Frank H. Clarke Research Department Pharmaceuticals Division CIBA-GEIGY Corporation Ardsley, New York 10502
I New Skeletal-Space-Filling Models I
A mode/ of an enzyme active site
One of the most exciting aspects of chemistry today is its ahilitv to nrovide a nictorial exolanation of hiochemical orocesses. T&, modein introductory chemistry texts ( I ) iead the student all the wav from the structure of water and methane to a m o ~ e c u ~ a r k x ~ ~ a n aoft ithe o n specificity of enzvmes. They succeed in this challeneine task hv the liberal use of graphic i;llustrations. Molecular models are necessary to obtain three dimensional perspective (2).The skeletal model gives substance to a line drawing on a printed page and illustrates conformation and to visualize how one molecule instereochemist&. teracts with another, especially with a macromolecule, the external shape of hoth must he visualized. Space-fillingmodels are needed for this purpose. With practice the chemist learns to "see" a conformational drawing as a three dimensional skeletal shape and, finally, as a space-filling molecule. In 1968. Anne Walton orovided a comorehensive review of the use of hoth skeletal and space-filling models (3),and a brief descriotion of newer additions was orovided bv Gordon in 1970 (4): Most of these are either skiletal, i.e., rei id in^, F.M.M. and Kendrew, or space filling, i.e., Corey-PaulingKolton (CPK), or Courtauld. Others such as those descrihed hv Petersen ( 5 ) and the Nicholson Molecular Models (6)are somewhat in between, being .'expanded models" with spheres tn represent the atoms hut with rods for the tn~nds.Ax pointed our hg Walton ( 3 )all modelsare imperfect. However, Rupley and Bruice (7) pointed out in 1968 the advantages of visualizing protein st&ctures by constructing the model first with Kendrew skeletal components and later adding adapted Courtauld space fillers. The concept has been commercialized in the DorMmith skeletal models with Atomunit SpaceFilline Envelones and in the Minit Molecular Model Buildine system with space-filling shells. Both models are large (2 cm per Angstrom) and expensive. Recently the late C. L. Stong described (8) smaller and inexpensive new plastic color coded skeletal (CCS) models (9)and their use for the construction of models of proteins. The new models will he commercialized hv Science Related Materials Inc. (P.O. Box 1422. Janesville, WI 53545). Not only are they color coded so that each indi: vidual atom is easily recognized hut they are versatile and practical for small and cohplex molecuies alike. Since the scale is the same (12.5 mm per Angstrom) the CCS models may he used in conjunction with the CPK models which were designed especially to represent hioorganic compounds (10). As descrihed in detail helow. a c o m ~ l e xCCS skeletal model is easily constructed and requires few supports because the com~onentsare lieht and held rieidlv hv friction so the model does not collapse& become distorted.-d he skeletal model is easily converted to its space-fillingcounterpart hy the addition of color coded spheres made of polystyrene. The CCS models are very versatile and besides models of simple and complex molecules they may he used to make models of strained ring compounds. The original version (8) has been modified so that they now consist of color coded plastic Minit atom centers and color coded plastic connector tubes that fit over the prongs of the atom centers. Color coded plastic sleeves fit over the tuhes to help identify the second atom in a hond between two different atoms. color coded metal sleeves are used when rotation is to he prevented or when strain might cause a tube to slip off the prong of an atom
s ow ever,
230 1 Journal of Chemlcal Education
Figure 1. Cyclohexane. CCS model.
Figure 2. Cyclohexane. (ten) CPK
model. (right) CCS model with space fill-
ers.
Figure 3. Oleic acrd ester of cholesterol, space filling version. center. The metal sleeves are easily crimped with diagonal cutters to provide a permanent hond that cannot he rotated. The space fillers i r e light weight expanded polystyrene spheres that fit over the color coded tuhes representing externally honded atoms such as a hydrogen atom or a carbonyl group. Each space filler is also color coded to identify the skeletal atom and the Van der Wads' envelooe of the external atom to which it is honded. As illustrated hilow, CCS models are uniaue in providine a soace filline model in which the skeleton remains visible.' Figure 1shows the CCS model of cvclohexane in the chair conf&mation with its axial and equa;orial hydrogens. Addition of the color coded spheres converts the skeletal model to its space-filling version,-~i~ure 2b. The latter is very similar in appearance to the CPK model, Firmre 2a, and a remarkable feature of the new models is that the partly painted spheres
'
The photographs do not show the new sleeves which are now available, but their appearance is very similar.
Flgure 4. Oleic acid ester of cholesterol, skeletal version
Flpvre 7. Cyclobutanol and benzene with space fillers.
F l g v s 8. Cyciobutanal and benzene, skeletal models.
F l w e 5 . Space fillers an B hydrogens at C,, C.. C,,, and on Cqsmethyl
Flgwe 6. Olelc acid ester of cholesterol, llne drawing.
create the illusion of space-filling carbon atoms. This feature is illustrated again in Figure 3 which depicts a model of the oleic acid ester of cholesterol. The corresponding skeletal version is seen in Figure 4. In Figure 5, the C19 methyl group has been made space filling as well as the B hydrogens at CI, Cn,and CI1to demonstrate the close proximity of these axial hydrogens with the C19 methyl group (see Fig. 6 for line drawing). The atom centers and bonds may be bent so that cyclohutane and cyclopropane rings are made easily. Figure 7 shows the space-filling version of cyclobutanol beside a model of benzene and Figure 8 shows the corresponding skeletal versions. The new models are thus eminently suited for illustrating both the conformation and the overall shape of small molecules. Since they will he relatively inexpensive they are very useful for the comparison of series of related compounds each of which is left assemhled. With CPK models such c o m ~ a r i sons require an expensive inventory of component atoms. Furthermore, the CCS models were designed especially to fill the need for inexpensive and practical components that could be used to study the interactions of smaller molecules with larger hioorganic compounds such as enzymes. The advantages of space-filling models of enzymes have been well demonstrated by Harte and Rupley (11) and the use of dihedral angles (12,13) provides a practical construction method. Haas applied the method to Kendrew skeletal models (14), and it is now the method of choice for the plastic Nicholson models (6). Models of entire enzymes are very instructive but usually it is the active site region that is most interesting to study. Unless one has access to the complete model it is very difficult to know from the published literature which atoms are necessary to provide a useful model of only the active site. We have built the complete model of chymotrypsin and used it
Flgure 9. Chymotrypsin active site. CCS model.
in constructing a smaller skeletal (15) and space-fillingversion of the active site region (8). The model has been so instructive and useful to us that we felt a complete description of its construction would he of eeneral interest. Chymotrypsin (16) is &e of the sertne proteases. a family recentlv discussed hv Robert M. of rotei in-cuttine enzvmes ~ t i o u d(17). These enzymes cleave the amide bonds of proteins by attacking the amide carbonyl with a serine hydroxyl located a t the active site. The attack is made possible by a charge relay system which effectively transfers the negative charge of an aspartate anion to the serine hydroxyl. The attacked carbonyl is held in exactly the right position through hydrogen bonds and hydrophobic interactions and its change from a trigonal to a tetrahedral configuration is assisted by hydrogen bond formation (18-20). A very detailed description of chymotrypsin has been provided by J. J. Birktoft and D. M. Blow (21). They have supplied a complete set of coordinates and torsion (dihedral) angles as well as the location of the hydrogen bonds. For the active site model an appropriate selection was made of 37 of the 240 amino acid units of the comnlete enzvme. The coordinate set (21) was converted from ~ n ~ s t r o m scentimeters to and the oriein moved from the noint (X= 40cm: Y = 25cm: Z = 12.5 c& at the lower left rront corner of the active site region. The completed skeletal model, Figure 9, shows clearly all of the atomic positions and hydrogen bonds in the region of Volume 54, Number 4, AprN 1977 1 231
Table 1.
Components for Alpha Chymotrypsin Active Site Modal Connectors
Bonds
Tuber Short. 12.5 m m 58 red
C 4 C=O C=N
54 blue
C-H N-H
22 black Medium. 14.5 m m 162 white 44 white
D-H
14 wnlte
C--C C-N
98 black 40 blue
C=C
I
8 red Long. 21.7 m m 2 yellow 6 yellow
0
1
Figure 10. Chymotrypsin active site. CCS modal withrpaceflllers. CPK model of N-formyltryptophanin front.
Metal Sleeves
S--5 C-S
20 20 44 44
Total6
black red black blue
38 black 10 black
162 black 34 blue
10 blue
10 white 4 red 4 white 196 black 40 black 40 blue 16 red
10 red
4 yellow 6 yellow ~
8 red 516
0
Plastic Sleeves
-
-~
16 red 480
Table 2.
254
Minit Atom Centers
Designation
NO.
Code
black tetrahedron black trigonal 120" black trigonal 114' blue trigonal 114' red divalent red trigonal 120' Yellow divalent aluminum pins
Table 3.
Figure 1 1 . CPK model of N-formyltryptophan in CCS model ol chymotrypsin active site.
the active site a s well as the charge relay system and the nature, size, and shape of the specificity cavity. The space-filling version of the model, Figure 10, may he used in conjunction with CPK models of substrates and inhibitors. In Figure 10 a CPK model of the virtual suhstrate, N-formyl-L-tryptophan is placed in front and below the model of the active site. In Figure 11 this substrate has been placed in the active site so that the indole portion occupies the specificity cavity. Although the photographs do not show it, the actual model clearly demonstrates just how neatly the substrate fits into the snecificitv cavitv. ~ l ; ecomp;nents'for building this model will he supplied as a kit bv Srienre Related Materials Inc. The techniaue of enzyme model building with the new CCS components has been described (8.9). . . . Construction of the resent model requires about two days once the components have been assembled. The model is huilt on a ~ l v w o o dbase. (% X 14 X 14 in.), which may be painted greenor pale blue. Lists of tubes and atom centers required for the skeletal model are provided in Tables 1and 2. Tahle 3 provides the notation for the individual atoms of the amino acids that are found in the active site model. The number of the respective amino acid units required for the model is also provided in the table. The 37 amino acid units are distributed among five segments of the main chain as noted in Table 4. Tahle 4 also provides the torsion angles, phi and psi, about each of the alpha carbon atoms. These angles are defined as in the IUPAC convention (22). Phi is the angle hetween the vacant arm of the nitrogen atom center (N) and the CA-CO bond as one sights from N to CA. With the CO atom center in the back pointed too' on the right, a positive angle is formed by rotating the vacant arm of the nitrogen atom center counterclockwise. Psi is the angle he232 1 Journal of Chemical Education
Amino Acid Units of Active Site Fleaion
Number of Units
Abbr
6 1 3
gly ala ual
glycine alanine valine
CA CA-CB CA-CB;CDl
1
ile
iroleucine
b ~ 2 CA-CB-CG1-CDl
1 4 8 3
met
rer thr
methionine cyrtine revine threonine
CA-CB-CG-SD-CE CA-CB-SG CA-CB-OG CA-CB-OG1
2
hi6
histidine
CA-CB-CG---NDl---CE1
2
asp
aspartate
h-NE.4 CA-CB-CG-OD1
2
am
arparagine
\OD2 CA-CB-CG;ODl
1
Phe
Phenylalanine
ND2 CA-CB-CG-CDI-CEl-CZ
Amino Acid
Symbols for Atoms of Side Chains
'CG2
CYI
'CD2-CE2 / 1
tyr
tyrosine
CA-CB-CG-CD1-CEl-CZ-OEH
1
try
tryptophan
'CD2.CE2 CA-CB-CG---CD1---NEl
/
LD2.CE2
/
CE3 1
pro
prollne
/
\,Zl
'Czz-cEH N.............. CA
/
/
rwen rhe CO-N bond and the CA-N bond as one sighu from CO tuCA H'irh the nitrogennrom crnrer in the back pointed toO0on the right, a posuive angle is formed by rotating the CO-N bmd counterclockwise (see diagram in reference (8)). Each of the chain segments is involved in hydrogen bonds with
Table 4. Features of Alpha Chvmotwpdn Active Sits Model Torsion Angler
Amino ~ Acids Segment 1
40hir 41 phe 42 C Y I 43 glY
7 phi
Dsi
6 7
118 -33
-114 -146
-87
157 167
are plaeedon the alpha hydrogens of d. Then each individual amino acid unit is assembled and the tarsion angles are fixed as provided in Tahle 4. 2) The five main chain segments are assembled by formingtrans, planar smide bonds between successive amino acid units. At this point the shapes of the chains are checked against the maps in which the positions of the alpha carbon atoms are indicated. The hydmgen bonds within the segments, in parentheses in Tahle 4, are also constructed at this stage. For some of these, aluminum pins must he bent aoorooriateh. 3, The side chains are cundrurted next using Table 3 ta determine how many of each kind are required. Serines 190.221,and 223 are fitted witlh "Y" shaped oxygen atom centers. One of the oxygen stoms of aspartic acid 102 has a "Y" shaped atom center. Histidine 57 has a short blue tube to accept a hydrogen bond. 4) Segment 5 is the first to he pasitioned on its three supports an the hase board. Map 4 shows the proper orientation. When segment 5 is in nosition. the side chains of cvstine 220 and cvstine 191 are attached with ihe help of maps 3 and 4 . The alpha carbun atom of glycine 196 ofsegment 4 is thenattached m its support rod. C,wtine 191and rwtine 220are ioined via the disullidelrnk and a hydrogen bond is &structed between the amino hydrogen of glycinel96 A d the carhonyl of d i n e 213. The side chains are now attached to segments 4 and 5 and all of the interconnectinghydrogen bonds are formed as indicated in Tahle 4. 5) Segments 2 and 3 are then attached to their supports and linked together via the hydrogen bonds to aspartate 102.The disulfide link between eystine 58 and cystine 42 is constructed and segment 1 is attached via this link and histidine 40. 6) The remaining side chains are finally plaeed in wition with the beln of the resnective maos. . . and the hvdroeen . - bond interconnections provided in Tahle 4. It is often necessary to superimpose one map over another to see the relationship of a particular group in one segment to its neighbor on a different segment. Sometimes it appears that a chain is out of shape but the plastic tubes bend and the hydrogen bonds and support rods bring the groups to their proper orientation. 7) Finally, the spa= fdem are added. The hydrogen atoms tillmost of the space while the carhonyls and sulfur atoms add color to the model. The base is complimentaryand shows off the bright colors to advantage.
Dirulfide Links and Hydrogen Bonds 7
NH
7
Side Chain
CO
CO 195
sermen* 2
~~~~~~
Sesmenr 3 99 iie 100arn 101 arn
102arp
76
-64
57
-79
10 150 47
76
O D I - N D I 57 OD?-NH 57 OD2-OG 214
-.*. ..-... . 189 rer 190 rer
OG-CO
191 C Y I 192 met 193 giy 194 a6p 195 rer 196 gly Segment 5 213val 214 rer
215tr~
OD1 (CO
CO -76 --97 -154
116 -77
176
146
224 --
thr ~
225 Pro 226 gly 227va1 228 tyr
(CO 227)
-87
-114 4 4 -70 -81 -105 -133
9 -14 133 141 146
i36
NH
40 43
NH
196
NE2
213
INH-2271 .
INH-220)
-36
8 27
( N H 194)
191)
-162 141
216 g l i 217 1 8 7 218 rer 219 t h ~ 220 cvr
222 th, 223 rev
194
(co 217) 00 217
00
189
( N H 224)
220
OG-OEH 228 5 5 220
ODI-NH 191 0 G - N E 2 57
OG-002 102 OG-NH 221 5-5 191 OG-NH 223 00-00 223
OG 22 (CO 221)
(CO
2151
N H 214
OEH-OG
137
190
Hydrogen bonds wlthin segments are in Parentheses. a neiehhorine seement and. occasionallv. with itself. These interactions are n&; in Tahle 4 using the notation of Tahle 3. Bonds between atoms of the same segment are in parentheses since these are constructed first. It can be seen from Tahle 4 for instance, that the NH of NE2 of histidine 40 forms a hydrogen bond with the CO of glycine 193.There is a disulfide link between cystine 42 of segment 1and cystine 58 of seement 2. The amide NH of elvcine 43 -. forms a hydrogen bond with the CO of serine 195. Tahle 5 contains the coordinates in centimeters of the alpha carbon and side chain atoms (excludinghydrogens). The height of an individual atom ahove the hase is provided by the X cwrdinate and supported atoms are indicated in the table hy italics. The horizontal location is provided by the Y and Z coordinates. Four maps are prepared with Y and Z as ordinate and abscissa, respeetively. The maps are designed so that they fit easily an 18 X 25-cm eraoh naoer. In nrenarine them the lower left corner has theeoor-
~. .
.
~~
-.
~
map 3. and Y = 0, Z = O for map 4. When the poinuare located un the maps. the alpha carbon pwitionsare numbered and joined with atrmght lines and the locations of atoms in the arde chains are connected with straight lines. The location of the support rods on the hase is marked on s 30 X 30 em grid. Holes are drilled at these locations and rods of the indicated heights are placed into the holes with a tight fit. When the ehains are constructed,some of the support rodsmust he bent st the top so that the rod will fit into the hole of the Minit atom center as it is oriented in the model. With all of the components on hand, actual aasemhly of the active site model is relatively simple and straightforward. The fallowing steps are involved = 2 for
1) Individual amino acid romwnents c..d.. and e..reoresentine the . amino, alpha carbon, and rnrhonyl portions, respectively, uf the ammo acid units. are prepared as described in the Scientific A m r r i r o n article ( 8 ) The . numbers of the individual amino acids
~~
~~
~
~
.~.~ ~C
~~~
T h e completed model is colorful and can be enjoyed as modern art scul~ture.On the other hand. i t is an accurate representation of the active site region of an enzyme and is useful as a scientific tool. As David Rlow minted out recentlv (16) chymotrypsin is probably the enzyme whose mechanism is best understood a t the Dresent time. It is thus very iustructive to follow the dis&tssion in detail with a phisical model. In our model, all of the features of the active site region stand out clearlv. The narrow s~ecificitycavity is located just to the right and a little below-serine 195. he charge relay system comprising aspartic acid 102, histidine 57, and serine 195 is easily visible. This system has been descrihed as delivering the negative charge of the carhoxylic acid of aspartic acid 102 to the hydroxyl of serine 195 (20,23).This latter attacks the carbonyl group of a substrate amide bond to form a tetrahedral intermediate. Formation of the tetrahedron is assisted by t h e ~ ~ g r o u of p sserine 193and serine 195 which are seen to he in iwt the rieht mition. The carhonvl of serine 214 is easily avahable to 6 r m ' a hydrogen bond &th the amino mouo.of a substrate molecule. and the s~ecificitvcavitv " -is iust . the right size and location to hold a phenyl ring of phenylalmine or the indole nucleus of tryptophan. A space-fdling CPK model of N-formyl-L-tryptophan fits perfectly into the receptor site area as shown in Figure 11. Models of substrates and inhibitors can be constructed also with the CCS components and their interactions with the enzyme can be studied. Earl Frieden ~ o i n t e dout recently that the com~lexityof macromolecules~haskept biology a largely empirical sciince (24).However, with the use of the inexpensive model building components described ahove, it is now possible for the student to demonstrate and interpret the non-covalent interactions which Frieden describes in his article. In fact. the model of the active site region of alpha chymotrypsin illustrates the hy~
~~
~
~~
~~
Volume 54, Number 4, April 1977 1 233
Table 5.
X
Y
Coordinates for Alpha Chymotryprin Active Site Modal"
z
Y
X
z
X
Segment I 4 0 higtidine
-
~
CA
28.4
Segment 3 99 iroleuclne
4 1 phenyialanine CA 31.4 CB 31.5 CG 32.1 CD1 30.9 CEl 31.4 CZ 33.0 CE2 34.3 CD2 33.8
CA
11.4
CB CGl CDI CG2
11.9 11.5 12.2 13.5
Map 56 alanine
CA
21.8
CB 21.0 57 hirtadine
59 glycine CA 6 0 vallno CA CB CG1 CG2
26.7 30.5 30.5 32.1 29.8
CA CB CG CDl CEl CZ CE2 CD2 OEH
Segment 56 226 plycine CA 10.4
12.9 13.4 13.6 12.4 12.7 14.3 15.7 15.3 14.5
Map 3
Segment 4 189 Ierine
192 methlonlne
194 aspartate
1 9 1 cyrtine
Moo 4 Segment 5. 213 valine CA CB CGl CG2 214 rerine
18.0 19.0 18.8 19.8
215 tryptophan CA 13.7 CB 12.6 CG 11.0 CD1 10.9 NEl 9.3 CE2 9.6 CZZ 6.5 CEH2 5.9 CZ3 7.1 CE3 8.2 CD2 9.4
a Supported atoms are itallcirea. There are C A of 56,99. 196. 220. 227 and CG of 225.
2 2 1 Serine CA 10.6 CB 8.9 OG 7.7 222 threonine CA 10.3 CB 11.1 OG1 10.5 OG2 13.0 223 Ierine
Y
z
drophobic and hydrogen bond interactions such as he describes which are res~onsihlefor the soecificitv of cbvmotrypsin. As Frieden points out. the interaction of a hormone with its receptor iniolves similar non-covalent interactions. New concepts are not reauired for the medicinal chemist to describ; the interaction of a drug with its receptor. Model building studies such as those described above are already assisting chemists to discover new medicinal agents.
Ill Diekemn. Richard E..and Gek, Iruing."Chemistry. Matterand the Univene: W. A. Benjamin, Ine.,Reading.Maus., 1975. F i e r . Louis F.. J. CHEM. EDUC.. 42,408 119651. Walton. Anne, Plopreas in Ster~orhernialry.4.335 (19681. Gc,rdon, Arnold J., J. CHEM. EDUC.. 47.30 (19701. Petersen,Quentin R.,J.CHEM. EDUC.,47,24 119701. Brochure: "Nichalson Molecular Mcdols: Labqmp, 18 R~uahlllPark Estate,Caversham, Reading RG4 8 XE. England. 17) Rupley, J.A..and Bruice, J. C.. J. Mol. Aiol.. 243.521 11968) 181 Stvnn.C. L..Sci. Arne?.. 234.124 119761. (2) 13) 1.0 15) 16)
Acknowledgment
The author is grateful to Drs. Louis J. Ignarro, Dr. Norbert Gruenfeld, Mr. Walter Munch, and especiallv Dr. Jeffrev W. H. watthey for their assistance in the con&uction of the original versions of the model described in this paper. The photographs of the models were kindly provided by Mr. Craig Cooper. The author is also indebted to Dr. David M. Blow of the Medical Research Council Laboratory of Molecular Biology, Hills Road, Cambridge, England for supplying a list of the torsion angles of chymotrypsin in advance of publication.
. . .
. .
(15) Clarke, Rank H.. and Watthey, Jeffrey W. H., Ahst.. 13th National Medicinal Chemistry S y m p ~ i u m . T h eUniversity of 1ows.Imw City, 10wa, Juns 1VZ.L972, p. 51. 116) Blow.David M.,Accntr. Chem. Raa., 9,145 lL976l. (171 SLroud, Roherf M., Sei.Amer..231.74 119741. 118) Steifz,T.A.. Honderaon, R.,and B1uu.D. M., J. M