Reader Comments

Post a new comment on this article

Referee Comments: Referee 1 (Kurt Anderson)

Posted by PLOS_ONE_Group on 30 Oct 2007 at 23:04 GMT

Reviewer 1's Review

-----

The MS from Goedhart et al presents a systematic study of fluorescent protein pairs used for the image-based detection of Fluorescence Resonance Energy Transfer (FRET). In an attempt to improve on the sensitivity of the standard FRET pair CFP/YFP, the authors have generated a series of fluorescent protein chimeras consisting of yellow or orange donors linked to red acceptors. In each case the length of the linker has been adjusted in an effort to maintain a constant separation between the fluorophores. Of particular interest are measurements in which fluorescence lifetime is shown to strongly depend on the orientation of the donor with respect to the acceptor using a series of five circularly permuted YFP constructs. For all constructs FRET is measured in living cells as a change in the fluorescence lifetime of the donor fluorophore using a custom built system for Fluorescence Lifetime Imaging (FLIM) in the frequency domain. One series of measurements are correlated with the FRET efficiency determined by acceptor-photobleaching in order to confirm the generality of results obtained using different methods. The serendipidous discovery of the photo-conversion of mKOrange from orange to green emission is also reported, which is associated with a dramatic change in fluorescence lifetime. Of the many combinations tested, mCherry-mKO is determined to be the best FRET pair on the basis of the largest reduction in donor fluorescence lifetime (1.3 ns). This FRET pair is finally compared to CFP/YFP for detection of p65 homodimerization within the NF-kB transcription factor complex and found to be superior.

General Points:
In my view these are extremely useful measurements which will help to guide the experiments of researchers in many different fields. The senior author is a well respected microscopist with a strong background in the development and use of both FRET and FLIM-FRET. The introduction is particularly well written and provides the relevant background information as rational for the current experiments. The data are of a very high technical standard, with average lifetime values generally based on measurement of between 20 and 30 cells. The author's approach to determining an optimal FRET pair is clear and highly systematic, although a few specific points need to be addressed (see below). The MS is written for a general audience, however in places a little more technical detail would be appropriate (see below). I'm sympathetic to the basic problem, which is that FRET, especially FLIM-FRET, has been developed by technical specialists but increasingly sought after by non-specialists such as molecular, cell, and developmental biologists. More technical language might frighten off some of the straight biologists, but if they want to use the technique they need to confront the principles involved.

Major Points:

1. Red/Green pairs
Authors have selected yellow-red and orange-red as donor acceptor pairs, but have neglected green-red, although there is already work in the literature demonstrating the utility of this pair (Tramier et al. Mic Res Tech, 2006; Peter et al, BiophysJ, 2005). The reference value of the MS would be substantially increased by the inclusion of at least one green-red pair in the author's data set. Baring this, the authors should at least state why they have not considered GFP-mCherry, when this combo has been shown to work in other experimental systems. In discussion (P15L16) authors even refer to their own previous use of tagRFP as an acceptor with GFP.

2. Linkers
Details of the linkers and distance between fluorophores need to be clearer. There is a sixth-power dependency of the FRET efficiency on the distance between donor and acceptor. To control for this and increase the comparability of their results, authors have attempted to use linker length to keep a constant distance between donor and acceptor. But their description of the linkers and constructs is unclear, eg. P6L8: "When mOrange and mKO containing pairs are compared, the distance between the donor and acceptor chromophore only differs by one amino acid" -but which is longer? the difference is not clear from Fig.1. How do the SYFP2 constructs compare? Authors should clearly map out the construct linkage by including a table (as supp info if necessary) listing donor molecule, donor terminus sequence up to a constant region within the FP, linker, acceptor terminus sequence up to a constant region within the FP, acceptor molecule. The assumption that distance between donor and acceptor has been held constant among the constructs is very important to authors final conclusions. However, see also P9L5: "probably due to the longer linker" why do authors blame the linker in this case, not properties of the fluorescent proteins? As mentioned below, additional background/discussion on what determines FRET efficiency would help.

3. Calculation of FRET efficiency
Authors should state, preferably in the methods section although it is currently not in the Supp Info either, how FRET efficiency is calculated for both lifetime and acceptor photobleaching approaches. In the photobleaching case, what corrections are applied? P13L20: Authors correctly note that different methods are used to calculate FRET efficiency but then do not describe the method they've used, noting only: P8L5: "The donor fluorescence before and after photobleaching was quantified, from which FRET efficiency was calculated." -yes, but how? what corrections, if any, were applied? In the lifetime case, a version of the formula E=(t1-t2)/t1 should appear somewhere.

4. FRET efficiency information
Additional discussion of the meaning of the FRET efficiency would be helpful, including discussion of the factors which determine the Förster's Radius. Personally I would move the discussion of Förster radii from SupInfo to the main text for reasons outlined above, or at least refer to the SupInfo more frequently in the main text. The danger of keeping this discussion in SupInfo is that no one will read it. The formula is perhaps ugly for non-technical readers, but at least by presenting it the basic terms are introduced so reader appreciates all the contributing factors to the Förster's Radius and FRET efficiency. At a minimum the abbreviation R0, which appears on P14, should be introduced in the main text. Conclusions such as P7L20: "The higher FRET efficiency for mCherry and mStrawberry acceptors is most likely due to the higher extinction coefficient..." are better supported by more discussion of other potential causes.

5. FLIM-FRET information
P7L17: "Both phase and modulation lifetime..." A bit more technical background would be helpful, including some mention of the difference between frequency and time domain FLIM. Frequency domain FLIM is probably the less well known of the two approaches and possibly the less intuitive (for those of us who unfortunately dont think in Fourier space). For instance, in Tables 1, 3, and 4 authors report lifetimes calculated according to both phase and modulation. The average reader probably wont know the difference. Furthermore, Table 1 reports phase and modulation lifetimes, as well as FRET efficiency calculated from both. But Tables 3 and 4 only report FRET efficiency calculated from the phase lifetime. Why only consider the phase lifetime? Note also the superscript '4' in Table 3 refers to "Etø average FRET efficiency calculated from tø or tm", when I think onlly tø has been considered. Authors nicely compare and contrast practical aspects of lifetime and acceptor photobleaching approaches (P13-14), but need also to consider the influences on FRET efficiency determined by FLIM. Some finer points of frequency domain FRET crop up in anyway, including p8L8-11: "This is a well known phenomenon, which can be explained by the fact that FLIM is biased towards the higher lifetimes of multi-exponentially decaying donors." Authors refer to their own previous work here, but one more sentence concerning this "well known phenomenon" would be appreciated. Why is FLIM-FRET biased towards higher lifetimes?

6. mKOrange photo-switching
P9: The section on photo-switching of mKOrange really disrupts the flow of the text. Would be helpful at least to warn reader a detour is coming by changing section title to include word "photo-conversion". The mixed population photo-conversion experiment is nice but also a little confusing. Wouldnt the same point be made more clearly by photo-converting cultured cells expressing either mKOrange or mOrange and measuring their emission and lifetime? Photo-conversion is an interesting point but it doesnt contribute to the final conclusions concerning sensitivity of FRET detection. Utility of photo-conversion would be more convincing if it were demonstrated with an laser capable of being coupled into a laser-scanning confocal for selective (region of interest) marking of cells or structure (such as a 442 nm HeCad). Author's conclusion that the protein can be used reliably as a FRET donor misses the danger associated with inadvertant photo-conversion, which authors correctly point out is problematic for CFP/YFP FRET applications (P4L17). Authors essentially show photo-conversion is equally problematic for mKOrange. See also P10L24: "Therefore, this protein can be used reliably as a donor in FRET studies." Because of the photo-switching and lifetime change it should rather be used with caution.

7. mStrawberry
Table 4, P11 top: Authors chose to make final comparison between mRFP1 and mCherry. Why was mStrawberry dropped? The conclusion from Table 1 is that Cherry and Strawberry are quite similar in FRET efficiency with SYFP2 as a donor, whereas mRFP1 is not as good. Where is the rational for dropping Strawberry? Do authors know its not as good as Cherry in combination with orange donors? See also P15L24: authors conclude mKO-mCherry is the best combo, but didnt test either mOrange or mKOrange with mStrawberry.

-----

N.B. These are the general comments made by the reviewer when reviewing this paper in light of which the manuscript was revised. Specific points addressed during revision of the paper are not shown.

RE: Referee Comments: Referee 1 (Kurt Anderson)

Joachim replied to PLOS_ONE_Group on 12 Dec 2007 at 07:45 GMT

Please find below our response to referee 1:

1. Red/Green pairs
The use of green-red is discussed in the text. We have focused on yellow and orange donors, since shifting the complete pair, including the donor, to higher wavelengths increases the R0. GFP is less attractive as a donor for mCherry/mStrawberry because of decreased overlap and hence decreased FRET efficiency. We refer to EGFP-tagRFP as the most optimal green-red pair. We have no experimental data on green-red FRET pairs and we feel that this is beyond the scope of this study.

2. Linkers
We have added a table in the supplementary info describing the amino acid sequence of the RFP C-terminus, linker and YFP/OFP N-terminus, which should clarify the used constructs and linkers. In addition, we will also publish the nucleotide sequence of all tandem constructs on the internet: http://wwwmc.bio.uva.nl/~...
A cautionary note on linker length is added, since the FRET efficiency may not directly relate to linker length. Addition or deletion of a few amino acids may change the relative orientation of donor and acceptor.

3. Calculation of FRET efficiency
The equations used to calculate the FRET efficiency by FLIM or acceptor photobleaching are now included in the methods section. For calculation of E based on acceptor photobleaching we did a background correction of the measured fluorescence intensities, which is also mentioned.

4. FRET efficiency information
We certainly agree that the discussion on Förster radii is “hidden” in the suppl. info, so we have moved the equations that are used to calculate R0 to the introduction. The calculated Förster radii, which is a very important result, are now presented in the results section. We also explain the contributing factors. The effect of higher extinction coefficients is now clear from equation 2, to which we refer.

5. FLIM-FRET information
We very briefly explain frequency domain FLIM and indicate that it yields two lifetimes, τϕ and τM, which in turn yield two FRET efficiency values. We included E based on τM in table 4 and 5. On the discrepancy between E determined from FLIM and acceptor bleaching we add a few words. For a more in-depth discussion we refer to Vermeer et al., since it is discussed extensively in this publication and it also includes some good examples.

6. mKOrange photo-switching
We can see that the mKO photoswitching disrupts the flow, but we think it is very important to discuss it in the main text. To increase readability we divided the section on orange proteins in two sections. The first section describes the lifetimes and photoswitching of orange fluorescent proteins. The second section describes FRET from orange to red fluorescent proteins.
The lifetimes of the photo-converted species of mKO are mentioned in the text and we think that the mixed cell experiment elegantly demonstrates the lifetime contrast between mOrange and mKO and also between mKO before and after photoconversion.
We agree that, because of possible photoconversion, mKO should be used with caution, and this point is made in the text.

7. mStrawberry
Since the fluorescence emission peaks of orange fluorescent proteins and mStrawberry are close together, it is rather difficult to separate donor emission from acceptor emission which is necessary for donor based FRET methods. Although the high overlap is beneficial for high FRET efficiency, we think that this is not very practical for donor-based FRET methods due to the narrow bandwidth that is left. Therefore, we did not consider mStrawberry as an acceptor for mKO/mOrange. We included this statement in the text.