haracterization of Crystals of Type 5 Adenovirus Hexon

GARY ~ORKICK,~AUL~.~IGLER

Department 6f Biophysks University of Chicago

Chicago, Ill. 60637, U. S.A.

AND

HARQLD S. GINSRERG

Department of Microbiology, The School of Medicine University of Pennsylvania

Philadelphia, Pa. 19104, U.S.A.

(Received 4 November 1970)

Tetrahedral crystals of type 5 adenovirus hexon have been studied by x-ray diffraction.Thecrystalsbelongto thecubicspacegroupP2,3witha=bb=c=149&. The protein in the asymmetric unit probably represents one-third of a hexon. These crystals differ from the description of the tetrahedral crystals of type 6 adenovirus hexon reported by Macintyre, Pereira &: Russell (1969) but are isomorphous with the bipyramidal-shaped crystals of type 2 adenovirus hexon studied by Franklin, Pettersson, Akervall, Strandberg & Philipson (1971j.

1. Introduction

A detailed analysis of the molecular structure of t,he viral coat components is of fundamental importance in understanding the basis of their assembly. Although crystalline viruses have been studied by X-ray diffraction, and indeed give excellent diffraction patterns, the size of the unit cells and the mass of the crystallographic subunits involved are so large as to present a substantial barrier to the study of the components at atomic resolution. On the other hand, crystals of the purified eapsid protein present a technically feasible system with which to study the structure of the coat components at atomic resolution. Some of the molecular interactions of capsid assembly may well be reflected in the molecular packing of the crystal. For example, the crystals of tobacco mosaic virus contain rings of 17 subunits which almost cer- tainly reflect the side-by-side packing of subunits in one shallow helical turn of the virus (Finch, Leberman, Yu-Shang & Klug, 1966).

In t,his study we report the X-ray diffraction properties of crystals of the hexon of type 5 adenovirus. Hexons are defined as the morphological subunits of the viral cspsid which have six nearest neighbors on the surface of the capsid and comprise the bulk of the icosahedral capsid of an adenovirus (Ginsberg, Pereira, Valentine & Wilcox, 1966). The remaining capsomers are the 12 pent,ons which have five nearest, neighbors and are centered on the icosahedral vertices of the virion. The hexon crystals produce clear diffraction maxima at 243 A resolution and therefore represent a

397

398 G. CORNICK, P. B. SIGLER AND H. S. GINSBERG

potential source of structural information at atomic resolution (Plate III), Results similar to ours for type 2 hexonst are reported by Franklin et al. (1971).

2. Materials and Methods The hexon antigen was purified as described previously (Velioer 8.~ Ginsberg, 1970).

The hexon was crystallized by a modification of the method of Pereira, Valentine & Russell (1968). Purified hexon was dialyzed at 4’C against 1000 to 2000 vol. of 0.01 M-phosphate buffer, pH 7.8, containing 0.001 M-EDTA. The protein solution was concen- trated to 10 to 15 mg/ml. and dialyzed at 4°C against 30 to 50 vol. of 0-S M-KH,PO, and 0.001 M-EDTA at pH 4.3. Dialysis was done without stirring and the fluid was changed daily; crystals began appearing within 1 day and increased in size over a period of 3 to 5 days. Approximately regular tetrahedral crystals were produced in the final stages of dialysis and typically measured 0.5 mm along the edge (Plate I). The crystals were sta- bilized in a protein-free solution of 0.8 M-KHsPO, , 0.001 M-EDTA, pH 4.3.

CrystaIs for X-ray diffraction were mounted and sealed in thin-walled quartz capillaries containing some of the stabilizing supernatant solution described above. Diffraction patterns were photographed at a specimen-to-film distance of 100 mm using a Beurger precession camera. A plane graphite monochromator, designed and fabricated by M. Navia & 5. Hanacek (manuscript in preparation), was used to provide well-collimated Culz~l radiation from the spot-focus of a Jarrel-Ash microfocus X-ray generator operated at an anode voltage of 38 kv and a beam current of 6 mA.

The photograph used for the measurement of t,he size of t,he unit cell was calibrated with the 111 line of a sample of powdered NaCI.

Crystal densities were determined in xylene-carbon tetrachloride and xylene-bromo- benzene density gradients which were approximately 23 ml. in volume, approximately 11 cm in height and generally spanned a density range of 1.1 to 1.3 g/cm3. These nominally linear gradients were calibrated with droplets of KHsPO, solutions of known density and were found to vary smoothly in the range of the crystal density. Crystals at least 0.5 mm on a side were allowed to equilibrate with the protein-free supernatant solution for at least 24 hr before making density measurements. Crystals in a minimum amount of supernatant solution were rinsed onto small discs of Whatman no. 3 filter paper, then quickly flicked into the gradient with a glass fiber. Densities of the calibrating solutions were determined pycnometrically.

3. Results

The characteristics of type 5 adenovirus hexon crystals are summarized in Table I. The crystals belong to the cubic space group P2,3 which has 12 equivalent positions.

The diffraction patterns shown in Plate II allowed the point group of the crystal to be assigned unambiguously. Since proteins prohibit mirror symmetry, t’he possible point groups were limited to the eleven purely rotational groups. The zero-level diffraction pattern shown on Plate II(a) has only a g-fold axis of symmetry which implies the presence of a 3-fold or a 6-fold crystallographic axis with no 2-fold axes normal to it. The patterns shown on Plate II(b),(c) and (d) all show plane symmetry 2mm which implies the presence of a-fold axes. The cubic point group 23 is the only group consistent with these observations.

Of t,he five space groups consistent with point group 23, the groups 123, I2,3 and J’23 were unequivocally eliminated since the systematic absences which arise from

t Type 2 and type 5 adenovirus are distinguished immunologically (Schlesinger, 1969). Electron microscopy of the virus particles and their respective hexons (Pettersson, Philipson & HGgland, 1967; Ginsberg et al., 1966), as well as amino-acid compositions (Pettersson et aE., 1967; H. S. Ginsberg C R. S&era, unpublished results) reveal. no significant differences between the two serotypes, :- ;:. ., .

E

.- 0 3 4 R

P ‘LATE 111. Diffraction pattern of type 5 adenovirus hexon crystal phutographod without layer- line screen wi ith a precession angle of 30’ and a specimen-to-film distance of 75 mm.

CRYSTALS OF TYPE 5 ADENOVIRUS HEXON

TABLE 1

Crystallographic data jar type 5 adenovirus hexon crystals

Edge of the unit cell 149.9hO.l A Crystal .density 1~173~0~004 g/cm3 Density of protein-free

supernatant solution 1*070&0*002 g/cm3 Mass of protein in the

asymmetric unit? 83,000-&8000 daltons Volume fraction of protein? 0.37&0.03 Crystal volume per unit

of protein masst 3.37hO.29 A3/dalton

7 A.ssuming a partial specific volume of 0.740&0.004 am3/g. See Note added in Proof.

centering were not found. Specifically, spots for which h + k + 1 # 2% and h + k f 2% were observed on adjacent layers which required no displacement to cause coinai- dence of lattice nets. This may be seen by inspection of the (hk0) plane (Plate II(b)) and the (hkl) (Plate II(c)). These photographs were taken consecutively from t,he same crystal. Spots along the h and k axes of the (hk0) plane and along the h axis of the (hkk) plane (Plate II(d)) app ear only at even values of these indices. Such systemat#ic absences in a primitive lattice implies that the 2-fold axes are in fact 2, screw axes. Thus the hexon crystal was assigned space group P2,3.

Crystal densities measured in both xylene-carbon tetrachloride and xylene-bromo- benzene gradients were 1.173 f0.004 g/cm3 and 1.172 &0*004 g/cm3 respectively. Using a conventional value of O-740 iO.004 cm”/g for the partial specific volume of crystalline protein (Matthews, 1968), we calculate the mass of protein in the asy.m- metric unit to be 83,000&8000 daltons and the volume fraction of protein to be 0.37 *0.03. The crystal volume per unit of protein mass, V,, is 3.37 iO.29 A3/dalton; which is considerably larger than for most crystalline proteins but well within the range observed for most crystalline icosahedral viruses (Matthews, 1968). Using a hydrodynamically determined value of O-727 cm3/g for the partial specific volume of the hexon of type 2 adenovirus (R. M. Franklin, personal communication), the mass of protein in the asymmetric unit is 78,000 17000 daltons, the volume fraction of protein is 0.34 10.03 and the crystal volume per unit of protein ma.ss is 3.59 $0.25. Both sets of figures for the mass of protein in the asymmetric unit and the volume fraction of protein are considerably less than those determined by Franklin et al. (1971) for an apparently isomorphous crystal of type 2 adenovirus hexon. (See Note added in Proof.:;

As a check of the method, the mass of the protein in the asymmetric unit of tetra- gonad y-ohymotrypsin was calculated from the crystal and solvent densities measured as described above, the published unit cell dimensions (Sigler et al., 1964), and an assumed partial specific volume of 0.74 cm3/g. The value obtained was 23,900 daltons which is 5.5 ‘A less than the known molecular weight of 25,300 daltons determined from the amino-acid sequence (Hartley, 1964).

4. Discussion and Conclusions (a) Structural relationship between type 2 and type 5 hexons

The space group and unit cell size of the type 5 hexon crystal and the type d hexon crystal (Franklin et al., 1971) are identical. Diffraction patterns of the two crystals are sufEciently similar to conclude that the crystal structure of type 2 and

2%

400 G. CORNICK, P. B. SIGLER AND H. S. GINSBERG

type 5 hexons are isomorphous. We are unable to explain the differences in mass of protein in the asymmetric unit and volume fraction of protein calculated for the two hexon crystals. It should be borne in mind that the density of the supernatant solution, no matter how accurately measured, is merely an approximation to the actual density of the interstitial liquid of the crystal and may give a false value for the mass of protein in the asymmetric unit.

Macintyre, Pereira & Russell (1969) have reported a tetrahedral crystal of type 5 hexon grown by dialysis against 0.5 M-KH,PO,. A body-centered space group, I212,3 or 123, was assigned on the basis of systematic absences but the published diffraction photograph was so poorly reproduced that these absences were simply not discernible. The length of the edge of the unit cell was found by Macintyre et al. (1969) to be 213 A which is remarkably close to d2 x 149.9 A, the length of the face diagonal of the P2,3 unit cell. The asymmetric unit was calculated to be one-half of the hexon.

(b) Relationship between space group symmetry and point symmetry of the hexon

The mass of the asymmetric unit is significantly less than any estimate of the mass of the hexon. These estimates range from 200,000 to 400,000 daltons (Franklin et al. (1971) give an exhaustive listing). Given the fact t’hat the 3-fold, axes are the only rotational (i.e. non-screw) axes of P2,3, the observation that the hexon has six nearest neighbors on the surface of the capsid and the assumption that the hexon in the crystal is intact, we suggest that the asymmetric unit is one-third of the hexon. In this case, the point group of the hexon would be 3 or a super-group thereof with the 3-fold or the higher order n.on-crystallographic axis colinear with a crystallographic 3-fold. Although there may be more than three protomers in the crystalline hexon (i.e. more than one protomer in the crystallographic unit), the point group symmetry of the hexon need not necessarily be higher than 3. This follows the fact that the set of operations relating any one protomer to all other protomers within the hexon need not have the properties of a group.

Although it is possible that tetrahedrons of Macintyre et a2. (1969) may represent a different crystal form than our crystals, we believe that speculations about possible higher point symmetry of the hexon required by their inference of possible 2-fold axes must be postponed until a more compelling exposition of the crystallographic data and space group assignment is made for their crystals.

We thank Dr Richard Franklin and his co-workers for the opportunity to review a draft of their manuscript before publication and Manuel Navia for his help in the operation of the monochromator.

This work was supported by grant no. GM 15225 to two oftheauthors (P. B. S. and G. C.) and by Public Health Service grant no. Al-95731 to the other author (H. S. G.) who was also supported by the U.S. Army Medical Research and Development Command, Depart- ment of the Army, under research contract DA-49-193-MD-2131 under the sponsorship of the Commission on Acute Respiratory Diseases of the Armed Forces Epidemiological Board. One of us (P. B. S.) is a Public Health Service Career Development Awardee and another of us (G. C.) is a predoctoral trainee of GM 780.

REFERENCES Finch, J. T., Leberman, R., Yu-Shang, C. & Klug, A. (1966). Nutwre, 212, 349. Franklin, R. M., Pettersson, V., Akervall, K., Strandberg, B. & Philipson, L. (1971). J.

Mol. Biol. 57, 376. Ginsberg, H. S., Psreira, H. G., Valentine, R. C. & Wilcox, TV. C. (1966). Virology3 28,

782.

CRYSTALS OF TYPE 5 ADENOVIRUS HEXON 401

Uiartley, B. S. (1964). Nature, 201, 1284. Nacintyre, W. &I., Per&a, H. G. & Russell, W. C. (1969). Nature, 222, 1165. Matthews, B. W. (1968). J. MOE. Biol. 33, 491. Pereire, II. G., Valentine, R. C. & Russell, W. C. (1968). Nature, 219, 946. Pettersson, V., Philipson, L. & Hogland, S. (1967). Virology, 33, 575. Schlesinger, R. W. (1969). Advancesin Virus Research, ed. by K. M. Smith & 51. A. Lauffer,

vol. 14, p. 1. New York: Academic Press. Sigler, P. B., Skinner, H. C. W., Coulter, C. L., Kallos, J., Braxton, II< & Davies, D. R.

(1964). Proc. Nat. Acad. Xci., Wash. 51, 1146. Velicer, L. F. & Ginsberg, H. S. (1970). J. Viral. 5, 338.

*Note added ,in proof: In an effort to understand the disparity between the protein content of the asymmetric unit reported by Fre&lin et al. (1971) and ourselves, we have measured the density of 13 large type-5 hexon crystals previously equilibrated with 0.5 M citrate buffer, pH 3.2 (density 1.050 & O-002 g/cm”) and shown to be well ordered by X.-ray. Bromobenzene-xylene gradients indicated 93,000 + 8000 daltons of protein per asymmetric unit. Two large, well blotted crystals were found to be isopyncnic with a dry bromobenzene-xylene mixture of density 1.187 & 0.002 g/cm3 indicatiig 104,000 & 9000 d&tons of protein per asymmetric unit. We feel this represents an upper limit essuming the partial specific volume is 0.740 * 0.004 cm3/g.