Molecular symmetry of the MoFe protein of nitrogenase. Structural homology/nitrogen fixation/x-ray crystallography.

X-ray diffraction data to 2.4-A resolution have been collected for native monoclinic crystals of the MoFe protein of nitrogenase from Clostridium pasteurianum. The MoFe protein is an alpha 2 beta 2 tetramer of 220,000 molecular weight with 1 molecule in the crystallographic asymmetric unit. A 6-A resolution rotation function shows the orientation of the crystallographic diad and pseudo mutually perpendicular diads representing 2-fold relationships between alpha and beta chains. Hence, at least at low resolution, there exists structural homology between these two polypeptide chains.

X-ray diffraction data to 2.4-A resolution have been collected for native monoclinic crystals of the MoFe protein of nitrogenase from Clostridiumpasteurianum. The MoFe protein is an a2/32 tetramer of 220,000 molecular weight with 1 molecule in the crystallographic asymmetric unit. A 6-A resolution rotation function shows the orientation of the crystallographic diad and pseudo mutually perpendicular diads representing 2fold relationships between a and /3 chains. Hence, at least at low resolution, there exists structural homology between these two polypeptide chains.
A class of bacteria exists which is capable of reducing atmospheric dinitrogen (N2) to ammonia. The enzyme which is common to all nitrogen-fiiing bacteria and plays the central role in the reduction of dinitrogen is nitrogenase. The reaction catalyzed by nitrogenase is where n is between 12 and 24 (1).
During catalysis, nitrogenase is composed of two proteins, the Fe protein and the MoFe protein. The molecular weight of the Fe protein is 59,700 for Clostridium pasteurianum. It is composed of two identical subunits and contains one FezS: center. The MoFe protein has a molecular weight of 220,000, contains two a and two /3 subunits, two Mo atoms, 28-32 Fe atoms, and 24-32 acid-labile sulfurs. The molecular weight of the a and /3 subunits is 60,000 and 50,000, respectively (2). The MoFe protein contains an extrudable MoFe cofartor, which has a metal to sulfur ratio of lMo:8Fe:6S (3).
In order to understand the mechanism of catalysis and its relation to structure, the three-dimensional crystal structure determination of the MoFe protein of nitrogenase from C. pasteurianum has been initiated. The crystals have the symmetry of a P2, space group with cell dimensions a = 69.99 A, b = 151.05 A, c = 122.04 A and /3 = l l O o 20' and with 1 molecule in the asymmetric unit (4). Native data were used to determine the orientation of the molecule within the cell and to search for a possible relationship between the a and / 3 subunits.
* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
f Present address: Department of Applied Chemistry, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464, Japan. To whom comespondence should be addressed.

EXPERIMENTAL PROCEDURES
Native data have been collected by oscillation photography using focused (5) CuKa radiation from an Elliot rotating anode X-ray generator. In order to assure that the Oz-sensitive MoFe protein crystals would remain in an anaerobic environment, mounting procedures were developed which excluded atmospheric oxygen from the mounted crystals (4, 6). The crystals were kept between 14 and 18 "C during data collection.
The crystal to film distance was 75 mm, permitting the recording of data to 2.4-A resolution. A complete set of data was collected rotating the crystals through a total angle of 9 0 ' about the a*-axis.
The crystals were oscillated through 1.25" for each film exposure and with an overlap of 0.25" between films. A total of 90 fim packs were collected from 22 different crystals.
The f h s were processed on an Optronics Pl OOO film scanner using a 1OO-pm raster step size and analyzed by techniques reported elsewhere (7,8). The final scaling included all reflections with intensities greater than 1 standard deviation. The overall R-factor, defined as where ZhZ is the intensity of reflection h on film i, f h is the mean of measurements for reflection h, and k, is the scale factor for reflections TABLE I Error in intensity measurements Internal agreement refers to error computed from differences of specific intensity measurement with respect to their mean. Counting statistics refer to the error estimate based on noise level on the films with respect to a mean background and the best fit of a profde. Precise definitiop for internal agreement and counting statistics were given in Ref. 8   Overloads refers to reflections which contained optical densities greater than an arbitrarily selected limit of around 2.  Mean background for resolution range 25-9 A was set at 0.0; root mean square deviation from mean was 7.5.

MoFe Protein Molecular Symmetry
on fim i, was 8.0% when intensities greater than 20 were included using full reflections and partial reflections greater than 0.5. Omitting all partial reflections, the overall R-factor decreased to 7.7%. A total of 92,505 full and partial reflections were measured and then reduced to 59,651 independent reflections. No account was taken of anomalous dispersion. Postrefinement (8,9) gave the cell dimensions stated above. An>nalysis of the error in the data as a function of the mean intensity, I, is given in Table I, and an estimate of the completeness of the data in 9 resolution ranges is given in Table 11. Rotation Function Calculations-The rotation function (10) compares the orientation of a molecule or a part of a molecule with another similar molecule or part thereof. A peak in the function implies similarity of structure either within a molecule or between molecules in an asymmetric unit of the same or a different crystal. Ir this molecule contains similar subunits, the rotation function gives a peak which determines the Orientation and rotational operation required to relate one subunit to the other. The orientation direction is given by a line joining the center of the sphere to a point on its surface defined by latitude (4) and longitude (4) relative to the orthogonalized crystal axes. The amount of rotation about this line is given by K . Hence, a search of the 4, + surface when K = 180" represents a search for molecular diads.
Self-rotation functions were calculated to determine the orientation of the molecular diad in the crystallographic asymmetric unit. The K = 180" plane was explored within the lo-A to 6-A resolution range.
Although a number of different conditions were used, the best results ( Fig. 1) were obtained when the radius of integration was set to 55 A, and 565 of the largest intensities were used to represent the second Patterson. The highest peak in the rotation function with 100.0 arbitrary units was the origin peak, the mean background level was set to 0.0 units, and the root mean square deviation from the mean was at 2.7 units (Table 111). The largest non-origin peak had a height of 27.0 representing 10.0 standard deviations above background. The largest noninterpretable peak was 1.9 standard deviations above background.
The second highest peak had a value of 12 at 4.4 standard deviations above background. Interestingly, it was orthogonal to the largest peak. A search for a peak orthogonal to the first two revealed a rather small peak only slightly above background (Fig. 1). The assumption was made that the major peak represented a molecular 2-fold axis (Table 111) between two aP dimers, whereas the smaller orthogonal peaks represented 2-fold axes between an (YZ and a P2 dimer. If the inferred homology between subunits was only approximate, the smaller orthogonal peaks should increase relative to the large peak when the resolution is decreased. Indeed, this was observed for peak 3 but not for the small general peak 5. The larger size of the pseudo diad on the equator is no doubt due to its special position on a mirror plane in the rotation function.

RESULTS AND DISCUSSION
The rotation function has shown the orientation of the molecular dimer axis between ap subunit pairs to be inclined at an angle of 20" to the b-axis in an orientation which projects onto the c-axis in the ac plane. The secondary peaks reveal a low resolution homologv between the a and / 3 subunits. shown. The X marks the position of peak 5 (Table 111).
. . ., arbitrary peak height units (Table 111). The a * , b, and c-axes are nitrogenase were not previously known to have any homology. The amino acid sequence of the a chain, as derived from the corresponding nucleotide sequence (11, 12) and the partial protein sequence, representing 20% of the total chains of the a and p subunits (13), had previously been determined. However, no obvious sequence homology between the a and / 3 subunits was observed among the various peptides at this stage. Homology of structure as shown by the current results does not necessarily imply observable homology of amino acid sequence, since the conformation of polypeptide folds are almost invariably more conserved than sequence (14).
The data collected here, and the information about the noncrystallographic symmetry, is essential for further progress in the structure determination of this enzyme. The molecular diad will be an aid in locating heavy atom sites and extremely useful for improving phases by use of molecular replacement averaging (15).