Reodnted from Journal of Mediciaal Chemistry' 1995' 38'Cop''righr @ 1995 by rh" A;;;;; til;i.J3*i"ii-..a-np.i"t"a by iermidion of the cop5night owner. 4179

Effective Inhibitors of l{emagglutination by Influenza virus synthesized ftom

Polymers Having Active Ester Groups. Insight into Mechanism of Inhibition

Mathai Ma:nmen, Georg Dahnann, and George M. Whitesides*

Department of Cfumistry, Harvard lJnioersity, 12 Orford. Street, Cambridge, Mossothusetts 02138

Receiued June 23, 19956

Highly effective sialic acid-containing inhibitors of influettzavirus X-31 were synthesized using

p.f"vflf-t"cryoyloxy)succinimiael (pNfu), a polyper preactivatedby incorporation of active ester

gt;;;r.' poiymers containing two and threL different components were prepared by sequential

i"actioo of pNAS with twJ and three amines, respectively. This preparation of co- and

terpolymers was synthetically more efficient than rnethods involvin_g _copolymerization of

different monomers and gave polymers that were more easily compared thal tlosgsgnerat-edby copolymerization. Poiymeir i" this study (pre_pat"-a tory a_sin-gle b-atct-r of pNAS) had a

.onrturf degree of polymerization (DP * 2000) indprobably h-ad a distribution of comp_onents

that was more random than analogous polymets prepared by copolym-erization. Use ofg-jtycosides of sialic acid made it poisiblsto investigate inhibition by diffeplt polymers at

temperatures ranging from 4 to'gg "C without artifacts due to the _hydrolytic actioll-of

neuiaminidase. The inhibitors were, in general, more effective at 36 oC than at 4 "C. The

hemagglutination (HAI) assay was used to measure the value of the inhibition constant d{AIeach polymer. The value of^(H for the two-component polymer containi,ng2|% sialic acid on

" p"fv.."ylamide backbone ui 4 'C was 4 nM (in terms of the sialic acid moieties present in

solution) and was approximately 50-fold more effective than the best inhibitors previously

described and 2b-fold^ *or. effective than the best naturally occurring inhibitor. The most

effective inhibifor sy"ttresized in this work containe d LTVo benzyl amine and 20Vo sialic acid

on a polyacrylamide backbone, and its value of {il was 600 pM at 36 "C. Approximately 100

pofytir*r tft"t differed in one or two componenis were assayed to distinguish between two

ii*itittg mechanisms for inhibition of the interaction between the surfaces of virus and

L"ytftt*ytes' high-;ffinity binding through polyvalency, and steric stabilization- The results

suilest itr"t bolh mechanisms pLv rn impottant role. The system comprising polyvalent

infriUitors of agglutination of earttrrocytes by influenza provides a-system that may b-e useful

as a model for-Inhibitors of other patLogen-host interictions, a large number of which are

themselves polyvalent.

(2-3 copies/100 nm2; or -600 copies per virus particle)

on the surface of influenza virus.l2 SA is a nine-carbon

sugar that occurs with high density (22-25 molecules/

10b nm2) on the surface of the mammalian cell.15-17 SA

terminates many of the extracellular saccharidic por-

tions of mammalian glycoproteins and glycolipids'

We assayed our inhibitors using the hemagglutination

inhibition (HAI) assay.l8 Interaction between virus and

erythrocytes in solution leads to the formation of an

extended, cross-linked gel (hemagglutination)' Trhe IIAI

assay enabled us to measure the ability of an inhibitor

to prevent hemagglutination by preventing the interac-

tion between the surfaces of virus and erythrocyte; we

quantitated the effectiveness of the inhibitor with an

inhibition constant, Iti*. We defin. ffN as the low-

est concentration of ligand that successfully inhibits

hemagglutination. All values of XfN reported here

are in terms of the concentration of SA groups in

solution (whether free in solution as monomeric units,

attached covalently to the backbone of a soluble poly-

mer, or associated noncovalently with a structure such

as a liposome).

We used SA-containing polymeric inhibitors (glyco-

polymers)ls that are based on polyacrylamide (pA) or

poly(acrylic acid) (pAA). We have described this strat-

Introduction

We describe a series of polymeric inhibitors designedto inhibit the attachment of influenza vin s X-31"1-4 tomammalian cells. The inhibitory potency of the best

inhibitor described here (/f,N : 0.6 nM; calculated onthe basis of sialic acid groups present in solution) isabout 2 orders of magnitude better than that of othersynthetic inhibitors (of which the best are liposomespresenting sialic acid on their exterior surface, I{N :

200 nM)s or natural inhibitors (of which the best is

equine crz-rn&croglobulin, Itj* : 100 nM)6 (Figure 1)-

Nature uses polyvalency-the cooperative associationof a receptor (or aggregate of receptors) containingmultiple recognition sites $rith a molecule containingmultiple compiementary ligands-in a number of bio-logical contexts, including many interactions betweenpathogen andhost.?'8 The attachment of influenzavimsto its target cell is a model system for such pollrualentinteractions; this attachment is mediated by multiplesimultaneous interactions between viral hemagglu-tinine-11 (IIA) and cellular sialic acid2'12'13 (Neu5Ac,la

SA). HA is a glycoprotein that occurs with high density

* Author to whom correspondence should be addressed.@ Abstract published inAduance ACS Abstratts, September 15, 1995.

0022-2623/95/1838-41?9$09.00/0 @ 1995 American Chemical Societv

4f80 Journal of Medicinal Chemistry, 7995, Vol. 38, No. 21

ery previously,5'20-22 and it has also been used explicitlyby others including Matrosovi ch,23,z+ Rny,2,-zt Paulson,28Bednarski,2e and Lee.30 The monomeric a-2-methyl-1tr-NeuAc (a-MeSA)l4 inhibits virally induced hemagglu-tination at 2 mM 0tT* - 2 mM). The value of ̂ 4N forthe most effective polymer reported in the present workis 600 pM. This highly effrcient inhibition of hemag-glutination reflects the ability of these polymers toprevent interaction of two biological surfaces: those ofthe virus and the erythrocyte. We believe that thismechanism of action reflects a pharmacologically rel-evant property,3l but we have not assayed these poly-mers for activity in uiuo.

Polymeric versus Monomeric tnhibitors. A sig-nificant body of work has developed around the syn-thetic modification of monomeric SA in order to improveits ability to inhibit virally induced hemaggluti-ttu1ion.20'32-34 The most successful of these attemptsgave derivatives of SA w"ith I4 equal to 3.7 pM,32 3orders of magnitude better than cr-MeSA.

Both natural and synthetic polymeric inhibitors (in-hibitors containing multiple copies of SA) are much moresuccessful at inhibiting hemagglutination than are theirmonomeric counterparts. A number of polymericsystems have yielded values of .fN less than 1ptM.20,22.2s'37 The mechanism by which a polymer actsas an inhibitor of hemaggiutination, and the reasonsthat it is more successful than its monomeric counter-part, is understood in broad outline,2l but not in detail.The monomeric inhibitor must compete with SA groupson the surface of the erythrocytes for SA-binding sitesof IIA on the surface of the virus; it is at an intrinsicdisadvantage in this competition since the interactionbetween virus and erythrocyte is polyvalent.

The polymer can inhibit hemagglutination by at leasttwo conceptually independent mechanisms (Figure 2).21First, it may act by high-affinity interactions based onthe possible entropic advantage of many SA groups onthe same polymer binding to the surface of the virus;these potentially cooperative interactions may lead toefficient competitive inhibition of viral attachment bycompeting with the polyvalent cooperative interactionsbetween virus and cell. Second, the polymers may actby steric stabiiization:3s SA groups on the same polymermay bind noncooperatively (with similar affinity to thatof monomeric SA) to the surface of the virus, but bringwith them a large, water-swollen polymer that maymake sites on the surface of the virus adjacent to thosebound to SAphysically inaccessible to SA on the surfaceof the cell. This second type of inhibition is a type ofmixed inhibition (a combination of competitive andnoncompetitive inhibition). The relative importanceand detailed mechanisms of these two types of inhibitionare currently not understood in sufiicient detail todesign new polymers that are more effective inhibitorsthan those already known.

We had four goals in this work: (i) to develop asynthesis for polymeric inhibitors based on pA and pAAthat is of more use than that used previouslV (copolym-erization of a mixture of acrylamide monomers); (ii) toexplore the role of temperature in inhibition of hemag-glutination by SA-containing polymers; (iii) to under-stand the relative importance of the two major mech-anisms of inhibition for these polymers through a guidedchoice of different side chains on the pA backbone; and

charged, e tc . )

o Irradiation (medium-pressure Hg arc lamp through a c.uar:zvessel for 6 h) of a mixture of substituted acrylamides and a rairca.initiator (AIBN) yields a substituted poiy'acrylamide -

Hea:ir: asolution ofN-(acryloyloxy)succinimide (NAS.r and AIB\ ir: be::zer=to 60 "C yields the preactivated "parent" pollrner p\AS surs.-quent sequential treatment with different, amj.nes 1.e.,:-- d:.N-substituted polyacryiamide.

(iv) 1e frnd inhibitors of hemagglutination that are moreeffective than those already known.

Using Activated Polymers to Construct Copoll'-mers. Various physical properties of the pollrcer(conformation in solution, conformation on the surfaceof the virus, conformational flexibility, net charse. size.degree of branching, degree of hydration, and h1'circ'phobicity) might, in principle, influence the success ofinhibition by polymers. We have previously attemptedto correlate some physical characteristics of thepolymer-especially the structures of the side charn-. ofthe polymer and of the group linking the SA to thebackbone-with their effectiveness in inhibiting hemag-glutination.22 This previous work was based on acopolymer of two or more differently N-substitutedacrylamides: one was l/-substituted with a spacer-group terminated with SA, 2; the other was.l/-substi-tuted w.ith the test-group, R2 (Scheme 1A). We defrne

S:.iNHso

lf,'ut the ratio of side chains containing Ri to thetotal number of side chains in the polymer (eq 1); forexample,.XsA is the proportion of side chains with linkersterminated in SA. The "standard polymer" is one inwhich /SA : 0.20. All comparisons between polymersare ultimateiy referenced to the standard polymer.

Scheme 1. Two strategies:(B) Preactivationb

A. COPOLYMERIZATION

B. PREACTIVATION

hv

A I B N

O N H S

I n1 ruH,

oX*r*,1\-i*r"rl"#\rl rux, J

Mammen et al.

(A) Copolymerizationo and

r ; n1rux,2) R2NH2

3) NH3

p N A S

o ^ o t t H s

oaVon\

t(o

o\

O NY

R ' N H 2 - o t h e r o r g a n i c A r o u p(hydrophobrc , bu lky ,

ot l

/vH

Effectiue Inhibitors of Hemagglutinatr'on Journal of Medicinal Chemistry, 7995, Vol. 38, No- 21 4181

-r-d> Increasing Etfectiveness

pA pA(10%SA)-C-glycoside pA(20%SA)-C-glycoside

o

0)

c

a

6=(uz

I a-rure-siarosioe I n1atr0%sA)-c-slvcoside I pA(eo%sA' 4'%benzvl)

I r divarentl -lo1,ro*sA)-o-srvcosidel

I r^'fr?:'iH;:lv,'.11i?'jf)"'

| i lonou"rent Inhibitorr t;bil;;l

I I Lioosomes(SA)

I I I

wmada vz r rr

a m.irin. o2rnl

guinea p ig o2M

Figure l' Values of Klul for natural and synthetic inhibitors of hemagglutilation based on SA.-Monomeric inhibitors (opeD

boies), previously studii)d polymeric inhibito." (h.t"hed bote"), and the blst polymeric inhibitors ftom the present study (solid

lor".j'i"" slown. Syntnetic polymeric inhibitors are based on liposomes, polyacrylic acid) (pAA) or_polyacrylauride (pA). Thewidth of the boxes ripresent ihe uncertainty in the values (generally + a factor of 2; that is, + I well iD serial dilution).

(Copolymerization)

characterizing different acrylamide monomers (most ofthe amines used were commercially available).

Exploring the Role of Temperature Using C-

Glycosides. Changes in the temperature at which the

HAI assay is performed might, in principle, change the

measured value of lffN. Temperature influences sev-eral variables of the HAI assay: the preferred distribu-tion of conformations of the polymer in solution;the rateof adsorption of the polymer to the surface of the virus;the rate of structure formation and the preferred

distribution of conformations of the polymer on the

surface of the virus; and the ability of a layer of polymer

on the surface of a particle to stabilize that particle

sterically. It has been difficult to examine the influenceof temperature on hemagglutination using substratescontaining O-glycosides of SA, because neuramini-dase38-40 (NA) cleaves these groups rapidly. Assays ofinhibitors that contain SA connected to the backboneof the polymer by O-glycosidic linkages could only beperformed at 4 'C (degradation of the inhibitor occurredacceptably slowly at 4'C). The physiologically morerelevant temperature is approximately 36-37 'C- Inthis study we used C-glycosides of SA (1) because theyare resistant to the action of NA. We collected paralleldata at 4,19, and 36 oC for each polymer.

The Hypotheses Guiding the Suryey of SideChains. There are at least two independent mecha-nisms that might contribute to the inhibitory action ofthese polymers: high affinity polyvalent binding andsteric stabilization (Figure 2). We synthesized polymersincorporating different side chains from different chemi-cal classes (charged, polar-uncharged, and nonpolar), totry to distinguish between these two mechanisms ofinhibition. No single structural modification allowed usto identify clearly which of the two mechanisms is moreimportant for inhibition of hemagglutination by poly-

mers. We expected that the results in aggregafe would,however, be informative. A guided survey of side chainsallowed us both to explore mechanisms of inhibition bypolymers and to determine the side chain (R2NHz,

Scheme 1) on a pA backbone that promoted the stron-gest inhibition.

We had five hypotheses: (i) Incorporating a numberof negatively charged groups along the backbone might

1O-3 1O-4 .10-s 1 0-6 rc-7 1O-B 10-9 10-10,'[rrHn fuurrrr iirlrrrr lrr*rrl llr+rlrt lrrrnrrr lrrlrrrr lxfAr tnl

1O-3 10-4 1O-5 10-6 10-7 10-B 10-9 1 0-1 0

T ma aa.iJtvttu.,or"o",n. lu. I | -l r.;inl

I bovine czM

fetuin human o,2M

_ tRNH: l

lR'NHzl + IR'NHd+ NH,

lPreactivation)

Copolymerization (Scheme 1A) possibly introducestwo uncontrolled variables into the experiment (Figure

3): (i) unknown differences in rate constants for copo-lymerization among differently N-substituted acryla-mides might result in differences in the distribution ofSA along the backbone of the polymer; (ii) the length,polydispersity, and tacticity of the polymer mightchange as the side chain was altered, due to differencesin rates and stereochemistry of polymerization.

In the present study, we used a different strategy toprepare polyrneric inhibitors based on addition of aminesto polymers having activated esters of carborylic acidgroups as side chains (Scheme 1B); this strategy hasbeen used previously,36'37 but with different motivations.We refer to the synthesis of different polymers by thisstratery as "preactivation". Polymerizing N-(acryloyl-oxy)succinimide yielded the preactivated polyrner poly-

lN-(acryloyloxy)succinimidel (pNAS). We then con-verted the activated esters on pNAS to amides bysequential reaction with two amines: frrst with 1, amolecule containing a SA group connected to a flexiblespacer (1 : R1NH2, Scheme 1); then with R2NH2, wherewe varied R2 depending on the hypothesis being tested.Preactivation was synthetically and mechanisticallymore attractive than copolymerization for several rea-sons: (i) the characteristics of the polymer backbone-length, polydispersity, tacticity-were kept constant (the

same batch of pNAS was used for all experiments); (ii)

the distribution of SA groups along the backbone,although not controlled, was the same for all pollnners,since introduction of the SA moiety was the first stepin modifoing pNAS; (iii) the distribution of SA groupsalong the backbone was probably more uniform thanfrom copolymerization (addition of the sterically bulkymolecule I to adjacent sites along the backbone of thepolymer was less likely than addition to nonadjacentsites); (iv) preactivation was more convenient syntheti-cally than copolymerization, since constructing manydifferent polymers did not require synthesizing and

oJ*R'

4182 Journal of Medicinal Chemistry, 7995, Vol. 38, No. 21

A. ATTACHMENT

f i( ui.r. $\9

SAwf f icen+

OV, Virus-CellOa, Association

olAHxz

€>nxn-.zo a

\/"y'HzN Propagation

Marnmen et aI.

i i ) Shor t po lymerbackbone

B. INIHIBITION OF ATTACHMENT

Monovarent a\ W'.:r[:r*:GW+G &{/\-Termination

POSSIBLE CONSEOUENCES

6ff\ id,> "\"rbr,r"lhg

OESIRED

Uniform distr ibution on along polymer backbone

A -

U N O E S I R E D

ffdfl",:,,,:'F G&Potyvarent f t

S \

;f*i[:'G&4Figure 2. (A) Influenza virus attaches to its host cell throughpolywalent interaction of HA with SA. (B) Inhibition: mono-meric molecules containing SA inhibit attachment by competi-tive inhibition, whereas polymeric molecules inhibit attach-ment by either (i) high-afftnity binding through polyvalency(competitive inhibition); andor (ii) steric stabilization of asurface (mixed inhibition).

change the conformation of the poiymers by favoringlong persistance lengths (thus favoring an extended,rigid-rod geometry).3s We expected extended polymersto be less effective than random-coil polymers in stericstabilization because these polymers have less confor-mational entropy. (ii) Placing bulky substituents on ornear the backbone might also lead to longer persistancelengths for the polymer and decrease steric stabiliza-tion.35 In addition, bulky substituents on sites adjacentto SA might 'crowd" the sialic acid and thus hinderaccess to the IIA on the viral surface. (iii) Adsorbing anegatively charged polymer onto the surface of a virusmight result in increased stabilization by electrostaticrepulsion of the negatively charged surface of theerythrocyte. Conversely, adsorbing a positively chargedinhibitor on the surface of the virus might promoteeiectrostatic attraction between the polymer-coatedvirus and the erythrocyte. (iv) Adding hydrophobic sidechains might increase the aftinity of the polymer for thevirus by inducing secondary hydrophobic binding inter-actions. (v) Using cross-linking agents in small propor-tions might increase steric stabilization. These agentsmight lead to branching of the polymer, or gel formation,and thus increase the effective degree of polymerization,DP. We have recently found that increasing DP gener-ally increases the effectiveness of the polymers.4l

Results and Discussion

Synthesis of Polymers. We prepared all polymersused in this study from a single batch of pNAS (Scheme1B). We characterized the molecular weight of this"parent" polymer by complete hydrolysis to pAAfollowedby gel filtration chromatography (GFC): the average

A. DIFFERENT RATES OF CHAIN PROPAGATION

i ) Non-un i fo rmdis t r ibu t ion

Figure 3. (A)Acrylamide substituted ri'rth a stencalil'bulkymolecule (for example,2)might add more slou'I1'to the growingpolymeric chain that does unsubsittuted acr-r'lamide. tB)Consequently, at least tu'o deviations frorn the desired polyner(a long one, with SA distributed evenll. alons the backbone)are possible: (i) the bulk1' substt:uted acrl'larnldes maypoiymerize into blocks after ail the ursubstttuted acrylamideis reacted; (ii) the bulkl' subsirtutei acn'ianrdes may formshort polymers when pol5menzed rr the absence of theunsubstituted acrylamides.

molecular weight of the hydroll-zed sample rvas 146 kD(M*) (DP t 2000). The distnbutron of moiecularweights was namow (Mr: 75-ir-D and ll/z = 279 kD).42

Sequential addition of tw'o different amines to pNASyrelded the finai functionalized pollmer rScheme 1B).We first acided 0.20 equiv per activated ester of 1. Inmost instances. we added the second amine. R:NH:. inexcess. In some instances, Rr\TI: s'as not added inexcess, and we quenched the residual. unreacted activeesters of the polymer with ammonia.

The number of equivalents of 1 determtned 7s'r in thefinal polymer. To establish the relationship between thenumber of equivalent of I and /sA, w€ performed a setof experiments in which pNA,S was reacted w-ith quanti-ties of I that varied from 0.05 to 1.2 equiv per activatedester, then was quenched with ammonia. We used twosorts of analysis on the resulting polymers (Figure 4):(i) lH NMR spectroscopy compared the integrated areaunder the peaks characteristic of the SA (e.g., theprotons of the acetylated nitrogen) to the area underthe peaks characteristic of the polymer (e-g., the-(CHCH2)- protons); and (ii) combustion analysisprovided estimates of relative values oflsA directly fromanalysis of sulfur.a3 For a small number of equivalents,

XsA was generally smaller than predicted theoretically;as the number of equivalents was increased, there wasstronger agreement between the measured value of UsAand that predicted theoretically. For most of thepolymers used in this study, we used 0.20 equiv peractivated ester; our analyses suggested that 1sA wasequal to -0.17-0.18 in these polymers.

Growing Chain

Effectiue Inhibitors of Hemagglutination

, ,o ' - - -

aa

aa

I neofe l lca l t O

\ ^ d 'V o

a

. ' O -

a'6

t6

e60

0 0.2 0.4 0.6 0.8 1 1.2

H,ri,:Hl, sA peractivated ester

Figure 4. We estimated zsA in the polymers using lH NMRanalysis (O) and combustion analysis for sulfur (O) as afunction of the number of equivalents of SA added peractivated ester in the polymer (see text for details).

Side View

Top View

pel l

,Y"'nii,,",', - I I i,, i,o. ria' ' t \ r l \

' | 'l 'l \ 'l '! lxr:rar '| 'l \ '

l-{-,a f.--7 f{2 N:/l| | f%:^\l ifu*-a: F {i''^' 'I r l f f i t l \usKl trFJlK2/<P@/g

Figure 5. Hemagglutination inhibition assay (HAI). See textfor detailed explanation. The 250-pL microtiter well is viewedboth from the side and the top. Hemagglutination (formationof a gel) is represented by hashed marks; a pellet is shown asa dark spot. The concentration of the inhibitor is 1 unit in thethird well. The inhibitor is sequentialiy diluted by a factor oftr,r'o in the wells to the right of the third one. The value forflt' is defined as the lowest concentration of inhibitor thatinhibits hemagglutination; in this ..s".KfN : 1/z unit.

The IIAI Assay. We assayed the effectiveness of thepolymers in inhibiting virally induced hemagglutinationusing the hemagglutination inhibition (HAI) assay.18'aaObtaining accurate and reproducible values for ICrequired attention to certain details of the assay andan understanding of the principles that underlie itsoperations. Influenza virus agglutinates erythrocytes(hemagglutination). The lowest concentration of inhibi-tor that prevents hemagglutination is defined a..(w(Figure 5). The dissociation constants (Ka) for thecomplexes between different rrLonorleric inhibitors andhemagglutinin have been measured using lH NMRspectroscopy and fluorescence and were found to equaltheir corresponding values of I4 Since half thereceptors are occupied when a ligand is at a concentra-tion equal to Ka, we infer that monomeric inhibitors of

Journal of Medicinal Chemistry, 1995, VoI. 38, No. 21 4183

xflfi, tor

0 .2

,/ \'/

@@@,@@@^1 ser\ t

Incubate for t hr

o@oo@@

1 . 0

0.8

0.6

0.4

0.2

0.0

1 . 0

0.8

0.6

1 0 - 9

*1at1u)

t o ' 8

Ir l

, r l t f? r l

I

I

I

I

I

I

z!f;"ror

to- 71 1 0

I r I l l l l l l | | r r r r . r l , , , , " , . 1

n 2 r O 3 r O 4

Time of Pre-incubation (s)

Figure 6. Using pA with /sA : 0.2, the value for /f,A {o),deireased with increasing time of preincubation (the time forwhich virus and inhibitor mix, prior to addition of erythro-cytes). The time of preincubation used in this study (shown

by a dashed line) was 1800 s (30 min).

SA need to bind approximately half the HA on the

surface of the virus in order to inhibit hemagglutination.

Measuring 4N requires three components in PBS

buffer: erythrocytes, virus, and inhibitor (at varyingconcentrations). As we performed the assay, virus andinhibitor were mixed and allowed to preincubate for 30min at a specified temperature. Erythrocytes were thenadded and mixed, and the system was allowed toincubate for t h. The order and timing of addition ofthe three components strongly influenced the results:if we added the inhibitor last, all inhibitors wereineffective; if we added the erythrocytes last, the timethat the polymer and virus were allowed to preincubate

strongly influenced AfAI Gigure 6). We have notexamined protocols in which the virus was added last.

The erythrocytes usually required at least 30 min ofincubation at 36 oC (or t h at 4'C) to form stable pelletsor gels; we allowed experiments at all temperatures toincubate for t h before reading the plates. At 4 oC, thepatterns of inhibition did not change as \Me varied thetime of incubation from 1-12 h. At higher tempera-

tures, however, the values for /f; generally de-creased over 12 h: at 19 oC, the values decreased by afactor of two, and at 36 oC, the values decreased byvarying amounts (from a factor of 2 to 8). We have notdefined the origin of these decreases, but we emphasizethat it was important to keep details of the experimentconstant-especially the time of preincubation of poly-mer and virus and the time of incubation after theerythrocytes are added-to allow comparison of differentpolymers.

The precision (reproducibility) in our experiments wasgenerally good G{AI mdtiple repetitions allowed pre-cision of less than a factor of 2) so long as we used asingle batch of blood, a single preparation of polymerand virus, and followed the timing of the variousadditions to within a few minutes. If we repeated anexperiment with different preparations of ail threecomponents of the assay, we could reproduce the value

for ^(u to within a factor of 3-4.Poly(acrylic acids) That Do Not Contain SA

Inhibit Hemagglutination. We assayed, as controls,a nurnber of polymers that do not contain SA (Table 1).

The values of .ft are given arbitrarily in terms ofconcentration of monomeric equivalents contained in thepolymers. These values have no clear_meaning, but they

allow comparison with values of lf;N obtained using

4184 Journal of Medirinal Chemistry, 7995, Vol. 38, No. 21

Table 1. Inhibition of Hemagglutination by Polymers That DoNot Contain SA

polymer afN qrurpAA 450bpAA 750bpAA 1200bpAA 30006pAA 40006pA 2006pA 200,6 l\Va acidMeO-PEG(2000)-OH'MeO-PEG(5000)-OH"MeO-PEG(5000)-NH2'

'14I"I is given in terms of total concentration of monomericunits. o MW (kD); pAA: poly(acrylic acid); pA = polyacrylamide.'Number of monomeric units shown in parantheses; MeO-PEG= o-methoxypoly(ethylene glycoi).

1 0-8

10-6

xfrrtrvrl

1o-4

L

Figure 7. I{I is plotted a function of the proportion of SAin the polymers, t'A. Data were collected for polymers (pre-pared by preactivation; Scheme 1) at three different temper-atures: 4'C (tr), 19 "C (o), and 36 "C (a). Data from previousstudies for polymers (prepared by copolymerization) at 4 "Care also shown (l). Incubation times were generally t h.Longer incubation times (12 h, O) yielded lower values of14I"I.

SA-containing polymers (see below). We found, as haveothers,22 that pA does not inhibit hemagglutination atany concentration up to 100 mM. The pAA, perhapssurprisingly, inhibits hemagglutination: pAA with lowermolecular weights were more effective than pAA withhigher molecular weights. These polymers may interactnonspecifically vrith the surface of both the vims andthe erythrocyte. fhey may then inhibit hemagglutina-tion mainly through steric stabilization.

Dependence of KI- on t'A: Qualitative Differ-ences from Previous Studies. We found previouslythat SA-containing polymers varied in effectiveness(4{AI) as a function of /sA in the polymer (Figure 7).The shape of the curve that describes the relationshipbetween /sA and AT* is characteristic of many of thepolymeric inhibitors of hemagglutination that we havepreviously examined: as fsA increases from 0, theeffectiveness increases rapidly, forms a plateau, thendecreases rapidly as.fsA exceeded -0.5. In this work,we varied the value of Xsa from 0.0 to 1.0 (Fieure 7).The pattern of inhibition was qualitatively differentthan before: the initial increase in effectiveness withincreasing XsA was not as sharp; the plateau was lesspronounced; and the decrease in effectiveness occurredat highervalues offl and was less pronounced (in somecases, it did not occur at all). Although we attribute

Mammen et al.

these differences to the different methods of preparingthe polymer, we can only guess the precise mechanisticorigin of these results: the highly effective inhibitionby polymers prepared by preactivation (for all valuesof XsA) rnay have been because of a more uniformdistribution of SA along the backbone of the polymer(that is, fewer wasted SA moieties) than was possibleusing copolymeization; the rapid decrease in effective-ness of the polymers prepared by copolymerization athigh values of Xsa rnay have been because the DP forthese polymers decreased with increasing /sA (Figure9 \.) ).

The polymers prepared previously by copolymenza-tion had values of .(u that decreased linearly withDP;+i a doubling of DP resulted in a halvrng of the valueof ̂ (ru. The average value of DP for these copolymerswas 4000,2r,22 where the average value of DP for thepolymers in the present work was 2000. If all else wasconstant (especially the temperature and the value of

.fsA), we expected the polymers from the present studyto have values of ^(H 2-fold higher (worse) than thepolymers prepared previously. All data indicate sig-nificantly more effective inhibition (values of KfNwere 100-fold lower) than obtained previously.

The differences between the values of .(il obtainedat different temperatures increased with increasingXsA.The lowest value of.fa we observed in this study was200 pM (at fa : 0.50 at 36 'C with al2-h incubation).These two observations suggest that kinetically slowprocesses of some sort (probably equilibration of thepolymer over the biological surfaces involved) increaseeffectiveness of the polymers in inhibiting hemaggluti-nation.

Effectiveness of the Polymers Decreases withIncreasing Charge. We performed a number ofexperiments in which we changed the side chains of SA-containing polymers (in all cases, XsA equals 0.20) andcorrelated these changes to differences in .ft Thestandard polyrner (7s't - 0.20, R2NHz - NHs) provideda point of reference. We divided the changes into groupsbased on the chemical nature of the side chain and thehypothesis we were testing. In the first experiments,we varied the charge of the side chain group (R2NHz,

Scheme 1) from *1 to -3 using the molecules shown inFigure 8. As the charge changed from 0 to -3, theeffectiveness of the polymers decreased; a change from0 to *1 resulted in an even sharper decrease. Since thegroups that introduced charge welg relatively large, theeffects of charge and size ot ^4' were not cleanlydecoupled by this one experiment. The different influ-ences of the *1 and -1 side groups, which are of similarsize, suggested, however, that size alone is not fullyresponsible for the observed differences between thesedifferently charged polymers.

In an attempt to distinguish the effects of steric bulkfrom those of charge, we sJrnthesized three sets ofpolymers in which the charge on the side chain wasvaried relatively independently of the size of the sidechain: (i) Using 0.10 and 0.20 equiv of charged sidechain per activated ester, we obtained polymers thatwere of similar size to the standard polymer, but stillsubstantially charged. (ii) Hydrolyzing the polymerrather than quenching it with ammonia (Scheme 1A)yielded polymers with carboxylates as side chains

100210230400

1000> 13000

300> 100000> 100000> 100000

t o-1

l o -20.4

o 3 6 o c , 1 2 h

a 3 6 o c , 1 h

! 1 9 o c , r h

t r a o c , r h

Effectiue Inhibitors of Hemagglutination

10-9

fi-7

x,Haltrul

10 -5

1o-3

Figure 8. Charged groups as side chains (CONHR2, whereR2 is charged) (Scheme 1). Various proportions of charged sidechains, XR', were assayed: 0.1 (O), 0.2 (a),1.0 (O).

rc-7

rfnl 0vr)

10-5

0 0.2 0.4 0.6 0.8 1

Number of equivalents of 1

Figure 9. EDC and NHS are used to couple 1 to pAA (MW750 kD). EDC, NHS and I were used in a 1:1:1 ratio. The valuefor ̂ 4w (o), is plotted a function of the number of equiva-lents of 1 per free carboxylic acid of pAA.

(primary amides and carboxylates are approximatelyisosteric). Alternatively, coupling pAA to I using EDCand NHS yielded similar SA-containing polymers \,\rithcarboxylates as side chains. (iii) Incorporating crownethers yielded polymers whose charge could be titratedby varying the concentration of the cation that the sidechain could specifically complex. The results of thesethree experiments follow.

Incorporation of lower proportions of the charged sidechains resulted in a less pronounced but similar rela-tionship between the charge and the value of ^f;N(Figure 9).

The parent polymer pNAS was reacted with 0.20equiv of I and then hydrolyzed to give a negativelycharged, isosteric analog of the standard polymer. Thevalue of ^(u of this charged analog was g00 nM,which was approximately 300-fold less effective than thestandard polymer. In an analogous experiment, wecoupled I to pAA (IVIW 750 kD) using EDC and .l/-hydroxysuccinimide. We did not characterize the actualextent of incorporation of I into the polymer. We variedthe equiv of I from 0.0 to 1.0 (Figure 9); the form of theplot relating the number of equiv of 1 and the value of4 was similar to that observed for the pA polymers(Figure 8). In this experiment, the introduction ofcharge introduced zo increase in size of the side chains

Journal of Medicinal Chemistry, 1995, Vol. 38, No. 21 4185

XrjAt (nn)

Na+ rich PBSbuffer

K* r i chbuffer

Figure 10. Polymers containing 18-crown-6 (selective for K+,O), 15-crown-5 (selective for Na+, O), or linear triethyleneglycol (l), and the standard polymer (tr) were assayed inphosphate buffered saline (PBS) or in buffers enriched in eitherNa* or K*. PBS contained 8 g of NaCl, 0-2 g of KCl, 1.1 g ofNa2HPO4, and 0.2 g of KHzPOa. Na+-rich buffer was made byreplacing all K* in PBS with Na+; K+-rich buffer was madeby replacing all Na+ in PBS with K+.

over pA; the negatively charged, isosteric polyrners wereless effective than the standard polymer.

In another experiment relating charge to .(ru, weincorporated Na+- and K+-selective crown ethers in theside chains (Figure 10), and compared the values ofAj* to that of a polymer containing linear triethyleneglycol (not selective for a specific cation). We usedbuffers enriched in either Na+ or K+. These bufferswere equivalent to phosphate-buffered saline (PBS) inionic strength and pH, but only a single type of cationwas used: Na+-rich buffer contained only Na+, and K+-rich buffer contained only K*. By examining the valuesof 4w for these polymers in the three buffers (PBS,Na+-rich, and K+-rich), we hoped to detect a change inthe inhibition that could be correlated with the selectiv-ity of the crown ether. We observed that in all instancesin which the polymer selective for Na+ or K+ wasassayed in Na+-rich or K+-rich buffer, respectively(conditions in which the polymer presumably had a netpositive charge), the polymers became less effective thanwhen assayed in PBS. Furthermore, when the polymerselective for Na- or K+ was assayed in K+-rich or Na+-rich buffer, respectively, the polymer became moreeffective than when assayed in PBS . Since we do notkrrow the dissociation constants for the complexesbetween the crown ethers and the cations, we do notknow the magnitude of the positive charge on thepolymer in the different buffers. These observations areconsistent qualitatively, however, with those from ex-periments in which positively charged side chains areincorporated into the polymer (Figure 10): positivelycharged polymers are much less effective inhibitors ofhemagglutination than neutral polymers.

Effectiveness of the Polymers Decreases withIncreasing Size of the Side Chain. We varied thesize of uncharged side chains using four differentfamilies of molecules: primary amides, linear ethyleneglycols, linear polyols, and cyclic ethers (Table 2). Thetrends for all four families were the same: the effective-ness of the polymers decreased with increasing size ofthe side chain (Figure 11). We represent the size of theside chain by the number of atoms (C, H, N, O)

o-

o-_NH

\ \? (

:^tt1

Yoo^o-

- t r - - - t r - -

l i neart - - C O N H , ^ O - M O x

,/'n^\ )o a( )

- > o v-CONHJ LJ

/-a I\ v - ]

, o )\ o

_ao^"_ro.*o_/

\

XRzo . 0 . 1r- 0.2

1 1 . 0

1 0 - 1 - 2

/ / "n"rn",on side chain

/ I I .p-ruH -

".fiHrur", *i, *"jo-

*t{o-

o^o-

4f86 Journal of Medicinal Chemistry, lgg1, VoI. 58, No. 21

Table 2. Inhibition of Hemagglutination by Polymers withPolar Side Chains, CONHRz, that Differ in Size

Mammen et al.

Molecule Number of Atomsd Identification(Figure 12)

-coNH2

o-coNH_,Ar.,rJ

i \ r 1 2

-CONH -,^On

-CONH-,^O.^--OH

-CONH -.,r. O ,-* O -.^OH

OH-coNH_A.orl

OH OH-coNH.-v\r^oHOH OH

-coNLJo

.CONH)o

r^orI

\ J I( o\ )

-CoNli-/ or----o -

ao b-, b

- )

\ o-6g1r111-,2^ oa,o J

Size of Side-Chain Small

Extent of Steric LowCrowding

K H A | ( M )

Medium

Medium

u

f

l 2

T2

r 9

26

1 6

28

rt, l -r l??

rt

i16

l 9

A 1T I

:-;" "^o 1 q u e

A , Q n i. v l

48

" The totai number of atoms in -CONHRT.

10's

- r a\ r o

10-8

1o-7

K | A r ( M ).-'+lj

l1

Jl

3

10-6

1 0 - s

- c t=__\l "-\___\_u g

o j \

0 1 0 2 0 3 0 4 0 5 0

Number o f A loms in CONHR2

Figure 11. As the size of the side chain CONHR2 (CON(R2)z)was increased the value of .dil increased. See Table 2 foridentifrcation of the labels. Foui families of neutral side chainsare shown: primary amides (O), cyclic ethers (C), polyethyleneglycols (l), and linear polyols (D). With all families of sidechains, the size is roughly proportional to the number of atomsin CONHR2.

contained in CONHR2: we did not calculate molecularvolumes. As the number of atoms in the side chain wasincreased, the value of KIN decreased. The backboneof polymers containing bulky side chains may be lessflexible (fewer available conformations) than those withsmall side chains. Furthermore, such polymers mayadopt more extended (rigid-rod) conformations. Ineither case, the ability of such poiymers to stabilize asurface sterically probably decreases. Since in these

1 0 - 4

aaSegment or Porymer

. ) 1,,^, )A,*+4f ffLftr

Figure 12. Extent of steric crowding around the sialic acid(cylinders) increases with size of the side chain (representedby curves of different length). Large side chains may introducesufficient steric crowding to hinder access to the HA receptorson the surface ofthe virus.

. ^ - q . =l u " - H

"l*.--^,O.-,,-.-^ O-et l H Z

. ^ - Ei - -

0 0 . 0 5 0 1 0 1 s c . 2

l '

Figure 13. The value of /(la increased as the proportion ofcross-linking agent 2 ,7'\ was increased. We assayed thepolymers at various temperatures: 4 "C (O), 19 "C (O), 36 'C(tr).

experiments, the effectiveness of the polymers decreasedas the size of the side chain increased, we tentativelyconclude that steric stabilization plays a role in inhibi-tion of hemagglutination b1- polymers.

Another possibly relevant consequence of increasingthe size of the side chains was to crowd the sialic acidstericaiiy and therebl'tunder access to the IIA receptor(Figure 12). We do not know if steric crowding was themajor mechanism by' which the effectiveness of thepolymers was decreased as a function of the size of theside chain. In an experiment reported previously,22 theiength of the linker connecting the SA to the backboneof the polymer was increased. This experiment at-tempted to decrease any effect of steric crowding. Theobservation was, however, that the polymers containinglong linkers were less effective than polymers qrithlinkers of intermediate length.

Low Levels of Cross-Linking Increase the Ef-fectiveness of Pol5rmers. We added various propor-tions (0.02-0.20) of diamine 2 to cross-link the esterswithin the same polymer and between polymers (Figure13). While low levels of cross-linking increased theeffectiveness of the polymers, high levels had theopposite effect. We reported previously that the ef-fectiveness of polymers correlates positively with DP.Introducing low levels of cross-linking may be equiva-lent to increasing the effective DP. Interestingly, as thetemperature increased, the differences between thepolymers differing in X'also increased. At this time,we do not know the precise mechanistic origin of theseresults.

Ilydrophobic Side Chains Increase the Effec-tiveness of the Pol5rmers. Different nonpolar (hy-

Hydrophobic side chain

Effectiu e Inhibitors of Hemagglutination

Table 3. Inhibition of Hemagglutination by Polymers withHydrophobic Side Chains, CONHR2

Kl"t (nM)Equivo A aa. tg oc 36 0C

-coNHz

-coNH^---*

-coNHo

-coNHCH2o 0.05 100 . 1 l 00.2 t00.4 10

2I/4

o

-coNHs

-coNHcHzq

o Number of equivaients of R2NH2 per activated ester usedin the synthesis (Scheme 1).

drophobic) side chains were incorporated into the poly-mer in an attempt to increase the affinity of the polymerfor the o'ir',rc.32'34'a5 We expected that interactionsbetween the hydrophobic groups and hydrophobic siteson the surface of the virus (including the lipid mem-brane) would enhance binding of the polymer to the viralsurface. Due to decreasing solubility with increasingextent of incorporation, we were limited generally(except in the case where R2NHz - benzyl amine, BA)to values of ,KfN less than -0.1. In the synthesis ofthe polymers (Scheme 1), addition of 0.1 equiv of thehydrophobic R2NH2 w&s followed by quenching w-ithammonia to convert all unreacted activated esters toprimary amides.

Adding certain hydrophobic side chains yielded poly-mers that were more effective than the standardpolymer (Tabie 3). With BA the solubility of thepolymer remained satisfactorily high with extents ofincorporation as high as 0.4: we varied 7BA from 0.0bto 0.40 and found that asXBA increased, the effectivenessof the polymer increased. Even polymers with XBA

--0.05 improved the effectiveness of the polymer by afactor of 2 atroom temperature. This increase was lowbut reproducible. The hydrophobic groups added bulkto the polymer and, in the absence of a balancing effect,were expected to decrease the effectiveness of thepolymer. The case of the BA group provided an exampleof a positive effect that more than balanced the negativeeffect of bulk.

The effectiveness of the polymers increased by a factorof 4-20 as the temperature was increased from 4 oC to36 "C. In contrast, the effectiveness of the standardpolymer increased by a factor of 2. As temperatureincreases, the. importance of hydrophobic interactions(which are ofrben entropically driven) generally in-creases: that the effectiveness of the polyrners increasedsignificantly with increasing temperature was thereforereasonable.a6

Measurements of Klw at Different Tempera-tures. We took parallel data for all side chains at 4,19, and 36 'C (the entire assay-preparation of virus,blood and buffer, preincubation, and incubation-was

1 . 93.8

1 n

3 8

r6

1 0

0.8

0 . 1

0 . 1

Journal of Medicinal Chemistry, 1995, Vol. 38, No. 21 4187

performed at a single temperatr:re). As the temperatureof the assay was increased, the effectiveness of thepolymer generally increased to varying extents. Theseobservations can be rationalized in two ways: (i) If theHAI assay measured equilibria between the virus,polymer and erythrocybe (that is, the systern was atthermodynamic equilibrium), then the dependence ontemperature suggests that there is a positive change inentropy during the process of inhibition. The value ofI{ varied, however, as a function of time of incuba-tion and time of preincubation, suggesting that thesystem was not at thermodynamic equilibrium duringthe assay. (ii) If the system was not at thermodynamicequilibrium, then we expected that the effectiveness ofinhibition should increase with increasing temperature(the rate of most processes increase with increasingtemperature). The dependence of the value of .IfN ontemperature did not follow similar trends with allpolymers. For example, the value of^44 decreased bya factor of two for polymer containing the 20Vo SA (ys^- 0.2) as the temperature of the assay was increasedfrom 4 'C to 36 oC, compared with a factor of 60 for thepolymer containing LA\Vo SA 12rsn - 1.0). These differ-ences in the temperature dependence of the values of4* for polymers containing different side chainsimply that there are differences in the enthalpy ofactivation for inhibition using different polymers.

Conclusions

This work confirms that incorporation of SA groupsinto the side chain of polyacrylamide strongiy enhancesits ability to inhibit hemagglutination mediated byInfluenza A X-31. Polymers prepared using the strategyof preactivation reproducibly gave values of ^4N ttratwere more than 2 orders of magnitude better thanpolymers prepared previously by copolymerization. Wehave suggested possible reasons for this enhancement(Figure 3), but we have not explored these hypothesesfurther in this work.

The effectiveness of the polymers generally increasedwith increasing temperature. Since the system wasprobably not at therrnodynamic equilibrium (the periodsof preincubation and incubation both influenced themeasured value of ,fN), this observation is consi.stentw'ith the increase in rates that are expected withincreasing temperature. Differences in the relation-ships between temperature and the value of ^(w fordifferent polymers imply that there are differences inthe enthalpy of activation for inhibition by these poly-mers.

Steric Stabilization Is an Important Mechanismin Preventing Interaction of Surfaces. Side chainsthat we expected would increase the likelihood of stericstabilization generally increased the effectiveness of thepolymer; side chains that we expected would decreasethe likelihood of steric stabilization generally decreasedthe effectiveness of the polymer. Taking clues from thearea of steric stabilization (also known as polymericstabilization) of coiloidal suspensioDS,3s two character-istics may be responsible for changing the effectivenessof a polymer in stabilizing a surface sterically (FigureL4). First, if the conformation that a polymer adoptsin solution becomes less random-coil and more rigld-rod (that is, if the ratio of its persistance length tocontour length approaches 1), then the polymer is less

l 0

10

0 . i

0 . 1

5

20.62A

2

aL

Virus/

Rigid Rod

>a-ffi@

Random Coil

+++Flexibi l i ty inSo lu t ion- +

4188 Journal of Medicinal Chemistry, 1995, VoL 38, No. 21 Mammen et aI.

effective at inhibiting interaction between virus and cell.Methods of increasing the affinity of the polymer for thesurface of the virus beyond that of the standard polymer

may include: increasing the affinity of a single SA-

HA interaction by using tighter binding derivatives of

SA; incorporating hydrophobic groups that bind adven-

titiousiy, or by design, to secondary sites on the surface

of the virus; incorporating tight-binding inhibitors of

NA. Methods of increasing the ability of the poiy'rner

to stabilize the surface of the virus stericaliy may

include: increasing DP, and incorporating a feu very

large, flexible and water-swollen side chains. Polyva-

tent inhibitors of the interaction between influenza and

erythrocyte provide a system that may be useful as a

model for inhibitors of other pathogen-host interac-

tions.

Experimental Section

Materials and Methods. All solvents and reagents werepurchased from Aldrich and used t'ithout further purificationuniess otherwise noted. Reaction mrxtures were stirredmagnetically and monitored by thin-iayer chromatography onsilica gel precoated glass plates (Merck). Flash column chro-matography was performed on silica gel60rzs+ (230-400 mesh,E. Merck) using the solvents indicated. Size-exclusion chro-matography was performed on Biogel P2 resin or SephadexG10 using distilled water as the eluent. Ion-exchange chro-matography was performed on Dowex 50W-X8 (H* form)cation-exchange resin. Vortexing of copolymer samples wasperformed with a Fisher vortex Genie II. Dialysis wasperformed using Spectra/Por Moiecularporous Membrane(28.6-mm cylinder diameter, 2-mL volume, molecualr weightcutoff -12 kD). The protected dipeptide GIU(OtBu)Glu(OtBu)-OtBuaT and compound 120 (mp 240 oC) were prepared aspreviously described.

All melting points were obtained using a Mei-temp ap-paratus and were uncorrected. Proton (lH) and carbon (13C)NMR spectra were measured on a Bruker AII-400 MHz or AIV[-500 MHz NMR spectrometer. Chemical shifts are reportedin ppm relative to TMS for the lH NMR spectra, and relativeto DMSO-de at 39.5 ppm for the 13C spectra.

N-Acetylneuramintc acid was obtained from extraction ofedible Chinese swiftlet's nest. Erlthrocytes from 2-week-oldchickens were purchased from Spafas Inc., and used within48 h of shipment. Influenza virus (X-31) was obtained fromthe laboratory of Professor John Skehel. The PBS used in theIIAI assays was prepared from 80 g of NaCl, 2 g of KCI, 11 gof NazHPOa, and 2 g of KHzPOT in 1 L of distilled HzO. This10x stock solution was diluted as needed with distilled HzOand then adjusted to pH 7.2 with 1 N NaOH.

N-(Acryloyloxy)succinimide. Acryloyl chloride (33.4 g,30 mL, 369 mmol) was added dropwise to a stirred solution of.l/-hydroxysuccinimide (42.5 g, 369 mmol) and triethylamine(4I.0 g,56.5 mL,409 mmol, 1.1 equiv) in CHCIg (300 mL, 1.23M) at 0 "C. The solution was aliowed to stir for 3 h at 0 oC,

then washed with water (2 x 300 mL) and brine (300 mL),then dried over MgSOa, and recrystalized from a solution ofethyl acetate/hexane (1:1) to give 46.1 g (273 mmol) colorlesscrystals in 72Vc yield (mp 69 "C; lit.36 mp 69.5 "C). 1H NMR(400 MHz, CDCIg): d 6.78 (:CHz, dd, 1H. ,J2s"^ = 17 -2 Hz,lr,",," = 0.78 Hz),6.40 (:CH2, dd, lH, .JBg"- = 17 .2H2, #.i" =

10.88 Hz),6.24 (CH:, dd, lH, J8.i, : 10.88 Hz, ,Fso,,.: 0.78Hz), 2.94 (CHzCHz, s, 4H).

PolyfN-(acryloylory)succini'nidel (pNN). A mixtureof N-(acryloyloxy)succinimide (3.15 g, 18.6 mmol) and AIBN(20 mg,0.007 equiv) in benzene (150 mL) was heated at 60'Cfor 24 h. After the solution was cooled to room temperature(rt), a white precipitate formed. This precipitate was frlteredand washed four times with tetrahydrofuran (THF, 30 mL).Dryrng in uacuo afforded poly[N{acryloyloxy)succinimide](3.08 g, 18.2 mmol,98Vo) as a white fluffy solid. The polymerwas taken up in dry THF (300 mL), vigorously stirred for 3

Collapsed

+++

+++Number of Sitesof Attachment + +

Steric Stabi l izat ion + +++ +

Figure 14. The conformation of the polymer on the surfaceof the colloidal particle (the virus is modeled here as a colloidalparticle) partly determines how effectively the polymer stabi-lizes a suspension of these particles. The polymers shown herehave equal length. Relative to a random-coil polymer with anintermediate number of sites of attachment, changing eitherthe conformation to a rigrd rod (and keeping the sites ofattachment constant), or increasing the number of sites ofattachment (and keeping flexibility constant), decreases theeffectiveness'of the polymer in stabiiizing a colloidal suspen-sion.

able to prevent a colloidal suspension from flocculating.Second, if the polymer binds to the surface of the colloidby a very large number of attachment points, then thepolymer is collapsed onto the surface and is again notable to prevent effectively a colloidal suspension fromflocculating. The polymers that are effective at stericstabiiization of a colloidal suspension are long (a highvalue of DP), well solvated, have a large number ofavailable conformations, and are random-coils.

High-AfEnity Binding Is an Important Mecha-nism in Preventing Interaction of Sur{aces. Sidechains that we expected would decrease the affinity ofthe polymer for the surface of the virus generallydecreased the effectiveness of the polymers; side chainsthat we expected would increase the afiinity of thepolymer for the surface of the virus generally increasedthe effectiveness. As discussed in the previous section,we expect that increasing the affinity of the pol5nnerfor the surface of the virus by increasing the number ofattachment points DBy, at some critical number ofattachment points, becomes an ineffective stratery inbuilding better inhibitors of hemagglutination" Wesuspect that between two polymers of the same lengthand having the same affinity for the surface of the vi:ms,the one with a lower proportion of tighter bindingmonomers will be more effective; that is we suspect thatincorporating small proportions of very tight bindinganalogs of SA will be an effective strategy of buildinghighly effective inhibitors of hemagglutination.

Some of the polymers from this study are the bestinhibitors of hernagglutination that have been reported.Of the polymers whose side chains are 20VoSA, our mosteffective inhibitor at 36 "C is one where l\Vo of the sidechains are benzyl groups (Ri : CHzPh, Scheme 1;

4AI - 600 pM). In conclusion, we suspect that poly-mers that interact strongly with the surface of the virus,or those that effectively stabilize that surface, will be

Effectiue Inhibitors of Hemagglutination

days, filtered, and dried in uacuo. IR36 (Nujol mull): 3340,3200,1730, 1660, 1210, 1070 cm-l.

Deter:nination of Molecular Weight of pNAS. A solu-tion of pNAS (2 mg) in 6 N HCI (aq, 1 mL) was heated at 105oC for 24 h in a sealed tube. After cooling to rt, the pH wasadjusted to 7.2 with 1 N NaOH and water (1 mL) was added.The solution was exhaustively dialyzed against buffer (pH7.2: L50 mM NazSOa, 10 mM NazHPO+). The solution of pAAthus obtained was analyzed by HPLC (Waters flltrahydrogelLinear) using the following standards: pAA of MW 130 kD,390 kD, and 1100 kD. Found: Mp - 129 kD, Mr.i :75 kD,Mw - 146 kD, Mz: 279 kD.

Preparation of Mixed Polymers Containing SA. Asolution of | (243 4mol) in triethylamine (TEA, 0.5 mL) wasadded to a stirred solution of pNAS (6.78 g, 1.22 mmol NHSester) in dimethylformamide (DMF, 20 mL). The solution wasstirred at rt for 20 h, heated at 65 'C for 6 h, and then stirredat rt for an additional 48 h. This procedure yielded tlrre stochsolution of preactivated polymer with ZsA : 0.20 (the stocksolution contained 2.0 pmol SA/mL and 8.0 pmol NHS/mL).AII other polymers that contained different proportions of SAwere prepared analogously by changing the amount of 1. Thenext step varied depending on the polymer.

For Polymers Containing SA on a pA Backbone(R2NHz : NHs). The stock solution (600 aL. 4.8 pmol NHS)was added dropwise to NHaOH (concentrated aqueous, 1.5 mL)and stirred at rt for 12 h.

For Pol5rmers Containing SA and One Other Compo-nent (R2NH2 = NHs). R2NHz (48 7.rmol, 10 equiv) and TEA(48 pmol, 10 equiv) were added to a stirred stock soiution (600pL,4.8 /mol NHS, 1 equiv). The resulting solution was heatedat 65 "C for 6 h. Any remaining NHS ester was quenched byaddition of NFIaOH (concentrated aqueous, 1.0 mL) follwed bystirring at rt for 12 h.

For Polymers Containing SA and TV*o Other Compo-nents. R2NHz (0.48 pmol for yBz, XR' :0.1) and TEA (0.48prmol) were added to a stirred stock solution (600 4L, 4.8 pmolNHS). The resulting solution was heated at 65 'C for 6 h.NH+OH (concentrated aqueous, 2 mL) was added and thesolution heated at 65 oC for 6 h and then stirred at rt for anadditional 12 h.

For all three classes of polymers above, the resultingmixture was then dialyzed exhaustively against distilled HzO,then NHaCI (1 M), again with distilled HzO, and then lyoph-ilized to yield a white powder. Tlpical recovery of SA was >80Vo.

Acknowledgment. The authors gratefully acknowl-edge many fruitful discussions with George Sigal. Wethank Professor J. J. Skehel for kindly providing us withInfluenza virus X-31. G.D. was supported by Fors-chungsstipendium of the Deutsche Forschungsgemein-schaft (DFG) and M.M. by an Eli Lilly predoctoralfellowship (1994). This work was supported by the NIHGrant GM30367 and GM 39589. The NMR facilities atHarvard were supported by NIH grant 1-S10-RR04870-01 and NSF grant CHE88-14019.

Supporting Information Available: A table of values ofIfN at 4,19, and 36 oC for all 100 polymers is available onrequest (6 pages). Ordering information is given on anycurrent masthead page.

References

(1) Krug, R. M. The Influenza Viruses; Plenum Press: New York,1989.

(2) Paulson, J. C. Interactions of Animal Viruses with Cell SurfaceReceptors. lnThe Receptors; Conn, P.M., Ed.; Academic Press:New York, 1985; pp 131-219.

(3) Kilbourne, E. D. BuIl.W. H. O. 1969, 41,643.(4) Kilbourne, E. D. Influenza; Plenum Medical Book Co.: New

York,1987.(5) Kingery-Wood, J. E.;Williams, K W.; Sigal, G. B.;Whitesides,

G. M. The Agglutination of Erythrocytes by Influenza Virus isStrongly Inhibited by Liposomes Incorporating an Analog ofSialyl Gangliosides. J. Am. Chem. Soc. 1992, 114,7303-7305.

Journal of Medicinal Chetnistry, 1995, Vol. 38, No- 21 4189

(6) Pritchett, T. J.; Paulson, J. C. Basis for the Potent Inhibition ofInfluenza Infection by Equine and Guinea Pig a2-Macroglobulin.J. Biol. Chem. L989. 264. 9850-9858.

(7) Cohen, J. A.; Williams, W. V. Molecular Aspects of Reovirus-Host Cell Interaction. Microbiol. Sci. 1988, 5, 265-270.

(8) Collono, R. J. Virus Receptors: the Achilles' Heel of HumanRhinoviruses. Adu. Exp. Med. Biol. 1992,312,6I-70.

(9) Boycot t , R. ;Klenk, H.D.;Ohuchi , M. Cel l Tropism of Inf luenzaVirus Mediated by Hemagglutinin Activation at the Stage ofVirus Entry. ViroLogy Lgg4, 203,313-319.

(10) Mazanec, M. B.; Coudret, C. L.; Fletcher, D. R. IntracellularNeutraiization of Influenza Virus by Immunoglobuiin A Anti-Hemagglutinin Monoclonal Antibodies. J. Virol. 1995, 69, 1339-1343.

(11) Wiley, D.C.; Skehel, J. J. The Structure and Function of theHemagglutinin Membrane Glycoprotein of Influenza Virus.Annu. Reu. Biochem. 1987, 56, 365-394.

(12) Weis, W.;Brown, J. H. ; Cusack, S. ; Paulson, J. C. ; Skehel , J . J . ;Wiley, D. C. Structure of the influenza Virus HemagglutininComplexed with its Receptor Sialic Acid. Nature 1988,333, 426-43r.

(13) Paulson, J. C.; Sadler, J. E.; Hill, R. L. Restoration of SpecificMyxovirus Receptors to Asialoerythrocytes by Incorporation ofSialic Acid with Pure Sialyltransferases. J. Biol. Chem. L979,254.2120-2124.

(14) The correct IUPAC name for this compound is 5-acetamido-2,6-anhydro-3, 5-dideoxy-2-C-[3-[( 2-aminoethyl )thio Jpro pyl]-o< ry thro -manno -nonulosonic acid.

(15) Seaman, G. V. F.; Knox, R. J.; Nordt, F. I.; Regan, D. H. RedCell Aging. I. Surface Charge Density and Sialic Acid Contentof Density-Fractionated Human Erythrocytes. Blood L977, 50,1001 - 101 1 .

(16t Van, B. W. P.: Smets, L. A.; Emmelot, P. Increased Sialic AcidDensity in Surface Glycoprotein of Transformed and MalignantCells. General Phenomenon. Cancer Res. 1973, 33,2913-2922.

(17) Levine. S.; Levine, M.; Sharp, K. A.; Brooks, D. E. Theory of theElectrokinetic Behavior of Human Erythrocytes. Biophys. J.1983.42, r27-r35.

(18) Rogers, G. N.; Pritchett, T. J.; Laue, J. L.; Paulson, J. C.Differential Sensitivity of Human, Avian and Equine InfluenzaA Viruses to a Glycoprotein Inhibitor of Infection: Selection ofReceptor-Specific Variants. Virology 1983, I 31, 39 4 - 408; W. H.O. Tech. Rep. Ser. L959,64.7-32.

(19) Roy, R.; Tropper, F. D.; Romanowska, A. New Stratery inGlycopolymer Syntheses. Preparation of Antigenic Water-SolublePoly(Acrylamide-co-p-Acrylamidophenyl B-lacto side). Bioconl u -gate Chem. 1992, 3, 256-261.

(20) Sparks, M.A.; Will iams, Ii W.; Whitesides, G. M. Neuramini-dase-Resistant Hemagglutination Inhibitors: Acrylamide Co-polymers Containing a C-glycoside of N-Acetylneuraminic Acid.J. Med. Chem. 1993.36, 778-783.

(21) Lees, W.J.; Spaltenstein, A.; Kingery, W. J. E.; Whitesides, G.M. Polyacrylamides Bearing Pendant a-Siaioside Groups StronglyInhibit Agglutination of Erythrocltes by Influenza A Virus:Multivalency and Steric Stabilization of Particulate BiologicalSystems. J. Med. Chem. 1994,37,3419-3433.

(22) Spaltenstein, A.; Whitesides, G. M. Polyacrylamides BearingPendant a-Sialoside Groups Strongly Inhibit Agglutination ofErythrocytes by Influenza Virus. J. Am- Chem. Soc. 1991, 113,686-687.

(23) Matrosovich, M. N.; Mochalova, L. V.; Marinina, V. P.; Byra-mova, N. E.; Bovin. N. V. Sl.nthetic Polymeric Sialoside Inhibi-tors of Influenza Virus Receptor-Binding Activity. FEBS Lett.1990. 272. 209-212.

(24) Matrosovich, M. N.; Gambaryan, A. S.; T\rzikov, A. B.; Byramova,N. E.; Mochalova, L. V.; Golbraikh, A. A.; Shenderovich, M. D.;Finne, J.; Bovin, N. V. Probing of the Receptor-Binding Sites ofthe Hl and H3 Influenza A and Influenza B Virus Hemagglu-tinins by Synthetic and Natural Sialosides.Virology 1993,196,1 1 1 - 1 2 1 .

(25) Gamian, A.; Chomik, M.; Laferriere, C. A.; Roy, R. Inhibition ofInfluenza A Virus Hemagglutinin and Induction of Interferonby Synthetic Sialylated Glycoconjugales. Can. J. MicrobioL 1991,37, 233-237.

(26) Roy, R.;Zanini, D.; Meunier, S.J.; Romanowska, A. Solid-PhaseSynthesis of Dendritic Sialoside Inhibitors of Influenza A VirusHemagglutinin. J. Chem. Soc., C hem. Commun. 1993, 213- 2\7 .

(27) Roy, R-; Andersson, F. O.; Harms, G.; KeIm, S.; Schauer, R.Synthesis of Esterase-Resistant 9-0-Acetylated Polysialosides asInhibitors of Influenza C Virus Hemaggiutinins. Angew. Chem.t992, 104, 1551-1554.

(28) Unverzagt, C.; Kelm, S.; Paulson, J. C. Chemical and EnzymicSynthesis of Multivalent Sialoglycopeptides. Carbohydr. Res.t994, 163, 167-170.

(29) Spevak, W.;Nary. J. O.;Charych, D. H.; Schaefer, M.E.;Gilbert,J. H.; Bednarski, M. D. Polymerized Liposomes ContainingC-glycosides of Sialic Acid: Potent Inhibitors of Influenza Virusin uitro Infectivity. J. Am. Chem. Soc. 1993, 115, 1146-1147.

(30) Lee, Y. C. From X-Ray Crystallography to Cure of Flu? TrendsGlycosci. Glycotechnol. 1993, 5, 309-310.

4190 Journal of Medicinal Chemistry, 1995, Vol. 38, No. 21

(31) Mochalova, L. V.; Tuzikov, A. B.; Marinina, V. P.; Gambaryan,A. S.; Byramova, N. E.; Bovin, N. V.; Matrosovich, M. IV.Synthetic Polymeric Inhibitors of Influenza Virus Receptor-Binding Activity Suppress Virus Replication. Antiuiral. Res.1994,23, 179-190.

(32) Knowles, J. R.;Weinhold, E. Design and Evaluation of a TightlyBinding Fluorescent Ligand for Infleunza A Hemagglutinin. J.Am. Chem. Soc. 1992. 114, 9270-9275.

(33) DeNinno, M. P. The Synthesis and Glycosidation of N-Acetyl-neuraminic Acid. Synfiesis 1991, 483-493.

(34) Toogood, P. L.; Galliker, P. K.; Glick, G. D.; Knowles, J. R.Monovalent Sialosides that Bind Tightly to Influenza A Virus.J. Med. Chem. 199L.34,3140-3143.

(35) Everett, D. H. Basic Principles of Colloid Sciznce; Royal Societyof Chemistry: London, 1988; Chapter 3.

(36) Pollak, A.;Blumenfeld, H.;Wax, M.;Baughn, R. L.;Whitesides,G. M. Enzyme Immobilization by Condensation Copolymeriza-tion into Cross-Linked Polyacrylamide Gels. J. Am. Chem. Soc.1980, 102,6324-6336.

(37) Schnaar, R. L.; Y. C., L. Polyacrylamide Gels Copolymerized withActive Esters. A New Medium for AIfinity Systems. Biochemistry1975 . 14 .1535-1541 .

(38) Hayden, F. G.; Rollins, B. S.; Madren, L. I{- Anti-Influenza VirusActivity of the Neuraminidase Inhibitor 4-guanidinoNeu5Ac2enin Cell Culture and in Human Respiratory Epithelium. AntiuiralRes. 1994, 25,123-137.

(39) Janakiraman, M.N.;White, C. L. ;Laver, W.G.;Air , G. M.;Luo,M. Structure of Influenza Virus NeuraminidaseBlLeel4O Com-plexed with Sialic Acid and a Dehydro Analog at 1.8-AResolution: Implications for the Catalytic Mechanism. Biochem-istry 1994, 33, 8I72-8L79.

(40) NA is a tetrameric glycoprotein found on the surface of influenzaat a density 1O-fold lower than that of HA (102-103 copies/4M2).Its natural enzymatic action is to cleave the glycosidic linkageconnecting SA to lactose. NA is, however, tolerant of many non-lactosyl groups.

Mammen et al.

(41) Sigal, G.; Whitesides, G. M. Manuscript in preparation, 1995.

@D M; = number-average molecular weight : IA',M,,4lNi Mw :

weight-average molecular weight : LN',M,2,E.N',M; M2 : l-average molecular weight : 2l'l,M,3DI'tr,M,2.

(43) The amount of water in the sample was calculated from theratios of carbon to hydrogen and nitrogen. Using the correctedmass for the polymers, we established a relation betrveen thepercentage of sulfur and fA. A similar relationship between thenumber of equiv of 1 and XsA was obtained using the ratio of Sto N from combustion analYsis.

(44) In the HAI assay a dilute suspension of chicken erythroclles rsadded to the wells of a 96-well microtiter plate. These n'ells havea volume of 2504L and a conically shaped baser the erythrocltessettle by gravity and concentrate to a point ta'pellet". easilyidentified visually). If influenza virus is present rn suspenslonwith the erythrocltes, the virus cross-links the erythrocl-tes bvinteraction of viral HA with cellular SA groups and forms a loosegel. The erythrocyces in such a gel do not settle to a pellet:- theyire "hemagglutinated" and appear homogeneously red. An

inhibitor that binas to the virus prevents formation of the gel;

it inhibits hemagglutination.(45) Sparks, M.A.;Will iams, K. W.; Lukacs, C.; Schrell, A.; Priebe,

G.; Spaltenstein, A.; Whitesides, G. M. Synthesis of^PotentialInhibitors of Hemagglutination by Influenza Virus: Chemoen-zymic Preparation of N-S Analogs of N-acetylneuraminic Acid'Tetrahedron 1993, 49, I-12.

(46) Privalov, P. L.; Gill, S. J. Stability of Protein Structure andHydrophobic Interaction. Adu- Prot. Chem. 1988, 39,79I-234'

(4?) Mammen, M.; Gomez, F. A.; Whitesides, G. M. Analysis of the

Binding of Ligands Containing the N-2,4-diqitrophenyl (DNP)

Group to Bivilent Monoclonal Rat anti-DNP Antibody UsingAffinity Capillary Electrophoresis. Azal. Chem.1995, in press'

JM9504666