Heart attacks are fatal when they dam-age more than a quarter of the heart’sleft ventricle — killing off about 109

heart cells in the process.In patients who sur-vive less severe attacks, dead heart cells arereplaced by cells from the connective tissuecalled fibroblasts, which divide and migrateinto the damaged area to form scar tissue. Inthese areas,the ventricular wall becomes thinand no longer contracts properly. Untilrecently, it seemed that the heart could notmend such an injury,but several reports nowsuggest that it might have some regenerativecapacity after all1–3. On page 647 of this issue, Laugwitz and colleagues4 providecompelling evidence that there are progeni-tor cells in the heart that might be able toengage in repair.

Like many specialized cells, fully devel-oped heart cells do not divide and so areunable to patch up damage. What’s neededare cells that are relatively unspecialized, sothat they can divide, maintaining their ownpopulation and providing cells that canmature into heart cells. The cell populationdescribed by Laugwitz and colleagues seemsto fit the bill.

These cells are distinguished by theirexpression of the islet-1 gene (that is, they areisl1�). By tracing the cells in a developingmouse embryo, Laugwitz et al.4 demonstrate

that the isl1� cells in the adult heart are rem-nants of a cardiac progenitor-cell populationfrom the heart of the developing fetus.More excitingly, the authors also find isl1�

cells in postnatal human hearts, and they can culture the mouse cardiac progenitors.Given an appropriate environment, the cultured cells divide and renew themselves,and they develop into what seem to bemature cardiac cells.

These cardiac progenitors are thus apotential source of cells for cardiac trans-plantation therapy. If they can be grownfrom human biopsies, then, in principle,tissue could be removed from a patient during surgery, and the cardiac progenitorcells could be amplified in culture and thenimplanted back into the patient’s heart. Theisl1� progenitor cells are found mainly in theheart atrium4,and taking atrial biopsies frompatients has had no demonstrable negativeeffects. Alternatively, it may be possible toencourage the proliferation of cardiac pro-genitors in situ in the heart, and to enhancean endogenous repair mechanism that doesn’t normally occur at a sufficient level toallow regeneration.

Cells derived from human embryos arebeing considered as another source for car-diac transplants5 because they can developinto almost any cell type. But use of such

news and views

NATURE | VOL 433 | 10 FEBRUARY 2005 | www.nature.com/nature 585

embryonic stem cells is controversial, andusing a patient’s own cells would pose fewerrejection problems. Three recent studies6–8

have independently identified other primi-tive cells from the adult heart that are capableof dividing and developing into mature heartand vascular cells. These cardiac stem cellsare distinct from the cardiac progenitorsdescribed by Laugwitz et al.4; both cell popu-lations divide and renew themselves, but theprogenitor cells are committed to becomingheart cells, whereas the stem cells have the potential to form several different celltypes (Fig.1).

Two of these studies found stem cells inthe hearts of adult rats6,7. These cells did notexpress isl1, but were isolated based on thepresence of cell-surface proteins (either c-kitor Sca-1) that are usually associated withstem cells derived from bone marrow. Thecardiac progenitors reported by Laugwitz et al. do not have these stem-cell markers,and so the cardiac progenitors and stem cellsare molecularly distinct.

In the first study6, c-kit� cells from rathearts were shown to be self-renewing andcapable of forming heart muscle cells andcertain vascular cells. Although the heartmuscle cells failed to contract spontaneouslyin culture, they did seem able to regeneratefunctional heart muscle when injected into a damaged heart. In the second study7,Sca-1�c-kit� cells from mouse hearts wereable to develop into heart muscle when givenintravenously after injury, although this wasin part by fusing with the host heart cells.

Cardiac stem cells can also be grown fromhuman biopsies as aggregates in suspensionculture8. These so-called cardiospheres con-tain a mixture of cardiac cell types, and,when transplanted into mice after an acuteheart attack, they formed vascular cells and heart muscle cells, albeit at a rather lowfrequency.

So, the race is on to find which cell typewill be the most useful for heart repair. Theoutcome will depend on whether the cellscan be produced in large numbers andinduce long-term functional repair withoutcausing irregular heart beats (arrhythmias).Fatal arrhythmias have occurred in patientswho received their own skeletal muscle pro-genitor cells in an effort to repair heart damage, and this was attributed to poorcommunication between the transplantedcells and host heart muscle9. The risk ofarrhythmia is minimal in mice because their fast heart rate (500 beats per minute)compensates for any disruption from trans-planted cells. However, the risk could be significant in larger mammals, includinghumans, where heart rates are at least five-fold slower. Experiments in large animalssuch as pigs will be the best way to predictpotential risks and assess the therapeuticvalue of cell transplantation into the heartbefore clinical trials are begun.

Cardiac stem cells(c-kit+ or Sca-1+)

Self-renewal

Self-renewal

Progenitorcells (isl1+) Cardiogenic

factors

Heart muscle cells

Vascular cells

Rightatrium

Leftatrium

Rightventricle

Leftventricle

Feeder cells

Figure 1 Cells for cardiac transplantation therapy. The cardiac progenitor cells described by Laugwitzet al.4 can renew themselves and develop into heart muscle cells; they are distinguished by thepresence of the islet-1 (isl1) protein. Cardiac stem cells6–8 can divide and develop into several celltypes. These cells express proteins typical of stem cells from other organs (c-kit and Sca-1), and donot express isl1.

Cardiology

Solace for the broken-hearted?Christine L. Mummery

The heart was thought to lack the capacity to regenerate after injury.But the identification of cells that can divide and mature into heartmuscle suggests that the heart has repair mechanisms after all.

10.2 n&v 583 NS 7/2/05 10:16 am Page 585

Nature Publishing Group© 2005

© 2005 Nature Publishing Group

Records of temperature during the pasttwo millennia provide clues to thenatural variation we might expect in

the future. They also support attempts topartition recent warming into natural andanthropogenic components, and to measurethe sensitivity of climate to greenhouse gasesin the atmosphere. Such long records comeonly from natural archives, for example treerings, ice cores and other ‘proxy’ evidence,and interpreting them has generated spiriteddebate. The crux of the issue is how muchwarmer or colder the average temperaturehas been in several century-long intervals ofthe past 2,000 years, the two most intensivelystudied intervals being the Medieval WarmPeriod (about AD 1200–1400) and the LittleIce Age (AD 1600–1850). Could we be in forone of these natural swings in the future? Ifso,how large would it be?

On page 613 of this issue1, Moberg et al.present a multi-proxy reconstruction ofannual temperature for the Northern Hemi-sphere for the past 2,000 years. Such a recon-struction is not new in itself, but it is uniquein the way that the authors consider theinformation encoded in proxies of differenttemporal resolution. They use a clever statis-tical approach, based on ‘wavelets’2, to com-bine the climate information preserved indifferent proxies at different scales, and gen-erate a reconstruction that reflects variabilityon scales ranging from the annual to themulti-centennial. In effect, the differentscales of information in the proxies are

voices of different frequencies, and the aim isto hear them clearly.

Wavelet analysis is a powerful tool,already in use throughout science and engi-neering, and it is used here to extract, andthen combine, the variance preserved in dif-ferent proxies. The approach is both elegantand appropriate, providing a means of sur-mounting known limitations of the palaeo-climate archive. High-frequency (multi-yearto multi-decade) variance is often blurred inall but the highest-fidelity ice or sedimentrecords, and low-frequency (century-scale)variance is not usually well preserved inhigh-resolution tree-ring reconstructions,particularly those assembled from shortoverlapping series (the ‘segment-length’curse3).

Using wavelets, the authors reconstruct atime series of Northern Hemisphere mean-temperature change from a suite of proxyrecords — hearing the high-frequency voices preserved in tree rings and the low-frequency voices retained in marine and lakesediments and other records. As a method oftime-series analysis, wavelets offer severaladvantages, and are notably free from theassumption of stationarity (unchangingmean and variance) that makes most meth-ods unsuitable for palaeoclimate time series.

The authors themselves recognize severalshortcomings. The limited number of longtree-ring series available provides only arough estimate of annual- to decadal-scaletemperature variability in the Northern

news and views

NATURE | VOL 433 | 10 FEBRUARY 2005 | www.nature.com/nature 587

100 YEARS AGOThe excavations [of Stonehenge] producedclear evidence touching the mode oferection… (1) The ground on the site it was to occupy was removed, the chalk rockbeing cut into in such a manner as to leave a ledge, on which the base of the stone wasto rest, and a perpendicular face rising fromit, against which as a buttress one side would bear when set up. From the bottom of this hole an inclined plane was cut to the surface, down which the monolith which had already been dressed was slid until its base rested on the ledge.(2) It was then gradually raised into avertical position by means first of leversand afterwards of ropes. The levers would be long trunks of trees, to one end of which a number of ropes wereattached… (3) As the stone was raised,it was packed up with logs of timber and probably also with blocks of stoneplaced beneath it. (4) After its upper end had reached a certain elevation, ropes were attached to it, and it was then hauled by numerous men into a verticalposition, so that its back rested against the perpendicular face of the chalk which had been prepared for it.From Nature 9 February 1905.

50 YEARS AGO“Symposium on Genetics of PopulationStructure.” Besides six papers, theProceedings of the Symposium contain short comments by Dobzhansky, Lerner and Epling, the conclusions by Buzzati-Traverso and the scholarly but delightfuladdress of thanks by Haldane… Scossirolireported on the results of selection forbristle number in Drosophila populationsafter heavy X-ray irradiations… thereoccurred a spectacular response toselection in the ‘high’ direction but not in the ‘low’ one… It could be that X-ray-induced inheritable variation is mainly in the direction opposite to that for whichnatural selection had to work harder— a point well worth investigation at thethreshold of the atomic age… The summing up… clearly showed how far the classic theoretical framework ofpopulation genetics has led… This change in outlook is essentially a shift of emphasis from the single gene to theintegrated systems of chromosomes, thegenotypes of the individuals and the wholegene pool of populations.From Nature 12 February 1955.

Adult bone-marrow cells were alsothought to be a source of replacement heartmuscle cells, but recently it was found thatthey cannot develop into cardiac cells10,11.However, several small-scale clinical studiessuggest that they may improve cardiac func-tion when transplanted immediately after aheart attack. Although it is unclear how thismight work, and most studies did notinclude a control patient group, no detri-mental effects were observed in patientsreceiving their own bone marrow in theheart. For this reason, there have been calls to set up large-scale, controlled clinical trials12,13 using bone marrow, without exten-sive experiments in animals first.

Regenerating heart muscle by cardiactransplantation therapy is an ambitious goal, but with the current developments,it holds more than an abstract promise. ■

Christine L. Mummery is in the HubrechtLaboratory and the Interuniversity CardiologyInstitute of the Netherlands, University MedicalCentre, Uppsalalaan 8, 3584 CT Utrecht,The Netherlands.e-mail: christin@niob.knaw.nl1. Nadal-Ginard B., Kajstura, J., Leri, A. & Anversa, P. Circ. Res. 92,

139–150 (2003).

2. Condorelli, G. et al. Proc. Natl Acad. Sci. USA 98, 10733–10738

(2001).

3. Laflamme, M. A., Myerson, D., Saffitz, J. E. & Murry, C. E.

Circ. Res. 90, 634–640 (2002).

4. Laugwitz, K.-L. et al. Nature 433, 647–653 (2005).

5. Kehat, I. et al. Nature Biotechnol. 22, 1282–1289 (2004).

6. Beltrami, A. P. et al. Cell 114, 763–776 (2003).

7. Oh, H. et al. Proc. Natl Acad. Sci. USA 100, 12313–12381

(2003).

8. Messina, E. et al. Circ. Res. 95, 911–921 (2004).

9. Menasche, P. et al. J. Am. Coll. Cardiol. 41, 1078–1083

(2003).

10.Murry, C. E. et al. Nature 428, 664–668 (2004).

11.Balsam, L. B. et al. Nature 428, 668–673 (2004).

12.Strauer, B. E. et al. Circulation 106, 1913–1918 (2002).

13.Wollert, K. C. et al. Lancet 364, 141–148 (2004).

Climate change

Let all the voices be heardD. M. Anderson and C. A. Woodhouse

It’s a tough job to excavate trustworthy records about past temperaturesfrom the palaeoclimate archives. The application of a fresh approach, inthe form of wavelet analysis of the data, is a step forward.

10.2 n&v 583 NS 7/2/05 10:16 am Page 587

Nature Publishing Group© 2005

© 2005 Nature Publishing Group

between potential substrates (Fig. 1a, b). Butenzymes that modify proteins and othermacromolecules need to distinguish betweensimilar (or even identical) sites within larger,dissimilar molecules. To do so, they must recognize the differences between substrates.That problem has been solved by diversifyingthe task of target recognition (Fig.1c).Where-as the motif to be modified (one or a fewamino acids in a protein, for instance) is rec-ognized by the enzyme’s active site, discrimi-nation between different substrates bearingthat motif is often accomplished throughspecific interactions between other sites onthe enzyme and substrate.

Cyclin-dependent kinases have appar-ently broken down this process even further.Whereas responsibility for recognizing thetarget motif (a serine or threonine followedby a proline) is delegated to a catalytic sub-unit (the CDK), both genetic and biochemi-cal studies suggest that exchangeableregulatory subunits (the cyclins) have a rolein discriminating between distinct proteinsubstrates (Fig. 1d). This is, perhaps, bestillustrated by baker’s yeast (Saccharomycescerevisiae), where the cell-cycle-regulatoryCDK, called Cdk1, can associate with ninedistinct cyclins — three G1 cyclins (Cln1–3)and six B-type cyclins (Clb1–6). Thesecyclins, in addition to activating Cdk1,directit towards distinct biological outcomes.

But although cyclins had been implicatedin substrate recognition, Loog and Morgan’spaper1 describes the first comprehensivestudy to compare the substrate specificity of purified CDK complexes that differ onlyin their cyclin. Their findings show thatClb5–Cdk1 and Clb2–Cdk1 complexes phos-phorylate most members of a group of 150previously confirmed Cdk1 substrates2 withroughly equal efficiency. However, 26 ofthose substrates are phosphorylated 2.5–800times as efficiently by Clb5–Cdk1. In con-trast, Clb2–Cdk1 does not preferentiallyphosphorylate any of the proteins.

The authors go on to extend previousstudies3–7 showing that a structural motif onthe surface of some cyclins, referred to as thehydrophobic patch (HP), specifically inter-acts with a so-called RXL or Cy motif foundon some CDK substrates and inhibitors. TheHP motif is important for the biologicalactivity of Clb5 (ref. 7). Loog and Morgan1

now establish that this motif is essential forenhancing the activity of Clb5–Cdk1towards its preferred substrates. Moreover,inactivating the Cy motif in the preferredClb5–Cdk1 substrates eliminates their pre-ferred status.

Strikingly, similar mutations in the Clb2HP motif do not affect the efficiency withwhich Clb2–Cdk1 phosphorylates any ofthe substrates, regardless of the presence orabsence of a Cy motif. That observation sug-gests that Clb2 does not use the HP motif forsubstrate recognition. In fact, Clb2 may not

confer substrate specificity upon Cdk1. Itmay simply activate it and leave substraterecognition entirely to the active site. Inkeeping with that interpretation, Archam-bault et al.8 have found that Cy-containingsubstrates depend upon the HP motif tointeract with Clb5 in an in vivo assay,but thatthose lacking Cy motifs interact equally wellwith HP-deficient Clb5 and Clb2.

So what is the role of the HP motif inClb2? Analysis of the relationship betweenthe six yeast B-type cyclins reveals that,although Clb5 and Clb2 are closely related interms of their overall sequence, their HPmotifs appear to be significantly different8.Given the known structure of a complexbetween human cyclin A3 and a Cy-motifpeptide3, the Clb2 HP motif seems to beincompatible with binding to the Cy motif 8.Nevertheless, it has been well conservedbetween different organisms,suggesting thatit is still important to Clb2’s function. Onepossibility is that it regulates a function ofClb2–Cdk1 other than its enzymatic activity.Indeed, mutation of the HP motif in Clb2impairs the protein’s export from thenucleus and its localization to at least one site in the cytoplasm9. Because Loog andMorgan’s analysis was performed largely invitro, using purified proteins, the impor-tance of subcellular localization in substrateselection was not evaluated.

Loog and Morgan’s study1 underlines theimportance of cyclins in recognizing appro-priate CDK substrates. The extent to whichsimilar mechanisms are exploited by othercyclins remains to be fully examined, butthere is ample evidence that other propertiesof cyclins are also important in substrateselection. Subcellular localization, alreadymentioned in the context of Clb2, is a well-established determinant of the biologicalfunction of yeast G1 cyclins10,11. Of equal oreven greater importance is the hallmark of

news and views

NATURE | VOL 434 | 3 MARCH 2005 | www.nature.com/nature 35

the cyclin proteins — their periodic accumu-lation during the cell cycle. Clearly, for a substrate to be phosphorylated it must bepresent in the cell along with the specificform of CDK that phosphorylates it.

So cyclins have a substantial role indirecting CDKs to specific substrates. Butthere are numerous mechanisms for doingso, more than one of which may be used by asingle cyclin. Ultimately, it is the combinedaction of these mechanisms that orchestratesthe orderly progression of events leading tothe faithful duplication of cells. ■

Curt Wittenberg is in the Departments of MolecularBiology and Cell Biology, Scripps Research Institute,La Jolla, California 92037, USA.e-mail: curtw@scripps.edu1. Loog, M. & Morgan, D. O. Nature 434, 104–108 (2005).

2. Ubersax, J. A. et al. Nature 425, 859–864 (2003).

3. Brown, N. R., Noble, M. E., Endicott, J. A. & Johnson, L. N.

Nature Cell Biol. 1, 438–443 (1999).

4. Kelly, B. L., Wolfe, K. G. & Roberts, J. M. Proc. Natl Acad. Sci.

USA 95, 2535–2540 (1998).

5. Sorensen, C. S. et al. Mol. Cell. Biol. 21, 3692–3703 (2001).

6. Takeda, D. Y., Wohlschlegel, J. A. & Dutta, A. J. Biol. Chem. 276,

1993–1997 (2001).

7. Wilmes, G. M. et al. Genes Dev. 18, 981–991 (2004).

8. Archambault, V., Buchler, N. E., Wilmes, G. M., Jacobson, M. D.

& Cross, F. R. Cell Cycle 4, 125–130 (2005).

9. Bailly, E., Cabantous, S., Sondaz, D., Bernadac, A. & Simon, M. N.

J. Cell Sci. 116, 4119–4130 (2003).

10.Edgington, N. P. & Futcher, B. J. Cell Sci. 114, 4599–4611

(2001).

11.Miller, M. E. & Cross, F. R. Mol. Cell. Biol. 20, 542–555 (2000).

Correction A misleading statement appeared in the News and Views article “Cardiology: Solace for thebroken-hearted?” by Christine L. Mummery(Nature 433, 585–587; 2005). The cardiacarrhythmias reported in reference 9 (P. Menascheet al., J. Am. Coll. Cardiol. 41, 1078–1083; 2003)were not the cause of fatalities in patients whoreceived their own skeletal-muscle progenitor cells as therapy for heart damage, as implied in the passage concerned.

Enzymes

Substrates

a b c d

Figure 1 How enzymes select their substrates. a, b, In general, enzymes recognize their targets through structural complementarity between the substrate and the enzyme’s active site (indicated here by the shape of the ‘pocket’). Small substrates (a) and relatively small modificationsites on proteins (b) can be recognized by this mechanism. c, Some enzymes make additional,specific contacts with the substrate that enable them to distinguish between proteins that haveidentical or related sites of modification. d, Loog and Morgan1 have compelling new evidence that cyclin-dependent protein kinases (CDKs) have relegated that function to the exchangeable cyclin subunit, enabling a single CDK catalytic subunit to exist in numerous forms with different specificities.

3.3 n&v 27 MH 25/2/05 5:27 pm Page 35

Nature Publishing Group© 2005