on September 24, 2015
` on September 24, 2015
` on September 24, 2015
` on September 24, 2015
` on September 24, 2015
`
`www.sciencemag.org
`www.sciencemag.org
`www.sciencemag.org
`www.sciencemag.org
`www.sciencemag.org
`
`Downloaded from
`Downloaded from
`Downloaded from
`Downloaded from
`Downloaded from
`
`19 August 1983, Volume 221, Number 4612
`
`SCIENCE:
`
`Solvent-Accessible Surfaces
`Proteins and Nucleic Ac
`
`of
`
`ids
`
`Michael L. Con
`
`reentrant surfaces and of Shrake and
`Rupley (12) for calculating solvent-ac-
`cessible area, I developed a numerical
`computer algorithm for placing dots over
`the solvent-accessible molecular surface
`of a protein (13, 14). Below, I briefly
`review the dot surface algorithm and
`present a new, analytical surface meth-
`od.
`
`Dot Surfaces
`
`Computer graphics has made the re-
`sults of x-ray crystallographic studies of
`proteins and nucleic acids more accessi-
`ble to biochemists and molecular biolo-
`gists. Traditionally, computer-generated
`images of molecular structures have con-
`sisted of lines for the chemical bonds (1-
`3) or spheres (4-7) and ellipsoids (8) for
`the atoms. I present an alternative repre-
`sentation, that of a smooth, continuous
`envelope in contact with the atoms that
`are accessible to solvent.
`Applications of this surface represen-
`tation include enzymology, rational drug
`design, the elucidation of molecular dis-
`eases such as sickle cell anemia, recogni-
`tion of specific DNA base sequences by
`proteins and drugs, and the location of
`possible antigenic determinants on virus-
`es.
`The historical basis for the smooth
`surface envelope method is the work of
`Richards (9) and colleagues on solvent-
`accessible area. Their emphasis was on
`chemical calculations measuring quanti-
`ties of hydrophobic and hydrophilic
`area, while the methods described below
`were developed primarily for the pur-
`pose of visualizing molecular structure
`and interactions. Nevertheless, these
`new methods also permit the measure-
`ment of area and volume in conjunction
`with the graphical display.
`
`Solvent-Accessible Area
`
`Solvent-accessible area was originally
`defined and computed by Lee and Rich-
`ards (10) as the area traced out by the
`center of a probe sphere representing a
`solvent molecule as it is rolled over the
`surface of the molecule of interest.
`These computational methods were in-
`vented as a tool for attacking the protein
`19 AUGUST 1983
`
`folding problem (9). The problem is that
`predicting
`three-dimensional
`of
`the
`structure of a protein given only its pri-
`mary sequence of amino acids.
`Simply measuring a quantity of area is
`insufficient for the study of many aspects
`of protein and nucleic acid function,
`such as substrate binding and catalysis,
`drug-nucleic acid interaction, and recog-
`nition by the immune system. A method
`for visualizing solvent-accessible
`sur-
`
`The basic approach in this method is
`to place a probe sphere, representing a
`solvent molecule, tangent to the atoms of
`the protein at several thousand different
`positions. For each probe position that
`does not experience van der Waals over-
`lap with the atoms of the protein, points
`lying on the inward-facing surface of the
`probe sphere become part of the pro-
`tein's solvent-accessible surface. The
`probe may be placed tangent to (i) single
`
`Summary. A method is presented for analytically calculating a smooth, three-
`dimensional contour about a molecule. The molecular surface envelope may be
`drawn on either color raster computer displays or real-time vector computer graphics
`systems. Molecular areas and volumes may be computed analytically from this
`surface representation. Unlike most previous computer graphics representations of
`molecules, which imitate wire models or space-filling plastic spheres, this surface
`shows only the atoms that are accessible to solvent. This analytical method extends
`the earlier dot surface numerical algorithm, which has been applied in enzymology,
`rational drug design, immunology, and understanding DNA base sequence recogni-
`tion.
`
`faces is needed. For this purpose, an
`alternative
`solvent-accessible
`surface
`definition, proposed by Richards (9), is
`appropriate. Unlike the original surface
`of Lee and Richards (10), this alternative
`molecular surface is not displaced from
`the van der Waals surface. Instead, it
`consists of the part of the van der Waals
`surface of the atoms that are accessible
`to the probe sphere (contact surface),
`connected by a network of concave and
`saddle-shaped surfaces (reentrant sur-
`face) that smooths over the crevices and
`pits between the atoms. This surface is
`the boundary of the volume from which a
`probe sphere is excluded if it is not to
`experience van der Waals overlap with
`the atoms.
`Improving on the algorithms of Greer
`and Bush (11) for calculating contact and
`
`atoms, creating a dot at the point of
`tangency, (ii) pairs of atoms, creating a
`concave arc of dots connecting the two
`points of tangency, and (iii) triples of
`atoms, creating a concave triangle of
`dots between the three points of tangen-
`cy. For each surface point generated, the
`numerical algorithm produces not only
`its coordinates but also an approximate
`solvent-accessible area associated with
`the point and an outward-pointing unit
`vector perpendicular to the surface at
`that point. The pancreatic trypsin-tryp-
`sin inhibitor complex (15) is shown in
`Fig. 1, with a dot surface for the enzyme
`only.
`
`The author is a Helen Hay Whitney postdoctoral
`fellow in the Molecular Biology Department, Re-
`search Institute of Scripps Clinic, La Jolla, Califor-
`nia 92037.
`
`709
`
`Par Pharm., Inc.
`Exhibit 1027
`Page 001
`
`

`
`isIa Surface
`
`This method has prod- %t4
`enzymology (16-20), mswn4'gy ( -I'
`A continuous molecular surface con-
`22), virology (23), molecular pathology
`tour is defined as the union of pieces of
`(24), and the study of protein-ligand (25)
`spheres and tori joining smoothly at cir-
`and protein-protein (26, 27) interactions.
`cular arcs. There are three kinds of
`Despite the many applications of the
`pieces: concave spherical triangles, sad-
`dot surface numerical algorithm, it was
`convex
`and
`rectangles,
`die-shaped
`necessary to invent an analytical surface
`spherical regions (Fig. 2).
`algorithm in order to generate high-reso-
`The computer algorithm proceeds in
`lution color raster display images and to
`three steps, one for each shape of sur-
`compute more accurate molecular areas
`face. First, a probe sphere is placed
`and volumes.
`
`Fig. 1. Stereo pair of the pancreatic trypsin-trypsin inhibitor complex. The enzyme is
`represented by a dot surface. The residues of the inhibitor in contact with the enzyme are
`represented by bonds. The part of the trypsin surface that is kept from contact with the solvent
`by the presence of the inhibitor is colored red.
`
`Fig. 2. Heme molecule drawn on a color raster graphics system. Green, convex surface; red,
`saddle surface; blue, concave surface. Surface pieces join at circular arcs.
`710
`
`tangent to every set of three neighboring
`atoms, and a concave triangle is generat-
`ed whenever the probe sphere experi-
`ences no collisions with any other atoms
`of the molecule. Each concave triangle
`has three concave arcs as edges. Next,
`the saddle rectangles are formed by con-
`necting adjacent concave arcs along the
`inner surfaces of tori (Fig. 3). The edges
`of each saddle rectangle consist of a pair
`of concave arcs and a pair of convex
`arcs. In the final step, the convex arcs on
`each atom are grouped to form closed
`circuits, or cycles, and the boundary of
`each convex face is defined by zero, one,
`or more cycles. The equations defining
`the surface and the details of the comput-
`er algorithm will be presented elsewhere
`(28).
`Since molecular areas and volumes are
`important physical chemical properties
`of molecules, efforts have been made to
`calculate them from x-ray coordinates.
`The areas of the convex faces are re-
`ferred to in the literature as contact
`areas, and approximate numerical meth-
`ods for their measurement have been
`developed (10, 29, 30). With the surface
`defined in an analytical fashion, it is now
`possible to calculate these contact areas
`exactly. This is done by using the Gauss-
`Bonnet theorem (31) from differential
`geometry. This theorem is traditionally
`used to study the relation of surface
`topology to integrals of curvature, but
`since the curvature of a convex spherical
`face is constant, the integrals simplify
`and the contact area may be expressed
`as a function of the atomic radius and the
`geometry and topology of the boundary
`cycles. The area of a concave face is
`calculated in a similar fashion. The area
`of a saddle face may be calculated by
`using integral calculus, since it is part of
`a surface of revolution, the torus.
`Molecular volumes have been calcu-
`lated from protein x-ray crystallographic
`coordinates, using a polyhedral defini-
`tion of the protein surface, and these
`calculated volumes have been compared
`to experimentally measured partial spe-
`cific volumes of proteins in solution (32).
`A smoothly curved definition of the pro-
`tein surface, such as the analytically
`surface,
`solvent-accessible
`defined
`should help provide a more accurate
`measurement of molecular volume. The
`volume enclosed by the solvent-accessi-
`ble surface may be calculated by parti-
`tioning this volume into simpler shapes
`whose volumes may be easily calculated
`by solid geometry and integral calculus.
`Most of the molecular volume is con-
`tained within an interior polyhedron
`whose vertices are the centers of the
`solvent-accessible atoms. Coating this
`SCIENCE, VOL. 221
`
`Par Pharm., Inc.
`Exhibit 1027
`Page 002
`
`

`
`Fig. 3 (left). Trajectory of probe rolling over a molecular surface. The trajectory arcs (red) connect positions where the probe is simultaneously
`tangent to three atoms. In a corresponding manner, saddle rectangles connect concave triangles. These reentrant surfaces (green) then define the
`Fig. 4 (right). Yeast phenylalanyl transfer RNA anticodon (GAA). The contact surface of the
`boundaries of the convex surfaces (magenta).
`three anticodon bases is shown. The contact areas in square angstroms are displayed next to the atom labels.
`
`polyhedron is a surface layer that is
`made up of one piece for each curved
`face of the analytical surface. This sur-
`face layer has an average thickness of
`about an atomic radius.
`As an example of an application of the
`area method, the contact areas of the
`atoms of the transfer RNA anticodon
`(33) have been computed and are shown
`in Fig. 4. The conjunction of both graphi-
`cal and area measurement methods
`makes it possible to see not only how
`much of an atom is accessible but also
`where the accessible regions are. For
`this anticodon and the DNA structures
`presented below, van der Waals radii
`with implicit hydrogens have-been taken
`from (30) and a probe with a radius of 1.5
`A has been used.
`To illustrate the ability of the analyti-
`cal method to measure small changes in
`area and volume, the room-temperature
`(34) and low-temperature (35) DNA do-
`decamer structures are compared. The
`molecular areas and volumes are 3631 A2
`and 6534 A3 (290 K, 1.9-A resolution)
`and 3623 A2 and 6514 A3 (16 K, 2.7-A
`resolution), respectively. The low-tem-
`perature structure is 0.3 percent smaller
`in volume. The room-temperature struc-
`ture is shown in Fig. 5.
`
`Computer Graphics
`
`The analytically defined surface, being
`continuous rather than discrete, is well
`suited to raster display. The input to a
`raster graphics system consists of a two-
`dimensional array of picture elements, or
`pixels, each of which has a color and
`shade value (36). In order to produce this
`19 AUGUST 1983
`
`pixel array from a three-dimensional
`curved surface, a hidden-surface elimi-
`nation algorithm is required (37). The
`analytical molecular surface representa-
`tion is substantially different from previ-
`ous curved surface representations, such
`as polygon mesh, parametric bicubic
`patches, and solid modeling (38, p. 506),
`so it was necessary to invent a hidden-
`surface algorithm for it, which will be
`published elsewhere (39).
`The use of stereo, in conjuction with
`hidden-surface elimination and shading,
`gives a vivid demonstration of protein
`topography (Fig. 6). The copper atom is
`seen to lie in a deep pit at the active site
`of Cu,Zn superoxide dismutase (40).
`One is not restricted to using spheres
`to represent individual atoms. For large
`molecular complexes, it is useful to mod-
`el a group of atoms with a single sphere.
`The 2.8-A structure of aspartate carbam-
`oyltransferase (41) has been modeled
`with each amino acid residue represent-
`ed by a sphere centered at the alpha
`carbon (Fig. 7).
`A method for smoothing the junctures
`between atoms by summing Gaussian
`densities for each atom and drawing sur-
`faces at various density contour levels
`was developed by Blinn (42). While the
`probe sphere method does a similar
`smoothing, its main effect is not the
`smoothing of crevices and pits, but rath-
`er the complete removal of the van der
`Waals surface of interior atoms. This
`interior surface removal is important,
`because most of a protein's van der
`Waals surface is in the interior and not
`directly involved in molecular interac-
`tions.
`Display of dot surfaces on a real-time
`
`color vector system is, in general, more
`useful than the raster surface representa-
`tion because (i) the dot surfaces are
`transparent, enabling chemical bonds
`and atom labels to be seen, and (ii) the
`image may be rotated and sectioned in
`real time. However, the raster system
`does have the advantage that it can show
`a larger region of surface at high resolu-
`tion. This is because raster systems typi-
`cally display a quarter of a million pixels,
`while real-time vector systems can han-
`dle only 10,000 to 20,000 vectors.
`In addition to the shaded-surface ras-
`ter representation, an analytical surface
`has a real-time vector representation,
`where each face of the surface is repre-
`sented by a set of concentric curved
`polygons (Fig. 3). These polygons may
`be calculated in a straightforward man-
`ner for concave and saddle faces, but
`convex faces have less regular shapes.
`For convex faces, concentric cycles bor-
`dering a shrinking contact area are gen-
`erated by progressively incrementing the
`radii of neighboring atoms.
`An interactive display program is re-
`quired to manipulate molecular surfaces
`on a vector graphics system. For the
`Evans and Sutherland Multi Picture Sys-
`tem, this need is satisfied by the general-
`purpose graphics program GRAMPS
`(GRAphics for the Multi Picture Sys-
`tem), developed by O'Donnell and Olson
`(43). GRAMPS may simultaneously dis-
`play any combination of curved polygo-
`nal surfaces, chemical bonds, atom la-
`bels, dot surfaces, and arbitrary geomet-
`ric figures, such as an icosahedron repre-
`senting a virus capsid (44). The various
`graphical objects are organized into a
`hierarchical tree structure and each ob-
`711
`
`Par Pharm., Inc.
`Exhibit 1027
`Page 003
`
`

`
`5.S DNA dodecamer with sequence:
`.--
`COCGAATrCGCG. The part of the van der
`Waals surface of each atom that is accessible
`to solvent is colored by atom type (red, oxy-
`gen; green, carbon; blue, nitrogen). Reentrant
`surface (white) smooths out the crevices and
`pits between the atoms.
`
`ject may be independently transformed
`and colored in real time.
`A primary value of the graphical dis-
`play of solvent-accessible surfaces is
`that it provides immediately comprehen-
`sible information about steric comple-
`mentarity. This is illustrated by the work
`of Blaney et al. (45), who used real-time
`color dot surface graphics to study the fit
`of the thyroid hormone thyroxin into the
`binding site of a blood transport protein,
`prealbumin. They noticed an empty
`pocket adjacent to one of the phenyl
`rings of thyroxin. Computer graphics
`modeling showed that naphthyl analogs
`of thyroxin would fit into the binding site
`and the larger naphthyl ring would fill
`this pocket. When a wide variety of
`thyroid hormone analogs were tested,
`
`Fig. 6. Stereo pair of Cu,Zn superoxide dismutase. Same color-coding as in Fig. 5, but with the
`contact and reentrant surfaces of sulfur and copper colored yellow and copper. The copper
`atom is part of the active site and interacts with the superoxide radical. Hydrogen atoms are
`given the color of the heavy atom they are bonded to. Self-intersecting surfaces create point and
`edge cusps and other artifacts in deep grooves.
`
`Fig. 7. Stereo pair of aspartate carbamoyltransferase. The top catalytic trimer is colored green,
`light green, and cyan. The bottom catalytic trimer is colored red, pink, and magenta. The
`regulatory dimers are colored yellow and white. Each amino acid residue is represented by one
`sphere 3 A in radius positioned at the alpha carbon, and a probe sphere 3 A in radius was used to
`calculate the surface of each subunit.
`712
`
`those with a naphthyl ring filling this
`pocket were found to bind better than
`those which left the pocket empty.
`Another use of this surface representa-
`tion has been to paint chemical informa-
`tion onto it. Weiner et al. (46) did this by
`coloring the surface dots of proteins and
`nucleic acids according to electrostatic
`potential. Interfacing surfaces in protein-
`protein, protein-ligand, and drug-nucleic
`acid interactions were seen to have not
`only topographic but also electrostatic
`complementarity. The electrostatic sur-
`face potential of DNA was seen to be
`strongly
`sequence-dependent.
`This
`method has also been used to study the
`binding of the negatively charged super-
`oxide radical to the enzyme superoxide
`dismutase (47). In the electrostatic meth-
`od, the potential is evaluated at the cen-
`ter of each probe sphere position that
`generates a surface point. That is, in
`addition to being a canvas for displaying
`chemical information, the surface can
`play a fundamental role in calculating
`that information.
`
`Conclusions
`
`The principal use of computer graph-
`ics by macromolecular x-ray crystallog-
`raphers has been in fitting the model to
`the electron density and in refining the
`structure (1-3). The methods described
`above will help crystallographers in the
`succeeding
`of interpreting
`the
`step
`solved structure. Scientists in related
`disciplines will also benefit, since the
`structures of more than 100 proteins,
`nucleic acids, and virus capsids have
`been deposited for general distribution at
`the Protein Data Bank at Brookhaven
`National Laboratory (48).
`While the display of solvent-accessible
`surfaces on real-time vector graphics
`systems is preferred for interactively ex-
`ploring a macromolecular structure, the
`color raster display of solvent-accessible
`surfaces made possible by the analytical
`algorithm is better able to communicate
`structural discoveries because of its
`higher resolution and greater visual real-
`ism.
`In time, the raster solvent-accessible
`surface display will acquire more of the
`capabilities of the vector display. For
`example, raster graphics methods for
`displaying transparent surfaces exist (49)
`and can be adapted to this system. Also,
`it should be possible to section away the
`front surface of a protein to display inte-
`rior pockets and cavities, since the hid-
`den-surface algorithm (39) uses a depth
`buffer (38, pp. 560-561), where the
`height of each pixel is stored.
`SCIENCE, VOL. 221
`
`Par Pharm., Inc.
`Exhibit 1027
`Page 004
`
`

`
`graphical methods
`Although these
`were developed to study the protein sur-
`face, they should also be useful in visual-
`izing the packing of alpha helices and
`beta sheets in the protein interior, simply
`by giving these structural elements indi-
`vidual surface contours. This will bring
`solvent-accessibility studies back full-
`circle to their original scientific problem,
`the understanding of the folding of the
`polypeptide chain to form protein ter-
`tiary structure.
`
`References and Notes
`1. R. Diamond, in Computational Crystallography,
`D. Sayre, Ed. (Oxford Univ. Press, Oxford,
`1982), p. 318; T. A. Jones, in ibid., p. 303.
`2. F. P. Brooks, Jr., in Proceedings of the 1977
`International Federation of Information Proc-
`essing, B. Gilchrist, Ed. (North-Holland, Am-
`sterdam, 1977), p. 625.
`3. C. D. Barry, C. E. Molnar, F. U. Rosenberger,
`Technical Memo 229 (Computer Systems Labo-
`ratory, Washington University, St. Louis, Mo.,
`January 1976); C. D. Barry, H. E. Bosshard, R.
`A. Ellis, G. R. Marshall, in Computers in Life
`Science Research, W. Siler and D. A. B. Lind-
`berg, Eds. (Plenum, New York, 1975), p. 137.
`4. T. Porter, Comput. Graphics 12, 282 (1978);
`ibid. 13, 234 (1979).
`5. R. J. Feldmann et al., Proc. Natl. Acad. Sci.
`U.S.A. 75, 5409 (1978).
`6. K. Knowlton and L. Cherry, Comput. Chem. 1,
`161 (1977).
`7. N. L. Max, Comput. Graphics 13, 165 (1979).
`8. C. K. Johnson, Oak Ridge Natl. Lab. Tech.
`Rep. 5138 (1976).
`9. F. M. Richards, Annu. Rev. Biophys. Bioeng. 6,
`151 (1977).
`10. B. Lee and F. M. Richards, J. Mol. Biol. 55, 379
`(1971).
`11. J. Greer and B. L. Bush, Proc. Natl. Acad. Sci.
`U.S.A. 75, 303 (1978).
`
`12. A. Shrake and J. A. Rupley, J. Mol. Biol. 79,
`351 (1973).
`13. M. L. Connolly, thesis, University of California,
`Berkeley (1981).
`14. __, QCPE Bull. 1 (1981), p. 75. The dot
`molecular surface program (MS) is written in
`Fortran and may be obtained by writing to
`Quantum Chemistry Program Exchange, De-
`partment of Chemistry, Indiana University,
`Bloomington 47405.
`15. R. Huber, D. Kukla, W. Bode, P. Schwager, K.
`Bartels, J. Deisenhofer, W. Steigemann, J. Mol.
`Biol. 89, 73 (1974).
`16. R. N. Smith, C. Hansch, F. H. Kim, B. Omiya,
`G. Fukumura, C. D. Selassie, P. Y. C. Jow, J.
`Langridge, Arch. Biochem.
`M. Blaney, R.
`Biophys. 215, 319 (1982).
`17. C. Hansch, R. Li, J. M. Blaney, R. Langridge,
`J. Med. Chem. 25, 777 (1982).
`18. S. Sprang, R. Fletterick, M. Stern, D. Yang, N.
`Madsen, J. Sturtevant, Biochemistry 21, 2036
`(1982).
`19. E. Goldsmith, S. Sprang, R. Fletterick, J. Mol.
`Biol. 156, 411 (1982).
`20. S. R. Sprang, E. J. Goldsmith, R. J. Fletterick,
`S. G. Withers, N. B. Madsen, Biochemistry 21,
`5364 (1982).
`21. R. A. Lerner, Nature (London) 299, 592 (1982).
`22. A. J. Olson, G. Cohen, D. Davies, Antibody
`Structure, film available from Byron Motion
`Pictures, 65 K Street, NE, Washington, D.C.
`20002.
`23. A. J. Olson, Virus Wars, computer-generated
`film presented at the International School of
`Crystallography, Conference on Crystallogra-
`phy in Molecular Biology, Italy, June 1982.
`24. R. E. Dickerson and I. Geis, Hemoglobin:
`Structure, Function, Evolution, and Pathology
`Calif.,
`Park,
`(Benjamin/Cummings,
`Menlo
`1983), p. 136.
`25. I. D. Kuntz, J. M. Blaney, S. J. Oatley, R.
`Langridge, T. E. Ferrin, J. Mol. Biol. 161, 269
`(1982).
`R. Langridge, T. E. Ferrin, I. D. Kuntz, M. L.
`26.
`Connolly, Science 211, 661 (1981).
`27. R. 0. Fox, Jr., and F. M. Richards, Nature
`(London) 300, 325 (1982).
`28. M. L. Connolly, J. Appl. Crystallogr., in press.
`29. T. J. Richmond and F. M. Richards, J. Mol.
`Biol. 119, 537 (1978).
`30. C. J. Alden and S.-H. Kim, ibid. 132, 411 (1979).
`
`31. M. P. do Carmo, Differential Geometry of
`Curves and Surfaces (Prentice-Hall, Englewood
`Cliffs, N.J., 1976), pp. 274-276.
`32. F. M. Richards, J. Mol. Biol. 82, 1 (1974).
`33. J. L. Sussman, S. R. Holbrook, R. W. Warrant,
`G. M. Church, S.-H. Kim, ibid. 123, 607 (1978).
`34. H. R. Drew, R. M. Wing, T. Takano, C. Broka,
`S. Tanaka, K. Itakura, R. E. Dickerson, Proc.
`Natl. Acad. Sci. U.S.A. 78, 2179 (1981).
`35. H. R. Drew, S. Samson, R. E. Dickerson, ibid.
`79, 4040 (1982).
`36. T. Whitted, Science 215, 767 (1982).
`37. W. N. Newman and R. F. Sproull, Principles of
`Interactive Computer Graphics (McGraw-Hill,
`New York, 1979), pp. 367-388.
`38. J. D. Foley and A. Van Dam, Fundamentals of
`Interactive Computer Graphics (Addison-Wes-
`ley, Reading, Mass., 1982).
`39. M. L. Connolly, in preparation.
`40. J. A. Tainer, E. D. Getzoff, K. M. Beem, J. S.
`Richardson, D. C. Richardson, J. Mol. Biol.
`160, 181 (1982).
`41. H. L. Monaco, J. L. Crawford, W. N. Lips-
`comb, Proc. Natl. Acad. Sci. U.S.A. 75, 5276
`(1978).
`42. J. F. Blinn, ACM Trans. Graphics 1, 235 (1982).
`43. T. J. O'Donnell and A. J. Olson, Comput.
`Graphics 15, 133 (1981).
`44. A. J. Olson, Tomato Bushy Stunt Virus, film
`available from Palmer Film Services,- 611 How-
`ard Street, San Francisco, Calif. 94105.
`45. J. M. Blaney et al., J. Med. Chem. 25, 785
`(1982).
`46. P. K. Weiner, R. Langridge, J. M. Blaney, R.
`Schaefer, P. A. Kollman, Proc. Natl. Acad. Sci.
`U.S.A. 79, 3754 (1982).
`47. E. D. Getzoff, thesis, Duke University, Dur-
`ham, N.C. (1982).
`48. F. C. Bernstein et al., J. Mol. Biol. 112, 535
`(1977).
`49. T. Whitted, Commun. ACM 23, 343 (1980).
`50. I thank T. J. O'Donnell and A. J. Olson for the
`use of their program GRAMPS, the Brookhaven
`National Laboratory Protein Data Bank for x-
`ray coordinates (48). M. Pique for assistance at
`the Computer Graphics Laboratory, University
`of North Carolina at Chapel Hill (NIH RR
`00898, F. P. Brooks, Jr., principal investigator),
`J.-P. Dumas for assistance at the Salk Institute,
`and the Helen Hay Whitney Foundation for a
`postdoctoral fellowship.
`
`Ground Water Contamination
`in the United States
`
`Veronica I. Pye and Ruth Patrick
`
`Ground water that is used by humans
`consists of subsurface water which oc-
`curs in fully saturated soils and geologi-
`cal formations. Nearly half the popula-
`tion of the United States use ground
`water from wells or springs as their pri-
`mary source of drinking water (1, 2); 36
`percent of the municipal public drinking
`water supply comes from ground water
`(I); and 75 percent of major U.S. cities
`depend on ground water for most of their
`supply (3). Total fresh ground water
`withdrawals in 1980 were estimated as
`88.5 billion gallons per day, of which 65
`percent were used for irrigated agricul-
`ture (4). Although ground water contami-
`19 AUGUST 1983
`
`lation of water between the oceans, at-
`mosphere, and land. It constitutes ap-
`proximately 4 percent of the water in the
`hydrologic cycle, second only to the
`oceans and seas, which account for
`about 94 percent (5). The volume of
`ground water in storage exceeds the vol-
`ume of fresh surface water in lakes,
`streams, and rivers. Approximately 30
`percent of the streamflow of the United
`supplied by ground water
`is
`States
`emerging as natural springs or other
`seepage areas (2). Ground water forms
`most, if not all, of the low water flow of
`streams during dry periods. The interre-
`lation between surface water and ground
`water is further indicated by the fact
`that, under certain conditions, surface
`water may recharge ground water aqui-
`fers.
`Aquifers may be composed of perme-
`able or porous geological material, either
`unconsolidated sand and gravel or con-
`solidated material such as carbonate
`
`nation has occurred for centuries, in-
`industrialization,
`population
`creased
`density, and agricultural activities have
`greatly exacerbated the problem in some
`areas. As our dependence on ground
`water increases, its quality becomes an
`ever more important issue.
`Ground water is not only important to
`man, it is also an integral part of the
`hydrologic cycle of the earth-the circu-
`Veronica I. Pye is Research Director of the Environmental Assessment Council, Academy of Natural
`Sciences, Philadelphia, Pennsylvania 19103. Ruth Patrick is Chairman of the Environmental Assessment
`Council, is Senior Curator of Limnology and occupies the Francis Boyer Research Chair at the Academy of
`Natural Sciences, Philadelphia, and is Adjunct Professor at the University of Pennsylvania. The report on
`ground water on which this article is based was prepared by the Environmental Assessment Council. Council
`members were Robert G. Dunlop, Caryl Haskins, Richard E. Heckert, Lane Kirkland, George Lamb,
`Charles F. Luce, Ruth Patrick, Glen Paulson, William Reilly, Laurance S. Rockefeller, Abel Wolman, and
`George Wills.
`
`713
`
`Par Pharm., Inc.
`Exhibit 1027
`Page 005