A dynamic basal complex modulates mammalian sperm movement

Reproductive success depends on efficient sperm movement driven by axonemal dynein-mediated microtubule sliding. Models predict sliding at the base of the tail – the centriole – but such sliding has never been observed. Centrioles are ancient organelles with a conserved architecture; their rigidity is thought to restrict microtubule sliding. Here, we show that, in mammalian sperm, the atypical distal centriole (DC) and its surrounding atypical pericentriolar matrix form a dynamic basal complex (DBC) that facilitates a cascade of internal sliding deformations, coupling tail beating with asymmetric head kinking. During asymmetric tail beating, the DC’s right side and its surroundings slide ~300 nm rostrally relative to the left side. The deformation throughout the DBC is transmitted to the head-tail junction; thus, the head tilts to the left, generating a kinking motion. These findings suggest that the DBC evolved as a dynamic linker coupling sperm head and tail into a single self-coordinated system.


Main Text
The sperm cell consists of a head and a tail linked by a neck that regulates sperm movement, as predicted by the basal sliding hypothesis 4,12 . The neck contains two centrioles (proximal centriole, PC, and distal centriole, DC, see Supplemental Glossary for term de nitions) (Fig. 1a) 13 . In most eukaryotes, the PC is a canonical cylindrical centriole 9,14 . However, in mammals, the DC, at the base of the agellum, is structurally atypical 15 . Since, similar to canonical centrioles, the atypical centriole functions postfertilization, the reason for its atypical structure remains unknown [16][17][18] . The PC and DC are embedded in a specialized mass of atypical pericentriolar material: the segmented columns (SCs) and the capitulum.
Distally, the SCs are continuous with outer dense bers associated with the microtubule doublets of the axoneme. Rostrally, the capitulum connects to the nuclear basal plate, forming the implantation fossa in the head-tail junction (Fig. 1a). How this multi-component assembly supports sperm movement is unclear, but it is usually modeled as a rigid structure that anchors the tail rmly, like a clamp, to the head, with little compliance allowed by the SCs 5,19 . Here, we show that the atypical DC and the surrounding atypical pericentriolar material form a dynamic basal complex (DBC) that acts as a mechanotransducer, coupling tail beating to coordinate head kinking.
The centriole luminal scaffold splits into two rods in the atypical centriole, increasing its compliance Two protein classes maintain the rigidity of canonical centrioles. First, A-C linkers connect the A-tubule of each triplet microtubule with the C-tubule of the neighboring triplet 6,20 . Second, a cylindrical luminal scaffold interconnects all triplets 21 . This scaffold includes the proteins POC1B, CETN1, POC5, and the two microtubule-binding proteins FAM161A and WDR90; mutating these proteins destabilizes centriolar structure 22,23 . In the spermatozoon, the DC consists of doublets instead of triplets and lacks the A-C linkers, while the doublets are splayed apart, suggesting increased DC compliance. Furthermore, the luminal scaffold proteins POC1B, CETN1, and POC5 reorganize into two rod structures found between loosely clamped microtubules 15 . Here, we show that the microtubule-binding proteins FAM161A and WDR90 colocalize with the luminal and rod protein CETN1, labeling both at the DC and PC in human, rabbit, and bovine sperm (Extended Data Fig. 1a-b). Like the other luminal scaffold proteins, they appear as two distinct rods in the DC (Extended Data Fig. 1c-e) and are enriched in the DC compared to the PC in all three species (Extended Data Fig. 1f). This common localization pattern suggests that FAM161A and WDR90 are new conserved components of DC rods. These observations suggest that the proteins forming the scaffold that stabilizes typical centrioles split into two rods in atypical centrioles. This splitting could be an evolutionary innovation for reducing centriole rigidity, thus facilitating basal sliding.
DC rods vary in size across the three mammalian species studied (Extended Data Fig. 1g-i). This difference suggests that the atypical centriole that appeared early in mammalian evolution 9 continued to evolve in mammals, creating structural and functional diversity. The atypical centriole is largest in bovine sperm, and the theoretical foundation of basal sliding was based on bovine sperm 5 ; therefore, we performed the remainder of our studies with bovine sperm.

The DC rods and bars have opposite asymmetry
To gain insight into how DC rod proteins are situated relative to other sperm structures, we oriented straight sperm images with the PC tip on the right side and implantation fossa on the left side 24 (Fig.   1a). We found that the rods are laterally asymmetric in bovine spermatozoa ( Fig. 1b-e). 3D Stochastic optical reconstruction microscopy (3D-STORM) imaging of FAM161A, POC1B, and POC5 showed that the left rods are consistently longer and thicker than the right rods. FAM161A and POC1B labeled much larger rods (about 50% longer, P<0.0001) than those labeled by POC5, suggesting different protein locations within the DC. To determine the relationship of the rods to the splayed microtubules, we measured tubulin staining width across the DC (Fig. 1e). We found that the width of the DC microtubule bundle is similar at the caudal end and 10% wider at the rostral end than that of the FAM161A and POC1B rods (P=0.2, and P<0.0001, respectively). FAM161A can simultaneously bind microtubules and other rod proteins 25 , suggesting a close rod-microtubule association at the DC lateral sides and that rods can act as scaffolds during basal sliding. This asymmetry agrees with the axoneme's structural and functional asymmetry, where the left four axonemal microtubules work against the right three axonemal microtubules, generating a stronger left torque 24,26 . Asymmetry in the neck is also observed in other mammals, suggesting that rod-asymmetry may be a more general feature of mammalian sperm 4,14,27-29 .
To de ne the structural organization of the sperm neck in detail, we imaged un xed, unstained bovine spermatozoa with cryo-electron tomography (cryo-ET) (Fig. 1f). We compared PC dimension measurements from cryo-ET and STORM and found them consistent with each other (Extended Data Fig.   2a). We observed two electron-dense bars in the center of the neck, as reported previously 24,30,31 . Our data shows that the bars are intimately associated with the DC's central pair microtubules and are anked by electron dense material that probably represent the rods (Fig. 1g). The bars are asymmetric and are made of 1-4 plates of varying sizes separated by electron-lucent inter-plate material, revealing an unexpected level of complexity (Fig. 1g). Unlike the rods' "V" shape, the bars are nearly parallel to each other (Extended Data Fig. 2b). The caudal edge-to-edge gap between the smallest DC rods (POC5) is much larger (43%) than the corresponding gap between the bars (Extended Data Fig. 2c). The rods and bars have opposite asymmetry: the right bar has more plates and is longer than the left (Extended Data g. Fig. 2d). Finally, the two bars are situated closer to the DC's right microtubules ( Fig. 1f and Extended Data Fig. 2d). Altogether, these differences suggest that the central bars scaffold the central pair, while the rods scaffold the DC lateral side microtubules.

DC rods and microtubules slide coordinately during leftbiased tail beating
To gain insight into a possible reason for the intricate but atypical architecture of the DC, we analyzed sperm that were snap-frozen while actively swimming. We used the rod asymmetry as a reference to describe the agellar waveform relative to the sperm head despite the cell's complex rolling motions 27 . We refer to this evaluation as centriole orientation-based sperm analysis (COSA). We classi ed the sperm images into four groups based on COSA, where the bigger rod is placed on the left and PC tip point to the right side of the head midline. We found that 15% had sharp left bends, 30% had mild left bends, 36% were straight, and 19% had slight right bend (Extended Data Fig. 3). Intriguingly, none of the sperm we analyzed (n=248) had a tail with sharp right bends. The observation that all sharp bends were to the left and none were to the right indicates that the sperm tail has bias towards the left side relative to the sperm head. A similar bias in the waveform was observed when staining for either FAM161A (Fig. 2a) or tubulin (Fig. 2b). The left-biased beat has structural origins in the inherent asymmetry of the neck, along with the increased rigidity imposed by the mitochondrial sheath that extends further rostrally on the right side of the neck (Fig. 1a and f) 32 .
We then examined the DC substructures in chemically xed sperm at distinct tail bending angles by 3D-STORM. In tails with sharp left bends, the right rod and microtubules are shifted rostrally relative to those on the left side (Fig. 2c-d and 2g-h). In contrast, the right rod and microtubules slide caudally relative to those on the left as the tail becomes straight or bent to the right (Fig. 2i). We also observed that the DC central microtubule protrudes rostrally, which likely represents the central pair (Fig. 2h). To examine the role of the central microtubule during movement, we measured the distance from a centerline drawn through the long axis of the PC to the left, center, and right of the DC microtubule bundle. As the tail bends from left to right, the DC's right microtubules move further away from the PC centerline, while the DC's left microtubules move closer (Extended Data Fig. 4). In contrast, the DC's central microtubules maintain the same distance from the centerline during tail beating. Consistent with this, cryo-ET found that the central singlets are closer to the PC than are the DC's left and right microtubules and have the least change in distance from the PC centerline (Extended Data Fig. 5b). Also, the left and right bars associated with the central singlets were relatively static relative to each other during tail beating (Extended Data Fig. 5c-e). These differential movements suggest that the DC central pair and bars form a rail-like tracking system along which the rods and peripheral DC microtubules slide.
Our data provide the rst direct evidence of microtubules sliding postulated by the basal sliding hypothesis 4 extending it to nanometer-scale shearing deformations in the neck. This hypothesis suggests that the tail's waveform is regulated by dynamic microtubules sliding at the axoneme base. We tested this hypothesis by comparing the sliding observed for DC microtubules and rods against three variables derived from the "sliding lament" hypothesis for agella movement 5,33 : the calculated average agellum beating amplitude (), microtubule sliding along the tail (), and the average waveform curvature (). As expected from the observed beating asymmetry, all three waveform characteristics were skewed towards negative values, the left side (Extended Data Fig. 6a-b). Basal microtubule and rod sliding show a strong correlation with the calculated averages of agellar beating amplitude, sliding, and curvature (Fig. 2j, Extended Data Fig. 6c-d, Extended Data Fig. 7a-b). Overall, from sharp left to slightly right bent tail, the DC rod and the basal ends of the microtubules are displaced 263-328 nm relative to each other.
This sliding has the calculated order of magnitude of the agellar control model ttings: 160 nm in bovine sperm 5 (Extended Data Fig. 7i-k). This similarity suggests that the DC sliding movement is the predicted basal sliding; however, the sliding is more complex and includes associated structures such as the rods.
A dynamic basal complex transmits the tail's microtubule sliding to the head The current dogma holds that the neck is cemented to the head, as the neck structures connect the tail to the head. Some movement was observed in the neck, but it lacked correlation with tail beating 24,34 .
Therefore, the sperm head is thought to passively follow tail swimming movement 35,36 . In contrast, we observed a dramatic coordinated motion of neck structures with a novel head movement, which we named "kinking" to signify a 2D head movement relative to the sperm long axis (Extended Data Fig. 8).
We found elastic deformation in the neck beyond the DC. The right SCs are displaced relative to the left SCs during sperm movement (Fig. 3a, Extended Data Fig. 9a). Also, the usually parallel segments of the SCs are bent between segments 8 and 9 in left curved sperm (Extended Data Fig. 9b). The motion in the neck causes the embedded PC to also move, and the PC angle relative to the neck midline changes ~24º during tail beating (Fig 3a and 3c, Extended Data Fig. 10a, Extended Data Video 1). The PC also changes its lateral position relative to the neck midline, moving 140-200 nm to the left (Extended Data Fig. 10b).
We found high to strong correlation of PC position change with tail variables, DC sliding, and SCs sliding (Extended Data Fig. 11b-g). Therefore, the neck deformation is due to concomitant displacements of DC, PC, and SCs during sperm tail beating.
Most signi cantly, we found a novel coordinated deformation inducing a head-neck kink, with the angle changing ~45º, causing the head to kink to the left when the tail beats to the left (Fig. 2e-f, Extended Data Fig. 11a, Extended Data Video 2, Extended Data Fig. 12). Similar kink was depicted in the past in live and reactivated bovine sperm 37,38 . Interestingly, we found that head kinking also correlates with tail variables and neck structure (DC, SCs, and PC) deformation during tail beating (Fig. 3b, Extended Data Fig. 11b-h), suggesting that the neck kinking is in coordination with other nanometric structural deformations of the neck during tail beating. This head kink is marked by a sharp angle between the tail's tangent angle at the neck, where both the head and tail bend to the left (Fig. 3c). The coordinated tail bending and head kinking suggests a dynamic structural modulation during swimming.
Sperm tail movement drives neck deformation and head kinking, as axonemal dyneins are the only active motor proteins during sperm swimming. Two mechanisms may translate forces from the tail: the axoneme's attachment to the DC and the tail's outer dense ber attachment to SCs. Signi cantly, exploratory factor analysis of three sets of data measuring a total of 21 variables during sperm beating indicated that 3 factors explained the underlying sperm behavior: a major tail-to-head coordinated movement, no movement of DC center and width, and a mixture of the two (Extended Data Fig. 13).
These data support a model in which the axoneme sliding that generates tail beating also deforms the neck, which subsequently kinks the head (Fig. 3c).

Conclusion
Centriole structure and function are conserved across ciliated cells, from protists to mammals. It is therefore surprising that the centriole found at the base of the agellum is structurally atypical in mammalian sperm [39][40][41][42] . We show that the atypical DC and its specialized pericentriolar matrix form the DBC. The DBC has two functions: (i) it shapes tail beating according to the basal sliding hypothesis; and (ii) it translates the axoneme's piston-like tangential movement into a cascade of multi-component shearing deformations culminating in a coordinated head kinking motion. A potential advantage of mechanically coupling tail curvature with the head attachment angle is that the head can impact tail movement to provide mechanosensory information, thus providing a way for the sperm to better navigate the various barriers in the female reproductive tract. This coupling may also help the sperm dig its way through the external protective shields surrounding the ovum. We show that in bovine sperm, this coupling associates with asymmetrical tail beating and head kinking to the left and may help achieve forward movement via swimming via sperm rolling motion motions 27 . However, small changes to neck components will likely result in distinct movement patterns, creating a spectrum of sperm behaviors in other animal species. Altogether, the DC's novel properties suggest that it evolved by repurposing centriolar proteins to assemble a transmission system (the DBC) that couples the agellar motor to the whole sperm, thereby enhancing sperm function.  The DC is Asymmetric. a) Schematics illustrating sperm cell asymmetries (top) with a magni cation of the neck (bottom). Note that sperm beating occurs in one plane, while head and neck asymmetry provide a basis for orientation in all gures, such that the top is rostral and the left side is on the left. b-e) Rod proteins and microtubules are asymmetric in the DC of bovine spermatozoa, as observed using 3D-STORM. Side view (left two panels); top view (right two panels). All sizes are in mean ± sd nm. The colors represent depths of 1178 nm in the Z-plane, with red on the bottom (bott.) in all gures. Statistical is shown in Extended Data Fig. 2e f-g) A computational slice (left) and corresponding 3D segmentation (right) of a Volta phase plate cryo-tomogram of a bovine sperm neck (f). A digital zoom of a cryotomogram illustrating bar asymmetry, complexity, and intimate association with the central pair (g). Throughout the paper, labels and their colors are the same: N, nucleus; IF, Implantation fossa; Bp, basal plate; Ca, capitulum; SC, striated column, grey; PC, Proximal centriole, yellow; DC, distal centriole, green; Ax, axoneme, red; M, mitochondria, brown; L, left side; R, right side; r, rod, C, central microtubules; B, bars, light yellow; CP, central pair, pink; Mts., microtubules, green in DC and red in axoneme.