Category Archives: blue protein

Roger Tsien – the scientist that colored our research

Roger Tsien died a few days ago, at the relatively young age of 64. He was a UCSD scientist, a Nobel laureate and he was one of the first to see the significance and usefulness of GFP.

I’ve never met him. But, I guess, this blogs owes him its existence.

I don’t want to discuss his body of work, his achievements, or awards he won (e.g. the Nobel award). Many wrote nice things about him, such as here, here or here and all over the internet, with nice pictures of fluorescent proteins used in research.

I thought it will be nice to look back at his first GFP paper.


His goal in this paper was to investigate the formation of the fluorophore of GFP. Specifically, he asked:

“What is the mechanism of fluorophore formation? How does fluorescence relate to protein structure? Can its fluorescence properties be tailored and improved-in particular, to provide a second distinguishable color for comparison of independent proteins and gene expression events?”

Already here he looked to utilize GFP – to improve it, to change it, so it can be useful for fluorescent studies in biology.

He used random mutagenesis of the GFP cDNA to screen for mutants with altered brightness and emission. A simple yet powerful method, still used today, to find new FPs with exciting and useful properties.

Here is an ex/em spectral analysis of some of the mutants:


One mutant, I167T, proved to be almost twice as bright as the WT GFP protein.

But the most exciting was the finding of a blue FP (Y66H):

blue mutant

To sum up in his words:

“The availability of several forms of GFP with such different excitation and emission maxima [the most distinguishable pair being mutant P4 (Y66H) vs. mutant Pll (I167T)] should facilitate two-color assessment of differential gene expression, developmental fate, or protein trafficking. It may also be possible to use these GFP variants analogously to fluorescein and rhodamine to tag interacting proteins or subunits whose association could then be monitored dynamically in intact cells by fluorescence resonance energy transfer (19, 20). Such fluorescence labeling via gene fusion would be site-specific and would eliminate the present need to purify and label proteins in vitro and microinject them into cells.”

He saw the future, and it was bright green.

ResearchBlogging.orgHeim, R., Prasher, D., & Tsien, R. (1994). Wavelength mutations and posttranslational autoxidation of green fluorescent protein. Proceedings of the National Academy of Sciences, 91 (26), 12501-12504 DOI: 10.1073/pnas.91.26.12501

An excelent new tool for comparing fluorescent protein properties

Screenshot of the new tool

Screenshot of the new tool

Two very useful tools for visualizations of many fluorescent and photoswitchable proteins have been developed by Talley Lambert and Kurt Thorn (UCSF). Continue reading

The microscope’s light may affect your experiment

The conditions used for microscopy are often not “physiological” conditions. If we are talking about live imaging, then the cells are usually in culture, placed on a glass surface and grown in an artificial media. In many cases, we use genetically encoded fluorescent markers, that are rarely inert. These are acceptable and known limitations of the system.
However, when we think about microscopy, we do not often consider to effect of the light itself. The light we use can have deleterious effect. For instance, UV light is known to cause damage to cells, e.g. creating reactive oxygen species (ROS), creating thymidine dimers, cross-linking macromolecules and probably more. That is why people tend to limit the use of fluorescent proteins that are excited by UV light (such as blue fluorescent proteins or some photoactivated proteins).

Yet, visible light is considered relatively harmless. Now, a new study suggests that visible light, particularly blue-green light (which is used to excite green-yellow fluorescent proteins), can affect the metabolic state of the cell.  How?

Well, all cells contain light absorbing molecules. Some molecules function as light sensors or light energy harvesting molecules: Chlorophyll, cryptochromes, phytochromes, photoreceptors, rhodopsins, and fluorescent proteins. Other molecules absorb light, e.g. pigments.

However, some cells do not have obvious light sensitive molecules. For example, the budding yeast. In this paper, Carl H. Johnson’s lab looked at the effect of visible light on yeast respiratory oscillations (YRO) and oxidative stress. Apparently, under some conditions, yeast show 1-6hr oscillations of oxygen consumption and metabolite productions. surprisingly, shining white, blue or green light (but not red) has shortened the cycle  and decreased the amplitude of oxygen consumption. There was also increased expression of antioxidant enzymes, and increased light sensitivity to oxidative-stress mutants.

The effects of different spectra of light on the YRO. Oscillations were initiated in the dark until stable oscillations formed (black line, left y-axis). Then 12-h treatments of red, blue, or green light were administered (colored lines matching color of light, right y-axis) with 12 h of darkness between treatments. After the application of colored light, two 12-h white light treatments were given. Light intensities of each treatment are shown on the right y-axis and are indicated by numbers under each of the colored or gray lines showing light treatment. Source: Robertson J B et al. PNAS 2013;110:21130-21135

The effects of different spectra of light on the YRO. Oscillations were initiated in the dark until stable oscillations formed (black line, left y-axis). Then 12-h treatments of red, blue, or green light were administered (colored lines matching color of light, right y-axis) with 12 h of darkness between treatments. After the application of colored light, two 12-h white light treatments were given. Light intensities of each treatment are shown on the right y-axis and are indicated by numbers under each of the colored or gray lines showing light treatment. Source: Robertson J B et al. PNAS 2013;110:21130-21135

Based on several other assays, the authors suggests the blue-green light harms cytochromes in the respiratory electron transport process. Photoinhibition of the electron transport causes accumulation of ROS and oxidative stress response. It is hypothesized that the shortened oscillation periods and reduced amplitude of oxygen consumption (= less respiration) is a cellular mechanism to reduce the photoinhibition to electron transport, thus reduce oxidative damage.

What does this mean to microscopists?

It only suggests that long term exposure to light (either in prolonged live imaging, or just the incubation conditions) can affect the metabolic state of the cell and therefor may have either deleterious effect on the health of the cell, or create false positive or false negative results, particularly in experiments that are designed to study respiration and oxidative stress. Obviously, if you have light sensors or light harvesting molecules, their absorbance wavelengths need to be taken into account.

Practically, it is always a good advice to minimize the exposure to light (which we do anyway to reduce photobleaching).

ResearchBlogging.orgRobertson JB, Davis CR, & Johnson CH (2013). Visible light alters yeast metabolic rhythms by inhibiting respiration. Proceedings of the National Academy of Sciences of the United States of America, 110 (52), 21130-5 PMID: 24297928

The red color dilemma: how to choose the right fluorescent protein

This post continues the previous post.

I encountered a serious dilemma in choosing the right protein. This is also a great opportunity to learn about the many properties of fluorescent proteins.

Let us start with the obvious: excitation and emission maxima and spectra.

We all know that EGFP has an excitation maximum at 488nm. That is, EGFP protein that is excited with photons at 488nm will give its maximum emission intensity at its emission maximum, 509nm.

However, we must remember that this is the maximum emission. The emission spectra is much wider, and for GFP it goes from ~470nm up to ~630nm. Figure 1 shows the EGFP excitation (dashed line) and emission (full green) spectra.

Fig. 1: GFP specrum with 488 excitation

And now, what if we want to look simultaneously at two colors, EGFP and dTomato?

Figure 2 (upper panel) shows you that excitation at 488nm (the max ex. of EGFP) also excite dTomato, leading to almost 30% emission at its peak of 581nm. Therefore, if we would collect all the emitted photons, we would detect the combined excitation of EGFP and dTomato.

Fig. 2: GFP dTomato spectra with 488 excitation. Lower panel: with filters.

We therefore use filters with a narrow band of wavelength (fig. 2, bottom panel). Thus, if we use the 510/20 filter, we would detect only photons emitted at the band, in this case only from EGFP. If we use the 580/30 filter, we would detect photons coming from dTomato and EGFP.

Now what happens if we add the long-stokes shift protein LSSmKate1 to the system? Figure  3 shows LSSmKate1 is maximally excited at 463nm (red dashed), and has max. emission at 624nm (full red) [note- I drew the LSSmKate spectra, based on prior publications]. However, 460nm also excites EGFP and dTomato. The EGFP signal, at 624nm is negligible. However, dTomato gives emission at 8-9% of its maximal emission. Is this negligible? We will soon learn.

Fig. 3: GFP dTomato LSSmkate1 spetra with 460 excitation

Now let’s look at mCherry. With excitation laser at 460, we have virtually no excitation of mCherry. We would therefore prefer to use mCherry in our 3-color system, instead of dTomato.

Fig. 4: GFP mCherry LSSmkate1 spectra with 460 excitation

This table summarizes the excitation & emission maxima:

FP Max ex. Max em.
EGFP 488 509
dTomato 554 581
LSSmKate1 463 624
LSSmKate2 460 605
mCherry 587 610

This table shows the relative emission of the different FPs with different excitations (the LSS data is estimated based on publications, since the BD

spectrum viewer does not include LSSmKates in its database):

  % from max emission % from max emission

Em   (nm):

509 580 605 610 624 509 580 605 624
EGFP Ex.488 100% 8% <5% Ex.460 65% 5%
dTomato 27% 20% 19% 15% 15% 11% 9%
LSSmKate1 50%? 100
LSSmKate2 50%? 100
mCherry 8%

However, excitation & emission spectra are not the only considerations. There are other parameters to consider.

Brightness, QY and EC

Different fluorescent proteins have different brightness: some are very bright, some are dim. As a matter of history, we usually consider the relative brightness, compared to EGFP (which is set to 1.00). You can see in the table below the relative brightness of the proteins discussed above.

The brightness is determined by two parameters: quantum yield (QY) and Extinction coefficient (EC).

QY is simply the ration between the number of photons absorbed to the number of photons emitted. If QY=1, then for each absorbed photon you get one emitted photon.  However, we never get 100% efficiency. The protein with the highest QY that I encountered is called ZsGreen with QY=0.91.

EC (or ε) relates to the formula A = εcl in which A is the absorbance, l is the path length (in cm, usually) and c the concentration in Molar units. ε, in M-1cm-1 units, is a measurement of the capability of a certain fluorescent protein to absorb light at a certain wavelength.

The formula ε*QY gives a measure of the brightness, so for EGFP, the brightness is 55,000*0.60=33,000

For ZsGreen, the EC is only 43,000 so the relative brightness is only 1.18. This example shows that a high QY does not necessarily mean a very high brightness.

This table summarizes al the brightness data:

FP QY EC(M-1 cm-1) Brightness Relative brightness
EGFP 0.60 55,000 33,000 1.00
dTomato 0.69 69,000 47,610 1.44
LSSmKate1 0.08 26,000 2080 0.06
LSSmKate2 0.17 31,200 5304 0.16
mCherry 0.22 72,000 15,840 0.48

Let’s go back to our dilemma: dTomato or mCherry?

We now know that with a laser excitation of 460nm, we get 100% emission from LSSmKat1 at 624nm, and only 9% emission from dTomato at 624nm.

However, dTomato is 24-fold brighter than LSSmKate1.  If we look only on the QY data, at 8% emission, dTomato will produce ~6 photons for each 100 photons (100*0.69*9%), whereas LSSmKat1 will produce 8 photons (100*0.08*100%). In other words, although we get very low emission of dTomato, relative to its maximum, it is almost as high as the LSSmKate1 emission in that wavelength. Since eventually, we just “see” the photons with no knowledge of their source, the signal that we will get will be ~40% from dTomato. This will create a problem if we wish to detect changes in fluorescent intensity over time due to changes in protein localization (or other changes) because we will not be able to know if the change is due to dTomato or LssmKate1.

That is why mCherry is a better choice than dTomato for this experiment.


Another parameter to take into consideration is the protein’s photostability.

Photostability is a measure (in seconds) of how long it takes for half of the number of proteins to bleach at the maximum excitation.  This matter is important for any experiment involving fluorescent proteins (or dyes) but is crucial when doing live imaging, particularly for time-lapse experiments.

LSSmKate1 is more stable than LSSmKate2 (see table below), and therefore, it might be a better choice than LSSmKate2 for my time-lapse experiment.

FP photostability
EGFP 174 sec
dTomato 98 sec
LSSmKate1 60 sec
LSSmKate2 44 sec
mCherry 96 sec

There are other parameters to take into account, some of them I discussed in earlier posts.

These include:

pH stability (and the effect of pH on the absorption and emission spectra).

Maturation rate – how long does it take your fluorescent protein to fold correctly and create the chromophore? If you are using your protein to measure expression rate, and the maturation time is longer than the time frame of your experiment, then you will not get the information you are looking for. Maturation rate  depends on protein characteristics, as well as oxygen levels and temperature. Several fast-folding FPs were developed in recent years.

Oligomerization state – many FPs are naturally monomeric. However some FPs particularly in the orange & red range) are dimeric or tetrameric.  Since in many cases the FP is genetically encoded as a fusion with a protein of interest, fusing a dimeric FP may cause your protein of interest to dimerize, through the FP, thus altering its biology. If your protein is naturally oligomeric, fusing it to a dimeric or tetrameric FP may create large protein aggregates. One solution for dimeric FPs is to create a tandem fusion proteins (i.e. two FPs are fused one to the other and to the your protein).

Generally, the letter m at the beginning of the FP name indicates that it is monomeric (e.g. mCherry, mKate), d indicates it is dimeric (dTomato) and td indicates tandem dimer (tdTomato). However, there are many FPs with no indication of their oligomeric state in their name (e.g. TurboRFP is dimeric). EGFP and its many derivatives are monomeric (except at high concentrations when they form weak dimers).

Phototoxicity  – although the FPs themselves do not emit harmful radiation, the excitation light can be harmful to the cell. For instance, UV light, that is used to excite BFPs, as well as some LSS FPs and photoactivatable FPs, can cause direct DNA damage, create reactive oxygen species that will cause an oxidative stress etc…  This is one reason why BFPs are the least popular FPs, particularly for live imaging.

All the colors of the rainbow

GFP was the first fluorescent protein to be dicovered, and subsequently used in biological research. However, by now, the biological community has found or developed an enormous number of fluorescent proteins of many colors.

According to my count (based on recent review papers)  there are over 90(!) differnt fluorescent proteins. These proteins can be classified based on several charactereisitcs:

Emission color: is the most obivous classification. The classification generally goes by: Blue (424-457nm), Cyan (474-492nm), Green (499-509nm), Yellow (524-529nm), Orange (559-565nm), Red (584-610nm) and Far-Red (625-650nm).

Bacteria expressing differnt FPs were plated to create a nice picture (source: Roger Tsien lab)

Oligomerization: many of the FPs are monomeric (i.e. fluorece as single molecules). Others may be dimeric (two) or tetrameric (four).

Photoactivation/photoconversion: some proteins can switch there color when activated by a specific excitation wavelength. This means that the emission wavelength can change from green to red, for instance. In a few cases, the initial state of the protein is non-fluorescent, thus allowing very low background level of fluorescence. This group can be sub-divided into reversible and non-reversible photoactivatable proteins.

Fluorescnet timers – These protein change their color over time. Therefore, these can be used as “timers” for cellular processes following their activation.

Large Stokes shift (LSS): Stokes shift (named after George G. Stokes) is the shift in wavelength from excitation to emission. For most FPs, Stokes shift is less than 50nm (usually much less).  For LSS proteins, the differnce is over 100nm (i.e. cells are excited by UV light or blue light and their emission is Green or Red light).

Natural vs. engineered: There is currently a lot of work invested in developing new colors and new activatable proteins by directed mutagenesis.

Three excellent review papers on the differnt kinds of FPs:

Stepanenko et. al. (2008) “Fluorescent proteins as biomarkers and Biosensors: Throwing color lights on molecular and cellular processes” Curr. Protein. Pept. Sci. 9(4):338.

Chudakov et. al. (2010) “Fluorescent proteins and their applications in imaging living cells and tissues”  Physiol. Rev. 90:1103.

Wu et. al. (2011) “Modern fluorescent proteins and imaging technologies to study gene expression, nuclear localization, and dynamics” Curr. Opin. Cell. Biol. 23:310.