Nature © Macmillan Publishers Ltd 1998

8

Such ligands might cluster L-selectin bymimicking either the mucins themselves orthe multivalent display of mucin epitopes atthe cell surface.

The L-selectin recognition element wasbased on the disulphated trisaccharide 38,6-disulpho Lewis x (Lex), an analogue of theGlyCAM-1 capping group 6-sulpho sialylLex4 (Fig. 1a). To assemble the recognitionelements into a multivalent saccharidearray6, we used ruthenium carbene-catalyzed ring-opening metathesis polymer-ization (ROMP)7, a technique that toleratesmultiple unprotected functionalities andcan produce polymers of consistent andcontrollable lengths. We produced syntheticligands composed of about 15 monomerunits, each substituted with the recognitionepitope 38,6-disulpho Lex (Fig. 1a)8.

scientific correspondence

30 NATURE | VOL 392 | 5 MARCH 1998

When we treated human neutrophilswith these neoglycopolymers, L-selectinwas lost from the cell surface in a dose-dependent manner (Fig. 1b). Soluble L-selectin was detected in the supernatant oftreated cells; therefore the decrease on thecell surface was due to shedding.

In contrast, the monovalent ligand 38,6-disulpho Lex(Glc)-b-OPr did not cause L-selectin release (Fig. 1c). And agalactose-substituted oligomer, which doesnot bind L-selectin, had no effect.

Cellular activators, such as phorbolesters and chemotactic peptides, also inducethe shedding of L-selectin; at the same timethey dramatically increase the surface con-centration of the b2-integrin Mac-1(CD11b/CD18) (Fig. 1d), a proteininvolved in the next step of leukocyte

FFiigguurree 11 Effects of neoglycopolymer on human neutrophils. aa,, Chemical structures of an L-selectin bindingepitope on GlyCAM-1 (Ac = acetyl) and the 3',6-disulpho Lex neoglycopolymer. bb,, Histogram overlay of L-selectin expression by untreated cells (grey) and cells treated with the 3',6-disulpho Lex neoglycopolymer(267 mM or 4 mM on a per saccharide basis; black). cc,, Histogram overlay of L-selectin expression by un-treated cells (grey) and cells treated with the monomeric ligand 3',6-disulpho Lex(Glc)-b-OPr (4 mM) (black).dd,, Histogram overlay of Mac-1 expression by untreated cells (grey) and cells treated with 10 nM PMA (black).ee,, Histogram overlay of Mac-1 expression by untreated cells (grey) and cells treated with 3',6-disulpho Lex

neoglycopolymer (267 mM or 4 mM on a per saccharide basis; black). Histogram depictions are representa-tive of at least three experiments monitored by flow cytometry using a fluorescein isothiocyanate-conjugatedanti-L-selectin antibody or a phycoerythrin-conjugated anti-Mac-1 antibody.

d

10

20

30

Fluorescence intensity

10 0 10 310 1 10 2 10 4

Rel

ativ

e ce

ll nu

mbe

r

0

e

Fluorescence intensity10 0 10 310 1 10 2 10 4

Rel

ativ

e ce

ll nu

mbe

r

0

10

20

30

b

10 0 10 310 1 10 2

Fluorescence intensity

Rel

ativ

e ce

ll nu

mbe

r0

10

20

30

10 0 10 310 1 10 2 10 4

Fluorescence intensity

c

Rel

ativ

e ce

ll nu

mbe

r

0

10

20

30

10 4

OO

OHO

OOSO3

—

OO

O

OHHO

OH

OHHO

—O3SOOH

Ph

HN

O

OHOH

HO

OO

NHAc

O

OHOH

OH

OSO3—

OO

CO2—

OH

OH

HHO

HAcHN

OH

O

3’,6-disulfo Lex neoglycopolymernave = 15

GlyCAM-1 epitope

a

OR

n

Protein shedding, the proteolytic release ofa cell surface protein, can serve a regulatoryrole by liberating soluble molecules into cir-culation while decreasing their concentra-tion on the cell surface1. We have created anew class of multivalent ligands, ‘neogly-copolymers’, which are designed to promotethe proteolytic cleavage of a cell adhesionmolecule involved in the inflammatoryresponse, L-selectin2. These synthetic lig-ands induce the release of the extracellularportion of L-selectin by appropriating anendogenous protease; such activities suggestnew strategies to generate anti-inflamma-tory agents and regulate the cell surface.

L-selectin mediates the process of leuko-cyte rolling3, an initial step in the inflamma-tory response, by recognizing carbohydrateepitopes that are present on endothelialcells. L-selectin is constitutively present onthe surface of most leukocytes, but a solubleform is also found in circulation.

The L-selectin ligands that occur physio-logically, such as GlyCAM-1 (ref. 4), aremucin-like proteins which contain clustersof O-linked saccharide chains. Such chainscould engage L-selectin in multivalent bind-ing at the cell surface. We hypothesized thatthe clustering of L-selectin, as a consequenceof ligand binding, leads to the proteolyticrelease of its soluble form from the cell5.

So we synthesized molecules that, likemucins, present multiple copies of saccha-ride epitopes on an extended backbone.

Synthetic ligands pointto cell surface strategies

Alternative models are worth consideringif they suggest new experiments3; this sug-gests several. First, it should be possible toreplace As by P (and perhaps even N) with-out a diminished anti-corrosive effect inbrass. Second, other non-metal pairs as wellas B/pnictide, such as C (or Si) with Si (orO), might behave similarly. Third, the solu-bility of B and As (or P) in brass may dependon each other’s concentration and on thedivacancy concentration. Novel quaternaryintermetallic compounds of B and As mighteven be found. Finally, the trick should notwork in alloys of metals below group 10which corrode with a divacancy mechanismif formation of metal boride or arsenidephases competes.P. J. van der PutLaboratory of Inorganic Chemistry, Delft Universityof Technology, 2628 BL Delft, The Netherlandse-mail: p.j.j.m.vanderput@stm.tudelft.nl

1. Cahn, R. W. Nature 389, 121 (1997).

2. Linnett, J. W. The Electronic Structure of Molecules. A New

Approach (Methuen, London, 1964).

3. Oreskes, N., Shrader-Frechette, K. & Belitz, K. Science 263, 641

(1994).

Nature © Macmillan Publishers Ltd 1998

8

3. Springer, T. A. Cell 76, 301–314 (1994).

4. Hemmerich, S., Leffler, H. & Rosen, S. D. J. Biol. Chem. 270,

12035–12047 (1995).

5. Walcheck, B. et al. Nature 380, 720–723 (1996).

6. Lee, Y. C. & Lee, R. T. Acc. Chem. Res. 28, 321–327 (1995).

7. Lynn, D. M., Kanaoka, S. & Grubbs, R. H. J. Am. Chem. Soc.

118, 784–790 (1996).

8. Sanders, W. J., Katsumoto, T. R., Bertozzi, C. R., Rosen, S. D. &

Kiessling, L. L. Biochemistry 35, 14862–14867 (1996).

9. Kishimoto, T. K., Jutila, M. A., Berg, E. L. & Butcher, E. C.

Science 245, 1238–1241 (1989).

10.Feehan, C. et al. J. Biol. Chem. 271, 7019–7024 (1996).

(Table 1). Dose distributions in settlementswith fewer than 12 thyroid measurementswere estimated from settlements contribut-ing many measurements by taking intoaccount contamination by caesium-137(137Cs) of the ground, and the settlement’slocation relative to Chernobyl3.

For children less than three years old liv-ing in contaminated areas, the dose receivedexceeded that to adults by a factor of five,while among evacuees the child dose wasabout 20 times greater than that received byadults. Individual doses show large variabil-ity, but this contributes only slightly to theuncertainty of average thyroid dose esti-mates because of the size of this study. Weestimate that a factor of 4 separates the 95%confidence boundaries of average thyroiddoses.

For Belarus, deriving 131I activity con-centrations in thyroids is more difficultbecause here the measurements of 131I activ-ities in the thyroid were performed withnon-spectrometric and uncollimateddevices. However, 131I activity measure-ments in milk and soil are available5,6. Thy-roid doses were assessed for 812settlements. In Bryansk, the most contami-nated region of Russia, 14,000 measure-ments of 131I activities in thyroids wereperformed. Count rates detected with non-spectrometric devices were corrected for thecontribution of radio-caesium in the body.Thyroid doses in 602 settlements were cal-culated by taking into account data on milkand soil contamination7.

Of 264 checked cases from Belarus andUkraine, western European pathologistsapproved the thyroid cancer classification in97% (ref. 8). In southern Ukraine, which weused as a control area, the thyroid cancerincidence among 9 to 18-year-olds increasedby 70% from the period 1986–1988 to theperiod 1991–1995. This probably reflectsboth greater thyroid surveillance and Cher-nobyl radiation. In principle there might bea higher screening effect in the contaminat-ed regions. However, the high frequency ofmetastases (these were found in lymphnodes in 66% of the children operated on in

scientific correspondence

NATURE | VOL 392 | 5 MARCH 1998 31

recruitment to the endothelium9. To determine whether a similar mecha-

nism was associated with neoglycopolymer-promoted L-selectin shedding, wemonitored Mac-1 levels in neoglycopoly-mer-treated cells. Surprisingly, the neo-glycopolymer had no effect on Mac-1expression (Fig. 1e). Moreover, hydroxamicacid-based protease inhibitors, whichdiminish the activation-induced sheddingof L-selectin10, do not prevent neoglyco-polymer-induced shedding (data notshown).

The disparity in protease inhibitor activ-ities raises the possibility that different pro-teases mediate the activation-dependentand independent events. These results indi-cate that molecules sharing aspects of physi-ological L-selectin ligands can induceL-selectin shedding, and that the releasemechanism is distinct from that occurringfollowing cellular stimulation.

Several unique features characterizeneoglycopolymer-promoted shedding of L-selectin. The neoglycopolymers, like otherselectin inhibitors, compete with naturalligands for selectin binding by non-covalentassociation. These agents can also facilitatea change in covalent bonding, thereby irre-versibly altering the cell surface. Specifically,molecules that cause L-selectin release willdiminish leukocyte rolling and thetransendothelial migration that can follow.Concomitantly, they increase the concen-tration of soluble L-selectin that can act asan inhibitor of the rolling process. More-over, when receptors are shed from the cellsurface, the affinity of the soluble receptorsfor the multivalent ligand should decrease,allowing the multivalent ligand to functionat substoichiometric levels.

Proteins that are shed from the cell sur-face form a diverse group. They includegrowth factors, cytokine receptors, celladhesion molecules and leukocyteantigens1. This diversity underscores thepotential applications of receptor sheddingin altering cellular responsiveness to specificligands, or promoting responses at distalsites. We have demonstrated the possibilityof devising molecules that selectively pro-mote the shedding of L-selectin, suggestinga new strategy for generating anti-inflam-matory agents. Because other proteins mayundergo ligand-induced shedding, theseresults have broad implications for theinvestigation and manipulation of the cellsurface and the extracellular environment.Eva J. Gordon, William J. Sanders, Laura L. KiesslingDepartments of Chemistry and Biochemistry,University of Wisconsin-Madison, Madison, Wisconsin 53706, USAe-mail: kiessling@chem.wisc.edu

1. Hooper, N. M., Karran, E. H. & Turner, A. J. Biochem J. 321,

265–279 (1997).

2. Kansas, G. S. Blood 88, 3259–3287 (1996).

The Chernobyl reactor accident was fol-lowed by a sharp increase in the incidenceof thyroid cancer among children and ado-lescents in Belarus (Belorussia) andUkraine1,2. Exposure to iodine-131 (131I)was responsible for most of the doses thataffected the thyroids of these children; how-ever, among evacuees, up to 40% of eachdose could derive from other incorporatedradionuclides and external exposures3.From the data set compiled after this inci-dent, we estimated the increased risk ofdeveloping thyroid cancer after exposure toradioactive iodine. The figure we obtainedfor most of the affected regions fell withinthe 95% confidence interval of a previousfollow-up of thyroid cancer after externalexposures.

In Ukraine, more than 150,000 measure-ments of 131I activity in the thyroid wereperformed from mid-May to mid-June1986. Collimators shielded against radia-tion from other parts of the body and fromthe environment during the taking of thesemeasurements. Background signals werecontinuously monitored and subtractedfrom the results.

In a re-evaluation of these measure-ments4, we reconstructed average thyroiddoses due to 131I exposure for three regionsclose to Chernobyl — Kiev, Zhytomyr andChernigov, containing 4,406 settlements(villages or towns) — and for evacuees

Thyroid cancer risk tochildren calculated

Table 1 Thyroid cancer risk among children born between 1971 and 1986

Area Children Average thyroid Thyroid cancer Observed/ Excess absolute risk(2 103) dose (Gy) cases 1991–1995 expecteda cases (per 104 person-year Gy)

UkraineZhytomyr oblastb 340 0.13 28 4 0.9Kiev oblast 399 0.18 47 6 1.1Chernigov oblast 273 0.09 33 6 2.2Kiev city 581 0.05 67 6 3.830 km zone (evacuees) 20 0.92 12 30 1.3

BelarusGomel/Mogilev oblastsc 76 0.73 89 56 3.2Minsk city 357 0.08 41 6 2.3Gomel city 113 0.40 72 30 3.1

RussiaBryanskc 169 0.12 31 9 2.7

a: Expected cases are calculated on the basis of the incidence in southern Ukraine of 4.2 cases per 106 person-yearsb: An oblast is an administrative regionc: Subarea of the region only