Analysis of motor-based transport in primary cilia by dynamic mode decomposition of live-cell imaging data

This study combines live-cell imaging with a novel dynamic mode decomposition-based kymograph analysis to demonstrate that the kinesin-3 motor KIF13B localizes to the ciliary membrane and regulates ciliary content without altering steady-state intraflagellar transport velocities, while uniquely maintaining transport stability even when retrograde dynein-2 function is inhibited.

The Gist

Imagine a cell as a bustling city, and the primary cilium as a tiny, antenna-like tower sticking out of that city. This tower is crucial for the cell to "hear" signals from the outside world. To keep this tower working, it needs a constant supply line of materials moving up and down its length. This supply line is called Intraflagellar Transport (IFT).

Think of IFT as a busy train system running on tracks (microtubules) inside the tower.

• Kinesin motors are the trains going up (delivering supplies).
• Dynein motors are the trains going down (taking out the trash/recycling).

For a long time, scientists knew about the main "Kinesin-2" trains. But recently, they found another motor protein, KIF13B, hanging out in these towers. The big question was: Is KIF13B just a passenger on the main train, or is it doing something totally different?

This paper is like a detective story where the researchers used a new high-tech camera trick to solve the mystery.

1. The New Camera Trick: "Freezing the Motion"

Looking at these tiny moving trains in real-time is messy. It's like trying to count cars on a highway while looking at them through a foggy, vibrating window. The video is full of moving dots, stationary blobs, and background noise.

The researchers invented a new way to analyze the video using a math technique called Dynamic Mode Decomposition (DMD).

• The Analogy: Imagine you have a video of a busy street. Some cars are zooming by (fast), some are parked (slow), and some are just background noise.
• The Magic: Their new math tool acts like a super-smart filter. It can separate the "fast cars" (the active transport trains) from the "parked cars" and the "fog" (the background).
• The Result: Suddenly, the blurry video becomes crystal clear. They can see exactly how fast the trains are moving and how often they stop, without getting confused by the noise.

2. The Mystery of KIF13B

Using this new "super-filter," they watched the KIF13B protein in action. Here is what they found:

• It's not a regular train passenger: The main IFT trains (carrying IFT172) move at a steady, predictable speed. KIF13B, however, behaves differently. It doesn't just ride the train; it seems to zip in and out of the tower in "bursts."
• It lives on the walls, not the tracks: While the main trains run down the center of the tower, KIF13B hangs out near the walls (the membrane). It's like a maintenance worker walking along the side of the tunnel rather than riding the train.
• It's a "Gatekeeper" at the base: KIF13B hangs out heavily at the very bottom of the tower (near the cell body). It seems to act like a bouncer or a loading dock manager, helping to grab materials and pull them out of the tower or prepare them for entry.

3. The "Drug Test" Surprise

To test if KIF13B was part of the main train system, the researchers used a drug called Ciliobrevin D.

• The Setup: This drug is like putting a speed bump on the "downward" tracks. In normal cells, this drug slows down the trains going down significantly.
• The Twist: When they used the drug on cells without KIF13B (a mutant version), the downward trains didn't slow down at all. They kept going at normal speed!
• The Conclusion: This was a huge surprise. It means KIF13B is somehow involved in how the cell reacts to traffic jams. Without KIF13B, the cell's "traffic control" ignores the speed bump. This suggests KIF13B isn't just a passenger; it's a regulator that talks to the traffic control system, but it does so in a way that doesn't depend on the main train tracks.

4. The Big Picture: What is KIF13B actually doing?

The researchers concluded that KIF13B is not part of the standard delivery train system. Instead, it has a different job:

• The "Recycling Crew": It seems to specialize in grabbing things from the tower walls and pulling them back down to the cell body for recycling (endocytosis).
• The "Scaffold": It acts like a construction scaffold at the base of the tower, helping to organize other proteins and vesicles (tiny bubbles carrying cargo) before they even enter the tower.

Summary

Think of the cell's antenna (cilium) as a busy construction site.

• The IFT trains are the elevators moving bricks up and down the center shaft.
• KIF13B is not an elevator. It's a foreman working on the scaffolding at the bottom and walking along the walls.
• The researchers used a mathematical "X-ray" to see that the foreman (KIF13B) doesn't ride the elevator. Instead, the foreman organizes the delivery trucks at the entrance and manages the recycling of materials from the walls. If you remove the foreman, the elevators still run, but the whole system loses its ability to react to traffic jams properly.

This study gives us a new, clearer way to watch these tiny cellular movies and proves that KIF13B has a unique, specialized role in keeping the cell's antenna healthy, separate from the main transport system.

Technical Summary

1. Problem Statement

Primary cilia are essential cellular signaling hubs requiring precise assembly and maintenance via Intraflagellar Transport (IFT). While the roles of canonical kinesin-2 motors in anterograde transport and dynein-2 in retrograde transport are well-established, the function of other kinesin family members, specifically the kinesin-3 motor KIF13B, remains unclear.

• Knowledge Gap: It is unknown whether KIF13B modulates the canonical IFT machinery or operates independently. Previous studies showed KIF13B regulates ciliary membrane content and endocytic retrieval, but its impact on IFT velocity and frequency was not quantified.
• Methodological Limitation: Traditional kymograph analysis (space-time plots) struggles to disentangle complex, multiscale dynamics in cilia, such as the simultaneous presence of active transport, diffusion, and stationary protein pools. Existing computational tools often fail to separate these modes effectively.

2. Methodology

The authors combined advanced live-cell imaging with a novel computational framework based on Dynamic Mode Decomposition (DMD).

• Cell Models:
  • Parental Cells: hTERT-RPE1 cells stably expressing IFT172-eGFP (a marker for the IFT-B2 complex).
  • Mutant Cells: CRISPR-Cas9 generated clones with a truncated KIF13B (retaining only the motor domain and part of the FHA domain, lacking C-terminal cargo-binding modules).
  • Dual Reporter: Cells co-expressing Halo-KIF13B (labeled with JF646) and IFT172-eGFP.
• Imaging Techniques:
  • Live-cell Confocal Microscopy: High-temporal resolution imaging to generate kymographs.
  • Super-resolution Microscopy: Structured Illumination Microscopy (SIM) and Stimulated Emission Depletion (STED) microscopy to determine sub-ciliary localization relative to the axoneme and membrane.
  • Pharmacological Perturbation: Treatment with Ciliobrevin D (dynein-2 inhibitor) and Forskolin (cAMP activator) to test IFT sensitivity.
• Computational Analysis (The Core Innovation):
  • Dynamic Mode Decomposition with Time-Delay Embedding (DMD-TDE): The authors applied DMD, a data-driven numerical method originally from fluid dynamics, to kymograph data.
  • Mechanism: By constructing a Hankel matrix (stacking time-shifted versions of the data), DMD-TDE approximates the underlying Koopman operator. This allows the decomposition of complex spatiotemporal data into distinct spatial and temporal modes.
  • Separation of Modes:
    • Low-rank approximation: Captures slow-moving or stationary background (diffusion pools, static proteins).
    • High-rank approximation: Captures fast-moving entities (active transport).
    • Subtraction: Subtracting the low-rank reconstruction from the original data isolates the actively transported population, significantly improving automated track extraction and velocity calculation.
  • Modeling: The authors used advection-diffusion equations with binding terms to simulate and validate the observed kymograph patterns (active transport vs. diffusion vs. transient binding).

3. Key Contributions

1. Methodological Advancement: Demonstrated that DMD-TDE is superior to standard DMD and traditional methods for analyzing ciliary transport. It successfully separates active transport from diffusion and stationary pools, enabling more accurate and automated velocity extraction from kymographs.
2. Functional Characterization of KIF13B: Provided quantitative evidence that a KIF13B truncation mutant does not alter the steady-state velocity or frequency of canonical IFT (IFT172) in untreated cells.
3. Discovery of Insensitivity to Dynein Inhibition: Revealed that while wild-type cells show slowed retrograde IFT upon Ciliobrevin D treatment, KIF13B mutant cells remain largely insensitive to this inhibition, suggesting a complex regulatory link between KIF13B and dynein-2 function.
4. Spatial Resolution: Utilized SIM and STED to map KIF13B localization precisely to the ciliary membrane and subdistal appendages (sDAs) of the centriole, distinct from the axonemal core where IFT172 resides.

4. Key Results

• DMD-TDE Performance:
  • Successfully reconstructed synthetic kymographs containing bidirectional transport, diffusion, and stationary particles.
  • Applied to experimental IFT172-eGFP data, DMD-TDE isolated the moving IFT trains from the background, increasing the number and length of extracted tracks compared to standard analysis.
• KIF13B Mutant Phenotype:
  • IFT Velocity: In untreated cells, the KIF13B mutant (Clone B4) showed no significant difference in anterograde (0.33 µm/s) or retrograde (0.22 µm/s) IFT velocities compared to parental cells.
  • Drug Response:
    • Forskolin: Did not significantly alter IFT velocity in either cell line.
    • Ciliobrevin D: Slowed retrograde IFT in parental cells (as expected) but failed to slow retrograde IFT in KIF13B mutant cells. This suggests the mutant cells have an altered sensitivity to dynein-2 inhibition.
• KIF13B Dynamics and Localization:
  • Motion: Halo-KIF13B exhibits "burst-like" bidirectional movement, rapid entry/exit, and diffusion. It does not co-localize with IFT172 tracks in kymographs, indicating independent motion.
  • Localization: Super-resolution imaging showed KIF13B is membrane-associated (external to the axoneme) and concentrated at the periciliary membrane and subdistal appendages (sDAs), below the distal appendage marker FBF1.
  • Export: KIF13B can be rapidly released from the ciliary membrane via a combination of diffusion and active transport (advection), fitting an advection-diffusion model with a transport velocity of ~0.92 µm/s.

5. Significance and Conclusion

• IFT Independence: The study concludes that KIF13B functions largely independently of the canonical IFT machinery. Its truncation does not disrupt the core transport of IFT trains, implying it regulates ciliary content through a different mechanism (likely endocytic retrieval and membrane trafficking).
• Regulatory Node at sDAs: The finding that KIF13B localizes to the subdistal appendages (sDAs) and influences sensitivity to dynein-2 inhibition suggests that sDAs act as a critical regulatory hub. KIF13B may coordinate the assembly or regulation of motor complexes (kinesin-3 and dynein-2) before they enter the cilium.
• Broader Impact: The DMD-TDE methodology provides a robust, generalizable tool for analyzing complex biological transport systems where multiple dynamic modes (diffusion, active transport, binding) coexist, overcoming limitations of current kymograph analysis tools.

In summary, this paper redefines the role of KIF13B in cilia as a membrane-associated regulator that operates parallel to, but distinct from, the core IFT system, while introducing a powerful computational method to dissect these complex dynamics.