DOI: 10.1126/science.1138375 , 186 (1975); 189Science et al.C Duve,

Exploring cells with a centrifuge

www.sciencemag.org (this information is current as of February 5, 2007 ):The following resources related to this article are available online at

http://www.sciencemag.orgversion of this article at:

including high-resolution figures, can be found in the onlineUpdated information and services,

http://www.sciencemag.org#otherarticles, 43 of which can be accessed for free: cites 69 articlesThis article

http://www.sciencemag.org#otherarticles 6 articles hosted by HighWire Press; see: cited byThis article has been

http://www.sciencemag.org/help/about/permissions.dtl in whole or in part can be found at: this article

permission to reproduce of this article or about obtaining reprintsInformation about obtaining

registered trademark of AAAS. c 1975 by the American Association for the Advancement of Science; all rights reserved. The title SCIENCE is a

CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by the

on

Feb

ruar

y 5,

200

7 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

tween cyclopolysilanes and aromatic hy-drocarbons. Further investigations of cy-clopolysilanes seem likely to provide thekey to understanding of controversialquestions of bonding in metalloid com-pounds even as studies of their carbonanalogs, the cyclic and cage hydrocarbons,have been crucial to present knowledge oforganic stereochemistry and reactionmechanisms.The reactions of cyclopolysilanes are not

only interesting in themselves, but haveopened the way to the synthesis of complexpolysilanes and thus to whole new areas ofstudy. Improved methods of synthesis andisolation are needed, but the number'andkinds of compounds that can be preparedseem almost limitless. Perhaps a polymetalchemistry comparable in breadth and vari-ety to carbon chemistry is now developing.

References and Notes

I. Organosilicon compounds of high nolecularweight are, however, well known in the form of thesiloxanes, whose catenated structures are based onaltemating silicon and oxygen atoms; these are thesilicone polymers of commerce.

2. The first polysilanes to be systematically studiedbore aromatic rings as substituents. In classic re-searches early in this century, F. S. Kipping pre-pared compounds such as hexaphenyidisilane(Ph,SiSiPh,) and octaphenylcyclotetrasilane[(Ph2Si)4]. The chemistry and physics of arylpoly-silanes tends however to be dominated by thearomatic rings, so that the special properties of thecatenated silicons are masked. For further infor-mation on these compounds, see H. Gilman and G.L. Schwebke, Adv. Organometal. Chem. 1, 89(I1964); J. Am. Chem. Soc. 86, 2613 (1964); J. Or-ganometal. Chem. 3, 382 (1965).

3. For a review of linear permethylpolysilane chem-istry, see M. Kumada and K. Tamao, Adv. Organ-ometal. Chem. 6, 19 (1968).

4. W. G. Boberski and A. L. Allred, J. Am. Chem.Soc. 96, 1244 (1974); J. Organometal. Chem. 71,C27 (1974).

5. B. G. Ramsey, Electronic Transitions of Organ-ometalloids (Academic Press, New York, 1969),pp. I 1 3 126.

6. The chemistry of cyclic permethylpolysilanes hasbeen briefly reviewed [R. West, Ann. N.Y. Acad.Sci. 239, 262 (l974)].

7. C. A. Burkhard,J. Am. Chem. Soc. 71, 963 (1949).8. U. G. Stolberg, Angew. Chem. Int. Ed. Engl. 2, 150

(1963).9. E. Carberry and R. West, J. Organometal. Chem.

6, 582 (1966); J. Am. Chem. Soc. 91, 5440 (1969).10. M. Ishikawa and M. Kumada, Chem. Commun.

(1970), p. 612; J. Organometal. Chem., in press.Photolysis of dodecamethylcyclohexasilane is anexcellent way to generate dimethylsilylene andhas been used with success for insertions intoheteropolar single bonds. A useful example is theproduction of dimethylchlorosilane by the photol-ysis of dodecamethylcyclohexasilane in the pres-ence of dry hydrogen chloride [M. Ishikawa andM. Kumada, Chem. Commun. (1971), p. 507].

11. H. L. Carrell and J. Donohue, Acta Cryst. B28,1566 (1972). The equatorial Si-C distances areequivalent averaging 1.935 ± 0.004 A, whereas theaxial Si-C distances are equivalent at 1.913 ±0.004 A. These differences are not certainly sig-nificant and, if real, may result from crystal pack-ing effects.

12. E. Carberry, R. West, G. E. Glass, J. Am. Chem.Soc. 91, 5446 (1969).

13. T. F. Hunter and M. C. R. Symons,J. Chem. Soc.(Lond.) Sect. A (1967), p. 1770.

14. J. R. Morton, Chem. Rev. 64,453 (1964).15. It should be noted, however, that one could also in-

voke a model using 3p orbitals which would implya "a* orbital model." This model is supported byRamsey in his elegant application of the Wood-ward-Hoffmann selection rules to the photolysis ofpolysilanes [B. G. Ramsey, J. Organometal.Chem. 67, C67 (1974)].

16. E. Kean and R. West, unpublished studies.17. C. G. Pitt, M. M. Barsey, P. F. Rogerson, J. Am.

Chem. Soc. 92, 519 (1970).18. H. Bock and W. Ensslin, Angew. Chem. Int. Ed.

Engl. 10, 404 (1971).19. The pattern of Si-Si bonding MO's in these mole-

cules is exactly like that observed for the pi energylevels in the aromatic species cyclopentadienide(C,H,-) and benzene. This coincidence arises be-cause the splitting of MO's is symmetry dependentand the symmetries of the polysilanes and thearomatic molecules are sufficiently similar (seeFig. 6).

20. V. F. Traven and R. West, J. Am. Chem. Soc. 95,6824 (1973). Charge transfer complex formation isnot limited to cyclic polysilanes; linear per-methylpolysilanes also serve as donors to TCNEand similar ir acceptors.

21. H. Sakurai, M. Kira, T. Uchida, J. Am. Chem.Soc. 95, 6826 (1973).

22. M. Ishikawa and M. Kumada, Chem. Commun.(1969), p. 567; Syn. Inorg. Metal-Org. Chem. 1,191, 229 (1971).

23. M. Kumada, Intra-Science Chem. Rep. 7, 121(1973).

24. The acid-catalyzed chlorodemethylation of cy-clopolysilanes can be thought of as analogous tothe electrophilic chlorination of aromatic hydro-carbons. As with aromatic compounds, polysubsti-tution can also take place. For example, the di-chlorination of (SiMe,), appears to produce atleast three major isomers, whose structures are asyet unknown. The situation in this field is ratherlike that in aromatic chemistry about a centuryago, when it was known that disubstitution of ben-zene gave three isomers but their identification asortho, meta, and para was not generally possible.

25. R. Parnell and R. West, unpublished studies.26. M. Ishikawa, A. Nakamura, M. Kumada, J. Or-

ganometal. Chem. 59, CII (1973).27. R. West and E. S. Kean, ibid., in press.28. A. Indriksons and R. West, J. Am. Chemi. Soc. 94,

6110(1972).29. C. Hurt and R. West, unpublished studies.30. M. Ishikawa, communication to the Third Interna-

tional Symposium on Organosilicon Compounds,Madison, Wisconsin, August 1972.

31. W. Stallings and J. Donohue, unpublished studies.32. W. Bukhtiyarov, S. P. Solodovnikov, 0. M. Nefe-

dov, V. 1. Shiryaev, Izv. Akad. Nauk SSSR, Ser.Khim. 5, 1012 (1968).

33. G. R. Husk, R. Wexler, B. Kilcullen, J. Organ-ometal. Chem. 29, C49 (197 1).

34. M. Biernbaum and R. West, ibid. 77, C 13 (1974).35. E. Hengge and F. Lunzer, Synth. Inorg. Metal.

Org. Chem. (1972), p. 93.36. E. Hengge and G. Bauer, Angew. Chen,. 85, 304

(1973).37. 0. M. Nefedov, M. N. Manakov, A. D. Petrov,

DokI. Akad. NaukSSSR 147, 1376(1962).38. 0. M. Nefedov, G. Garzo, T. Szekely, V. 1. Shir-

yaev, ibid. 164, 822 (1965). A compound believedto be (Me,Ge)4 was also isolated from the reactionof methylmagnesium bromide or methyllithiumwith trichlorogermane; see 0. M. Nefedov, S. P.Kolensnikov, V. 1. Sheichenko, Angew. Chem. 76,500 (1964). More recent studies (E. Carberry)show that it is most unlikely that the substance ob-tained was actually (Me,Ge)4.

39. E. Carberry, B. Dombek, S. Cohen, J. Organ-ometal. Chem. 36, 61 (1972).

40. E. Carberry and B. Dombek, ibid. 22, C43 (1970);E. Carberry and D. Jones, unpublished report.

41. Ge-Si compounds whose reduction was also at-tempted include dodecamethylgermacyclohexasil-ane and dodecamethyl- 1 ,4-digermacyclohexasil-ane [see (38)].

42. T. L. Brown and G. L. Morgan, Inorg. Chem. 2,736 (1963).

43. Supported by the Air Force Office of Scientific Re-search (NC), Office of Aerospace Research,USAF, grant AF-AFOSR 70-1904; research atSouthwest Minnesota State College has been sup-ported by partial grants from the Research Corpo-ration and the donors of the Petroleum ResearchFund, administered by the American ChemicalSociety.

Exploring Cells with a Centrifuge

Christian de Duve

In one of her masterpieces, Nobel Lau-reate Selma Lagerlof tells how the littleboy Nils Holgersson visited the whole ofSweden, from Skane to Lappland, on thewings of a friendly white gander. I too havemade a wonderful journey, using, like NilsHolgersson, an unconventional mode oftravel. For the last 25 years, I have roamed

186

through living cells, but with the help of a

centrifuge rather than of a microscope.On these trips I was never alone. I want

to mention this at the onset, since I owe

much to my traveling companions. Someof their names will come up as my tale un-

folds; but there are so many of them that Iwill be quite unable to mention them all.

My debt goes also to my early mentors inscience: Joseph Bouckaert, Joseph Maisin,Hugo Theorell, Carl and Gerty Cori, andEarl Sutherland. Four of them have pre-ceded me on this podium. Three, unfortu-nately, are not with us any longer.

The Development of AnalyticalCell Fractionation

Thirty years ago, much of the living cellstill remained virtually unexplored. Thereasons for this are simple. Morphologicalexamination was limited downward in thescale of dimensions by the resolving powerof the light microscope, whereas chemicalanalysis stopped upward at the size of thesmaller macromolecules. In between, cov-

ering almost two orders of magnitude, laya vast terra incognita, impenetrable withthe means of the day. Invasion of this terri-

SCIENCE, VOL. 189

on

Feb

ruar

y 5,

200

7 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

tory started almost simultaneously on itstwo frontiers, after electron microscopybecame available to morphology and cen-trifugal fractionation to biochemistry.When, in 1949, I decided to join the little

band of early explorers who had followedAlbert Claude in his pioneering expedi-tions, electron microscopy was still in itsinfancy. But centrifugal fractionation, thetechnique I wanted to use, was already wellcodified. It had been described in detail byClaude himself (1), and had been furtherrefined by Hogeboom, Schneider, and Pa-lade (2), and by Schneider (3). Accordingto the scheme developed by these workers,a tissue, generally rat or mouse liver, wasfirst ground with a Potter-Elvehjem ho-mogenizer, in the presence of either 0.88M(2) or 0.25M (3) sucrose. The homogenatewas then fractionated quantitatively bymeans of three successive centrifugationsand washings, under increasing centrifugalforce x time integrals, to yield "nuclei,""mitochondria," "microsomes," and a fi-nal supernatant. The fractions, as well asthe original homogenate, could then be an-alyzed for their chemical composition, en-zyme content, and other properties.

All these details were available in the lit-erature, and there seemed little more for usto do than to acquire the necessary equip-ment and follow instructions carefully, es-pecially since our interest in cell fractiona-tion itself was rather peripheral at thattime. All we wanted was to know some-thing about the localization of the enzymeglucose 6-phosphatase, which we thoughtmight provide a possible clue to the mecha-nism of action, or lack of action, of insulinon the liver cell.

Fortunately, this is not exactly howthings happened. Working with me on thisproject was Jacques Berthet, still a medicalstudent at that time, but with an unusuallymature and rigorous mind. He went aboutthe job of setting up the technique in acareful and systematic fashion, paying spe-cial attention to all physical parameters. Afew practical tips from Claude, who hadjust returned to Belgium, were also helpful.

Particularly important, I now realize inretrospect, was the fact that we took sometime to study the theory of centrifugation,as beautifully exposed in the classical bookby Svedberg and Pedersen (4).

Although separating mitochondria andmicrosomes might appear worlds apartfrom the determination of the molecularweight of macromolecules, certain con-cepts were common to the two operationsand could be usefully transposed from thelatter to the former. One was that of sedi-mentation coefficient (Fig. 1), which obvi-ously was applicable to any particle, irre-spective of its size. Another was that ofpolydispersity, which, owing to biologicalvariability, was likely to be a property ofthe populations made up by subcellular or-ganelles. This meant that the centrifugalbehavior of such populations could be de-scribed only by a frequency distributioncurve of sedimentation coefficients (Fig. 2),not by a single s value as for most molecu-lar populations. A third important pointrelated to the resolving power of differ-ential sedimentation, which some elemen-tary calculations revealed to be surpris-ingly low (Fig. 3).There was much insistence in those days

on the various artifacts that complicatecentrifugal fractionation, such as, for in-stance, breakage or agglutination of par-ticles, and adsorption or leakage of solubleconstituents. But these were only acci-dents, no doubt serious but amenable toexperimental correction. The problem, as

(A

I/I

II

Ii

II

it appeared to us, was a more fundamentalone. What we were doing was trying toseparate populations that, owing to over-lapping polydispersities, might at best beonly partly separable from each other. Inaddition, we were using a poorly discrimi-nating method for this purpose.

I cannot claim that all this was immedi-ately clear to us. But considerations of thissort undoubtedly colored our approachfrom the start (5). We fully expected cen-trifugally isolated fractions to be impure,while suspecting that populations of cellorganelles might be difficult, if not impos-sible, to resolve quantitatively. Consciousalso of the severe limitations of light mi-croscopic examination of the fractions, wetried to extend the biochemical inter-pretation as far as possible. Instead oflooking at each fraction separately and fo-cusing on its enzyme content, as was usu-ally done, we looked rather at each individ-ual enzyme and contemplated its distribu-tion between all the fractions.

In order to permit a comprehensive viewof enzyme distribution patterns, I in-troduced a histogram form of representa-tion (Fig. 4). In this figure are shown thedistribution patterns of three of the firstenzymes we studied, on the left as deter-mined by the classical four-fraction

Sedimentation velocity

dx/dt = s 2 x

x = Radial distance (cm)

c = Angular velocity (rad x sec-1)

s = Sedimentation coefficient of

particle (sec)

Copyright © 1975 by the Nobel Foundation.The author is Andrew W. Mellon Professor at Rock-

efeller University, New York 10021, and professor atthe Universite Catholique de Louvain, Louvain, Bel-gium. This article is the lecture he delivered in Stock-holm, Sweden, on 12 December 1974 when he receivedthe Nobel Prize for Physiology or Medicine, a prize heshared with Albert Claude and George Palade. The ar-ticle is published here with the permission of the NobelFoundation and will also be included in the completevolume of Les Prix Nobel en 1974 as well as in the se-ries Nobel Lectures (in English) published by the Else-vier Publishing Company, Amsterdam and New York.The lectures by Claude and Palade will appear in laterissues.

18 JULY 1975

For spherical particle of radius r (cm) and of density pp (g x cm-3)

In medium of density Pm (g x cm-3) and of viscosity rj (poises)

s = 2r2(pp -Pm )/977

Fig. 1. The Svedberg equation and its application to a spherical particle.

187

on

Feb

ruar

y 5,

200

7 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

% Sedimented

100

80F

60F

40

20

Sedimentation coefficient 10 0.1Relative volume

Fig. 2 (left). Image of a polydisperse population of particles. Owing to individual difor density (or both), different members of the population do not have the samecoefficient. The centrifugal properties of the population as a whole are depicted by a

tribution curve of sedimentation coefficients. Size or density distributions can be !sented. Frequency is usually defined as dn/Ndx (or, in the case of histograms, An/l(dn/N) (An/N) is the fraction of total particles having an abscissa value comprisedx+ dx (Ax). Instead of relative number, similar diagrams may be drawn in terms orelative enzyme activity, and so forth. Fig. 3 (right). The percentage of particlessediment as a function of relative particle volume. Particle density is assumed to beparticles. The meniscus of fluid in the rotating centrifuge is assumed to be halfway band the bottom of the tube or cell.

Classical schemeI

M Cytochror

0 204040 60 80 10

Acid pt

M

0 20 40 60 80 101fr

4Modified scher

me oxidase M

L2 -

1N

0 20 40 6010 rhosphatase

8-9 -

7^

2

1

,6 0

0 20 40 60 80 100

Dsphatase PLAN L

0 20 40 60

scheme, and on the right as determined bythe modified five-fraction scheme that weworked out in an effort to elucidate the sig-nificance of the small difference in distribu-tion observed between acid phosphataseand cytochrome oxidase (6). This differ-ence, as can be seen, is very much magni-fied by the modification in fractionation

0.01 0,001 scheme.

These histograms turned out to be veryfferences in size *sedimentation revealing, by more or less automaticallyafrequencydis- conveying the notion of polydispersity,

similarly repre- illustrated in Fig. 2. In fact, since theN Ax) in which fractions are aligned along the abscissabetween x and in order of decreasing sedimentation

f relative mass,s recovered in a coefficient, one may, In a very crude fash-the same for all ion, look at the abscissa as a deformedetween the axis scale of sedimentation coefficients, and at

the histograms as correspondingly de-formed frequency distribution histogramsof sedimentation coefficients. The logical

me next step in this line of reasoning was to as-similate enzyme distributions to particledistributions, and therefore to interpret, atleast tentatively, significant differences inthe distribution patterns of two enzymes asreflecting association of the enzymes withdistinct particle populations.

S Extrapolation from enzymes to particles80 100 could not, however, be made without some

sort of assumption concerning the relation-ship between relative enzyme activity, thenumerator in the ordinate of Fig. 4, andrelative particle number, the numerator inthe ordinate of Fig. 2. The simplest, and atthe same time most plausible, such as-sumption was that members of a givenparticle population have essentially thesame biochemical composition, largerparticles simply having more of everythingthan smaller particles. Within the limits ofvalidity of this assumption, which I havecalled the postulate of biochemical homo-geneity, the histograms of Fig. 4 could nowbe likened to distribution diagrams of total

s particle mass or protein (not of actualparticle numbers, it should be noted, al-

80 100 though further conversion to numericaldistributions can be made with some addi-tional information). We had to assume, ofcourse, that the enzyme distributions werenot grossly distorted by translocation arti-facts, or to correct for such artifacts asmuch as possible.Another postulate we made was that

s each enzyme is restricted to a single intra-_ ~ cellular site. This postulate of single loca-

80 l1o tion is less essential than that of biochemi-

Nitrogen content (% of total nitrogen)

Fig. 4. Enzyme distributions represented in histogram form. The relative specific enzyme content (percentage of activity divided by percentage of pro-

tein) of the fractions is plotted against their relative protein content, inscribed cumulatively from left to right in their order of isolation (decreasingsedimentation coefficient): nuclear (N), mitochondrial (M), microsomal (P), and supernatant (S), in classical four-fraction scheme; and nuclear (N),heavy mitochondrial (M), light mitochondrial (L), microsomal (P), and supernatant (S), in modified five-fraction scheme (6). Although very crude, the

similarity with frequency distribution curves of polydisperse populations can be recognized. Distinction between three populations, now known to con-

sist of mitochondria (cytochrome oxidase), lysosomes (acid phosphatase), and endoplasmic reticulum fragments (glucose 6-phosphatase), is enhancedby use of five-fraction scheme. [Source: (44)]

cStoo00-

Swe

ENcS

0

U>

U

OR

4.

S

S

._

cL

._00

41

6

.5

4

3

2

.1

01055

-phc.4.

Glucose-(p

NXL

u

a>craL)U-

4.i

I

?I -ILIO

SCIENCE, VOL. 189188

on

Feb

ruar

y 5,

200

7 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

cal homogeneity, since bimodal or multi-modal distributions are amenable to thesame kind of interpretation. In practice,however, single location made a useful ad-dition to biochemical homogeneity, sup-porting the use of enzymes as markers oftheir host particles.

First used empirically as pure workinghypotheses, the above considerations wereprogressively validated, as more enzymeswere studied and a limited number of typi-cal distribution patterns began to emerge.Actually, as shown by the results of Fig. 5,things were not quite as simple, and a num-ber of complications of various sortstended to blur the picture. But most ofthese could be dealt with satisfactorily byancillary experiments (7).

In these studies, a second line of evi-dence based on enzyme latency, provedvery useful. Owing to impermeability ofparticle membranes to one or more of thesubstrates used in the assay of enzymes,many particle-bound enzymes fail to dis-play activity "in vitro" as long as the mem-brane surrounding them is intact. Variousmeans, mechanical, physical, or chemical,can be used to disrupt the membrane andto release the enzymes, as we first showedfor rat liver acid phosphatase (Fig. 6). Iftwo or more enzymes are present togetherin the same particles, they will be releasedtogether in this kind of experiment; if indifferent particles, they may come out sep-arately (Fig. 7). In our hands, such studieshave been very useful, providing an inde-pendent verification of the significance ofthe similarities and differences revealed bycentrifugation experiments.By 1955, our results were sufficiently ad-

vanced to allow us to propose with a cer-tain measure of assurance the existence ofa new group of particles with lytic proper-ties, the lysosomes, and to hint at the exis-tence of another group of particles, the fu-ture peroxisomes (7). At the same time, wehad, from the mixture of theoretical con-siderations and experimental results that Ihave just briefly recalled, derived a certain"philosophy" of centrifugal fractionation,which I subsequently elaborated in greaterdetail in several publications (8). The keyword here was "analytical." Basically, wefelt that our approach was no more than anextension of the classical Svedberg tech-nique from the molecular to the sub-microscopic and microscopic level.A major difficulty at this stage, however,

was that available techniques did notmeasure up to the kind of information wewere hoping to extract. The answer to thisproblem was provided by density gradientcentrifugation, which was introduced in theearly 1950's. This new technique offeredprospects of improved resolution; it al-lowed the use of density, as well as of sedi-18 JULY 1975

mentation coefficient, as a separation pa-rameter; and, finally, its analytical charac-ter was unmistakable (Fig. 8). In fact, asshown as early as 1954 by Hogeboom andKuff (9), it could even be used successfullyfor the determination of molecularweights.

Here again, we devoted some time totheoretical studies (10). In this, Berthetand I were joined by another young co-

0 20 40 60 80 100I If I ,1 IF I if

. . .

.0a

0

0

0

U.

4-

0._

0._Z

U

0

0-0

d._

(_

-

4-U

Q._

4-

CK

O _

4 -

3 -

2 -

o _

o L

3 _

4_3 _2 _

1 L

4 _-

I I

0

Cytochrome o*idase (17)Pattern I

worker, Henri Beaufay, whose skills as aself-taught engineer proved particularlyvaluable for the design of various accesso-ries, culminating in the construction of acompletely automatic rotor (11), differentin principle from the zonal rotors built byNorman Anderson (12), and particularlyadapted to rapid isopycnic separation atminimum hydrostatic pressure. The im-portance of the latter advantage has been

0 20 40 60 80 100[ I I I I I I I I I I

Acid phosphatase (19)Pattern IlIl

RibonuclSuccinate-cytochrome creductase (3)

Rhodanese (3) Deoxyr

TPNH-cytochrome creductase (3)

Fumarase (2)

DPNH cytochrome creductase (6)

Glucose-6-phosphatase (9)

Pattern II

a

I I II I I I I 120 40 60 80 100

Catheps

lease (8) S

33I 2- ~~~~0

Sbonuclease (6) _ 4

3I 2_S

,n (9) 43I 2

- ~~~~~0

/3-Glucuronidase (4)

Uricase (S)Pattern IV

I I I 1 1 1 1 1 1 1

2

98

-7

6

S

-43

-2

0

0 20 40 60 80 100

Total nitrogen (%)Fig. 5. Distribution patterns of enzymes in rat liver fractions separated by five-fraction procedureshown in Fig. 4. Pattern I, shared by three enzymes, represents the distribution of mitochondria; pat-tern 11 (glucose 6-phosphatase), that of microsomes. In between, in the left column, are complexcombinations of patterns I and 11. Pattern III is shared by five lysosomal acid hydrolases, except for,B-glucuronidase which has an additional microsomal component. Pattern IV belongs to the per-oxisomal urate oxidase. The numbers of determinations are given in parentheses. Details are given inthe original report. [Source: (7)]

189

on

Feb

ruar

y 5,

200

7 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

Intact granules

Enzyme bound

10o 1-

Low activity > .= 80K- Acid

°2 60 phosphatase

M-Catalase

Glycerophosphate 8

0 0.5 1.0 1.5 2.0

Waring blenderHypotonic media

|Freezin g an d th awingDetergents

//Acid

hs ,/-Glycerophosphote

Enzyme soluble High activity

Injured granules

emphasized by my former collaboratorRobert Wattiaux (13).

Particles sedimenting through a densitygradient are apt to undergo a progressiveincrease in density, due to inflow of soluteor outflow of water or both, depending on

the number and permeability properties oftheir membranes and on the nature of thesolute (or solutes) and solvent used to

make the gradient. These factors we triedto incorporate in a theoretical model ofparticle behavior (10, 14), and at the same

time to take into account in the design ofour experiments. It appeared from our the-oretical considerations that the sucrose

concentration of the medium might be a

particularly important variable, and thatdifferent types of particles might responddifferently to changes in sucrose concen-

tration. We therefore subfractionated largegranule fractions from rat liver in isosmo-tic glycogen gradients prepared with su-

crose solutions of different concentrationsas solvent, as well as in sucrose gradientsprepared with either H20 or D20 (15).The results of these experiments con-

Digitonin (mg/ml)

Fig. 6 (left). Model of latency of rat liver acidphosphatase, as proposed in 1951 (58). [Source:(59)] Fig. 7 (above). Differential release ofthe lysosomal acid phosphatase and of theperoxisomal catalase by increasing concentra-

tions of digitonin. [Source: (59)]

firmed and extended our earlier findings,establishing the existence of three distinctgroups of enzymes, as defined by their cen-

trifugal behavior. There was little doubt inour minds that these observations reflectedthe occurrence of three distinct popu-

lations of particles in the large granulefraction. By fitting our results to the theo-retical equation, we were even able to eval-uate a number of physical parameters foreach putative particle population and to

construct, from purely biochemical data, a

sort of "robot picture" of the particlesthemselves (Table I). Due to heterogeneitywithin the population, the data given inthis table for the lysosomes are of ques-

tionable significance. On the other hand,those listed for mitochondria and per-

oxisomes agree very well with measure-

ments made by other techniques.Although analytically satisfactory, the

results described so far still fell short of de-finitive proof, since they had unfortunatelyconfirmed our fear that distinct popu-

lations of subcellular particles might prove

intrinsically inseparable quantitatively due

to overlapping of size or density distribu-tions, or both. It was possible to obtainpure samples by cutting off non-

overlapping parts of the populations, butthis introduced the danger of biased sam-

pling. A means of almost complete separa-

tion, although under somewhat artificialconditions, was provided in 1962 by Wat-tiaux, Maurice Wibo, and Pierre Baudhuin(16), when they discovered that treatment

of the animals with Triton WR-1339causes a selective decrease in the density ofsubsequently isolated lysosomes, due to ac-

cumulation of the Triton within these par-

ticles (Fig. 9). Thanks to this finding and to

the Beaufay rotor, large-scale separationof the three populations has now becomepossible, allowing a variety of biochemicaland functional studies that were not fea-sible before (17).While the biochemical approach I have

outlined was being developed in our labo-ratory, electron microscopy was makinggreat strides of its own, soon becomingavailable for the examination of sub-cellular fractions. For obvious reasons we

were very anxious to take a look at our

purest fractions, in order to test our con-

clusions and eventually identify our hypo-thetical particles. Already in 1955, thanksto the expert collaboration of Alex Novi-koff and to the facilities of Claude in Brus-sels and of Wilhelm Bernhardt in Paris, wewere able to do this for lysosome-rich frac-tions, which were found to contain densebodies, surrounded by a membrane and ofabout the size predicted for lysosomes (18).Later, we were able to acquire an in-

strument of our own, and Beaufay taughthimself another skill, which he later per-

fected under the guidance of GeorgePalade. With Baudhuin, he confirmedthe identification of lysosomes as "peri-canalicular dense bodies" and showed that

the peroxisomes correspond to the parti-

Table 1. Typical physical properties of rat liver particles [Source: (59)]

Reference enzyme

Parameter Mitochondria Lysosomes Peroxisomes

Cytochrome Acid Acid deoxy- U rate d-amino acidoxidase phosphatase ribonuclease oxidase Catalase oxidase

Dryweight(Mg) 10-7 2.7 x 10-8 3.6 x I0-8 2.4 x 10-#Dry density 1.315 1.300 1.331 1.322 1.319 1.315Osmotically active solutes

(milliosmole/g, dry weight) 0.157 0.128 0.334 0 0 0Water compartments (cm3/g, dry

weight)Hydration 0.430 0.256 0.212 0.214 0.295 0.296

Sucrose space 0.905 1.075 0.330 2.51 2.68 2.54Osmotic space in 0.25M sucrose 0.595 0.485 1.265 0 0 0

Total in 0.25M sucrose 1.930 1.816 1.807 2.724 2.975 2.836Sedimentation coefficient in 0.25M

sucrose (Svedberg units) 104 4.4 x 103 5 x 103 4.4 x 103Diameter in 0.25M sucrose ( Mm) 0.8 0.51 0.56 0.54Density in 0.25M sucrose 1.099 1.103 1.100 1.095 1.088 1.090

190 SCIENCE, VOL. 189

-6 G-

on

Feb

ruar

y 5,

200

7 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

cles known as "microbodies" (19). Thus,the gap between biochemistry and mor-phology was finally bridged, after some 15years of research.More recently, Baudhuin has adapted

quantitative morphometric methods to theexamination of subcellular fractions, mak-ing it possible to compare measurementsderived from biochemical data with thoseobtained by direct mensuration (20). Inseveral instances, excellent agreement hasbeen found between the two sets of data(17, 20-22).

Applications to Biology

I have chosen to dwell at some length onour theoretical and technical studies, be-cause they were, I believe, the key to what-ever achievements were made by ourgroup. I know that others have accom-plished important advances by the alterna-tive process of first purifying a subcellularcomponent and then analyzing it. For ex-ample, nuclei, secretion granules, plasmamembranes, and Golgi elements have beenlargely characterized in this fashion. Butpurification is generally a laborious proce-dure, it is difficult to control, and it israrely quantitative. The advantage of theanalytical approach is that it is widely ap-plicable, and it can provide a considerableamount of quantitative information, evenwith a relatively poor resolving power. Theimportant point is that with this kind ofmethodology, we derive the informationnot from the properties of specific frac-tions believed to approximate a given in-tracellular component, but from the man-ner in which properties are distributed overa large number of fractions, which togetherrepresent the whole tissue.

In our laboratories, this general ap-proach has been applied to a variety of bio-logical materials and for the study of manydifferent problems. In continuation of thework on liver, already described, it hassupported a number of studies concernedwith the functions of lysosomes, includingthose of Wattiaux on intralysosomal stor-age (23), of Pierre Jacques on pinocytosis(24), of Russell Deter on autophagy (25),of Jack Coffey, Nick Aronson, and StanleyFowler on lysosomal digestion (26), and ofAndre Trouet and Paul Tulkens on the ef-fects of antibodies against lysosomes (27).It has also allowed Brian Poole, FedericoLeighton, Tokuhiko Higashi, and PaulLazarow to make a searching analysis ofthe biogenesis and turnover of peroxisomes(21, 28). In recent years, a large teamgrouped around Beaufay and Berthet, andincluding Alain Amar-Costesec, ErnestFeytmans, Mariette Robbi, Denise Thines-Sempoux, and Wibo, has launched a major18 JULY 1975

attack on microsomal and other mem-brane fractions with the aim of character-izing physically, chemically, enzymically,and immunologically the various types ofcytomembranes occurring in these frac-tions (29).

In its applications to other mammaliantissues and cell types, analytical cell frac-tionation has allowed Baudhuin and Pooleto recognize peroxisomes in kidney (30);Gilbert Vaes to carry out a thorough studyof bone lysosomes, leading to very reveal-ing observations on the role of these parti-cles in bone resorption (31); Bill Bowers tomake a comprehensive biochemical dis-section of lymphoid tissues and lympho-cytes, as a preliminary to an analysis ofcell-mediated immune cytotoxicity (32);Marco Baggiolini to characterize the twotypes of granules present in neutrophilpolymorphonuclear leucocytes (33); Rich-ard Schultz and Jacques to unravel someof the complexities of placental tissue (34);Tim Peters to fractionate aortic smoothmuscle cells (35) and enterocytes (36); andTulkens to do the same for cultured fibro-

Den sity Befo re

blasts (37), a system also used by Pooleand Wibo for investigations of proteinturnover (38).Under the leadership of Mikl6s MUller,

a series of fascinating studies have beenperformed in New York on a number ofdifferent protozoa. In Tetrahymena pyri-formis, MUller was able to identify twotypes of lysosomes, which discharge theirenzymes, one in phagocytic vacuoles andthe other in the outside medium (39). Incollaboration with Baudhuin and later withJim Hogg, he has shown the existence inthe same organism of peroxisomes that,like plant glyoxysomes, contain enzymesof the glyoxylate cycle (30, 40). More re-cently, with Don Lindmark, he has charac-terized in Trichomonas a completely newtype of cytoplasmic particle, with the ca-pacity of converting pyruvate to acetate,carbon dioxide, and molecular hydrogen,the hydrogenosome (41).

Other studies have dealt with the role oflysosomes in tissue regression, notablythose of Denise Scheib-Pfleger and Wat-tiaux on MUllerian ducts in chick embryos

After Frequency

p

1. Differential sedimentationGradient: ShGllow stObl/Izlng, m < mm"',~ mox. "CIp 1min.

Centrifugatimon IIncomplete sedomentotlon

Abscissa of frequency distribution Sedimentoffon coefflcienf

2. Density equilibrationGradient Steep, ma > m"mox /p mov<

Centrifugat ion: Prolonged, high speed

Abscissa of frequency distribution. Equli//brilum densifyFig. 8. Schematic representation of density gradient centrifugation, with initial top layering of thesample. Two forms, based on differences in sedimentation coefficient and density, respectively, areshown. Diagram at the right pictures frequency distribution of particles or markers as a function oftube height. Conversion to frequency distributions of sedimentation coefficients or densities gener-ally requires readjustment of ordinate and abscissa values, leaving surface area of each block (per-centage of content in fraction) unchanged. For details of calculations, see (10). [Source: (59)]

191

on

Feb

ruar

y 5,

200

7 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

(42), and those of Yves Eeckhout on thetail of metamorphosing tadpoles (43).

It has been my good fortune to partici-pate in most of these investigations, some-times actively and sometimes simply in anadvisory capacity, and to watch at thesame time the growing interest of otherlaboratories in similar problems. Aftertrying, with increasing difficulty, to reviewthe field of lysosomes at regular intervals(44, 45), I welcomed with some relief theappearance in 1969, under the editorship ofJohn Dingle and Honor Fell, of the multi-author treatise Lysosomes in Biology andPathology (46), of which volume 4 is nowin press. The literature on peroxisomes andrelated particles has grown more slowly,but has now also reached an appreciablesize (47).

It must be pointed out that many ofthese advances have been made by meansof morphological rather than by biochemi-cal methods, or by a combination of both.In this respect, the development of cy-

tochemical staining reactions for enzymespreviously identified biochemically as spe-cific particle markers has been an in-valuable aid, thanks to the pioneeringwork of Novikoff, Stanley Holt, WernerStraus, Fritz Miller, Sidney Goldfisher,Marilyn Farquhar, and many others.

Applications to Pathologyand Therapeutics

In recent years, we have become increas-ingly concerned with the possible medicalapplications of our findings. The possi-bility that lysosomes might accidentallybecome ruptured under certain conditions,and kill or injure their host cells as a result,was considered shortly after we got ourfirst clues to the existence of these par-ticles. We even made a number of attemptsto test this hypothesis in ischemic tissueand in the livers of animals subjected tohepatotoxic treatments or to carcinogenic

diets (48). But we became discouraged byproblems of interpretation (45). Eventoday, clear-cut demonstration of the so-called "suicide bag" hypothesis remainsvery difficult, although there seem to be atleast a few authenticated cases involvingthis mechanism of cell death. Much moreclearly documented is the mechanism oftissue injury through extracellular releaseof lysosomal enzymes, a field that has beenpioneered by Fell and her co-workers.The two mechanisms mentioned above

rely on the plausible instance of lysosomalenzymes exerting their lytic effect at ab-normal sites. What we did not suspect inthe beginning was that the failure of ly-sosomal enzymes to act at their normalsite could also cause serious diseases. Thisfact was brought home to us in a rathersurprising fashion through the work of mycolleague Gdry Hers, who in 1962 diag-nosed glycogen storage disease type II asbeing due to a severe deficiency of a lysoso-mal enzyme (49). This finding initiated a

0

0

C-)

0

11)

0~

ItO0 1.15 1.20

Equilibrium density.25

Fig. 9 (left). Influence of a previous injection of Triton WR-1339 on the

equilibrium density of rat liver particles equilibrated in an aqueous _sucrose gradient. Upper graph shows overlapping of lysosomes (A. ° 50 0 50 100Pase, acid phosphatase; A. DNase, acid deoxyribonuclease) with mito- % Volumechondria (Cyt. ox., cytochrome oxidase) and peroxisomes (Ur. ox., urate oxidase). Four days after intravenous injection of 170 mg of Triton WR-1339to the animals, the density of the lysosomes has decreased drastically, whereas that of mitochondria and peroxisomes remains unchanged; graph con-

structed from results of Wattiaux et al. (16). [From: (59)] Fig. 10 (right). Influence of cholesterol feeding on density of aortic smooth muscle celllysosomes. Graphs show distribution patterns of enzymes after density equilibration (Fig. 8) in sucrose density gradient depicted by "staircase" on top.Starting material was a postnuclear supernatant of rabbit aortic cells, brought to a density of 1.26 and layered initially at outer edge of gradient (dottedarea). Broken lines give distributions in normal preparations; solid lines, those in preparation from a rabbit showing grade IV atheroma as a result ofcholesterol feeding. Note extensive shift to the left of five acid hydrolases, indicating lowered density of lysosomes due to lipid accumulation. Distribu-tion of protein, 5'-nucleotidase (plasma membranes), and mitochondrial cytochrome oxidase (not shown) was unchanged. [Source: (52)]

u

CL

0)

192 SCIENCE, VOL. 189

on

Feb

ruar

y 5,

200

7 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

series of fruitful investigations on otherstorage diseases, in which Francois VanHoof played a major part (50). It also pro-vided useful guidelines to the chemists andpathologists who, in various parts of theworld, were trying to unravel the patho-genesis of hereditary lipidoses and muco-

polysaccharidoses. Today, with more than20 distinct congenital lysosomal enzyme

deficiencies identified, this mysteriouschapter of pathology has been largelyelucidated (51).

According to some results obtained over

the last few years by Peters, MUller, Tat-suya Takano, Bill Black, and Helen Shio,with the collaboration of Farquhar, lipidaccumulation in arterial cells during thedevelopment of atherosclerosis could wellbe due to a mechanism similar to that in-volved in congenital lipidoses. At least incholesterol-fed rabbits, there is strong evi-dence, both biochemical and morphologi-cal, that the lysosomes of the aorticsmooth muscle cells are the main site of in-tracellular cholesterol ester accumulation,and there are indications that a relativedeficiency of the lysosomal cholesteryl es-

terase may be responsible for this phenom-enon (35, 52, 53). Figure 10 shows some ofthe biochemical evidence: after cholesterolfeeding, lysosomes become considerablyless dense due to lipid accumulation. Thisfigure also illustrates the sensitivity of our

present techniques. These fractionationswere performed on a total of about I milli-gram of cell protein. Similar experimentshave been successfully performed on a

needle biopsy.Other interesting applications of the

lysosome concept are in pharmacology andtherapeutics. In line with the "suicide bag"hypothesis, early investigations in this area

focused on "labilizers" and "stabilizers"of the lysosomal membrane (54). One out-come of this work has been the suggestionthat certain anti-inflammatory agents,such as cortisone and hydrocortisone,might owe at least one part of their phar-macological properties to their effect on

the lysosomal membrane.More recently, we have extended our in-

terest to the various substances that are

taken up selectively into lysosomes andowe some of their main pharmacologicalproperties to this phenomenon. These "ly-sosomotropic" agents are surprisingly nu-

merous, including such variegated com-

pounds as neutral red, chloroquine, strep-tomycin, dextran, polyvinylpyrrolidone,Triton WR-1339, and trypan blue (55).Particularly interesting is the use of certainlysosomotropic agents as carriers fordrugs. In Louvain, Trouet has applied this

principle to leukemia and cancer chem-

otherapy, by using DNA as carrier for the

18 JULY 1975

drugs daunorubicin and adriamycin. Ex-perimentally, these DNA complexesproved less toxic and more effective on

L1210 leukemia than the free drugs (56).Clinical trials under way over the last 2years in several hospitals have given veryencouraging results (57).

Conclusion

In the conclusion of his Nobel lecturedelivered in 1955, Hugo Theorell asked:"What is the final goal of enzyme re-

search?""The first stage," he answered, "is to in-

vestigate the entire steric constitution of allenzymes.... In the second stage," he con-

tinued, "it is a matter of deciding how theenzymes are arranged in the cell-struc-tures. This implies, as a matter of fact, thefilling of the yawning gulf between bio-chemistry and morphology."The gulf still yawns today. But it is a

particular pleasure for me to be able to tellmy old friend Theo that it yawns a littleless. In our efforts to narrow it, my co-

workers and I have been privileged to con-

template many marvelous aspects of thestructural and functional organization ofliving cells. In addition, we have the deepsatisfaction of seeing that our findings donot simply enrich knowledge, but may alsohelp to conquer disease.

References and Notes

1. A. Claude,J. Exp. Med. 84, 51 (1946); ibid., p. 61.2. G. H. Hogeboom, W. C. Schneider, G. E. Palade,

J. Biol. Chem. 172, 619 (1948).3. W. C. Schneider, ibid. 176, 259 (1948).4. T. Svedberg and K. 0. Pedersen, The Ultracentri-

fuge (Clarendon, Oxford, 1940).5. C. de Duve and J. Berthet, Int. Rev. Cytol. 3, 225

(1954).6. F. Appelmans, R. Wattiaux, C. de Duve, Biochem.

J. 59, 438 (1955).7. C. de Duve, B. C. Pressman, R. Gianetto, R. Wat-

tiaux F. Appelmans, ibid. 60, 604 (1955).8. C. de Duve, J. Theor. Biol. 6, 33 (1964); in Enzyme

Cyto ogy, D. B. Roodyn, Ed. (Academic Press,New York, 1967), pp. 1-26; J. Cell Biol. 50, 20D(1971).

9. G. H Hogeboom and E. L. Kuff, J. Biol. Chem.210, 33 (1954).

10. C. d - Duve, J. Berthet, H. Beaufay, Progr.Biop ys. Biophys. Chem. 9, 325 (1959).

11. H. B aufay, La Centrifugation en Gradient deDens te (Ceuterick, Louvain, Belgium, 1966).

12. N. G. Anderson, Ed., Natl. Cancer Inst. Monogr.21(1 966).

13. R. W ttiaux, Mol. Cell. Biochem. 4,21 (1974).14. H. B aufay and J. Berthet, Biochem. Soc. Symp.

23, (1963).15. H. B aufay, P. Jacques, P. Baudhuin, 0. Z. Sel-

linge J. Berthet, C. de Duve, Biochem. J. 92, 184(1964.

16. R. Wattiaux, M. Wibo, P. Baudhuin, in CibaFoundation Symposium on Lysosomes (Churchill,London, 1963), pp. 176-196.

17. F. Leighton, B. Poole, H. Beaufay, P. Baudhuin, J.W. Coffey, S. Fowler, C. de Duve, J. Cell Biol. 37,482 (1968).

18. A. B. Novikoff, H. Beaufay, C. de Duve, J.Biophys. Biochem. Cytol. 2, 179 (1956).

19. P. Baudhuin, H. Beaufay, C. de Duve, J. Cell Biol.26,219 (1965).

20. P. Baudhuin, L 'Analyse Morphologique Quan-titative de Fractions Subcellulaires (Ceuterick,Louvain, Belgium, 1968).

21. B. Poole, T. Higashi, C. de Duve, J. Cell Biol. 45,408 (1970).

22. M. Wibo, A. Amar-Costesec, J. Berthet, H. Beau-fay, ibid. 51, 52 (1971).

23. R. Wattiaux, Etude Expnrimentale de la Surchargedes Lysosomes (Duculot, Gembloux, Belgium,1966).

24. P. Jacques, Epuration Plasmatique de ProteinesEtrangieres, Leur Capture et Leur Destinte dansl'Appareil Vacuolaire du Foie (Librairie Universi-taire, Louvain, Belgium, 1968).

25. R. L. Deter and C. de Duve, J. Cell Biol. 33, 437(1967); R. L. Deter, P. Baudhuin, C. de Duve,ibid. 35, ClI (1967).

26. J. W. Coffey and C. de Duve, J. Biol. Chem. 243,3255 (1968); N. N. Aronson, Jr., and C. de Duve,ibid., p. 4564; S. Fowler and C. de Duve, ibid. 244,471 (1969); Biochim. Biophys. Acta 191, 481(1969).

27. A. Trouet, Caracteristiques et Proprietes An-tigeniques des Lysosomes du Foie (Vander, Lou-vain, Belgium, 1969); P. Tulkens, A. Trouet, F.Van Hoof, Nature (Lond.) 228, 1282 (1970).

28. F. Leighton, B. Poole, P. B. Lazarow, C. de Duve,J. Cell Biol. 41, 521 (1969); B. Poole, F. Leighton,C. de Duve, ibid. p. 536; P. B. Lazarow and C. deDuve, Biochem. Biophys. Res. Commun. 45, 1198(1971); J. Cell. Biol. 59, 491 (1973); ibid., p. 507.

29. D. Thines-Sempoux, A. Amar-Costesec, H. Beau-fay, J. Berthet, J. Cell Biol. 43, 189 (1969); H.Beaufay, A. Amar-Costesec, E. Feytmans, D.Thines-Sempoux, M. Wibo, M. Robbi, J. Berthet,ibid. 61, 188 (1974); A. Amar-Costesec, H. Beau-fay, M. Wibo, D. Thines-Sempoux, E. Feytmans,M. Robbi, J. Berthet, ibid., p. 201; H. Beaufay, A.Amar-Costesec, D. Thines-Sempoux, M. Wibo,M. Robbi, J. Berthet, ibid., p. 213; A. Amar-Cost-esec, M. Wibo, D. Thines-Sempoux, H. Beaufay,J. Berthet, ibid. 62, 717 (1974).

30. P. Baudhuin, M. MUller, B. Poole, C. de Duve,Biochem. Biophys. Res. Commun. 20, 53 (1965).

31. G. Vaes and P. Jacques, Biochem. J. 97, 380(1965); ibid., p. 389; G. Vaes, ibid., p. 393; Exp.Cell Res. 39, 470 (1965); La Resorption Osseuse etI'Hormone Parathyroidienne (Warny, Louvain,Belgium, 1966); J. Cell Biol. 39, 676 (1968).

32. W. E. Bowers, J. T. Finkenstaedt, C. de Duve, J.Cell Biol. 32, 325 (1967); W. E. Bowers and C. deDuve, ibid., p. 339; ibid., p. 349; W. E. Bowers, J.Exp. Med. 136, 1394 (1972); J. Cell Biol. 59, 177(1973); J. Immunol. 113, 1252 (1974).

33. M. Baggiolini, J. G. Hirsch, C. de Duve, J. CellBiol. 40, 529 (1969); ibid. 45, 586 (1970); M. Bag-giolini, C. de Duve, P. Masson, J. F. Heremans, J.Exp. Med. 131, 559 (1970); M. G. Farquhar, D. F.Bainton, M. Baggiolini, C. de Duve, J. Cell Biol.54, 141 (1972).

34. R. L. Schultz and P. J. Jacques, Arch. Biochem.Biophys. 144, 292 (1971).

35. T. J. Peters, M. MUller, C. de Duve, J. Exp. Med.136, 1117 (1972).

36. T. J. Peters, in Peptide Transport in Bacteria andMammalian Gut, A Ciba Foundation Symposium(American Elsevier, New York, 1971), pp. 107-122.

37. P. Tulkens, H. Beaufay, A. Trouet, J. Cell Biol. 63,383 (1974).

38. B. Poole and M. Wibo, paper presented at a sym-posium on intracellular protein catabolism, Frie-drichroda, German Democratic Republic, May1973; M. Wibo and B. Poole, J. Cell Biol. 63, 430(1974).

39. M. Muller, P. Baudhuin, C. de Duve, J. Cell Phys-iol. 68, 165 (1966); M. MUller, Acta Biol. Acad.Sci. Hung. 22, 179 (1971); J. Cell Biol. 52, 478(1972).

40. M. MUller, J. F. Hogg, C. de Duve, J. Biol. Chem.243, 5385 (1968).

41. M. MUller, J. Cell Biol. 57, 453 (1973); D. G. Lind-mark and M. Muller, J. Biol. Chem. 248, 7724(1973); J. Protozool. 21, 374 (1974); J. Biol. Chem.249,4634(1974).

42. D. Scheib-Pfleger and R. Wattiaux, Dev. Biol. 5,205 (1962).

43. Y. Eeckhout, Acad. R. BeIg. Cl. Sci. Collect. Oc-tavo Mem. 38 (No. 4) (1969).

44. C. de Duve, in Subcellular Particles (Ronald, NewYork, 1959), pp. 128-159.

45. C. de Duve, in Ciba Foundation Symposium onLysosomes (Churchill, London, 1963), pp. 1-3 1; inInjury, Inflammation and Immunity (Williams &Wilkins, Baltimore, 1964), pp. 283-311; Fed. Proc.23, 1045 (1964); and R. Wattiaux, Annu.Rev. Physiol. 28,435 (1966).

46. J. T. Dingle and H. B. Fell, Eds., Lysosomes in Bi-ology and Pathology (North-Holland, Amster-dam, 1969), vols. I and 2; J. T. Dingle, Ed., ibid.(1973), vol. 3.

47. C. de Duve and P. Baudhuin, Physiol. Rev. 46, 323(1966); Z. Hruban and M. Rechcigl, Microbodiesand Related Particles (Academic Press, NewYork, 1968); C. de Duve, Proc. R. Soc. Lond. Ser.B. Biol. Sci. 173, 71 (1969); J. F. Hogg, Ann. N.Y.

193

on

Feb

ruar

y 5,

200

7 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

Acad. Sci. 168, 209 (1969); N. E. Tolbert, Annu.Rev. Plant PhysioL 22, 45 (1971).

48. L. Deckers-Passau, J. Maisin, C. de Duve, ActaUnio Int. Contra Cancrum 13, 822 (1957); C. deDuve and H. Beaufay, Biochem. J. 73, 610 (1959),H. Beaufay et al., ibid., p. 617.

49. H. G. Hers, Biochem. J. 86, 11 (1963); N. Lejeune,D. Thinds-Sempoux, H. G. Hers, ibid., p. 16; P.Baudhuin, H. G. Hers, H. Loeb, Lab. invest. 13,1140 (1964).

50. F. Van Hoof, Les Mucopolysaccharidoses en tantque Thesaurismoses Lysosomiales (Vander, Lou-vain, Belgium, 1972).

51. H. G. Hers and F. Van Hoof, Lysosomes and Stor-age Diseases (Academic Press, New York, 1973).

52. T. J. Peters and C. de Duve, Exp. Mol. Pathol. 20,228 (1974).

53. H. Shio, M. G. Farquhar, C. de Duve, Am. J. Pa-thol. 76, 1 (1974); T. Takano, W. J. Black, T. J. Pe-ters, C. de Duve, J. Biol. Chem. 249, 6732 (1974).

54. C. de Duve, R. Wattiaux, M. Wibo, Biochem.Pharmacol. 9,97 (1962).

55. C. de Duve, T. de Barsy, B. Poole, A. Trouet, P.Tulkens, F. Van Hoof, ibid. 23, 2495 (1974).

56. A. Trouet, D. Deprez-De Campeneere, C. deDuve, Nat. New Biol. 239, 110 (1972); A. Trouet,

D. Deprez-De Campencere, M. De Smcdt-Malen-greaux, G. Atassi, Eur. J. Cancer 10, 405 (1974).

57. G. Sokal, A. Trouct, J. L. Michaux, G. Comu, Eur.J. Cancer 9, 391 (1973); G. Cornu, J. L. Michaux,G. Sokal, A. Trouct, ibid. 10, 695 (1974); J. Long-ueville and H. Maisin. in Adriamycin Review,Proceedings, Second international Symposium,Brussels (European Press, Ghent, Belgium, 1975),p. 260.

58. J. Berthet and C. de Duve, Biochem. J. 50, 174(1951); J. Berthet, L. Berthet, F. Appelmans, C. deDuve, ibid., p. 182.

59. C. de Duve, Harvey Lect. 59,49 (1965).

Causality and Anticipation

Analysis of the concept of anticipation can

contribute to the philosophy of biology.

J. M. Burgers

The purpose of this article is to renewdiscussion of the problem whether the phe-nomena of life can be satisfactorily ana-lyzed and explained on the basis of thelaws discovered in the physical sciences, orwhether more is needed. When mentioningthe physical sciences, I have in mind thephysical laws as they are formulated atpresent, with the trend of thinking thatforms their present background. Otherwisethe problem would become indefinite. Iwish to consider the thesis that the featuresof life involve relations not covered by thepresent formulation of the physical laws,relations which, although not amenable toquantitative analysis, nevertheless play adecisive part in many reactions of livingorganisms. The problem is, on one hand,how to put this in appropriate terms, andon the other, to analyze some con-sequences of the thesis. It is useful to startwith a brief recapitulation of what may becalled the central doctrine of the laws ofphysics, namely the idea of causal relation-ship. This will be given in the next section.The principal argument concerning theneed for extension to another form of rela-tionship is presented in the third section ofthe article. It is taken from features of ourhuman mental life (1).

The author is emeritus research professor at the In-stitute for Fluid Dynamics and Applied Mathematics,University of Maryland, College Park, Maryland20742.

194

Causal Relationship

Our ideas concerning causal relationshipare a central feature of the physical laws. Itis not necessary to present an extensive de-scription of these ideas and a summarystatement will suffice.We are convinced that all natural phe-

nomena occurring in systems where thereis no indication of life are related to thepast of the systems in such a way thatknowledge of any situation gives us infor-mation on which we can base more or lessadequate predictions concerning sub-sequent situations."Knowledge of a situation" means the

complex of data that we can obtain bymaking observations and measurementsaccording to a scheme accepted and elabo-rated in the physical and related sciences.In making these observations we eliminateour personal involvement. We do not in-troduce such qualifications as "beautiful"or "ugly," "good" or "bad." Neither dowe make reference to any purpose or in-tention; there is no reference to the futureas a determining agent. In many cases we

apply dissection of complicated phenom-ena into more simple events, in the con-

viction that the separate effects can be con-

sidered as self-contained. Recombinationof features follows at a later stage, and al-though reciprocal influences are taken intoaccount, it is assumed that these influences

will not refer to a particular intention, in-volved in a "whole" and in some way tran-scending the partial effects.

It is in this way that the "past"-that is,those aspects of past phenomena which areamenable to measurement-has come toappear as the "cause" of the present. It is aconcise statement summarizing the accu-mulated experience obtained by observingthe behavior of nonliving bodies and sys-tems, collected since the beginning of mod-ern science.

In this century it has become evidentthat in the atomic and electronic domainthe measurable data are not sufficient for acompletely definite prediction of succeed-ing states; there is dispersion in the devel-opment and statistical predictions are themost that can be made. Thus, causal rela-tionship is partly deterministic, partly sta-tistical. In the investigation of these rela-tions no evidence has been found for the ef-fectiveness of "finalistic causes," that is,causes directed to the future. All predic-tions that can be made are still statementswhich relate the present to the past. It hasalso become clear that in the atomic andelectronic domain every observation ormeasurement disturbs the system underobservation.

There is no need to dwell on the powerthe scientific method of observation hasgiven to mankind. However, it should notbe overlooked that this enormous successhas also depended on the type of problemsstudied. Many questions occupied the hu-man mind during the Middle Ages, for ex-ample, the problem whether man's destinywas to adjust himself to a cosmic order,embracing both the moral and the materialworld. The new science of Leonardo, Gali-leo, and their followers was directed awayfrom such problems and substituted a newset of questions. Pushed in this new direc-tion, Western thought came more andmore to rely on the assumption that every-thing in the Universe is determined bywhat has occurred in its past. It is here thata warning note is in order. There is no jus-tification for enforcing this concept ofcausality on the entire Universe as the onlypossible form of relationship. In particu-lar, while many phenomena exhibited byliving beings can be foreseen on the basis

SCIENCE, VOL. 189

on

Feb

ruar

y 5,

200

7 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from