FRET-based cyclic GMP biosensors measure low cGMP concentrations in cardiomyocytes and neurons

Several FRET (fluorescence resonance energy transfer)-based biosensors for intracellular detection of cyclic nucleotides have been designed in the past decade. However, few such biosensors are available for cGMP, and even fewer that detect low nanomolar cGMP concentrations. Our aim was to develop a FRET-based cGMP biosensor with high affinity for cGMP as a tool for intracellular signaling studies. We used the carboxyl-terminal cyclic nucleotide binding domain of Plasmodium falciparum cGMP-dependent protein kinase (PKG) flanked by different FRET pairs to generate two cGMP biosensors (Yellow PfPKG and Red PfPKG). Here, we report that these cGMP biosensors display high affinity for cGMP (EC50 of 23 ± 3 nM) and detect cGMP produced through soluble guanylyl cyclase and guanylyl cyclase A in stellate ganglion neurons and guanylyl cyclase B in cardiomyocytes. These biosensors are therefore optimal tools for real-time measurements of low concentrations of cGMP in living cells.

5. At several instances n numbers are too low. e.g. "3-4-cells from 2-3 animals" which is in my opinion not solid enough. pls add some more data points in such cases.
Reviewer #2 (Remarks to the Author): In this paper, Calamera and colleagues describe the development of FRET-based cGMP biosensors which are suited to detect the changes in low nanomolar cGMP concentrations. The authors have accomplished this purpose by using the cGMP binding domain D of Plasmodium falciparum PKG which is known for high cGMP affinity, and monitored subtle cGMP productions in ganglion neurons and cardiomyocytes using one of the resulting sensors, Yellow PfPKG. I have found two novel points in this paper: i) PfPKGs are the first sensors with the protozoan parasite's cGMP binding domain. Mammalian PKGs or PDEs have been used in the previous reported cGMP biosensors; ii) Yellow PfPKG is the first sensor based on CFP-YFP FRET which has the high cGMP affinity as well as Red cGES-DE5. Although its selectivity for cGMP/cAMP is obviously lower than of Red cGES-DE5, the dynamic range is larger. The new varieties of cGMP biosensors would be welcome in the research community for cell signaling.
(1) It is unclear how to calculate FRET ratio for T-Sapphire-Dimer2 sensors in live cell imaging. In Materials and Methods section, the authors describe "Cells expressing the red sensors were excited at 422±15 nm and 572±15 nm" (lines 197-198) but "Cells expressing the yellow sensor were excited at 436±10 nm" (lines 205-206). Why were two excitation wavelengths needed for the red sensors? The Dimer2 emission excited at 572 nm shouldn't be used as the denominator of FRET ratio because this emission doesn't depend on cGMP concentrations. If this emission was used for correction for "direct excitation at 422 nm (1%)" (lines 202-203), the yellow sensors should also be excited at ~515 nm for the correction for a fair comparison.
(2) Although there are many figures for time course of live cell FRET imaging in this paper, Figs. 4a-b are the only panels which have the vertical axis with "% of max". I think that this axis should be changed with the axis with "CFP/YFP Ratio (normalized)" as well as the other figures because the actual changes of FRET ratio are important for signal detection in this case. Also, the result of a statistical test should be shown for quantitative comparison of Fig. 4a described as "As predicted, BNP stimulation increased FRET significantly more for the Yellow PfPKG than for the cGi-500 biosensor" (lines 331-332).
(3) As the authors also noticed, 100-fold selectivity for cGMP/cAMP is not high in high affinity sensors. cAMP affinities of PfPKGs (the Yellow sensor and the Red sensor with EC<sub>50</sub> 4.6 μM and 1.9 μM, respectively) are comparable to of some cAMP biosensors. 2) The FRET signal change of Red cGES-DE5 finally reached ~18% in Fig. 5f, but I can't find the plot in Fig. 5g (Iso+CNP+IBMX).
3) Although the manuscript says "However, the dynamic range of the cGi-500 was larger than the PfPKG sensor ( Figure 4c)." (lines 431-432), I think that basal cGMP levels in cells might reduce the maximal FRET responses of Yellow PfPKG which has the high affinity. 4) Since the dynamic range of Red PfPKG seems small, comparable to of Red cGES-DE5, I think that the representation of "a large dynamic range" in lines 455-456 ("To conclude, we have generated two novel biosensors for cGMP with high affinity and a large dynamic range.") is not appropriate.
Reviewer #3 (Remarks to the Author): The manuscript by G. Calamera et al. presents two new genetically-encoded sensors (Yellow PfPKG and Red PfPKG) that are designed for measuring nanomolar concentrations of cGMP. Only one sensor was available so far that could provide similar sensitivity (Red cGES-DE5). Yellow PfPKG shows higher dynamic range and ~2.5 higher affinity than Red cGES-DE5 in cardiomyocytes. Yellow PfPKG also performs well in stellate ganglion neurons, when compared to another known sensor, cGi-500. These data demonstrate the significance of the study. The paper is well written, sound, and contains necessary control experiments. Together with positive experimental results, the authors also describe the experiments, where their sensors failed to surpass the performance of the known sensors. This information is important because it can save time for potential users. A serious improvement of the paper can be expected if the authors would measure the cGMP concentration dependence in purified solutions. The cell homogenates that they use now contain unknown basal concentration of cGMP and also non-negligible autofluorescence and scattering. These effects can introduce systematic and random errors in their measurements of the EC50 and dynamic range. Several minor issues should also be addressed. 1. The authors should specify what protein corresponds to an abbreviation Dimer 2. What red FP is a monomer in this dimer? Is this tandem dimer or physically associated dimer without peptide link? 2. It would be easier for the reader to see all EC50 values presented in the same units, namely either nM or uM (lines 88-90); 3. Please spell out EC50 and add it to the list of abbreviations; 4. Please add the abbreviation of stellate ganglion (SG) to the list as well; 5. It seems that the method of SG neurons transduction is missing; 6. When comparing the performance of Cp173Venus-Venus with Yellow PfPKG (lines 277-278), the authors say that they found no alteration of affinity to cGMP, although EC50 of Cp173Venus-Venus is ~ 2 times lower than that of Yellow PfPKG (46 vs 22 nM). Is this correct? 7. It also looks like Yellow PfPKG has ~2.5 times higher affinity compared to Red cGES-DE5 in intact cardiomyocytes (lines 322-323). That makes Yellow PfPKG even a better sensor. However, the authors statement is that they have equal sensitivity. Is the 2.5 x difference insignificant statistically? This should be clarified. 8. When comparing sensitivity of Yellow PfPKG to cAMP and cGMP it would be helpful to show the titration curves on the same plot (For example, combining data of Figs. 1c and 5d). 9. Concentration of forskolin should be given in line 347. 10. In Discussion, lines 381-383, different dynamic ranges of Yellow and Red sensors are attributed to different efficiency of transferring energy from donor to acceptor. It is rather different relative change of energy transfer upon binding cGMP that explains different dynamic ranges. 11. In lines 388-389, it is better to say that longer excitation wavelength, instead of emission wavelength, can penetrate deeper, etc. The bluer emission wavelength will experience more scattering events, but still can rich the detector with the high enough numerical aperture. 12. Lines 401-402: See comment 7.

Reviewer #1 (Remarks to the Author):
This is a nice report describing the development of characterization of the new high affinity FRET biosensors for cGMP based on D domains of cGMP dependent proteins kinase from Plasmodium falciparum. This a a nice development for the field since it adds a new (or even two new) high affinity cGMP biosensors to our toolbox which had contained just one red cGES-DE5 that far, capable of measuring cGMP in low-middle nanomolar range. Here, the authors present a clear rationale and a straight forward "rational design" approach and thoroughly characterize new sensors in vitro and in live cells (HEK293 line an primary cardiomycytes and SG neurons), two latter cell type contain endogenously low levels of cGMP. The paper was well written. Authors are lauded for the development of new sensor and for presenting clear data, and for their honestly dealing with issues of challenging cGMP measurements and sensor development for low nm range. Since it is a clear advance for the field I would recommend acceptance with minor (mostly text revisions): 1. Figure 2b  We have now performed more experiments to increase the number of animals and number of cells.

Reviewer #2 (Remarks to the Author):
In this paper, Calamera and colleagues describe the development of FRET-based cGMP biosensors which are suited to detect the changes in low nanomolar cGMP concentrations. The authors have accomplished this purpose by using the cGMP binding domain D of Plasmodium falciparum PKG which is known for high cGMP affinity, and monitored subtle cGMP productions in ganglion neurons and cardiomyocytes using one of the resulting sensors, Yellow PfPKG. I have found two novel points in this paper: i) PfPKGs are the first sensors with the protozoan parasite's cGMP binding domain. Mammalian PKGs or PDEs have been used in the previous reported cGMP biosensors; ii) Yellow PfPKG is the first sensor based on CFP-YFP FRET which has the high cGMP affinity as well as Red cGES-DE5. Although its selectivity for cGMP/cAMP is obviously lower than of Red cGES-DE5, the dynamic range is larger. The new varieties of cGMP biosensors would be welcome in the research community for cell signaling.
(1) It is unclear how to calculate FRET ratio for T-Sapphire-Dimer2 sensors in live cell imaging. In Materials and Methods section, the authors describe "Cells expressing the red sensors were excited at 422±15 nm and 572±15 nm" (lines 197-198) but "Cells expressing the yellow sensor were excited at 436±10 nm" (lines 205-206). Why were two excitation wavelengths needed for the red sensors? The Dimer2 emission excited at 572 nm shouldn't be used as the denominator of FRET ratio because this emission doesn't depend on cGMP concentrations. If this emission was used for correction for "direct excitation at 422 nm (1%)" (lines 202-203), the yellow sensors should also be excited at ~515 nm for the correction for a fair comparison.
In the original manuscript, the correction of "direct excitation at 422 nm (1%)" is based on 422nmexcitation of Dimer2 being 1% of its 572nm excitation. Please note that excitation at 422nm and 572nm was performed sequentially but with the same illumination times to allow this correction. The same correction could not be performed with the Yellow PfPKG biosensor since direct excitation of Venus (515nm), at the illumination times chosen for the considerably weaker CFP, reached saturation, thus precluding meaningful correction of direct excitation of Venus. The Donors (T-sapphire and CFP) and acceptors (Dimer2 and Venus) are in a 1:1 stoichiometry and the direct excitation of acceptor at donor excitation (422nm and 436nm) will be similar in each cell and throughout each experiment and therefore this correction is not necessary for comparing changes in FRET, similar to other labs using similar methods 1,3 Figure 2e, 3i-k and 5g. The reanalysis did not result in large changes in data or conclusions from the experiments.

. Since correction of direct excitation of acceptor (at donor excitation) was only performed with the Red biosensors, we therefore agree with the reviewer that a direct comparison of the maximal FRET obtained for the Red and Yellow PfPKG was not a fair comparison. To make a fair comparison, we have therefore analyzed the Red PfPKG and Red-cGES-DE5 responses as ratios of T-sapphire over Dimer2 with Dimer2 emission only corrected for spillover of T-sapphire into the Dimer2 channel (15%). This has now been corrected in lines 219-222 and
(2) Although there are many figures for time course of live cell FRET imaging in this paper, Figs. 4a-b are the only panels which have the vertical axis with "% of max". I think that this axis should be changed with the axis with "CFP/YFP Ratio (normalized)" as well as the other figures because the actual changes of FRET ratio are important for signal detection in this case. Also, the result of a statistical test should be shown for quantitative comparison of Fig. 4a described as "As predicted, BNP stimulation increased FRET significantly more for the Yellow PfPKG than for the cGi-500 biosensor" (lines 331-332).
We originally presented the traces as % of max because we thought that it would be a better comparison between the two sensors since they display a different dynamic range (cGi-500 having a larger dynamic range than the Yellow PfPKG; Figure 4h). However, we agree with the reviewer that consistency between figures is better for the reader and we have therefore modified the figure to show CFP/Venus ratio (normalized); for Figure 4a, d or CFP/YFP ratio (normalized); for Figure 4b, e. We have also added the quantification of the individual responses and we have added the data points into bar graphs. The statistical difference between PfPKG and cGi-500 subjected to BNPstimulation are now performed and indicated in Figure 4c and changes are made in the text (lines 357-9).
(3) As the authors also noticed, 100-fold selectivity for cGMP/cAMP is not high in high affinity sensors. cAMP affinities of PfPKGs (the Yellow sensor and the Red sensor with EC50 4.6 μM and 1.9 μM, respectively) are comparable to of some cAMP biosensors. Thank you for the suggestion. We agree that a single-wavelength cAMP biosensor would be the best option for this purpose. We have now included this possible use of the sensor in the discussion (lines 482-485).
Minor comments. 1) Emission spectra with and without cGMP of PfPKGs are informative for users. Adding it into Fig. 1 (or Supplementary Figure) might be helpful.
Thank you for the suggestion. We have measured the emission spectra for both PfPKG sensors with and without cGMP and added into Figure 1 as suggested. We agree that this adds useful information about the sensors.

Traditionally, FRET measurements have been calculated as F(Acceptor)/F(Donor). Since this produces a decrease in FRET with increasing concentration of cGMP for all the sensors used in this study, we decided to plot all traces as F(Donor)/F(Acceptor) for pedagogical purposes (increases in cGMP would translate into increased ratio), as performed by 1, 2 and others . Unfortunately, the bar graphs represented the percentage change in F(Acceptor)/F(Donor). We have now reanalyzed all data as described in Materials and Methods and plotted this as F(Donor)/F(Acceptor). This changed the magnitude of the responses but not the conclusion from each experiment.
3) Although the manuscript says "However, the dynamic range of the cGi -500 was larger than the PfPKG sensor (Figure 4c)." (lines 431-432), I think that basal cGMP levels in cells might reduce the maximal FRET responses of Yellow PfPKG which has the high affinity.
We also considered if basal cGMP levels within the measurable range of this biosensor would limit the maximal FRET responses. However, when we applied ODQ to inhibit basal sGC activity in cardiomyocytes (figure 3c) there was not a decrease in FRET, suggesting that cGMP levels, at least in cardiomyocytes, were below the range of detection with this biosensor.

4)
Since the dynamic range of Red PfPKG seems small, comparable to of Red cGES-DE5, I think that the representation of "a large dynamic range" in lines 455-456 ("To conclude, we have generated two novel biosensors for cGMP with high affinity and a large dynamic range.") is not appropriate.
We agree and thus modified the text (lines 486-7).

Reviewer #3 (Remarks to the Author):
The manuscript by G. Calamera et al. presents two new genetically-encoded sensors (Yellow PfPKG and Red PfPKG) that are designed for measuring nanomolar concentrations of cGMP. Only one sensor was available so far that could provide similar sensitivity (Re d cGES-DE5). Yellow PfPKG shows higher dynamic range and ~2.5 higher affinity than Red cGES-DE5 in cardiomyocytes. Yellow PfPKG also performs well in stellate ganglion neurons, when compared to another known sensor, cGi -500. These data demonstrate the significance of the study. The paper is well written, sound, and contains necessary control experiments. Together with positive experimental results, the authors also describe the experiments, where their sensors failed to surpass the performance of the known sensors. This information is important because it can save time for potential users. A serious improvement of the paper can be expected if the authors would measure the cGMP concentration dependence in purified solutions. The cell homogenates that they use now contain unknown basal concentration of cGMP and also non-negligible autofluorescence and scattering. These effects can introduce systematic and random errors in their measurements of the EC50 and dynamic range.
We thank the reviewer for the good evaluation and suggestion to improve the paper. We have now performed purification of the sensor using a protein purification protocol based on His-tag trapping column. We tested the purified sensor simultaneously with the cell homogenate expressing the Histagged sensor for comparison and used the same protocol as previously. The affinity for cGMP (EC 50 ) was the same for the sensor in pure solution as in homogenates and comparable to that of the Yellow PfPKG sensor without the His-tag. The dynamic range of the His-tagged purified sensor and the Yellow PfPKG sensor in homogenates was not different (42.3±0.4% vs. 39.4±3.5%). We present these results as supplementary data but describe them in lines 279-284. Based on these data, we think that the potential influence of endogenous cGMP from HEK293 cells on the determined cGMP affinity of our biosensor is negligible.
Several minor issues should also be addressed. 1. The authors should specify what protein corresponds to an abbreviation Dimer 2. What red FP is a monomer in this dimer? Is this tandem dimer or physically associated dimer without peptide link?
A monomeric Dimer2 was taken from the established Red-cGES-DE5 biosensor 4 . Dimer2 is evolved from the tetrameric DsRed (through 17 mutations) and each monomer of Dimer2 can potentially form dimers 5 . To avoid confusion in the specification of the fluorescent protein we used in the red biosensors, we have now specified what corresponds to Dimer2 in the Material and Methods section (lines 142-3).