionaltelomeres is eagerly anticipated.

One drawback of STELA is that itreveals little information concerning theG-rich overhang (see figure), which isemerging as a critical determinant fortelomere capping. However, used in

combination with the telomere PCRmethod (Table 1), which generates prod-ucts that correspond to the telomericduplex plus the entire G-rich overhang,these two methods have the potential toprovide a much clearer view of telomericDNA architecture. �

1. Baird, D. et al. Nat. Genet. 33,197–202 (2003).2. Hemman, M.T. et al. Cell 107, 67–77 (2001).3. Counter, C.M. et al. EMBO J. 11, 1921–1929 (1992).4. Zijlmans, J.M. et al. Proc. Natl. Acad. Sci. USA. 94,

7423–7428 (1997).5. Rufer, N. et al. Nat. Biotechnol. 16, 743–747 (1998).6. Förstemann, K. et al. Nucleic Acids Res. 28,

2690–2694 (2000).7. Griffith, J.D et al. Cell 97, 503–514 (1999).

©20

03 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://w

ww

.nat

ure

.co

m/n

atu

reg

enet

ics