Architecture of CTPS filament networks revealed by cryo-electron tomography

2.1. Cryo-ET analysis of dmCTPS filament network

In our previous research, we discovered that the purified CTPS protein of Drosophila can spontaneously assemble into filament structures in vitro under the presence of substrates or products [34]. During this process, we incidentally observed that these filaments could assemble into larger complex structures.

Using negative-stain electron microscopy, we screened the sample conditions. In subsequent experiments, we identified using CTP to mimic the product condition of CTPS (referred as CTP-state) and UTP, ATP, the glutamine analog 6-diazo-5-oxo-norleucine, and the allosteric regulator GTP to simulate the substrate condition of CTPS (referred as DON-state). We prepared cryo-EM samples of the dmCTPS filament network in both CTP-state and DON-state: dmCTPS was first incubated with either substrates or products, then mixed with an equal volume of 10 nm colloidal gold particle solution, and finally rapidly frozen using a Vitrobot. The hole diameter of grid is 1.2 μm, and the center-to-center distance between holes is 1.3 μm. When we collect data, we select objects located within the ice layer of the hole, so the length and width of the study objects are usually below 1 μm. We collected cryo-ET data using a dose-symmetric data collection strategy. After 3D reconstruction and analysis of the data, we found that in both CTP-state and DON-state, dmCTPS filaments form structural bundles, which then further form networks.

In the reconstructed 3D data (Fig. 1A), we manually selected and tracked filaments longer than 15 nm with continuous density, obtaining the spatial coordinates of the filaments (Fig. 1B). We focused our analysis on the bundles formed by these filaments.

For comparison, we processed the data within a defined spatial length and width, choosing the bundles with the highest number of filaments in the volume for analysis (Fig. 1C and D). Preliminary statistical analysis showed that a width of 150 nm and a length of 400 nm could maximally include structural bundles from different datasets, so we used these dimensions for statistical analysis of the structural bundle regions.

We randomly selected four datasets from both CTP-state and DON-state data (Fig. 1A and B) and compared and analyzed the spatial distribution of the filaments from multiple perspectives (Sup Fig. 1). The results showed that CTPS filaments primarily assembled in a parallel mode to form structural bundles, and these bundles would then form larger complexes in space, either in a roughly parallel or staggered manner.

2.2. dmCTPS filament network under CTP-state

The statistical analysis reveals that within the selected area (dimensions 150 nm × 400 nm), the average number of filaments per bundle in the CTP-state is 55 (Fig. 2A). CTPS filaments form bundles through mutual parallel and overlapping arrangements, with an average height of 46.75 nm. These bundles occupy an average cross-sectional area of 5413.5 nm², with an average width of approximately 115 nm. On a representative cross-section of the sample (Fig. 2B), the filaments exhibit a characteristic pattern of wide lateral and thin vertical alignment, with a higher count of filaments in the lateral dimension (5-12) compared to the vertical dimension (1–4).

The minimum, maximum, median, and mean center-to-center distances between CTPS filaments were measured at 1.4 nm, 384.8 nm, 51.9 nm, and 56.8 nm respectively, indicating an overall approximately left-skewed normal distribution of filament spacing (Fig. 2C).

Analyzing the relative positional relationships of CTPS filaments in three-dimensional space reveals a median relative deviation direction of 13.4° and a mean of 18° between filaments (Fig. 2D). This suggests that within CTP-state bundles, despite some angular deviations between filaments, there is an overall tendency towards a consistent orientation.

Subsequently, we performed statistical analysis correlating filament distances (in nanometers) with their relative local orientations (in degrees) (Fig. 2E). The results indicate that under CTP-state, most filaments are arranged parallel with distances around 20 nm, with a smaller fraction arranged at approximately 40 nm intervals. Based on previous research, CTPS filaments are known to have a width of approximately 12 nm and a thickness of around 7 nm. Therefore, distances less than 10 nm suggest close proximity or direct contact between filaments. For example, in our sample, we observed instances where filaments appeared to branch, which might explain the minimal distance of 1.4 nm between them (Sup Fig. 2). This phenomenon could be due to estimation errors in identifying filament starting points during statistical analysis, causing one filament’s start point to be located within another filament. Alternatively, it might indicate the presence of two filaments sharing the same endpoint within the structural bundle. Additionally, the two-dimensional correlation plot reveals a minor presence of filament spacings at 60 nm and 80 nm. Filaments with angles less than 20° likely represent distant filaments within the same bundle, while those with angles greater than 20° may correspond to filaments from other bundles within the network.

2.3. dmCTPS filament network under DON-state

According to the results of statistical analysis, within the specified area (dimensions 150 nm × 400 nm), the average number of filaments per bundle in the DON state is 51, which is comparable to the CTP-state under same protein concentration. However, the structural bundles of DON-state exhibit an average height volume of 36.25 nm, occupying an average cross-sectional area of 3302.5 nm2, with an average width of approximately 91 nm. On representative cross-sections of the samples, filament distribution similarly displays characteristics of wide lateral and thin vertical alignment, with a higher number of filaments in the lateral dimension (approximately 6–7) compared to the vertical dimension (approximately 2–3) (Fig. 3A and B).

The minimum, maximum, median, and mean center-to-center distances between DON filaments and adjacent filaments are 1.1 nm, 231.7 nm, 48.3 nm, and 53.7 nm, respectively, also exhibiting an approximately left-skewed normal distribution, indicating a denser arrangement of filaments under DON conditions (Fig. 3C).

Analysis of the relative positional relationships between dmCTPS filaments in three-dimensional space shows a median relative direction (degrees) of 13° and a mean of 15.7°. This suggests that within the structural bundles in the statistical sample space, the DON-state structural bundles also tend to exhibit a consistent trend of filament aggregation, with smaller deviations in angles between filaments (Fig. 3D).

Correlating the distances between filaments (measured in nanometers) with their relative local directions (in degrees), it is observed that DON-state filaments predominantly arrange in parallel with distances around 20 nm (Fig. 3E). Compared to bundles in the CTP-state, filaments in the DON-state exhibit closer and more uniform spacing.

2.4. Comparison of two dmCTPS filament networks

To further contrast the spatial distribution of filaments within bundles under two different conditions, statistical analysis was conducted on relative filament angles, filament lengths, and filament curvatures. Initially, a second-order polynomial fitting was applied to coordinates of all filaments within the structure to establish a reference direction (axis) for the overall bundle. This fitting method helps capture the curvature and variations in filamentous structures, facilitating subsequent analysis and understanding of mechanical characteristics. Subsequently, by measuring and statistically analyzing the angle between each filament’s end-to-end vector and the reference direction, the distribution of relative filament axial angles can be obtained.

In the statistical sample space, the average filament length under CTP-state was 149.3 nm, longer than the 119 nm observed under DON-state (Fig. 4A). Distribution analysis of individual filament lengths shows that both CTP and DON filament lengths are predominantly concentrated within 200 nm, accounting for 77.31 % and 88.82 % of the total, respectively, with a minority extending between 200 nm and 400 nm, accounting for 10.42 % and 10.66 %, respectively.

Comparison statistics of relative filament angles reveal that under both CTP and DON conditions, filaments within the same structural bundle exhibit similar orientations (Fig. 4B). Approximately half of the filaments deviate from the reference direction by less than 11°. Specifically, 56.6 % of CTP-state filaments are concentrated within 10°, and 85.46 % within 30°, while 50.79 % of DON-state filaments are within 11°, and 90 % within 30°. Although there are differences in distribution between CTP and DON conditions, with 39.69 % of CTP-state filaments and 25.46 % of DON-state filaments having deviations ≤ 5°, and 13.31 % for CTP-state and 21.82 % for DON-state between 9 and 13°, this suggests varying modes of filament binding or interaction within structural bundles under CTP and DON conditions. Specifically, filaments in CTP-state tend to align more axially, while filaments in DON-state exhibit greater relative axial angles.

Filament curvature was defined as the ratio of filament length to the distance between its two ends, where a curvature of 1 indicates a straight line, and values greater than 1 indicate curvature, with higher values indicating greater structural curvature. Approximately half of the filaments under CTP and DON conditions approximate straight lines, comprising 48 % for CTP-state and 53 % for DON-state (Fig. 4C). Curvature exhibited a negative correlation with relative abundance; higher curvature corresponded to fewer filaments. A minority of filaments exhibited significant curvature, with approximately 10 % of CTP filaments having curvatures above 1.05, and approximately 4 % of DON filaments having curvatures above 1.05.

In previous in vitro studies, the CTPS filaments resolved in the DON and CTP states exhibited different structures. Under the DON-state, the tetramer is in a close conformation with more tightly packed stacking, theoretically making DON-state filament less prone to bending [35]. On the other side, in vivo experiments have shown that the addition of DON not only promotes the formation of cytoophidia but also increases the proportion of highly curved, ring-shaped cytoophidia among all cytoophidia [4,51].

This is consistent with the results of our analysis, where single filaments in the DON-state indeed have less curvature. Additionally, larger relative axial angles may provide a possible explanation for the observed increase in the proportion of ring-shaped cytoophidia/all cytoophidia upon the addition of DON in vivo.

2.5. Filament distribution within the structural bundles

In addition, we calculated the distribution of filaments along the axis of the entire structure. Averages and tracking were performed along the axis of the entire bundle. By calculating the cross-sectional areas of structures under CTP and DON conditions, it was observed that the cross-sectional area under CTP conditions (ranging from 3986.57 to 7246.64 nm²) was greater than under DON conditions (ranging from 1645.23 to 4757.36 nm²) (Fig. 5A).

To measure structural density, we defined the volume ratio as the ratio of the total volume of all filaments to the total volume of the bundle. The CTP-state ratio was 0.54, while the DON-state ratio was 0.67, indicating that the structural density of bundles in DON-state is higher than in CTP-state (Fig. 5B). Specifically, under equal statistical spatial volumes, the number of filaments in DON conditions exceeds that in CTP conditions by 20 %.

Additionally, given that the constituents of the bundles are identical, we utilize the moment of inertia of the filaments with respect to the y-axis in the cross-section to describe the structure’s resistance to bending in the z-direction, which we refer to as the vertical bending stiffness. Stiffness of bundles under CTP and DON conditions was computed (Fig. 5C and D). It was found that the vertical bending stiffness of the structure under CTP conditions was 9138.44 nm², compared to 3526.72 nm² under DON conditions. Similarly, the transverse bending stiffness under CTP conditions was 23876.5 nm², whereas under DON conditions it was 19361 nm². This indicates that bundles in the CTP-state exhibit greater resistance to bending in both vertical and transverse directions compared to those in the DON-state. Specifically, in the vertical direction, the CTP-state demonstrates more than double the bending resistance of the DON-state, assuming equal filament strength.

Overall, these findings indicate that CTP filaments form a loosely packed parallel bundle structure with greater filament spacing and relative angles, yet exhibit strong overall resistance to external bending, particularly in the vertical direction. In contrast, DON structural bundles consist of filaments tightly packed in parallel, with smaller filament spacing and relative angles, but weaker overall resistance to external bending, especially in the vertical direction.

4.1. Protein purification and sample preparation

The protein was purified using the same method as before. A concentration of 50 μM CTPS was used to prepare the sample solution. For the DON-state and CTP-state samples, we added 60 μM DON, 2 mM ATP, 2 mM UTP, 2 mM GTP, and 10 mM magnesium ions or 2 mM CTP and 10 mM magnesium ions to the system, respectively. Ligands and protein are incubated at room temperature for 30 min. Subsequently, an equal volume of gold nanoparticles (Aurion Gold Nanoparticles – Carboxyl Functionalized-10nm) was added to the system. After thorough mixing, 3 μL of the mixture was applied onto a hydrophilized grid(No. M024-Au300-R12/13), blotted using a Vitrobot, and rapidly plunged into liquid ethane to obtain the cryo samples.

4.2. Cryo-ET data collection

Cryo-ET data was collected using a Titan Krios G4 300 kV electron microscope equipped with an energy filter and a Falcon4 detector (Thermo Fisher Scientific). The Tomography software facilitated the acquisition of tilt series. To achieve high-resolution reconstruction, a dose-symmetric tilt scheme [53] was used during data collection with a magnification of 85,000x, a pixel size of 1.22 Å, and a defocus of 2 μm. A tilt series spanning ±30° in 2° increments was collected, comprising 31 angles, with an electron dose of 4.5 e/Ų per angle and a cumulative dose of 140 e/Ų per tilt series.

4.3. Image preprocessing

Motion correction and Contrast Transfer Function (CTF) estimation were conducted on each image using Relion4.0 [54] to compensate for sample movement and focusing errors. The processed single images were then reassembled into new tilt series based on their respective tilt angle information.

4.4. Tilt-series alignment

Tilt series alignment was performed using IMOD [55] to correct for sample movement during tilting. A fourfold binned three-dimensional reconstruction was carried out using Weighted Back Projection (WBP). Three-dimensional data were processed with a twelvefold binning using IMOD, and image signal enhancement was performed using the ISONET [56] software to achieve better image contrast.

4.5. Filament identification and statistical analysis

Four distinct datasets for each of CTP and DON were randomly chosen for comprehensive analysis. Utilizing the Dynamo [57] software, filaments were meticulously identified through manual selection with the filament along axis module. Filaments exceeding 30 nm in length and exhibiting discernible continuous density were categorized as individual entities.

To facilitate further analysis, custom scripts were developed to convert the resulting coordinate files from.mat format to.txt, enabling seamless integration with the MATLAB-based Computational Toolbox [58]. This conversion allowed for a comprehensive quantitative analysis of the data.

Employing the cropping.m script, we extracted individual structural bundles, each measuring 150 nm by 400 nm. The Analysis_Filopodia.m script was then executed to precisely delineate the X-axis orientation of these bundles.

Visualization of the three-dimensional structure and cross-sectional profiles of the filaments was achieved using the Plots_Single_Cell_Filopodia.m script. This enabled us to calculate key statistical metrics, including the total filament count, average cross-sectional area, and cumulative volume for each structural bundle.

Finally, the Plots_Group_Cell_Filopodia.m script was engaged to scrutinize the interactions between CTP and DON. This analysis encompassed a range of parameters such as filament angles, lengths, and cross-sectional dimensions, providing a nuanced understanding of the interplay between these two components.