Stabilization of an electro-optic modulator for quantum communication using a low-cost microcontroller

Author: George Iskander
Mentor: Maria Spiropulu
Co-Mentors: Neil Sinclair, Cristián Peña, Si Xie
Editor: Hana Kim


An electro-optic modulator modifies an optical signal in response to an applied electrical signal. These modulators are essential components for many applications ranging from internet telecommunications to sensing to quantum information science.1 A Mach Zehnder electro-optic modulator is used to vary the intensity of a continuous-wave laser beam in order to produce pulses.2 The amplitude of the pulses is quantified by an extinction ratio (in units of dB),

r = 10 \log_{10} {\frac{I_{on}}{I_{off}}}, (1)

in which the intensity of the output light is I_{on} (I_{off} ) when the modulator is configured to produce maximum (minimum) light intensity. A Mach-Zehnder modulator relies on interferometric stability to achieve a constant extinction ratio, and in-turn a stable pulse amplitude (see Sec. II). Temperature variations of a Mach-Zehnder modulator can lead to variations in r due to phase advances or delays.3 Stabilization against temperature variations is often performed by measuring the output of the modulator and adjusting an applied electrical signal in response.
Here we describe a simple and robust feedback method for controlling the laser intensity from a commercially available 20 GHz-bandwidth Mach-Zehnder electro-optic modulator using a low-cost microcontroller.

Specifically, we program an open-source Arduino UNO R3 microcontroller with a 9 kHz-bandwidth gradient-ascent algorithm to stabilize and maximize r of GHz-bandwidth pulses by detecting light at the modulator output. The algorithm is designed to minimize the average amount of output light using a DC voltage drive to the modulator, an approach that ensures maximum r for low duty cycles. We measure an extinction ratio of 22 dB over several hours, equivalent to the extinction ratio measured over minute-long timescales with manual control of the DC voltage. We also discuss the applicability of our stabilized modulator for generation of high-fidelity time-bin quantum bits (qubits) for quantum communication, in which our extinction ratios the generation of qubits with 99% fidelity.

Operating Principle Of A Mach-Zehnder Modulator

Figure 1. Simplified schematic of a Mach-Zehnder modulator. Black (blue) lines represent optical waveguides (electrodes).

A Mach-Zehnder modulator is an interferometer that features phase control of the light by the electro-optic effect, see Fig. 1.2 Laser light enters the modulator using a fiber optics cable where it is subsequently split into two paths using a 50:50 mode coupler that acts in analogy to a free-space 50:50 beam splitter. The light is guided in an electro-optic material, such as LiNbO3. Each path directs light near an electrode. One electrode is driven with a high-bandwidth (AC) signal while the other features a lowbandwidth (DC) signal. The electric field emitted by an electrode causes a phase variation of the light propagating in the nearby waveguide due to the electro-optic effect, i.e. the index of refraction of the waveguide is dependent upon the applied electrical signal. Light from each path is then combined again on a second 50:50 coupler that features two outputs which direct light into independent optical fibers. The accumulated phase difference between light traversing each path dictates the relative intensity of the light that leaves each output port due to wave interference. Specifically, a linear variation of the relative phase results in a sinusoidally-varying light intensity at each output, see Fig. 2. In the case where the relative phase difference is an integer multiple of \pi(2\pi), all of the light will exit port 1 (2) through constructive (destructive) interference. However, this is never observed due to device imperfections, e.g. the splitting ratio of the couplers is not exactly 50:50. The extinction ratio quantifies the impact of these imperfections by the ratio of light intensities under the constructive and destructive interference conditions.

Figure 2. Light intensity at each output of the modulator. Output 1 (2) is shown in blue (orange) while the relative phase is varied by slowly varying the applied DC voltage. The green line indicates zero optical power. Energy conservation ensures that each \pioutput is out of phase. The extinction ratio is 22 dB and calculated from a fit. The dotted lines indicate Ion and Ioff. The measurement timescale is a few seconds.

A Mach-Zehnder modulator can be used create pulses from a continuous-wave laser input in the following way. First, a slowly-varying voltage is applied to the DC port to determine the voltage for which minimal light is produced at one output. In other words, the plot shown in Fig. 2 is produced and used to determine the DC voltage setting. This voltage is set to counteract the effect of any temperature-induced phase offset as mentioned in Sec. I. Next, a short voltage pulse is applied to the RF port such that its amplitude corresponds to varying the relative phase accumulation between \pi and zero, i.e. we ensure that maximum light intensity is briefly output of one port of the modulator. This results in the production of a short light pulse.

Methods and Setup

The principle of our stabilization scheme is to measure the second output port of the modulator and to apply a slowly-varying voltage to the DC port of the modulator to ensure that the first (second) port features maximum (minimum) light intensity. A key point is that our scheme will always ensure a high-extinction ratio when the modulator is used to generate high-bandwidth pulses at a low rate, i.e. the modulator is operated at a low duty cycle. This is due to the low operating bandwidth of the photodetector and microcontroller, which only responds to the time-averaged light intensity. For low duty cycle operation, the time-averaged intensity is determined mainly by light that is detected when the modulator is set to produce minimum light intensity, i.e. when the modulator produces Ioff.

Our experimental setup and feedback loop is shown in Fig. 3, while a photograph of our Arduino microcontroller setup shown in Fig. 4. Our system utilizes continuous wave laser light of 1536 nm wavelength and power of about 1 mW that is sent into a 20 GHz-bandwidth Mach Zehnder modulator. Light used for stabilization is detected by a low-bandwidth photodetector (i.e. a power meter), while the second output of the modulator is sent to a DC-coupled 20 GHz-bandwidth amplified photodiode and oscilloscope for measurement of extinction ratio and optical pulse amplitudes. The low-bandwidth photodetector outputs a voltage that is proportional to light intensity, and is sent into the Arduino, which interprets the signal, and generates a response signal that is applied to the DC port of the modulator. Light pulses are created by driving the RF port of the modulator with electrical pulses that are generated by an arbitrary waveform generator (AWG). The short-term, or intrinsic, extinction ratio of the modulator is determined by slowly varying the voltage applied to the DC port of the modulator. We measure the output of the modulator with the low-bandwidth photodetector, see Fig. 2, and fit the result with a sinusoidal curve to determine the short-term extinction ratio to be 22 dB, a value that is consistent with factory specifications.

Figure 4. Arduino, left, and its associated PCB, right. The PCB contains the analog-to-digital and digital-to analog conversion components that are necessary for interfacing with the low-bandwidth photodetector and the modulator.

We employ a gradient ascent algorithm to lock the modulator to a phase that corresponds to maximum light detected by the low-bandwidth photodetector, see Fig. 2. The voltage x^* corresponding to the peak is a critical point of the response function R(x). For small deviations (\epsilon) around the peak we have \frac{dR}{dx} (x^* - \epsilon) >0 and \frac{dR}{dx} (x^* + \epsilon) <0. Using this feature of the response function, the Arduino will find and lock to the maximum with the following procedure. The Arduino initializes by generating an initial voltage V. It applies a small increase to this voltage, so that the total voltage applied is V + V_{\delta}. It records the corresponding light intensity output from the modulator I^+. Then, it applies a small decrease such that the voltage applied is V - V_{\delta}. It records the intensity I^-. A check is performed to determine which is greater: either I^+ or I^-. If I^+> I^-, then V corresponds to a part of the response curve in which the slope is positive, and consequently V is less than the voltage that is needed to lock to the maximum. In response, the Arduino increases the voltage it applies. Conversely, if I^+ < I^-, then V corresponds to a part of the response curve that has a negative slope. In turn, the Arduino decreases the applied voltage. The procedure iterates at a rate of about 9 kHz until the maximum is found.

Gradient ascent is chosen due to its versatility in locating an optimum without defining a setpoint, compared to, e.g., often-used PID (proportional-integral-derivative) algorithms. We note that a gradient ascent algorithm is known to create oscillations in a small region about the maximum, since the program does not stop iterating. However, the sinusoidal response near each maximum and minimum is sufficiently small and, since V_{\delta} is also small, any oscillations are negligible.

To benchmark our results, we compare the performance of our Arduino-based (cost 50 USD) system to the unstabilized system and to a commercially-available feedback controller from YYLabs (cost 1000 USD). The YYLabs controller utilizes a PID algorithm and pilot tone frequencies.4

Results and Analysis

If no voltage is applied to any port of the modulator, we observe the output of the modulator to vary over several minutes, see Fig. 5.

Figure 5. Intensity of the output of the modulator without any applied voltage. The intensity is measured with the low-bandwidth photodetector and Arduino digital-to-analog converter. Fluctuations are measured out of the 216 bits available on the Arduino.

With the gradient ascent algorithm engaged, the extinction ratio is maximized and is stable over several hours. We measure the output intensity I_{off} of the second arm using an oscilloscope. Using I_{on} as determined from Fig. 2 and Eq. 1, we calculate an extinction ratio of 22.05 ± 0.13 dB. To confirm this result, we generate a 100 ps-duration pulse every 100 ns using our AWG and, with the feedback engaged, measure an extinction ratio of 22 dB, see Fig. 6.

Figure 6. Oscilloscope screenshot of a light pulse produced using our Arduino-stabilized 20 GHz Mach-Zehnder modulator. Photodetector is negatively coupled. Ripples are due to capacitive overshoot.

We repeat the procedure using the YYLabs controller, measuring a lesser extinction ratio of 20.5 ± 0.4 dB. Error is due to the limited resolution of the Arduino analog-to-digital and digital-to-analog converters.

Towards Generation of Quantum Bits for Quantum Communication

One application of our modulator setup is to generate qubits for quantum communication.5,6 Among the different ways a qubit may be encoded into light, one method involves the possibility of a photon arriving early, late, or in coherent superpositions of early and late, with respect to a pre-defined time. This so-called time-bin qubit.7 can be created by temporal (and phase) modulation of laser light that is attenuated to the single photon level.8 Specifically, an early time-bin qubit state corresponds to a photon arriving earlier than if the late time-bin qubit state is encoded, see Fig. 7. Note that, strictly speaking, this approach generates a quasi-qubit due to a non-zero two-photon probability from Poissonian statistics of laser light, an effect that can be accounted for in experiments.9

Figure 7. Temporal distribution of probability amplitudes (red) for early and late time-bin qubit states. (a, b) An ideal time-bin qubit corresponds to the case in which the probability amplitude is well-localized in time. (c, d) A low extinction ratio gives rise to a non-zero probability amplitude at all times, e.g. late when early is desired, or vice versa.

The quality of qubit generation is determined by a measurement fidelity F =  \psi_{\parallel}/( \psi_{\parallel} + \psi_{\perp}), in which \psi_{\parallel(\perp)} refers to the measurement basis oriented parallel (orthogonal) to the state that is intended to be generated. For time-bin qubit states that are encoded into early (late) states, the extinction ratio is r =10 \log_{10} {[F_{e(l)}/(1-F_{e(l)})]} if no other imperfections are present. See Fig. 7 and its caption. A similar calculation can be performed for early and late superposition states. Therefore, an extinction ratio that is maximal and constant over the duration of an experiment will produce time-bin qubits of the highest fidelity. Our Arduino system suggests F = 0.9938 \pm 0.002 for early and late qubit states, which is sufficient for quantum communication experiments.6 We also note that in addition to high fidelity, the compact and low-cost design of our feedback system makes it ideal for operating a Mach-Zehnder modulator outside of the laboratory environment (i.e. in a remote location) for real-world quantum communication experiments.9


We stabilize the intensity of the output of a high-bandwidth Mach-Zehnder electro-optic modulator by using a gradient-ascent algorithm encoded into an Arduino. We measure an extinction ratio of 22 dB over several hours, consistent with the specifications of the modulator. Our result is also consistent with the extinction ratio we measure over minute-long timescales using manual tuning. Our system, which costs approximately 50 USD improves over the performance of a YYLabs commercial feedback controller, which costs 1000 USD. Finally, we predict our system to generate early and late time-bin qubits of fidelity 0.9938 \pm 0.002, which is suitable for several quantum communication tasks, such as quantum key distribution or quantum teleportation.6,9

Further work involves exploring other feedback algorithms, such as PI control,10 which would require linearization of the response, dithering and lock-in demodulation, finer-resolution gradient-ascent, or using a detector with higher dynamic range, such as a superconducting nanowire.11 Additionally, other low-cost open-source microcontrollers warrant research, such as the more extensible, yet more complex, Raspberry Pi.


We thank Jason Trevor and Dr. Yewon Gim for their experimental assistance. G.I. acknowledges support by the WAVE Fellows program and Southern California Edison, N.S. and C.P. acknowledge support by the Alliance for Quantum Technologies’ Intelligent Quantum Networks and Technologies research program. This work is partially supported by the DOE/HEP QuantISED program grant, Quantum Communication Channels for Fundamental Physics, award number DE-SC0019219.


  1. E. L. Wooten, K. M. Kissa, A. Yi-Yan, E. J. Murphy, D. A. Lafaw, P. F. Hallemeier, D. Maack, D. V. Attanasio, D. J. Fritz, G. J. McBrien, and D. E. Bossi, “A review of lithium niobate modulators for fiber-optic communications systems,” IEEE Journal of Selected Topics in Quantum Electronics 6, 69–82 (2000).
  2. Tetsuya Kawanishi, “Integrated mach–zehnder interferometer-based modulators for advanced modulation formats,” in High Spectral Density Optical Communication Technologies, edited by Masataka Nakazawa, Kazuro Kikuchi, and Tetsuya Miyazaki (Springer Berlin Heidelberg, Berlin, Heidelberg, 2010) pp. 273–286. 
  3. J. D. Zook, D. Chen, and G. N. Otto, “Temperature dependence and model of the electro-optic effect in linbo3,” Applied Physics Letters 11, 159–161 (1967),
  4. E. I. Ackerman and C. H. Cox, “Effect of pilot tone-based modulator bias control on external modulation link performance,” in International Topical Meeting on Microwave Photonics MWP 2000 (Cat. No.00EX430) (2000) pp. 121– 124.
  5. J. Brendel, N. Gisin, W. Tittel, and H. Zbinden, “Pulsed Energy-Time Entangled Twin-Photon Source for Quantum Communication,” Physical Review Letters 82, 2594–2597 (1999).
  6. Nicolas Gisin, Gregoire Ribordy, Wolfgang Tittel, and Hugo Zbinden, “Quantum cryptography,” Rev. Mod. Phys. 74, 145–195 (2002). http://doi.org10.1103/RevModPhys.74.145.
  7. I. Marcikic, H. de Riedmatten, W. Tittel, V. Scarani, H. Zbinden, and N. Gisin, “Time-bin entangled qubits for quantum communication created by femtosecond pulses,” Physical Review A – Atomic, Molecular, and Optical Physics 66, 6 (2002).
  8. Wolfgang Tittel and Gregor Weihs, “Photonic entanglement for fundamental tests and quantum communication,” Quantum Information & Computation 1, 3–56 (2001).
  9. Raju Valivarthi, Marcelli Grimau Puigibert, Qiang Zhou, Gabriel H. Aguilar, Varun B. Verma, Francesco Marsili, Matthew D. Shaw, Sae Woo Nam, Daniel Oblak, and Wolfgang Tittel, “Quantum teleportation across a metropolitan fibre network,” Nature Photonics 10, 676–680 (2016).
  10. L. R. Hofer, D. B. Schaeffer, C. G. Constantin, and C. Niemann, “Bias Voltage Control in Pulsed Applications for MachZehnder Electrooptic Intensity Modulators,” IEEE Transactions on Control Systems Technology 25, 1890–1895 (2017).
  11. F. Marsili, V. B. Verma, J. A. Stern, S. Harrington, A. E. Lita, T. Gerrits, I. Vayshenker, B. Baek, M. D. Shaw, R. P. Mirin, and S. W. Nam, “Detecting single infrared photons with 93% system efficiency,” Nature Photonics 7, 210 EP – (2013).


The Arduino control loop makes use of the following hardware: the Arduino UNO R3, Adafruit MCP4725 12-bit digital-to-analog converter (DAC), and Adafruit ADS1115 16-bit analog-to-digital converter (ADC).

Without the ADC and DAC, the analog input precision of the Arduino is limited to 10 bits, and analog output is not possible. To circumvent these limitations, the DAC is used to enable analog output, and the ADC is used to increase input precision. These two devices are breakout boards which utilize the I2C communication protocol.

The code libraries used include the ADC and DAC libraries from Adafruit, the ResponsiveAnalogRead library (RAR), and the Wire library. RAR implements exponential moving averages to reduce input noise from the ADC. The Wire library allows for I2C communication with the
breakout boards.

The following circuit diagram shows how the boards connect to the Arduino.

Note that the VOUT pin on the MCP4725 connects to the intensity modulator, and any one of AIN pins on the ADS1115 can be connected to the output of the low-bandwidth photodetector. For this code and project, we use pin 0 (AIN0).

The code running on the Arduino is as follows:

#include <ResponsiveAnalogRead.h>
#include <Adafruit_MCP4725.h>
#include <Adafruit_ADS1015.h>
#include <Wire.h>
Adafruit_ADS1115 adc;
Adafruit_MCP4725 dac;
ResponsiveAnalogRead analog(0, true);
double vStep = 5.0;
double vLess;
double vMore;
double voltage = 0;
int i;
void setup() {
void loop() {
dac.setVoltage(voltage - vStep, false);
vLess = adc.readADC_SingleEnded(0);
vLess = (analog.getValue()/8.0);
dac.setVoltage(voltage + vStep, false);
vMore = adc.readADC_SingleEnded(0);
vMore = (analog.getValue()/8.0);
if(vMore > vLess) {
voltage += vStep;
if(vMore < vLess) {
voltage -= vStep;

The variable ”vStep” encodes the step-size of the voltage. Trial-and-error may be needed to determine a suitable value. Too small of a value, and increasing or decreasing the output voltage by the step size will produce no voltage change. Too large of a value, and oscillations will be observed around the maximum.

Note that the ADC maps voltages from [0V, 5V] to [−215, 215]. The output precision of the Arduino is limited to 12-bits, so in order to simplify calculations, any readings the Arduino makes of the power measurements are divided by 8 so that all calculation is performed over 12-bits.

Interview with Professor Aaron Ames

Interviewer: Sherry Wang

Can you give me an overview of your research?

Broadly speaking, what I do is study how robots walk. That’s kind of the core disciple we consider in my lab; that is, specifically achieving walking behaviors on robots that are as dynamic and dexterous and fluid as humans walking. So this actually involves quite a bit of math to try to understand and characterize what walking is, and once you can do that, you can apply it to a lot of application domains, especially domains where humans walk, or humans walking is part of it. What I mean is, robotic assisted devices, prosthetics, exoskeletons, etcetera. So I have a wide spread of robotic applications, but the real focus is dynamic locomotion.

How did you get involved in robotics, and in particular dynamic walking and dynamic motion?

It’s an interesting story that is probably not wholly unique from a Caltech perspective. I was introduced to science fiction as an undergrad, so later in life than most people. I started reading Asimov and all these other authors, and I just got totally hooked. I was just fascinated with this idea of science fiction in general, but a big theme of a lot of Asimov’s books were robotics, and humanoid robots, and all the robots playing a part in our daily lives, and thinking about what that would really mean if we could really realize those behaviors. In undergrad, I did mechanical engineering and math, and I wanted to go to grad school to really understand this better. I think what appealed to me the most was just this deceptively simple thing that we do. Humans walk around all the time, they make dynamic movements all the time, and it seems like something we should be able to understand. Yet under the surface it’s such a complicated problem I find it intriguing. It’s one of those cases where you actually know a solution exists. Feasibility is guaranteed, because humans do it all the time. You just have to be smart enough to try to find that. I was purely theoretical in grad school. I didn’t touch a robot. I didn’t touch an experiment, I didn’t build anything, I just proved theorems and theorems and theorems. I think that made me a better person, but because my passion was robots–it was something I always really loved–I ended up coming back to it as a professor. It was a couple years into being a professor, and I had been doing some walking stuff and some simulation and everything, and I realized no one listens to me, and everyone probably thinks I’m crazy because the math I was doing was really out in left field. The only way I can actually convince people is if I get stuff to work. If I get physical hardware to do awesome stuff, then maybe they’ll listen to the math. I started doing that with simple robots, and making that one walk, and then I started getting addicted and making more and more robots; I worked with NASA on a humanoid. So I started getting really into the robotics side, while at the same time not losing the math that got me excited about this stuff and keeps me going to this day. Intrinsically, it was science fiction that led to studying a whole bunch of math, that led to me actually making robots to do stuff to demonstrate the math.

What are some difficulties and successes that you’ve seen in your research?

Difficulties always stem from the fact that the real world is hard. I always say to my students, and it’s a pretty silly quote but descriptive nonetheless, which is “hardware is hard.” There’s a “hard” in the “hardware” for a reason. The reality is that however smart you think you are–however clever you think you are–you go to put something on a robot and Mother Nature lets you know very quickly that you’re stupid, and you don’t know anything, and the world’s really hard. And you constantly are reminded of this. It’s not so much a difficulty; I view it almost as an opportunity, but it’s humbling. You’re constantly humbled, and what it takes to make anything work is so incredibly complicated. It’s exactly this difficulty that gives you opportunity, and what’s exciting and leads to your success story. Especially when you try to do what’s unique to my lab, (which is not to say that there’s no other labs that do it, but it’s a fairly unique thing) which is trying to do real theory–theorems and proofs and mathematically justifiable things–and then put that on hardware. So the goal is not so much to make the hardware do something, but the goal is to understand something at a deep mathematical level, and then make the hardware demonstrate that understanding. Recently, we worked with a French startup company called WonderCraft on their powered lower-body exoskeleton. And this is specifically aimed at paraplegics. There has never before this time been a powered lower-body exoskeleton that allowed paraplegics to walk dynamically. That is, walk without crutches, move their hands freely, and have a dynamic gait. And they [WonderCraft] took math that was developed by me–my lab, with collaborators and everything–but they took theorems that were in papers that we wrote, and they put it on the exoskeleton, and they actually got paraplegics to walk for the first time ever with a powered assisted device. So that was another one of these moments. You’re writing the papers, you’re proving the theorems–in this case, it was a paper that appeared in 2014–you would never think that in four years that that math would be helping a paraplegic to walk. Those are the moments that everything comes together, and it all looks perfect. And then you realize, “time to go back and try to do something new.” Now we want to have the next behavior, and nature starts smacking you down again. Then nature starts reminding you that you have a lot to learn. The success is great, but there’s always problems out there–which I find exciting. There’s never a lack of problems to solve.

What behaviors are you investigating now?

On the legged robots side, we really want to get robots out in the real world more and more. We want to have them do things that are impressive and dynamic, but in a really unstructured environments. Examples are, we want to take a robot on a hike. We want it to be able to hike one of the trails up in the mountains near Caltech. To actually be able to handle all these different terrain types autonomously, rather than just a pre-prescribed [route]. And not just walking; we’d love to do running and jumping. We’ve made a robot jump in our lab, which was amazing, but we’d like to get dynamic behaviors outside, in the real world. So on the walking robot side, that’s what it’s all about to us. It’s handling more and more complex terrain in a robust and dynamic way.

And in the same light, translating those behaviors that we’ve achieving on walking robots to robotic assisted devices. We actually have one of these exoskeletons by WonderCraft in our lab at Caltech now. We’re putting our new, more advanced stuff on it and testing it out, and we’d love to get that device walking outside, and walking on sand and dirt. Right now it’s walked dynamically, but in clinical settings. Everything we’re doing is pushing outside the lab. Start to handle the complexity of the real world. And that’s where a lot of interesting theoretical problems come from too. How do you quantify that uncertainty? How do you potentially bring in machine learning to make sense of that uncertainty, but use it in the context of the models and the theories that governs the motions of the systems. That’s where we’re pushing: trying to get robots into the real world, and have them try to do fun things.

Do you see the field progressing toward the “singularity” that science fiction authors write about?

I’d love to think that we’re that close, but no, we’re way off from any sort of “singularity”. We can’t even make robots do simple stuff still. If you think about the million years of evolution that it took for humans to do the things that we do, we haven’t recreated a fraction of it. We’ve come an incredibly long way. When I started doing experimental robotics, I had really simple robots and just making them walk dynamically was a big deal. This was before Boston Dynamics days. At the time, Boston Dynamics had quadrupeds that were cool, but no walking robots–no bipeds–and that was only like eight years ago, so it wasn’t that long ago. And now, we have humanoid walking robots. I take our biped from our lab outside and have it walk on grass and dirt, completely untethered, 3D walking dynamically. We’ve made a huge progression towards getting robots out of the lab and into the “real world”, but it’s still very scripted. When we have our walking robots walk on grass or dirt, we get it in various specific scenarios, and we tune the robot right, and we can’t have it go walk some mountain trail, or go walk downtown Pasadena and pick me up a cup of coffee and bring it back to me. These higher level behaviors and reasoning are an enigma to this day. And in some ways, the world of robotics is also bifurcated at this point. There’s a whole wealth of people that are working on getting robots to do dynamic things, and then there’s a whole wealth of people working on machine learning and AI, and those are different people. The things that [machine] learning is good at are things that can’t, right now, be translated to actual robots. And conversely, the things that we can make robots do doesn’t currently doesn’t inform machine learning. I think that’s starting to change, and people are looking for that intersection. [However,] to talk about a singularity is to talk about a robot that can do everything from walk effortlessly, to actually reason about its environment, much less become conscious. We’re just not even close. Right now, if I want to lose my robot, I’ll walk into a volleyball court and it will fall on its face, I guarantee–any robot will. Soft sand, or go in some snow, and it’s over. Until we can do those basic things that animals do so easily, much less humans, we’re a long ways off from achieving a “singularity”.

What are some of your interests outside of science?

I still love science fiction to this day. Books, movies, all of it, I find it really exciting. There’s nothing like a good, corny science fiction movie when you’re burned out mentally, or a good sci-fi book to rekindle that excitement in science. Little things just around LA. We try to take hikes–I have a dog, so I love going on walks with my dog and my wife, and going on hikes around Pasadena.

Do you have any career advice for students interested in robotics?

It’s an exciting area with a lot of potential. It’s interesting because the robotics push has been going on for a while now. It was a “dead” area for a while–very small–and it just bloomed research-wise in the last decade. And it doesn’t look like it’s slowing. We’re seeing more and more robots going into the real world. One of my students is the CTO of a startup that’s doing cooking robots–Miso Robotics–and he’s getting these robots into Dodger Stadium and cooking burgers. So you’re starting to see robots in the real world, so it’s an exciting time to be in the domain. My advice for students that are interested is to, at the undergraduate level, get involved with research labs. There’s no replacement for getting your hands on hardware and starting to play with these things early on in your undergraduate studies. Go to grad school because that’ll give you a depth in robotics that’s hard to emulate. Get involved in research because that’s where the exciting stuff is still happening. That’s not to say that industry doesn’t have exciting things, but the next generation [is] being developed [in research labs]. Get involved early, stay involved, and be part of some of these fun projects. The great thing about robotics is that it doesn’t take years of training to do something. You get in with an undergraduate degree–the standard courses you have–and pretty quickly make something happen that can motivate you to go deeper and deeper into the area.

Phenotypic Analysis of Autism-associated mutations in C. elegans

Author: Judy Kim
Mentor: Dr. Paul Sternberg
Co-Mentor: Sandy Wong
Editors: Sherry Wang and Jonathan Chan


Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder with a large genetic component. Among these rare variants that are related to ASD, missense mutations (a type of point mutation where a single nucleotide change results in a codon that codes for a different amino acid) make up nearly 30% of the rare variants.1 Because missense mutations are so abundant, they are usually difficult to study in most organisms; nearly five decades ago, however, biologist Sydney Brenner developed the nematode Caenorhabditis elegans as a genetic model for understanding developmental biology and neurobiology.2 C. elegans is a model organism to study mutations because it is easy to genetically manipulate, has a life cycle of three days, is transparetnt, a well-annotated genome, etc.4 Properly understanding missense mutations in C. elegans might help us better understand Autism’s mechanism and effects in humans.       

Patients with syndromes affected by BRAF mutations were often found to exhibit autistic features, such as social impairment and internalizing and externalizing problems, as measured by commonly used ASD-related diagnostic tools.5,6

Similarly, TRIO has been found that when expressed in neurons, these mutations produce a wide range of alternations in glutamatergic synapse function that is similar to those observed in current animals models of ASD. Among the human genome, there are two genes, BRAF and TRIO, which are well-known oncogenes (genes that can transfer into tumors). BRAF and TRIO were chosen because these are the human orthologs of lin-45 and unc-73, respectively.  BRAF codes for the B-raf protein, which is involved in sending signals inside the cells that are key in cell growth, was shown to be mutated in some human cancers.7 TRIO, coding for a guanidine nucleotide exchange factor, was frequently amplified and abundantly expressed in soft tissue sarcomas.8 Extensive research has focused on the cancer-inducing effects of these genes, but little is known about the genes’ neuronal functions. Molecular genetics research has been done with BRAF and TRIO on other animals such as rabbits,9 dogs,10 and mice.11 A previous study conducted on C. elegans indicated that there is interaction between lin-45 and unc-73 after using a RNA interference method in C. elegans to test approximately 65,000 pairs of genes for their ability to interact genetically.12

Chemosensory cues can lead to chemotaxis, rapid avoidance, and changes in movement; these behaviors are mainly regulated by chemosensory organs, which contain eleven pairs of chemosensory neurons.13 AWA and AWC are type types of olfactory neurons that sense volatile odors and is tested through chemotaxis to volatile chemoattractants in C. elegans. The lin-45 mutant is known to be phenotypically defective for these neurons during chemotaxis, meaning the worms cannot sense the odor.14 Olfactory signaling is initiated by interactions between odorants and olfactory receptors.

odr-10 was used as a positive control used because these mutants have a specific defect in chemotaxis to diacetyl, of several odorants detected by AWA olfactory neurons.14 Since there is little known about these genes and their impact on neuronal function, we wanted to investigate this function in C. elegans, specifically by comparing the established wild-type strain to the newly generated double mutant worms by genotyping and setting up multiple crosses.

These studies inform the various types of neuronal functions that preexist in these genes. A study that investigated the neuronal functions of lin-45 and unc-73 demonstrated that although the nervous and immune systems influence each other, the complex nature of these systems in mammals makes it difficult to determine how neuronal signaling influences the immune response.15 Another study also found that serotonergic chemosensory neurons modify the C. elegans immune response by regulating G-protein signaling in epithelial cells.15  Finally, we also know that unc-73, signals through nucleotide exchange factors such as RhoGEF-2 to regulate pharynx and vulva musculature and to modulate synaptic neurotransmission.16

In this study, we investigate whether diacetyl is an attractant to N2 wild-type and PS double mutant strains that were generated to ensure that the chemotaxis assay was working properly. The purpose of the chemotaxis assay is to investigate whether the worms are attracted to diacetyl. The odr-10(ky225) gene impair C. elegans so these mutants served as a chemotaxis to low concentrations of the odorant diacetyl.13 Our null hypothesis held that the odr-10(ky225) mean of the Region of Interest value is equal that of N2. After confirming the assay was working, we investigate whether diacetyl was an attractant to the double mutant strains like it was for N2 wild-type. The null hypothesis of the question was that the mean of the wild-type, mean of unc-73(sy898);lin-45 (sy875), mean of unc-73(sy896);lin-45 (sy875), and mean of unc-73(sy892);lin-45 (sy875) were all equal to each other. If our null hypothesis is void, then we know that there is statistical significance among the means.


The C. elegans used in this project were all synchronized at the L4 stage. After genotyping the F1(first filial) and F2 (second filial) crosses and finding new double mutant strains by identifying mutants that were homozygous for both unc-73 and lin-45, chemotaxis was performed to examine the worms’ attraction towards diacetyl and ethanol (Figure 1). Chemotaxis is the movement of an organism in a direction corresponding to a gradient of increasing or decreasing concentration of a particular substance such as diacetyl, an attractant that has been used consistently in previously studied literature.

Figure 1. Cross Overview. This diagram highlights the ideal F1 and F2 after each cross to ultimately generate a new double mutant strain.


The cross overview highlights the F1 and F2 crosses that were performed to find the new double mutant strains. In the first cross, we placed +/+; +/+ males in to a dish of +/+; lin-45/lin-45 hermaphrodites. From this cross, males were picked to use in the next cross. In the second cross, +/+; lin-45/+ males were crossed to unc-73/unc-73; +/+ hermaphrodites. After about three days, worms from the F1 progeny were then genotyped. For the F2 progeny, we repeated a similar process to find homozygous mutants in both lin-45 and unc-73 to generate new double mutant strains.

To find the double mutant strains, the F1 and F2 were genotyped as following:

Worms were lysed adding 25uL of proteinase K from stock (10mg/mL) to 100uL of lysis buffer for use in PCR. In worm lysis, the lysis buffer solution was first prepared. Lysed worms were placed in wells such that one well in each 8-strip tube was used for the negative template control and another well was used for the N2 wild-type worms. Afterwards, the worms were centrifugated, then frozen for 15 minutes at -80°C. Finally, lysis was performed on the thermocycler for a period of 1 hour at -80°C.

PCR MasterMix was performed to amplify the lysis product. Once the lysis reaction was completed, DNA was also added. Two sets of PCR for lin-45 and unc-73 were performed. After the PCR was completed, lin-45 and unc-73 restriction enzyme in buffer to DI water and 5uL of DNA from the PCR were added to cut the DNA in the expected regions after incubating at 37°C for 1 hour.

Gel electrophoresis (1 hr, 130 V) was run to locate the wells that had the homozygous mutants for lin-45 and unc-73. Then, we distinguished the wells that were common in both strains to generate new double mutant strains. If, in the F2 stage, we could not find wells that were common in both strains, we had to separate the F2 and run another set of genotyping.

Chemotaxis assay

Approximately 20 L4 stages worms of these 3 double mutant strains, with approximately 5 worms per strain, in addition to N2 wild-type strains, and a mutant of odr-10, with five worms per plate were prepared. Four acrylic plates of each strain with a total of 12 plates were prepared for chemotaxis. After a few days, the worms were then used in chemotaxis. First, diacetyl (1:10,000 in ethanol) was prepared and placed on a pre-drawn template (Figure 2).

Figure 2. Chemotaxis diagram with indicated regions of ethanol and diacetyl.

1uL of 1M sodium azide was added on each spot to paralyze the worms with its toxicity and arrest worms within a centimeter of where it is placed. For each plate, 1uL of ethanol was pipetted on the spots labeled as “X” and 1uL of diacetyl on “O”. The worms were then washed multiple times using S-basal buffer. After the final supernatant removal, about 100 worms were transferred to the center of the chemotaxis plate under the microscope. The worms were chemotaxed for 1 hour, then burned.


Generation of unc-73 and lin-45 double mutants     

After genotyping the F2 progeny, three double mutant strains, unc-73(sy898);lin-45 (sy875), unc-73(sy896);lin-45 (sy875), and unc-73(sy892);lin-45 (sy875), were generated. An example of a F2 gel image is shown in Figure 3, which highlights the homozygous, heterozygous, and wild-type wells. The double mutant strains are those that are homozygous for both unc-73 and lin-45.

Figure 3. Gel Image example of F2 #115/80. In unc-73, there was aband length of 436 for wild-type and a band length of 187, 249 for the mutant.

Verification of chemotaxis assays

As seen in previous literature, the odr-10 mutant had a higher chemotaxis ROI index than did N2, confirming that more worms were concentrated in one region and more attracted to the diacetyl than they were to the ethanol. The final image of the worms after the worms were burned on the Bunsen burner is shown in Figure 4.

The plates were captured and the worms were counted using ImageJ.The different regions on the chemotaxis plates are illustrated in Figure 2. The chemotaxis ROI index was calculated as the following:

For statistical testing, a statistic software (SPSS) was used to perform a one-way ANOVA with post-hoc Tukey HSD. A one-way ANOVA is used to determine if there are any statistically significant differences between the means of two or more independent groups. When there was a statistically significant difference in group means, a post-hoc Tukey HSD tested whether significant differences occurred between groups.

Figure 4. N2 Chemotaxis image taken. After 1 hour of chemotaxis, the worms were flamed using a Bunsen Burner.  
Figure 5. Chemotaxis ROI result for N2 and CX 3410. Bar graph shows the ROI index for both the wild-type strain and the positive control strain. CX3410 is the strain that has the odr-10(ky225) gene with the diacetyl.

Examination of unc-73 and lin-45 double mutants in chemotaxis assay

After verifying our chemotaxis assay, we used it to test the diacetyl chemosensory functions in the double mutants of unc-73 and lin-45. Region of Interest (ROI) value, standard deviation, and standard error of the double mutant strains are shown in Table 1.

Table 1.  Chemotaxis results of each strain.

Figure 5 demonstrates our chemotaxis assay was consistent with the results from previous studies that observed a higher ROI value for N2 than that of CX3410. This means that more wild-type worms were attracted towards diacetyl than they were towards ethanol. With an α- level of .05 and a t-test, odr-10 mutant shows statistically lower values in chemotaxis score compared to N2.; t(1), p < .0001. Thus, we rejected the null hypothesis that the mean of the wild-type is not equal to the mean of odr-10(ky225) and therefore expect an equal distribution in both regions 1 and 5, as shown in the plate template.

Figure 6. Chemotaxis ROI results for N2 vs. PS7876, PS774, and PS7878. Bar graph shows the ROI index for both the wild-type strain and the double mutant strains.

Figure 6 demonstrates that mutants are differently attracted to diacetyl. With an α- level of .05 and a one-way ANOVA test, there was a statistically significant difference between the means of the double mutants N2, unc-73(sy898);lin-45 (sy875), unc-73(sy896);lin-45 (sy875), unc-73(sy892);lin-45 (sy875); p < .05.However, the post-hoc Tukey test results show that there is no statistical significance among the means of the double mutant strains to the wild-type strain (p < .01).

Table 2. Chemotaxis strain information with the human amino acid change.


Our result showed that N2 was significantly different from the diacetyl defective odr-10 mutant, proving the validity of our assay. However, the chemosensory index of double mutants did not differ from wild-type, indicating the unc-73 and lin-45 double mutants can sense diacetyl as well as wild-type.

Our results did not support the hypothesis that unc-73 and lin-45 double mutants exhibit diacetyl sensory defects. This may due to the huge variations in some double mutant strains such as PS7876, PS7878, and PS777. It may also be due to the missense mutants on these two genes located in domains that do not interact with each other.12, 13, 14  C. elegans only has 32 chemosensory neurons, but chemosensation is its most complex sensory measure and the main mechanism by which the animal makes judgments of its environment. Since there is a chemical specificity of odr-10 and the small number of sensory neurons, each cell might need to express many receptors for the animal to recognize the full spectrum of salient odorants.12

Future studies could explore a larger sample size and better control maturity and synchronicity of worms. Even though there was no statistical significance among the comparison of each double mutant to the mean of the wild-type ROI, we hope to eventually use this method in the human gene alignment to characterize new missense alleles for ASD patients (Table 2). In future studies, with a longer period of time and a larger sample size, more double mutant strains will be able to generated, chemotaxed and studied. This information will ultimately contribute to implications of medical practices and genetic studies. 


To investigate the genetic interaction of BRAF and TRIO, we use C. elegans equivalents of human missense mutations genes, unc-73 and lin-45. In this study, we generated unc-73 and lin-45 double mutants and examined the diacetyl chemosensory functions in these mutants. First we determined if diacetyl is an attractant to N2 wild-type and PS double mutant strains that were generated to make sure that the chemotaxis assay was consistent with previous literature. After confirming the assay was working, we investigate whether diacetyl was an attractant to the double mutant strains like it was for N2 wild-type and found that there was a statistically significant difference among the means.


I would like to thank my mentors, Sandy Wong and Professor Paul Sternberg for their guidance and support throughout the duration of this study. I would also like to thank the rest of the Sternberg Lab for their support. Lastly, I would like to thank the Caltech Student-Faculty Programs for giving me the opportunity to participate in research and the Simons Foundation Autism Research Initiative (SFARI) for awarding me as a SURF Fellow recipient with funding towards this project.


  1. Simons Foundation Autism Research Initiative. (n.d.). Retrieved from
  2. Brenner, S., 1988 Foreword. The Nematode Caenorhabditis elegans, edited by W. B. Wood, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.
  3. Brenner, S., 2002 The worm’s turn. Curr. Biol. 12: R713. Abstract Article
  4. Corsi, A. K., Wightman, B. & Chalfie, M. A. Transparent window into biology: a primer on Caenorhabditis elegans. Genetics 200, 387–407 (2015).
  5. Alfieri P, Piccini G, Caciolo C, Perrino F, Gambardella ML, Mallardi M, Cesarini L, Leoni C, Leone D, Fossati C, Selicorni A, Digilio MC, Tartaglia M, Mercuri E, Zampino G, Vicari S (2014). Am J Med Genet A. 164A(4):934-42.
  6. Adviento et al., 2014
  7. Davies H, Bignell GR, Cox C, et al. Mutations of the BRAF gene in human cancer. Nature 2002; 417:949-954
  8. Adamowicz, M. et al. Frequent amplifications and abundant expression of TRIO, NKD2, and IRX2 in soft tissue sarcomas. Genes Chromosomes Cancer 45, 829–838 (2006).
  9. Yoon, H. , He, H. , Nagy, R. , Davuluri, R. , Suster, S. , Schoenberg, D. , Pellegata, N. and Chapelle, A. d. (2007), Identification of a novel noncoding RNA gene, NAMA, that is downregulated in papillary thyroid carcinoma with BRAF mutation and associated with growth arrest. Int. J. Cancer, 121: 767-775.
  10. Mochizuki H, Kennedy K, Shapiro SG, et al. BRAF mutations in canine cancers. PloS One 2015; 10:e0129534.
  11. Shimamura, M., Shibusawa, N., Kurashige, T., Mussazhanova, Z., Matsuzaki, H., Nakashima, M., Yamada, M., … Nagayama, Y. (2018). Mouse models of sporadic thyroid cancer derived from BRAFV600E alone or in combination with PTEN haploinsufficiency under physiologic TSH levels. PloS one13(8), e0201365.
  12. Lehner, C. Crombie, J. Tischler, A. Fortunato, A. G. Fraser, Systematic mapping of genetic interactions in Caenorhabditis elegans identifies common modifiers of diverse signaling pathways. Nat. Genet. 38, 896 (2006). 
  13. Hirotsu, T., Saeki, S., Yamamoto, M., & Iino, Y. (2004). Corrigendum: The Ras–MAPK pathway is important for olfaction in Caenorhabditis elegansNature, 432(7017), 653-653.
  14. Sengupta, P., Chou, J. H. & Bargmann, C. I. odr-10 encodes a seven transmembrane domain olfactory receptor required for responses to the odorant diacetyl. Cell 84, 899–909 (1996).
  15. Anderson, A., Laurenson-Schafer, H., Partridge, F. A., Hodgkin, J., & McMullan, R. (2013). Serotonergic chemosensory neurons modify the C. elegans immune response by regulating G-protein signaling in epithelial cells. PLoS pathogens9(12), e1003787.
  16. Steven, R., Zhang, L., Culotti, J., & Pawson, T. (2005). The UNC-73/Trio RhoGEF-2 domain is required in separate isoforms for the regulation of pharynx pumping and normal neurotransmission in C. elegans. Genes & development19(17), 2016-29.

Culture Conditions Influence Iridescence in Marine Microbes

Author: J Livingston
Mentor: Dr. George O’Toole, Dr. Dianne Newman,
Co-Mentor: Lynn Kee
Editor: Grace Xiong


Microbiology is the study of microorganisms: how they grow, where they are found, and how they interact with each other and their environment.1,2 The study of microbiology allows researchers to develop a better understanding of the basic processes of life and has led to breakthroughs in medicine, agriculture, and industry.1,2 Some of the most important discoveries in the field of biology have been made by studying and isolating seemingly insignificant microbes. For example, the widely used technique of PCR (Polymerase Chain Reaction) was developed by studying the extremophilic bacterium Thermus aquaticus,3 isolated from hot springs in Yellowstone National Park.  The discovery of CRISPR-Cas9, a revolutionary gene-editing mechanism, was made while working with bacteria in industrial yogurt making.4.

Iridescence is a widespread phenomenon in the natural world, appearing in animals ranging from fish to butterflies.5,6,7 However, it is poorly characterized in single-cell organisms like bacteria, particularly because of historical misunderstandings of the term.8 Some older publications use “iridescence” interchangeably with “fluorescence,” although they are distinct phenomena.8 While fluorescence is the emission of light, iridescence is the interaction of reflected light waves that result in different observed colors based on the angle of view or illumination.7,8 In the past decade, work has been done to characterize and define bacterial iridescence. Different categories of iridescence have been defined, including rainbow, metallic, and “glitter-like.”7

Bacterial iridescence, as described in marine bacteria of the family Flavobacteriaceae, results from photonic crystals formed by periodically ordered cell populations in colony biofilms.9 Bacterial iridescence has also been linked to gliding motility, which is a form of movement used by bacteria to move across a surface without a propulsive structure.10,11 While previous studies described and characterized iridescence in members of the genera Flavobacteria and Cellulophaga,5,6,8,9,10,12 iridescence in the Tenacibaculum genus had not been described well. Recent work at the Marine Biological Laboratory (Woods Hole, MA) has begun to decipher both the physical and genetic basis of iridescence in Tenacibaculum.13,14,15,16,17

The biological function of iridescence in bacteria remains unknown. Some hypothesize that the structure that gives rise to iridescence may give the bacteria a selective advantage in the dynamic marine environment in which they live. Iridescent bacteria are found in rocky shore environments that are characterized by significant fluctuations in temperature, dryness, and salinity, so these iridescent structures might help them survive in changing environmental conditions.  It is also possible that the iridescence itself is an unimportant characteristic associated with an important biological structure, only by coincidence appearing as iridescence when the bacteria is grown in the lab. To better understand the functions of biological iridescence, we isolated and classified several new strains of iridescent Tenacibaculum, and characterized their iridescence in different culture conditions. By defining the conditions that promote iridescence, we hope to uncover the biological significance of this phenomenon.

Results and Methods:

To begin this project, we isolated iridescent bacteria from seawater samples collected between the rocks in a jetty at Stony Beach in Woods Hole, MA. We serially diluted each sample and spread them onto seawater complete (SWC) agar plates. Each plate was then incubated overnight at 30 ˚C and grew many different bacterial colony morphologies.

Iridescent colonies were isolated from these plates by using plastic inoculation loop to pick up the colony of interest and drag it on a new plate of SWC. This was repeated until the microbial growth appeared the same, indicating that the isolates were genetically identical. The iridescent microbes resembled those used in an ongoing project, which have been identified as Tenacibaculum discolor. We named this first strain CA-IR1 (Figure 1). Following this isolation, more samples were taken from Stoney Beach and three more strains of iridescent microbes were isolated using the same procedure, named CA-IR2, CA-IR3, and CA-IR4. We determined that all the strains of iridescent microbes were rod-shaped and between 1-6 microns in length by viewing the cultures under a microscope at 100X magnification.

Figure 1: Images of CA-IR1: CA-IR1 streaked on both (a) standard and (b) black SWC agar at room temperature and spotted on black SWC agar after (c) three days and (d) seven days. The iridescence is more obvious on the black agar plate but is present in both. The microbe appears to vary in color and level of iridescence based on the thickness of the growth in a particular location, with thin films being clear, thicker films being green and iridescent, and the thickest areas being orange. The “glitter-like” element of the iridescence is especially obvious in the spot colonies in (c) and (d).

To identify the genus and species of each iridescent strain, we sequenced the 16S ribosomal RNA gene and compared the sequences of our isolated strains with those of known bacteria. 16S sequences were obtained using colony PCR, in which a colony of the strain was boiled to lyse the cells, then the resulting lysate was used as a template in a PCR reaction designed to amplify the 16S rRNA sequence, and the PCR product was sent for Sanger sequencing. We used sequence similarity to determine the taxonomic relationship between our newly isolated strains and known strains using the NCBI (National Center for Biotechnology Information) 16S ribosomal RNA database. Additionally, genomic DNA of CA-IR1 was extracted, sequenced, and annotated, and the 16S sequence was extracted from the full genome. CA-IR1, CA-IR3, and CA-IR4 were identified as most similar to Tenacibaculum mesophilum strain “NBRC 16307,” indicating that these strains are closely related, despite being isolated from different environmental samples. A definitive identity for CA-IR2 was not obtained. In addition to the four strains isolated, in this study four strains (NH, RLM, LK1, and EP2) were used from previous projects (Figure 2).

Figure 2: Strains used in this study: The top row of strains were isolated during this project, while the bottom row of strains were isolated in previous projects at the Marine Biological Laboratory. All the strains used in this study were strains of Tenacibaculum, though iridescence has been observed in other labs in other marine bacteria in the family Flavobacteriaceae. All the known iridescent strains in this family have “glitter-like” iridescence that does not appear in trans-illumination. The relative abundance of this “glitter-like” iridescence in strains in this family suggests that there may be a common genetic mechanism for iridescence between the different strains.

Following the isolation and identification of our iridescent strains, we used light microscopy to analyze the effect of several culture conditions on iridescence to better understand its biological significance. Images were taken daily of colonies grown in different culture conditions using light microscopy at 7.5X magnification, and visual iridescence was recorded subjectively, recording the apparent amount of color change and the level of brightness of iridescence for each colony. The goal of these experiments was to identify which conditions had a visible effect on iridescence, either promoting, inhibiting, or changing the color and type of iridescence. The culture conditions that we varied were pH, agar concentration, salinity, and nutrient availability.

  1. pH

pH was observed to affect the growth rates of the colonies, with no growth occurring on SWC agar at pH 5 and slower growth occurring at pH 6 and 9 than at pH 7 and 8. Iridescence was observed across all conditions in which growth occurred, indicating that pH does not affect iridescence (data not shown).

Figure 3: CA-IR1 cultured on SWC agar at different agar concentrations
Low agar concentration, especially the 0.25% agar,  inhibited iridescence, to the point that there was no noticeable iridescence on the 0.25% agar. In the 0.5% and  0.7% agar concentrations distinct colony morphologies appeared, with odd winding patterns appearing in the colony. At the highest agar concentration the colony was thinner, but significantly iridescent.
  1.  Agar Concentration

The concentration of agar in the medium was used to define the firmness of the growth substrate, with greater concentrations corresponding to greater firmness. Low agar concentrations inhibited iridescence (Figure 3), indicating that a solid surface is required for the strains to display iridescence. Growth was inhibited at higher agar concentrations, with colonies appearing thinner while displaying intense iridescence even earlier than the other conditions, possibly indicating that the lower cell density allowed the cells to form a regular pattern more easily. Distinct colony morphologies, with winding patterns and a rough texture, were observed at 0.5% and 0.7% agar concentrations compared to the control 1.5% agar concentration. In this condition, limited iridescence was observed, but was difficult to discern on clear agar, so the experiment was repeated with black ink added to the medium to add contrast. With the addition of black ink we observed iridescence after four days of growth in all agar concentrations except 0.25% and we continued to see the distinct colony morphologies at 0.5% and 0.7% agar. Different strains responded in different ways to the limited agar concentration conditions. The most striking response was by strain NH on 0.5% agar (Figure 4), which spread to a much larger colony size than on any other media type, overtaking the other colonies on the plate. Both NH and LK1 displayed this spreading morphology at 0.5% agar. This could indicate that LK1 and NH have more efficient gliding motility than the other strains, allowing them to move through the agar faster.

  1. Salinity

We tested the effect of salinity on iridescence through two related experiments. First, we tested the effect of high salinity by growing spot colonies on black SWC with different amounts of added NaCl. Second, we tested the effect of low salinity by growing spot colonies on clear SWC made with a diluted seawater base (SWB). High salinity (≥ 40 g/L NaCl) appeared to inhibit growth, but did not appear to inhibit iridescence beyond the effects of the low growth levels. The color of the iridescent colonies was shifted from green to red at higher salinities than the control, but this may have also been due to the inhibited growth. Lower salinity appeared to inhibit iridescence, as iridescence was observed to be limited and dull on SWC made with 25% SWB. This may indicate that a threshold salinity is required for iridescence or growth. The lower salinity medium was less solid than that of the control, possibly contributing to the lower levels of iridescence. A minimum salinity is required for Tenacibaculum growth, as no growth was observed when strains were plated on media with no salts or with salt levels meant to replicate fresh water. What iridescence did appear on SWC made with 25% SWB appeared to be less green and more red and was much less “glitter-like”  than colonies on the control.

Figure 4: NH after 48 hours on black SWC 0.5% agar
Unlike other strains, NH9 spread easily on the soft agar condition and covered most of the plate after four days. LK1 also showed this spreading trait in this condition, while the other strains did not.
  1. Nutrient Availability

We tested the effect of nutrient availability on iridescence by growing colonies on several different types of media (Figure 5). Strains were grown on SWC, a nutrient-limited form of SWC called SWC-GC (for growth curve), SWB without added carbon sources, marine agar (MA), black marine agar (BMA), and artificial seawater (ASW), supplemented with yeast (ASW-YE), tryptone, (ASW-T), and a combination of yeast, tryptone, and additional salts (ASW-CYT). We expected to see similar colony morphologies and levels of iridescence in all the complete media conditions (i.e. SWC, ASW-CYT, and MA) so we were surprised to see different colony morphologies, colors of iridescence, and even types of iridescence (“glitter-like” and dull)  in the different conditions. ASW-based media yielded colonies with a winding, wave-like morphology, similar to the morphology displayed by CA-IR1 and CA-IR2 in 0.5% agar concentration SWC. Interestingly, no glitter was observed in the first 4 days in any of the colonies on ASW-YE, even though an intense color change was observed. This suggests that tryptone may be responsible for the development of the “glitter-like” bright reflection typically seen in the iridescent colonies. This is supported by the fact that the ASWT condition had the most intense iridescence after 4 days, being so bright that the “glitter-like” reflection was visible when the lamp was pointed away. Colonies on MA appeared more red than those on the other types of media, and were smaller in size. Colonies on MA also had consistent, bright, “glitter-like” iridescence across the entire colony, without any dark spots or patterns of reflection. Colonies on ASW without added carbon sources took much longer to grow than other conditions, but developed after a week and color change was visible. Significant growth was not observed on SWB. Strains on SWC-GC appeared more green than on standard SWC in the first two days and developed iridescence as intense as standard SWC after a week.

Figure 5: NH cultured on different types of media

No glitter was observed in the ASW-T condition after 24 hours or in the ASW-YE condition after four days, though a distinctive iridescent color change was observed in both. ASW-T had the strongest iridescence of any media after four days, to the point that color change and bright, “glitter-like” reflection was visible even when the lamp was not pointed at the colony. Different colony morphologies were observed depending on which salt mixture was used as a base. Colonies on marine agar were smaller than those in other conditions, had iridescence everywhere in the colony rather than having a dark spot in the middle, and were more red than green. Colonies on ASW-based media had an irregular, bumpy morphology similar to the morphology in .5% agar SWC. Strong iridescence was observed on SWC, SWC-GC, ASW, and ASW-T media. No significant growth was observed on SWB media, which was expected due to the lack of carbon sources.

Conclusions and Future Work:

In this work we investigated the effect of culture conditions on iridescence. The only culture conditions in which iridescence was inhibited were low salinity and low agar concentration, both of which had the effect of making the medium less solid. This suggests that iridescence in a colony biofilm requires that the biofilm grow on a more rigid surface. The iridescence observed on agar could be the result of the individual cell-cell interactions found in biofilms being maintained and repeated due to the unmoving solid surface. So what benefit do bacteria gain from growing in a highly ordered biofilm that appears iridescent? The structure might help the biofilm maintain its shape in the highly agitated waters of the marine environment where these bacterial live. While we know these bacteria form aggregates in liquid, they have not been observed displaying  iridescence in their natural environment, so it is possible that the iridescence does not appear in the wild.  Comparing the iridescent properties of biofilms grown in liquid to those grown on solid surfaces could provide insight into the relationship between the iridescence observed on agar and the behavior of these bacteria in their natural environment.

Our work supports prior findings of a relationship between gliding motility and iridescence, as we found that higher levels of iridescence were observed in broader colonies such as those that formed by NH on 0.5% agar. Future studies may include the difference between the strains with higher gliding motility and those with lower gliding motility, such as between the fast-spreading NH strain and the slower-spreading strain of the same species CA-IR1. A comparative genomic experiment could give us a better sense of the genes that regulate both gliding motility and iridescence in these strains.

The variation in color and level of iridescence among strains grown on different types of media is surprising and warrants further study, especially the lack of “glitter” in the ASW-YE condition. A broader experiment testing the effect of different carbon sources on iridescence should be performed to better understand this phenomenon. The different intensities and colors of the iridescence could be due to differences in the properties of the colony biofilms. Growing the bacteria on a wide range of media with different types or concentrations of carbon sources and taking TEM (Transmission Electron Microscopy) images of the resulting biofilms to assess the level of organization could give us a better idea of the effect of carbon sources and nutrient availability on iridescence. TEM allows for the imaging of the physical structure of the colonies on the level of individual cells, so the TEM experiments will provide a better idea of the physical underpinnings of the different forms of iridescence observed in Tenacibaculum biofilms.

We have shown that the appearance of iridescence is affected by changes in dryness, salinity, pH, temperature, and nutrient availability, so it is important to control for all of these things in future experiments to determine which change causes which effect. Ideally, this project should be repeated with the addition of time-point transcriptomics to determine whether changes in iridescence are caused by changes in gene expression. Iridescent colonies should also be analyzed using TEM, to determine what changes in the spatial configuration of the bacteria are causing the differences in iridescence between conditions. A strain-strain interaction condition should also be tested, as some strains have been observed to lose iridescence when mixed with another strain. Finally, a condition varying the growth surface will be essential to verify whether the iridescence in Tenacibaculum is plausibly able to occur in a marine environment, as to date it has only been observed on agar in a laboratory setting. Attempting to grow iridescent bacteria on fish carcasses, sea sponges, or seaweed will shed light on whether the iridescence we observe in laboratory conditions could reflect the strain’s behavior in the marine environment.


  1. Madigan, Michael T., et al. Brock Biology of Microorganisms. 14th ed., Pearson, 2015.
  2. Wainwright, M, and J Lederberg. “History of Microbiology.” Encyclopedia of microbiology 2 (1992): 419–437. Web.
  3. Ishino, Sonoko, and Yoshizumi Ishino. “DNA Polymerases as Useful Reagents for Biotechnology – The History of Developmental Research in the Field.” Frontiers in Microbiology 5.AUG (2014): 1–8. Web.
  4. Patrick, Hsu D., Lander S. Eric, and Feng Zhang. “Development and Applications of CRISPR-Cas9 for Genome Engineering Patrick.” Cell 157.6 (2014): 1262–1278. Web.
  5. Kientz, Betty et al. “Glitter-Like Iridescence within the Bacteroidetes Especially Cellulophaga Spp.: Optical Properties and Correlation with Gliding Motility.” PLoS ONE 7.12 (2012): n. pag. Web.
  6. Kientz, Betty, Pauline Marié, and Eric Rosenfeld. “Effect of Abiotic Factors on the Unique Glitter-like Iridescence of Cellulophaga Lytica.” FEMS Microbiology Letters 333.2 (2012): 101–108. Web.
  7. Doucet, Stéphanie M., and Melissa G. Meadows. “Iridescence: A Functional Perspective.” Journal of the Royal Society Interface 6.SUPPL. 2 (2009): n. pag. Web.
  8. Kientz, Betty et al. “Iridescence of a Marine Bacterium and Classification of Prokaryotic Structural Colors.” Applied and Environmental Microbiology 78.7 (2012): 2092–2099. Web.
  9. Kientz, Betty et al. “A Unique Self-Organization of Bacterial Sub-Communities Creates Iridescence in Cellulophaga lytica Colony Biofilms.” Scientific Reports 6.July 2015 (2016): 1–11. Web.
  10. Kientz, Betty et al. “Glitter-Like Iridescence within the Bacteroidetes Especially Cellulophaga spp.: Optical Properties and Correlation with Gliding Motility.” PLoS ONE 7.12 (2012): n. pag. Web.
  11. Chapelais-Baron, Maylis et al. “Colony Analysis and Deep Learning Uncover 5-Hydroxyindole as an Inhibitor of Gliding Motility and Iridescence in Cellulophaga lytica.” Microbiology 164.3 (2018): 308–321. Web. 28 Sept. 2018.
  12. Johansen, Villads Egede et al. “Genetic Manipulation of Structural Color in Bacterial Colonies.” Proceedings of the National Academy of Sciences 24 (2018): 201716214. Web.
  13. Perry, E. “Pleomorphism and iridescence in Tenacibaculum geojense. Microb. Divers.” Course mini-project, (2017).
  14. Mickol, R. “Making a Monster into a Mutant: Isolation and Analysis of a Tenacibaculum sp.” Microb. Divers. Course mini-project, 13 (2016).
  15. Nguyen, C. “Shining the Light on the Structural Color of Tenacibaculum ectocooler.” Microb. Divers. Course mini-project, 15 (2017).
  16. Herrera, N. “Characterization of the iridescence and motility mutants in Tenacibaculum sp. strain ECSM.” Microb. Divers. Course mini-project, (2017).
  17. Kee, L. “Characterization of Gliding and Iridescent Mutant in marine Tenacibaculum discolor.” Microb. Divers. Course mini-project, (2016).

Further Reading:

For more information on the Microbial Diversity Program at MBL:

For more information about the phenomenon of bacterial iridescence, a seminal paper that defined the terms used to describe prokaryotic structural color:

Kientz, Betty et al. “Iridescence of a Marine Bacterium and Classification of Prokaryotic Structural Colors.” Applied and Environmental Microbiology 78.7

For more information about the genetic basis of bacterial iridescence, a recent article uncovering the genes underlying structural color through a mutant analysis:

Johansen, Villads Egede et al. “Genetic Manipulation of Structural Color in Bacterial Colonies.” Proceedings of the National Academy of Sciences 24 (2018): 201716214.

For more information about the physical basis of bacterial iridescence, an article uncovering the structural components of biofilms that give rise to iridescence:

Kientz, Betty et al. “A Unique Self-Organization of Bacterial Sub-Communities Creates Iridescence in Cellulophaga lytica Colony Biofilms.” Scientific Reports 6.July 2015 (2016): 1–11.


I would like to thank my mentors, George O’Toole and Dianne Newman, for their help and instruction as well as for the opportunity, Lynn Kee, Callie Chappell, and Janet Sheung who all worked with me on bacterial iridescence, Rachel Whitaker, Scott Sanders, Gabriela Kovacikova, Kurt Hanselmann, Dominique Limoli, Sarah Guest, and all the rest of Microbial Diversity course faculty for their help, support and instruction, my fellow course assistants Rebecca Wipfler and Deaja Sanders for working with me and helping with my project, our course funding sources including the Simons Foundation, Promega, the Agouron Institute, the Gordon and Betty Moore Foundation, the Howard Hughes Medical Institute, NASA, the National Science Foundation, the U.S. Department of Energy, and Zeiss, and the MBL and the Caltech Student-Faculty Programs for providing me with this opportunity and with funding.