The cell envelope of Synechococcus sp. WH8113 has multiple layers (Fig. 1). Proceeding from inside, the cell envelope consists of cell membrane (∼ 10 nm thick), peptidoglycan layer (∼ 15 nm thick), outer membrane (∼ 10 nm thick), and surface layer (∼ 35 nm thick). The surface layer evident in Fig. 1 is not evident in conventional, chemically-fixed preparations (data not shown, see Ref. 16). Chemical fixation appears to cause detachment of the surface layer from the outer membrane whereas cryopreservation and freeze-substitution (data not shown) retain this layer.

Convex fractures of the inner cell membrane exposing its inner leaflet (Fig. 2a) as well as concave fractures exposing its outer leaflet (Fig. 2b) are densely covered by ball-shaped intramembrane particles that likely represent a panoply of membrane proteins and channels. These particles range in diameter from 5 nm to 13 nm. There is no obvious pattern in their distribution on either face, although a subset may represent the roots of the spicules described below. Pits complementary to the intramembrane particles on both convex and concave fractures of the cell membrane were infrequent [12].

Convex fractures of the outer membrane exposing its inner leaflet (Fig. 3) are covered by pits. The pits range in diameter from 3 to 18 nm. The smaller pits occur in greater density than the larger pits. There is no obvious pattern in the distribution of pits. Concave fractures of the outer membrane exposing its outer leaflet appear rough and uneven, lacking clear structures like particles or pits (Fig. 2b).

A surface layer uniformly covers the outer membrane. In cross section, the surface layer appears as cylindrical pillars of diameter ∼ 8 nm that extend radially ∼ 18 nm from the outer leaflet of the outer membrane (Figs. 1, 2). The pillars are separated by gaps of ∼ 4 nm that are revealed by freeze-etch, indicating that they were formerly filled with water. Beyond ∼ 18 nm, the adjacent pillars connect laterally forming a fused region ∼ 17 nm thick. In cross sections of the cell envelope, this region sometimes appears as a congealed layer (see Fig. 1) but sometimes appears as a row of balls (see upper-left quadrant of Fig. 2a). This variability may reflect cleavages along different axes of the surface layer, or simply different degrees of plastic deformation and damage during fracture. The surface layer is a crystalline lattice, as is evident in aerial views. Such views reveal balls packed in a rhomboid organization with spacing ∼ 12 nm and obtuse angle ∼ 110° (Fig. 4a, 4b). These balls have the same size and spacing as those balls identified in cross-sections.

Spicules arise in profusion over the surface of the bacterium and extend up to 150 nm into the surrounding medium. The smallest observed separations between adjacent spicules are 12 nm and 24 nm, which corresponds to the spacing of the surface lattice. Spicules have uniform thickness of ∼ 5 nm along their length. Spicules are flexible; we have measured curvatures with radii as small as 30 nm.

It is difficult to trace the contour of a spicule through the entire cell envelope. Convex fractures of the outer membrane are characterized by a quantity of spicules projecting toward outer layers (Fig. 3). We frequently find spicules protruding outward between adjacent balls in cross-sections of the surface layer (arrow, Fig. 2a; arrowhead, Fig. 4a). We also frequently find fibers of 5 nm thickness that arise from the cell membrane before entering higher layers that could represent the lower stems of spicules (arrow, Fig 2a). These facts taken together suggest that the spicules extend outward through channels in the surface layer and extend inward through underlying layers to contact the cell membrane.

Pitta et al. [11] found that depriving Synechococcus of calcium inhibited motility and that the addition of calcium chelators EDTA or EGTA in excess of the initial concentration of calcium irreversibly arrested motility. We found that resuspension of Synechococcus in calcium-free seawater supplemented with 10 mM EDTA severely disrupted both the surface layer and internal thylakoid membranes. The surface layer lost its regularity, and would sometimes fall off entirely. Thylakoid layers that are ordinarily tightly apposed (see Fig. 1) separate by as much as 50 nm. Such severe damage (data not shown) could explain the abolition of motility by any of multiple causes, including the removal of the protein SwmA, an outer membrane protein which has been implicated in motility. Similar EDTA treatment is used to isolate SwmA from motile Synechococcus [3].

In the absence of flagella, the cell surface of Synechococcus sp. WH8113 must generate thrust. Theoretical work suggests that a cyanobacterium could swim at the observed speeds by oscillating its surface in the manner of a traveling wave [6,13]. Whereas motility is powered by an ion-motive force across the cell membrane [17] and the protein SwmA that is required for propulsion is associated with the outer membrane [3], the traveling wave must be formed at the very exterior of the cell body, outside the crystalline surface layer. Since the spicules contact the cell and outer membranes and protrude into the surrounding medium, they could be utilized to transmit motions in the cell membrane or outer membrane into the surrounding medium. If motors concealed in the cell membrane or outer membrane are capable of contracting and expanding in a regular manner (perhaps in analogy to the motor protein embedded in the basolateral membrane of cochlear outer hair cells that contract and expand to set the length of the cells [5]), these oscillations would be transduced into a rowing motion of the spicules. In this case, the crystalline surface layer could serve as oarlocks for the spicules, converting smaller motions at the bases of spicules into larger motions at the tips.

The motile mechanism of Synechococcus sp. WH8113 may be related to that of bacteria that glide over surfaces. Type IV pili (Tfp) have been implicated in the gliding movement of a species of the cyanobacterium Synechocystis [2], social gliding of Myxococcus, and twitching motility of Pseudomonas aeruginosa [14]. The spicules of Synechococcus sp. WH8113 could be related to Tfp, but comparison reveals differences. Tfp are generally much longer (1-2 μm), and stiffer than the spicules of Synechococcus. Tfp also tend to form bundles, playing roles in bacterial aggregation or attachment to surfaces. The spicules of Synechococcus do not tend to attach to one another, nor do cell bodies of Synechococcus tend to aggregate.

The motility apparatus of Synechococcus sp. WH8113 may be related to that of filamentous gliding cyanobacteria [6]. The cell envelopes of the gliding Oscillatoriaceae cyanobacterial species do have crystalline surface layers that resemble the outermost layer of Synechococcus, but the crystalline surface layer of Synechococcus is considerably thicker (∼ 35 nm thick compared to ∼ 20 nm thick). Structures like the spicules of Synechococcus have not been described in Oscillatoriaceae. Arrangements of helical fibers with diameters of 8-12 nm lie flat along the crystalline surface layer of Oscillatoriaceae. No such fibers are evident on Synechococcus. The surface of Oscillatoriaceae continuously extrudes slime while moving, whereas the surface of Synechococcus does not.

Synechococcus sp. WH8113 is but one of a large number of closely related species of Synechococcus, both motile and non-motile, that have been isolated. Comparative analysis of a selection of species could illuminate the motile mechanism of the cell envelope and the role of the structures described here.

Inactivation of the abundant outer membrane protein SwmA abolishes swimming motility [3]. However, when either motile wild-type or immotile SwmA cells collide with and become attached to a cover slip, cells will rotate about the point of attachment at rates ∼ 1 Hz [4]. This suggests that part, but not all, of the motility apparatus is disrupted in the absence of SwmA.

We uncovered no clear structural flaw in the mutant SwmA that would conclusively explain its motility phenotype or specify a role in motility for the structures described here. No images of spicules extending into the surrounding medium analogous to Fig. 2a have been obtained for SwmA. However, fibers crossing from the cell membrane to higher layers that we interpret as the lower stems of spicules are evident for SwmA, and spicules are evident on some convex fractures of the outer membrane of SwmA (data not shown). It remains possible that the cell envelope of the SwmA mutant has subtle flaws consistent with a partial disturbance of its motility apparatus.

The organelle that drives the motility of swimming cyanobacteria has not yet been identified. The spicules identified here could comprise part of the motility apparatus as they extend to both the site of energy transduction (the cell membrane) and the site of mechanical work (external to the surface layer). In any case, our elucidation of the surface structure (summarized in Fig. 5) provides new constraints on theoretical models for motility.