COMMUNICATIONS

Porous Solids by Design : [Zn(4,4'-bpy)2(SiF,)1,*x DMF, a Single Framework Octahedral Coordination Polymer with Large Square Channels** S. Subramanian* and Michael J. Z a w o r o t k o *

The concept of crystal engineering"' has in recent years pro- vided numerous examples of rationally designed one- (1 D), two- (ZD), and three-dimensional (3D) polymeric structures. Of par- ticular potential interest to inclusion chemistry are 2D and 3D coordination polymers which, in principle a t least, can be de- signed to have very specific pore size and type. In such a context, there already exist examples of 2D square"] and hexagonal/ honeycornbC3l grids generated from metal ions in square-planar or trigonal environments, respectively, coordinated to linear bifunctional ligands such as 4.4'-bipyridine or pyrazines. Fur- thermore, 3D diamondoid coordination polymers based upon tetrahedral or S, coordination environments have been generat- ed by either coordinate covalent bonds or hydrogen bonding.[41 However. given the ubiquity of octahedral metal environments. it is somewhat surprising that simple 3D octahedral polymers[51 remain largely unexplored. Indeed, to our knowledge p-cyano polymers such as Fe&I[Fe"(CN),],, Prussian Blue,[61 the poorly characterized [Cd(CN),(q~inoxaline)],,,['~ and triply interpene- trated [Cd(pyrazine)(Ag,(CN),)(Ag(CN),)llsl represent the only examples of such structures.[9a1 Recently a coordination polymer from silver ions and pyrazine was synthesized and structurally characterized.[gb1 In this contribution we detail a potentially general strategy for generation of such polymers.

Scheme 1 illustrates how cross-linking of cationic square grids with suitable linear bifunctional anionic ligands could in prin-

0 oktaedrisch umgebenes Metallzentrum

"Spacer"-Ligand - linearer

0'- 0'- 0'- 0'- 0'- 0' zweiter "Spacer"-Ligand,

4 vorzugsweise anionisch

Schcmc 1 . Schematic illustration of how a three-dimensional coordination polymer might he gcncrated by cross-linking of square grids.

cavity in 1 is self-included with hexafluorosilicate ions, which are bound through hydrogen bonds to water molecules, and a second independent [Zn(4,4'-bpy),] square grid. It oc- curred to us that the anhydrous form of 1 could be expected to fit the hypothetical network proposed in Scheme 1 since SiFi- ions are capable of acting as a linear bridge between transition metal moieties such as [Co(N-vinylimida~ole)~]~+ [lo]

and [Cu(S-phenylpyra~ole)~]~ + . [ I 'I We therefore synthesized the anhydrous form of 1, namely 2, and herein report its remarkable crystal structure.

[Zn(4,4-bpy),]SiF6 2

The compound 2.x D M F crystallizes in the tetragonal space group P4/mnzm with Z = 1 and therefore sits around a crystallo- graphic 4/mrnrn position."" This alone is quite a remarkable feature since the space group symmetry manifests the maximum D,, point group symmetry that can be sustained by an MX,Y, species. The space group symmetry and cell dimensions are therefore entirely rational or indeed predictable based upon the chemical composition of 2. However, an even more remarkable feature of 2 is that it eschews interpenetration'I3] and therefore, as Figure 1 reveals, large square channels are generated parallel

ciple afford neutral octahedral polymers with porosity con- trolled by the length, volume. and chemical type of the spacer ligands. Coinpound 1 polymerizes to afford square grids with- sides of length 11.43 A,['] Unfortunately, unlike [Cd(4,4'- bpy),](NO,), , which eschews the large

[*] Prof Dr M. J. Zaworotko Ikpartmeiit of Chemistry, Saint Mary's University Halifax. N o b a Scotia, B3H 3C3 (Canada) Telefax: Int. code + (902)420 5261 c-iiiail: m i a w orotiir science.stmarys.ca n r . S. Suhrninanim Department of Chemistry, Sir Wilfred Grenfell College Memori'il University of Newfoundland Corner Brook. Newfoundland, A2H 6P9 (Canada)

[**I This woi-l wits supported by the N . S. E. R. C. (Canada) and the Environmen- tal Science and Technology Alliance Canada. bpy = bipyridyl, DMF = dimethylf~)riiiaiiiide.

Fig. 1. ORTEP (a) and space-filling (b) views of the uh plane in 2. The square channels have the same dimensions as the unit cell (1 1.396 x 11.396 A) and therefore measure about 16 A diagonally. The effective cross-sectioiiiil area IS given by sub- tracting the van der Waals radius of an aromatic ring (ca. 1.7 ,&) from each Pdce of the channel, thereby affording a void that should hold molecules with a cross-sec- tional area of ahout 8 x 8 ,&.

COMMUNICATIONS

to the crystallographic c axis. As would be anticipated, these channels have essentially the same dimensions as those seen in the interpenetrated dihydrate, 11.3959(11) x 11.3959(11) A, and the effective size of their pores (8 x 8 A) compares well with the pore sizes of large zeolites.['41 The volume of the channels is about 50 % of the total volume.r151 The p-SiFi - ions form per- fectly linear bridges, and the Zn-F distances of 2.082(10) A, are shorter than the M- F distances encountered in previous studies concerning p-SiF; units.[". The Zn-N distances (2.1 57(8) A) are also within the expected range. That there is no interpenetration in 2 can possibly be attributed to the orienta- tion of the 4,4'-bipyridine ligands (Fig. 2). These ligdnds are

Experiniental Procedure

2: [Zn(OH,),]SiF, (0.31 g, 1.0 mmol) was suspended in 1,4-dioxane (25 mL) and benzene (25 mL), and the mixture was heated to reflux in a Dean-Stark apparatus to azeotropically remove the water. The powder of ZnSiF, thus formed was dis- solved by adding D M F (25 mLj to the above mixture. Most of the benzene and 1.4-dioxane were removed by rotoevaporation at this point. The mixture was cooled and a solution of 4.4'-bipyridine (0.31 g, 2.0 mmolj in 1.4-dioxane (10 mLj was added. The solution was heated to reflux for 30 min. The resulting pale gold;yellow solution was cooled and allowed to stand at room temperature for 12 h. This afford- ed colorless crystals of 2 (0.48 gj. Crystals appear to lose solvent and crystallinity within ininutes if left to stand exposed to a solvent-free atmosphere and dissolve readily in water. IR (nujol mull): a = 2985, 2872 (bpy C-H): 1668 ( D M F C=O): 1609. 1457, 1379. 1236. 1086. 833. 770, 635 cm-' .

Received' May 30, 1995 [Z8046IE] German version: Angew. Chem. 1995, 107. 2295-2297

Keywords: crystal engineering - hydrogen bonds * porous mate- rials * zinc compounds

Fig. 2. A view of the ac plane in 2. The 4,4'-bipyridine moieties are aligned with the ac plane and therefore preclude porosity or interpenetration in this direction. The a-C-H moieties of the 4,4'-bipyridine ligands are hydrogen-bonded to the fluorine atoms of the SiFi ~ ions.

disposed such that the pyridine moieties are almost coplanar with each other['61 and the crystallographic ac and bc planes, respectively. The former is quite unusual since such a conforma- tion for 4?4'-bipyridine is sterically disfavored, and the latter is important since it precludes porosity and interpenetration along either the crystallographic a or b directions. We attribute this conformation and orientation to the existence of C-H . . F hy- drogen bonds between c(-C-H hydrogen atoms and fluorine atoms of the SiFg-

Although 2 does not compare to other zeolitic solids in terms of stability in aqueous environments or potential catalytic activ- ity, we consider the following features of this study to be partic- ularly salient: 1) Compound 2 represents, to our knowledge, the first example of a neutral octahedral coordination polymer and, therefore, there are no counterions occupying the microchan- riels.['*] 2) Compound 2 is inherently modular and all three components are in principle interchangeable. 3) The dimensions of the channels are precisely what one would expect based upon the geometric features of the three components. 4) The channels in 2 are hydrophobic. 5) The space group of 2 can be regarded as a direct manifestation of the point group symmetry (i.e. the structure of 2 can be regarded as having been "crystal engi- neered" even from the perspective of space group symmetry). We consider 2 to be the prototype of a potentially wide range of porous octahedral structures and are actively investigating the generality of this class of solid and its ability to incorporate medium-sized hydrophobic molecules of cross-sectional area consistent with the dimensions observed in 2.

[l] C. B. Aakeroy. K. R. Seddon. Chem. Sac. Rev. 1993, 397; G. R. Desiraju, Crystal Engineering. The Design ofOrganir So/;& Elsevier, Amsterdam. 1989; M . C. Etter. Arc. Chem. Res. 1990, 23, 120.

[2] M. Fujita, Y J. Kwon, S. Washizu, K. Ogura. J Am. Chem. Sor. 1994, 116, 1151; R. W. Gable. B. F. Hoskins, R. Robson, J Chem. Soc. Chem. Commun. 1990, 1677.

[3] L. R. MacGillivray. S. Subramanian, M. J. Zaworotko. J Chem. Soc. Chem. Commun. 1994. 1325; L. Carlucci, G. Ciani. D. M. Proserpio, A. Sironi, J. Am. Chrm. Soc. 1995, 117.4562.

[4] M. J. Zaworotko. Chem. Sor. Rev. 1994, 23, 284; X. Wang, M. Simard, J. D. Wuest, J Am. Chem. Soc. 1994, 116, 12119.

[5] For the purpose of this study, we define an octahedral polymer as an infinite 3D framework that is sustained by octahedral metal centers that are cross- linked by linear bifunctional ligands. We exclude infinite frameworks which consist of both octahedral metal centers and those with a different coordina- tion geometry.

[6] H. J. Buser, D. Schwarzenbach, W. Petter, A. Ludi, Inorg. Chem. 1977. 16. 2704, and references therein

[7] B. F. Abraham, M . J. Hardie, B. F. Hoskins, R. Robson, E. E. Sutherland, J Chem. Soc. Chem. Commun. 1994, 1049.

[S] T. Soma, H. Yuge, T. Iwamoto, Angen,. Chem. 1994. 33, 1746; Angew Chem. Int. Ed. Engi. 1994, 33, 1665.

[9] a) Although Hoffmann type clathrates contain octahedral metal centers their architecture is based upon 2D p-cyano sustained grids: T. Iwamoto, Inclusion Compounds, Vol. 1 (Eds.: 1. L. Atwood, J. E. D. Davies, D. D. MacNicolj, Academic Press, London, 1984, Chap. 2, p. 29; b) L. Carlucci. G. Ciani, D. M. Proserpio, A. Sironi, Angew Chem. 1995, 107, 2037-2040; Angew. Chem. Int. Ed. Engl. 1995.34. 1895- 1898.

[lo] R. A. J. Driessen. F. B. Hulsbergen, W. J. Vermin, J. ReediJk. Inorg. Chem. 1982, 21, 3594.

[ I l l F. S. Keij. R. A. G. de Grdaff. J. H. Haasnoot, J. Reedijk, E. Pedersen, Inorg. Chini. Actu 1989, 156, 65.

[12] X-ray structure of 2 . x D M F : C,,H,,F,N,SiZn.xDMF, colorless cubes (0.30 x 0.30 x 0.40 mm), tetragonal, space group P4/mmm, with u =11.3959(11j.c=7.6775(9jA, V=997.05(15jA3.Z=l ,p = 0 . 8 7 M g m - 3 (octahedral polymer only) or 1.27 Mgm-3 (including disordered solvent). Data were collected on an Enraf-Nonius CAD-4 diffractorneter with Mo,, radiation (i = 0.71073 A). Non-hydrogen atoms of the coordination polymer were refined with anisotropic thermal parameters. The solvent molecules were observed to be disordered in a manner that could not be readily resolved. Solvent atoms were therefore treated as carbon atoms and refined with fixed isotropic thermal parameters and variable site occupancy. Values of R = 0.063 and R, = 0.061 were obtained for 426 out of 569 reflections with I > 2.5a(l) and 48 parameters. Hydrogen atoms of the 4,4'-bipyridine moieties were placed in calculated positions with d,-, = 1.00 A. Further details of the crystal struc- ture investigation may be obtained from the Director of the Cambridge Crys- tallographic Data Centre. 1 2 Union Road, Cambridge CB2 I E Z (UK) on quoting the full journal citation.

[I31 Self-inclusion by interpenetration invariably occurs when cavities would be larger than 50% of the volume in 3D or 2D frameworks. For example, in diainondoid systems up to sevenfold interpenetration has been observed: K. Sinzger, S. Hunig, M. Jopp, D. Bauer, W. Bietsch, J. U. von Schutz, H. C. Wolf, R. K. Kremer. T. Metzenthin, R. Bau, S . I. Khan, A. Lindbaum, C. L. Lengauer, E. Tillmanns, J Am. Chem. Soc. 1993, l15, 7696.

[14] W. M. Meier, D. H. Olson, AtlasofZeolite Structure f i fes . 3rd revised edition, Butterworth-Heinemann. Boston, 1992.

2128 6 VCH VeriagsgesrN.srhufi mhH. Do-69451 Wrinheim, tYYS 0570-08331Y5/3419-2128 .S 10.00 + ,2510 Angel,,. Chem. Int. Ed. Engl. 1995, 34, No. IY

COMMUNICATIONS

[ I S ] The density of 2 in the absence of solvent is just 0.87 Mgm-- ' compared to 1.856 Mgm in the corresponding dihydrate 1. The density was not deter- mined cxprrimcntally because of the tendency of the crystals to lose solvent rapidly

[16] As noted by 'I reviewer and revealed by Figure 1 a, there is considerable distor- tion of the thermal ellipsoids associated with carbon atoms 2,3,5.6,2',3',5' and 6' o l the 4.4'-bipyridine ligands. This observation is consistent with slight twist- ing of the the pyridine moieties and either static or dynamic disorder between two orient11tions.

[I71 C-H" F and C . . F contacts, 2.342 and 3.429(9) A, respectively. are well within ranges expected for significant C-H ' . . F interactions. 1. Shimoni, H. 1. Carrell. J. P Glusker, M. M. Coombs, J. Am. Chrm. Soc. 1994, ff6. 8162.

[IS] Compounds with comparable or larger pores have recently been reported but cither cations (T. J. McCarthy, T. A. Tanzer, M. G. Kanatzidis, J . Am. Chem. So( 1995. 117. 1294) or anions are necessarily present in the pores (B. F. Ahrdhams. U. F. Hoskins, D. M. Michall, R. Robson, A'urure, 1994,369, 727).

Rational Design of Orthogonal Receptor - Ligand Combinations** Peter J. Belshaw, Joseph G. Schoepfer, Karen-Qianye Liu, Kim L. Morrison, and Stuar t L. Schreiber*

Signal transduction in cells occurs primarily by two mecha- nisms, one involving allostery and the other involving proximi- ty.['] The role of ligand-induced allosteric change in proteins has been appreciated for many years, and has led to the synthesis of numerous molecules that either promote or inhibit the allosteric change required to transduce information in cells. More recent- ly, cell biological studies have illuminated the role of regulated protein dimerization or oligomerization as a means of informa- tion transfer. Such an event can promote a proximal relation- ship between an enzyme and its substrate or a receptor and its ligand, thereby facilitating a molecular interaction leading to a signaling evcnt. Examples include the dimerization of growth factor receptors,[*] oligomerization of antigen and dimerization of transcription factors.[41 These insights have cre- ated new opportunities for chemists to synthesize molecules with two protein-binding surfaces and thereby to induce protein association. Such chemical inducers of dimerization (CIDs) have the potential to activate many cellular processes, including ones of biological and medical significance. We recently de- scribed a method to inducibly control the association of proteins in cells.[', 61 This was accomplished through the expression of chimeric proteins in mammalian cells, consisting of a dimeriza- tion domain fused to a protein or protein domain of interest. By treating these cells with a synthetic, cell-permeable CID that binds to thc dimerization domain, self-association of the protein occurs and a signal is transmitted (Fig. 1).

Although in theory many receptor-ligand systems can be used, the immunophilins FKBP12 and cyclophilin A (CyP) and their ligands FK506 and cyclosporin A (CsA) were selected for this purpose. The ligands are cell permeable and can be modified

[*I Prof S L Schreiber, P. J. Belshaw. Dr. J. G. Schoepfer, K-Q. Liu, K. L Morrison Howard 11 ughes Medical Institute, Department of Chemistry Harvard Uniwrsity, Cambridge, MA 02138 (USA)

e-mail . bclshawfn slsiris.harvard edu TCkfaX' + (617) 495-0751

[*"I This research was supported by a grant from the National Institute of General Medical Sciences (GM-52067). We thank theNSERCfor a postgraduateschol- arahip awarded to P. J. B., and the Swiss National Science Foundation, the Roche Kcsearch Foundation, and Ciba-Geigy Jubilaums Stiftung for fellow- ship) to J G S

Fig. 1. Regulated intracellular dimerization with cell-permeable synthetic ligands named CIDs.

synthetically in a rational way to remove their intrinsic immuno- suppressive and toxic properties. This requires modifying the calcineurin-binding (but not the immunophilin-binding) do- main of these naturally occurring CIDs. FKBP12 and CyP bind their ligands with high affinity (K,, = 0 . 5 n ~ and 5 n ~ , respec- tively), are monomeric, have no discernible effect when ex- pressed in cells, and their small size (12 kDa and 18 kDa, respec- tively) facilitates the incorporation of their cDNA into expression vectors. A shortcoming is that the natural im- munophilins are expressed at high levels in many cells and can therefore diminish the potency of CID ligands toward im- munophilin fusion proteins by forming nonproductive recep- tor-ligand complexes (Fig. 2a). The resulting loss in specificity is expected to be especially troublesome in whole organisms, where cells expressing the fusion proteins may be relatively few in number, thus magnifying the buffering effect of natural im- munophilins.

Fig. 2. Improved receptor-ligand pairs for the regulated inlracellular protein asso- ciation with synthetic ligands a) Possible protein associations occurring in cells expressing wild-type immunophilins as dimerization domains. b) Protein associa- tions predicted to occur in cells by using modified ClDs and iinniunophilin~ with compensatory mutations.

To solve this problem we envisioned the creation of new re- ceptor-ligand pairs with interacting surfaces that ensure a high degree of specificity. Immunophilin ligands were designed to contain substituents that clash sterically with amino acid sidechains in the immunophilin receptor, thereby abolishing their interaction with endogenous immunophilins. Compensa- tory mutations in the receptor were sought that would remove the offending interaction, thereby creating a unique receptor that would be used as a dimerization domain (Fig. 2b). Herein, we report the implementation of this strategy using CyP-CsA as a test system. In addition to providing an effective solution to the specific research problem outlined above, we propose that the creation of new receptor-ligand pairs in this manner will result in new experimental systems for testing our understand- ing of molecular recognition involving protein receptors.

An,qw Cliem. Inr Ed. Engl. 1995. 34, No. 19 A, VCH Ver-lu~sge.s~~llschrrfi mhH, D-694if Weinheim, 1995 O571~-0833/95;34fY-ZIZY $ / O flOT .Z5X 2129