Columbia Logo

The Hunt for the Great White Tubule

Hello visitor #web counter free!

Welcome to the Project Wiki for Tube Tech! We're a team based out of Hess Laboratory at Columbia University's Biomedical Engineering Department.

This summer, we designed a protocol that produces EXCEPTIONALLY LONG MICROTUBULES! For details on our team, please check out the links below:

Who We Are! How It Began! What We Did! What We Got! Our Lab Notebook!

Introduction - So What's the Scoop on Microtubules?

Microtubules are polymers found within our cells that serve as components of the cytoskeleton. They are filamentous polymers composed of subunits known as tubulin heterodimers and are involved in a variety of cellular functions ranging from cell division to intracellular transport.

Under in vivo conditions, tubulin heterodimers are formed from α- and β-tubulin molecules spontaneously binding together. These heterodimers are organized into functional microtubules within the cell through nucleation and polymerization. Microtubule organizing centers, or MTOCs, contain another type of tubulin known as γ-tubulin. They form zones within the cell from which microtubules can grow by forming protofilaments which can then combine to form microtubules. The energy for these heterodimers binding onto one another is provided through the hydrolysis of guanosine tri-phosphate (GTP) to guanosine di-phospate (GDP).

Growth from Heterodimers
Polymers of HeterodimersImage Source:[4]

In vitro, an adequate supply of GTP leads to nucleation and subsequent polymerization; however, kinetic barriers involved in the reaction process maintains a low rate of polymerization. Nucleating agents, such as di-methyl sulfoxide (DMSO) or glycerol, help overcome these barriers.

Microtubules possess a property known as dynamic instability; they shift between growing and shrinking phases. Microtubules are also capable of combining together if they are capable of meeting within the solution they are present in. These characteristics allow for several interesting interactions within populations of microtubules and free tubulin heterodimers, including length redistribution, seed growth and annealing.

Dynamic Instability
Dynamic Instability Image Source:[4]

These are interactions that would allow us to grow LONGER microtubules if manipulated correctly. But why are we interested in longer microtubules? We stated before that microtubules form a part of the intracellular cytoskeleton. More specifically, they form a network of tubules that span the cell and provide it structural stability. This network also serves as a highway through which cellular loads can be shuttled around the cell by travelling along the length of the microtubule.

It achieves this transport by working in tandem with kinesin, a motor protein capable of attaching itself onto a target molecule. Once bound on to a microtubule, kinesin can take “steps” along the length of the microtubule by hydrolyzing adenosine tri-phosphate (ATP). Any payload bound to the kinesin is carried along as well!

Kinesin Walking
Kinesin Walks Across Microtubules Image Source:[5]

This particular property could prove to be extremely useful when engineering on the nanoscale. Microtubules could potentially provide a highway for shuttling molecular cargo within devices. This would require very long microtubules that would span the size of the device.

Project Video - But Where Is the Great White Tubule?

But wait! Before you go further...

You probably guessed where we're going with this...

I mean, we did title the project after Moby-Dick...

You're right! We wanted to make a procedure to develop long microtubules!

Here's a short funny video laying out the groundworks of what we did!

Well, It's a Problem Worth Solving...

As we stated before, microtubules serve as highways for kinesin molecules to shuttle cellular loads under in vitro conditions. In advancing the field of nanotechnology, this could be particularly useful as microtubules can be used as tracks for carrying molecular cargo within devices designed and specified at an extremely small scale. This would require us being capable of producing microtubules long enough to serve as these tracks; the microtubules would need to be long enough to span the length of a given channel and this may stretch to the scale of several hundred of microns. In order to achieve this goal, our team spent the past summer working on developing a procedure for consistently developing long microtubules in vitro.

Approach - It Always Starts With the Seeds!

Our first intuition in attempting to solve this problem came from a paper our principal investigator told us about in a lab meeting! A paper co-authored between Henry Hess and Yolaine Jeune-Smith and published by the Royal Society of Chemistry in 2010 claimed that shearing could drive the annealing of microtubules, leading to longer microtubules! The first phase in our experiments explored the role of shearing and annealing in developing longer tubules and whether this was a feasible solution to our problem!

The second phase of our experiments explored the role of seeding and growing microtubules from seeds as a viable solution. This was based upon work done by Mitchison and Kirschner, which described a significant increase in average microtubule length when the tubules were grown from stabilized seeds.[2][3][6]

To understand seeding, let’s model the dynamics of microtubule polymerization:


dTp/dt = konTFM - koffM
TF + M ↔ Tp
Tp = TTotal - Teq
TF at Equilibrium = Teq = kon/koff [From solving Equation 1 when dTp/dt = 0 at Equilibrium]
L = Tp/M = (TTotal - Teq)/M
Tp = Number of Tubulin dimers in polymerized microtubules
TF = Number of Free Tubulin dimers in the system
TTotal = Total Number of Free Tubulin dimers in the system
M = Number of Nucleation Sites
koff/on = Rate Constants for Polymerization/Depolymerization from Microtubules
Teq = Free Tubulin at Equilibrium
L = Average Number of Tubulin Dimers per Microtubule

In the above equations, L is a measure of the average length of the microtubules. This is measured in units of tubulin heterodimers per microtubule, where one dimer is 8 nm long.[6][8]

From the equations above, we can see that increasing the average length depends upon:
(1) Reducing the number of nucleation sites.
(2) Decreasing the equilibrium constant or
(3) Increasing the total tubulin heterodimers in the system

We tested for (1) and (3) by carrying out experiments involving continuous successive polymerization and seeding in the second phase of our experiments. Method (2) was factored into our experiments when comparing the efficacy of the various types of tagged tubulin used.

Please review the appendices for the specific log of the lab notebook on each day of our testing!

Methods - So What Did We DO Exactly?

Shearing and Annealing

A standard polymerization buffer was first prepared by combining stock solutions of standard Brinkley Resuspension Buffer with 80mM PIPES solution (BRB80 buffer), 100 mM Mg2+ solution, 4 mg/mL GTP solution and 5% DMSO solution to produce final concentrations of 4 mM Mg2+, 0.16 mg/mL of GTP and 0.0024% of DMSO.

6.25 µL of this polymerization buffer was then added to an aliquot of tubulin to produce a final tubulin concentration of 3.2 mg/mL within the solution. This solution was then incubated in a water bath for 30 minutes at 37 ͦC.

At the end of this incubation period, the microtubule solution was stabilized by adding paclitaxel and more BRB80; they were then sheared by passing through a 25-gauge needle over a 10 second period. The sheared microtubules were then replaced within the water bath to allow annealing to take place. These were then imaged.[3]


Several nucleating agents were used in these experiments to test their efficacy in forming stabilized microtubule seeds. Please note that what follows is a general outline of the procedures utilized in the seeding experiment as the exact nucleating agent used, as well as the volumes and concentrations of specific reagents, were modified to produce the largest microtubule possible.

In general, as before, a standard polymerization buffer was first prepared by combining stock solutions of standard BRB80 buffer, 100 mM Mg2+ solution, 4 mg/mL GTP solution and 5% DMSO solution to produce final concentrations of 4 mM Mg2+, 0.16 mg/mL of GTP and 0.0024% of DMSO.[3]

6.25 µL of this polymerization buffer was then added to an aliquot of tubulin to produce a final tubulin concentration of 3.2 mg/mL within the solution. This solution was then incubated in a water bath for 30 minutes at 37 ͦC.[3]

These microtubules were then stabilized by adding paclitaxel and more BRB80. The stabilized microtubules were then imaged to see whether microtubules viable to serve as seeds were present. When stable candidates were observed, another aliquot of microtubules were reconstituted by using the same polymerization buffer WITHOUT any nucleating agent i.e. no DMSO, glycerol or guanosine-5’-[(α,β)-methyleno]triphosphate (GMPCPP). This aliquot was then combined with the seeds prepared from before and replaced within the water bath to incubate for at least 30 minutes before imaging.

On Imaging

Please note that all imaging in both sets of procedures took place using a Nikon TE2000 epifluorescence microscope equipped with an Andor Technology Zyla sCMOS camera. The tubulin aliquots utilized were purchased from Cytoskeleton Inc. and were tagged with rhodamine and/or hilyte GFP Fluor 488.


In addition to experiments, we developed a simulation in Python to model the growth of many microtubules within a small volume of solution. We used a stochastic simulation algorithm. There is a certain random timestep between reactions, and that timestep follows an exponential distribution. Either a polymerization or a depolymerization reaction occurs after each timestep, and the probability of either one occurring is determined by the propensity of each reaction, which is roughly equivalent to the reaction rate. These simulations were then compared to the behavior of microtubule growth found from the experiments in order to glean a future direction of experimentation to be taken.

Lab Notebook

Access the lab notebook here: Lab Notebook records.

Results - So what did we get?

Microtubules grown during the shearing experiments did not exceed the control average length of 18 - 20 µm.

The seeding experiments produced long and stable microtubules. Images taken with seeds and polymerization tubulin labelled with separate fluorophores suggest that the tubules grew from the seeds as shown below:


Microtubules generated from rhodamine tagged seeds and hilyte free tubulin during post-seeding polymerization. The seeds are bright segments on the microtubules

GMPCPP seeds were stable but did not form seeds longer then the average of the regular motility assay protcol. Seeds with glycerol were successful in producing longer microtubules. In fact, the largest microtubule we found all summer, measured at 144 µm, was found from a glycerol seeding experiment. The averages for these tubules were found to be approximately between 34 to 36 µm, which was higher than microtubules grown from a normal motility assay protocol, which were found to range between 18 to 20 µm.

Image 1: Seeds from Glycerol.

Image 2: Seeds from GMPCPP.

Image 3: Glycerol Seeds after second polymerization.

Image 4: GMPCPP Seeds after second polymerization.

Well what about the simulations? With the simulation prepared, we were able to plot the lengths of individual microtubules over time. We also constructed a kymograph to visualize a single microtubule growing over time. The relevant images summarizing the results of the simulation are given below. These are summarized in the next section.

Image 7: Lengths of 33 simulated microtubules over time. Traces of individual microtubules are shown in gray, and the population average is in red. The graphs show either A) traces for the entirety of the simulation or B) enlarged portions of the traces from A, taken from after the population average has reached steady-state.

Image 8: Kymograph of a single microtubule over time. The seed is marked in blue, the positive end is in red, and the minus end is in pink.

Image 9: Average, standard deviation, and length histograms over time for a population of 100 microtubules. Red points on the histograms represent the population averages. Note that simulation time is dimensionless.

Image 10: Effects of varying the parameters A) number of nucleation sites, B) tubulin equilibrium constant, and C) total tubulin concentration on the lengths of microtubules. All data is taken at the simulation time point t = 10^7. The parameters are normalized such that the default values (equivalent to the parameters used in Image 3) are equal to 1 and changes to each parameter are given as fold-changes. Red points on the histograms represent population averages.

Discussions - Well, What Does This All Mean?

Annealing requires time! (Or a lot of tubules!)

As we stated earlier, the first phase of experimenting was inspired by a paper written by Professor Hess and Dr. Jeune – Smith, first published in 2010, in which they explored the process of shearing and annealing microtubules.
Shearing refers to the process by which polymerized microtubules were broken down by exposing them to a shear force; this results in a cleavage of the microtubule as a result of the high mechanical force they are exposed to. Conversely, annealing refers to a process by which smaller microtubules join to form a longer whole; this involves them joining end-to-end, which means that the process is heavily dependent on the length and concentration of the microtubules present within a given population.[1] The effects of these processes in controlling the length distribution of a given population of microtubules are highlighted below:


Effects of Shearing and Annealing on Microtubule PopulationImage Source:[1]

The microtubules that we grew from our annealing experiments did not demonstrate the expected level of growth highlighted in the literature utilized. We expect that this is due to the concentration dependence of the annealing process; due to limitations in stocks of tubulin aliquots, the experimental concentration of tubulin utilized was below the volume-concentration of tubulin used within the literature we referred to.

Seeds seem to work!

Although annealing by itself did not, the microtubules grown from our glycerol seeds seem to have been extremely successful in growing. In order to test this, we imaged microtubules grown from free tubulin and seeds prepared with different fluorophores. When the resultant microtubules were imaged, we found that two separate, and distinct fluorescence patterns were observed pertaining to the area of the seeds and to the area of tubulin attached on to the existing seeds. Due to overlaps in the excitation spectrum of hilyte and rhodamine, both fluorescent molecules could be imaged with the Zyla Rhodamine excitation setting on the microscope to differing levels of fluorescence. The bright dots within the tubules are the seeds prepared when they were imaged with the Zyla Rhodamine excitation setting, since the seeds were prepared from rhodamine tagged tubulin.

Seeding experiments carried out from GMPCPP produced stable seeds but these seeds did not polymerize successfully. Clusters of tubules were found to form from our seeding protocol with GMPCPP when imaged under the microscope; all microtubules observed were found to be smaller than the normal motility assay protocol. Seeding experiments carried out with glycerol as the nucleating agent were largely successful in producing microtubules polymerized from the seeds. In fact, the largest microtubule we found all summer, measured at 144 µm, was found from a glycerol seeding experiment.

Once more, with simulations.

With this simulation, we were able to plot the lengths of individual microtubules over time, shown in Image 7. This demonstrates that some microtubules grow longer than others due to the randomness in the polymerization/depolymerization processes. We also constructed a kymograph to visualize a single microtubule growing over time, shown in Image 8. Growth from the positive end of the microtubule is more rapid than growth from the minus end.

We also observed how the distribution of lengths across the microtubule population changed with time by plotting a histogram of lengths at different time points throughout the simulation. Histograms of 100 microtubules for 5 different time points are shown in Image 9, as well as the average and standard deviation of the population over time. Early in the polymerization process, the mean length increases over time. Eventually the mean reaches a steady state, but the distribution continues to change shape after that point and the standard deviation increases. This represents a “dynamic equilibrium” in which tubulin continues to redistribute between different microtubules even after the total amount of polymerized tubulin has reached a steady state. Thus, the simulation allows us to examine the dynamics of individual microtubules, which provides information that we can’t get from just observing the population average.

Our mathematical model consists of 3 tunable parameters: the number of nucleation sites (M), the tubulin equilibrium constant (Teq), and the total tubulin concentration (Ttotal). We were also able to manipulate these parameters both experimentally (see experimental results section) and within the simulation (Image 10). The simulated data matches our mathematical model and the experimental results. Specifically, we were able to produce overall longer microtubules in the simulated populations by decreasing the number of nucleation sites, decreasing Teq, and increasing Ttotal.

The mathematical model and simulated data allow us to determine in more detail how the population changes with each parameter change. Note, for example, that average length increases linearly with Ttotal, while average length does not increase linearly with the other two parameters. The average length increases as the number of nucleation sites(M) and free tubulin at equilibrium(Teq) decreases. This information could help us in designing future experiments to further increase microtubule length.


1. Hess, H., Jeune-Smith, Y., R.S.C. (2010). Engineering the length distribution of microtubules polymerized in vitro
2. Mitchison, T., Kirschner, M., Nature (1984). Dynamic Instability of Microtubule Growth
3. Mitchison, T., Kirschner, M., Nature (1984). Microtubule assembly nucleated by isolated centrosomes
4.Rice University, Structures and Functions of Microtubules
5. Wikipedia, Image of Kinesin Walking On Microtubule, Kinesin
6. Wieczorek, Michal, et al. "Microtubule-associated proteins control the kinetics of microtubule nucleation." Nature cell biology 17.7 (2015): 907-916.
7. Idan, O., Lam, A., Kamcev, J., Gonzales, J., Agarwal, A., & Hess, H. (2011). Nanoscale transport enables active self-assembly of millimeter-scale wires.Nano letters, 12(1), 240-245.
8. Lodish, Harvey, et al. Molecular Cell Biology. New York: W.H. Freeman, 2013. Print.



Sasha Zemsky, Undergrad, Caltech BioE
Sasha Zemsky is a senior in bioengineering at Caltech and is from Mount Kisco, New York. She enjoys drawing, playing water polo, and summoning demons with her pals in Ricketts House.

Azraf Anwar, Undergrad, Columbia BME
Azraf is a third year BME Major at Columbia University hailing from Dhaka, Bangladesh. He enjoys reading, coding, and randomly rapping Shakespearian soliloquoys with Darnel.

Brandon Cuevas, Undergrad, Columbia BME
Brandon Cuevas is a first year student at Columbia University who intends to major in BME and is from Yonkers, New York. On his spare time, he loves to watch YouTube videos, play the piano, play basketball and spend time with friends.

Darnel Theagene, Undergrad, Columbia BME
Darnel Theagene is a third year BME Major from Hightstown New Jersey who likes to rock climb, play the saxophone and tell whale jokes.


Stas Tsitkov, Graduate Student, Columbia BME
Stas is a PhD Candidate at Columbia University. Outside of lab, he is a clarinetist in a chamber group and a sailor. In his free time he ponders on ways to discover the tachyon, terraform Pluto, and save the live action DC Extended Universe from Rotten Tomatoes...all using long microtubules.

Neda Bassir-Kazeruni, Graduate Student, Columbia BME
Neda is a PhD Candidate at Columbia University from Paris, France. When not in the lab, she enjoys reading, spending time with her friends and doing yoga.

Professor Henry Hess, Principal Investigator
Henry Hess is a Professor at the Biomedical Engineering Department at Columbia University. His research interests include nanobiotechnology, biomaterials science, and synthetic biology. A particular focus of his work is the engineering of hybrid nanodevices which integrate biological and synthetic components.

Follow: biomod