Systematic Enzyme Mapping of Cellular Metabolism by Phasor-Analyzed Label-Free NAD(P)H Fluorescence Lifetime Imaging

Round 2

Reviewer 1 Report

The authors have adequately addressed my concerns from my first review.

Reviewer 2 Report

The authors significantly improved the manuscript.

There is one comment: p2/r65 "...more recently (?),..." Perhaps, there is something missing in the text?

I suggest to accept the manuscript for publication after minor correction.

Answer: I'm very sorry for making you confused. What we meant was that first technologies to measure fluorescence lifetime in bulk solution (without spatial Information) were developed and, later on, these technologies were applied to microscopy to perform (spatially resolved) fluorescence lifetime Imaging (FLIM). We will change the words.

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

REVIEWER 1

The authors in the manuscript entitled: “Systematic enzyme mapping of cellular metabolism by phasor-analyzed label-free NAD(P)H fluorescence lifetime imaging” presented systematic study and analysis of NAD(P)H-FLIM data. The work and presentation of the results are very nicely presented. I have only few minor comments.

We thank the reviewer for the encouraging opinion and for the valuable suggestions.

2 r. 47: There is a statement that the oxygen consumption measurements can be performed only ex vivo and not on single-cell level. The authors should consider, that there exists a possibility to study oxygen consumption of cells even in vivo with oxygen-sensitive probes as it was demonstrated in work: In vivo measurement of tissue oxygenation by time-resolved luminescence spectroscopy: advantageous properties of dichlorotris(1,10-phenantroline)-ruthenium(II) hydrate, JBO 2014 77004.

We thank the reviewer for pointing out this important approach we missed to mention: measuring pO₂ in tissue using ruthenium-organic fluorophores. We included the citation as well as a review of Dmitri Papkovsky and Ruslan Dmitriev in the introduction. We also better discussed in this section the relevance of not only measuring the oxygenation level of the tissue but also the oxygen use in biochemical reactions involving NADH and/or NADPH, catalyzed by enzymes – both necessary pieces of information in order to fully retrieve tissue function.     

2 r. 56 A potential application of FLIM in glioblastoma to monitor cell metabolism was recently demonstrated in work Tomkova S. et al. In vitro identification of mitochondrial oxidative stress production by time-resolved fluorescence imaging of glioma cells, BBA – Mol Cell Res 2018, 1865:616-628. This reference should be amended.

We apologize we missed this relevant reference and included it to the revised version of the manuscript.  

3 r. 144 What is “e” in 1/e. Did you mean the time?                           

We meant by ”e“ the Euler’s number, since the fluorescence decay follows an exponential function. We included this explanation to the revised manuscript.

4 r. 157 some characters are missing: …”( BG):”                                                                              

We thank the reviewer for indicating us these typos: here „sigma" for standard deviation is missing.

4 r. 167 it was not introduced what is “Re” and “Im”. Probably it was explained in the previous papers. It will be helpful for the reader to have short description. It is similarly further in the text.  Please, check all shortcuts.                                                                                                          

We better state in the revised manuscript that the phasor vector is a complex number:  z = Re + Im*i, with i² = -1. „Re“ represents the real part of complex number, „Im“, the imaginary part of the same. We also checked all other shortcuts and verified their explanation in the text.  

Figure 1 A, the character “i” is missing in the figure.                                                      

We included the character to Figure 1A.  

What is the aim of the asterisk in Figure 1D*?                                                                   

The asterisks indicate the data sets used to calculate the SNR (Suppl. Fig. 3) as well as to calculate τ₂ to mark the half circle. We included the information in the figure caption.

Similar in 3F

Fi)* is the area marked by * in Cii). We erased the potentially confusing asterisks from the figure, because the explanation is already stated in the figure caption. The same was done for the asterisk in Fig 4D, which had the same meaning.  

8 r. 276 Please, check equation 2. Is it really root of square?                                                     

In Eq. 2 we aimed to determine the absolute values of the variation, as indicated in the revised version of the manuscript. One method to determine the absolute value is to calculate the square and then apply the square root. For the sake of simplicity, we rewrote the equation using the more common mathematical operator |...|.  

8 r. 309 …” of two pixel ( =2)” Something missing in the bracket.                                           

Also here, „sigma" (standard deviation) was missing and have been now included to the text.

REVIEWER 2

The authors describe the utilization of the phasor method, a fit free analysis of fluorescence lifetime, used to examine the contribution of metabolic enzymes to the NAD(P)H fluorescence emission. Herein, the authors describe a method to isolate the contributions of individual proteins and to determine subcellular populations and distributions. The importance of providing a detailed analysis of metabolic properties cannot be understated as deviations within cell homeostasis is the foundation for many life-threatening diseases, such as cancer and diabetes. The overall manuscript is interesting with potentially important information technical advances that could be used to further screen cellular disease states; however, there are some major issues that need to be addressed before it can be considered for publication.

We thank the reviewer for agreeing with us on the relevance of NAD(P)H metabolism in understanding various diseases and for his comments and suggestions that helped us to make the manuscript more convincing and, thus, more useful for a broad audience.

Major Issues:

1. The authors do not which variants of enzymes they are working with (e.g. Mitochondrial vs. cytosolic MDH).

We thank the reviewer for bringing up this important point. In the revised version of the manuscript we indicate which enzyme variants we used. For instance, we measured mitochondrial MDH. In the section „Material and Methods“, we listed the source of all investigated enzymes which is directly linked to the cellular compartment and species they have been isolated from. We mentioned in the revised manuscript the fact that enzymes are not only isolated from different cellular compartments but also from different species. However, it is known that these very important enzymes are highly conserved across the species and across cellular compartments – a central evolutionary feature. In this sense, the NADH or NADPH binding site on the enzymes is also conserved across species and cellular compartments. Since the NAD(P)H binding site is mainly determining the NAD(P)H fluorescence lifetime when bound to enzymes, we expect the fluorescence lifetime to be similar for the same enzyme isolated from different species or cell compartments. A thorough discussion in this respect was added to the rewritten manuscript.

2. Furthermore, they report “benchmarking” of the phasor using samples under in vitro aqueous conditions and provide their phasor lifetimes as that of which is found in live cells. Do the authors know these conditions used are identical to those found within the cell? These conditions can have drastic effects on the fluorescent properties of NAD(P)H. For example, in 1970 Dr. Weber’s group was able to show that LDH could have multiple lifetimes depending upon specific interactions. This is major observation that needs to be addressed as it could undermine the validity of the authors’ analysis.

We thank the reviewer for pointing out that we missed in the previous version of the manuscript to explicitly comment on the “ground truth” that the fluorescence lifetime of NAD(P)H depends on many environmental factors (Gordon Scott et al, 1969) such as refractive index (Strickler&Berg, 1962), pH, ion concentrations, etc. – as the lifetime of all fluorophores does – and on how we accounted for these effects when designing our “benchmarking” approach. In order to exclude major effects on the fluorescence lifetime of NAD(P)H which were not related to binding to specific enzymes, we performed all our in vitro studies using media resembling similar solvent polarity, pH, ion concentrations (and also similar refractive index) as those found in cells and tissues. Following the criticism of the reviewer, we included and commented on the early work from the lab of Gregorio Weber (Gordon Scott et al, 1969) regarding the influence of various factors on the lifetime of NADH bound to selected enzymes (also LDH). In this early work, the fluorescence lifetime of NADH bound to LDH in aqueous solution – similar but not identical to cell culture media – was found to be 1.5 ns. This is in good agreement with the value later determined by other FLIM labs: 1.6 ns and also by us, since we expect that the pure fluorescence lifetime of NADH when bound to a single enzyme (such as LDH), in a homogenous media is mainly determined by the steric geometry of the corresponding binding site. Possible longer fluorescence components of the decay curve (> 10 ns) cannot be resolved using pulsed lasers at repetition rates of 80 MHz since the decay measurement time window is limited to 12.5 ns.  While the fluorescence lifetime of fluorophores may vary if it is not measured under the magic angle, since the molecular diffusional rotation impacts on it,  we expect that the fluorescence lifetime of NAD(P)H, as a small molecule, will not be influenced by this phenomenon since it is much shorter (hundreds of picoseconds to few nanoseconds) than the diffusional rotation time if bound to enzymes of several 100 kDa, either freely rotating or bound to membranes (few µs to few ms). We included all this information in the revised version of the manuscript.

3. In Fig. 1, the authors show an example of their phasor data for MDH and LDH under their ideal conditions. However, the phasor trajectory is clearly off the line for both cases which indicates a deviation in expected summation of lifetimes (possibly having a third lifetime present or an interaction with NADH that is altering its emission). This representation clearly has errors within it and needs to be addressed.

We agree with the reviewer and thank him for this comment because it made to us the explanation of our phasor plots was not complete and, thus, misleading. It is absolutely true that the phasor trajectory is even under optimal conditions slightly parallel to the line. This becomes more evident if the fluorescence signal is even lower, in cells or tissues. The main reason for this is that not only free and enzyme-bound NAD(P)H contribute to the fluorescence decay but also the noise of background and signal, since the experimental system itself, especially the detector, introduces uncertainty (noise) to any measured signal. Even if no signal is detected, Poisson distributed electronic noise will be registered. This always adds to unspecific background from the environment and to the true signal, leading to super-Poisson distributions (as also indicated in literature e.g. from the lab of Prof. Gratton). From a time-resolved perspective, the (noisy) background represents the overlap of undamped oscillations of various frequencies, hence, being characterized by an infinite decay lifetime. Since this component is intrinsically contained in all measured data, it shifts the experimental data off the expected trajectory, being a third lifetime impacting on the data. As the physics laws of detection do not allow us to experimentally remove this phenomenon, we determined the correlation between the signal-to-noise-ratio (a normalized measure of background noise related to the multiple frequencies of background oscillation) and the results of optimal time-resolved NAD(P)H fluorescence data. We found that above an SNR value of 5, the lifetime results using the phasor approach are similar to each other and, thus, are comparable – without being perfect. In cells and tissues, we pay particular attention to evaluate only data being characterized by an SNR above 5 in order to minimize the effect of the „background noise lifetime“. Moreover, in cells and tissue we expect the high variety of enzymes to which NADH and NADPH may bind to have an additional impact on the phasor representation of the experimental FLIM data. These aspects add to previously investigated parameters which change the fluorescence decay curve of fluorophores to different extents such as refractive index (Strickler, Berg, 1962), ion concentrations (Na+, Ca2+, Mg2+, Cl-), pH, pO2 etc. However, since the microenvironment in cells and in media are similar as far as these parameters are concerned, we expect that they do not have an impact on the fluorescence lifetime of NAD(P)H, neither free nor bound to enzymes. We discussed these aspects in the revised version of the manuscript and made the statements clearer.             

Minor Issues:

Line 60: “Currently, the time-correlated single-photon counting (TCSPC), which requires pulsed excitation as delivered by two-photon microcopy, is the method of choice to comprehensively acquire the molecular complexity within living cells and tissues despite of being rather slow (1–10 s/frame) [9, 26].” Why is TCSPC the method of choice?? This is the authors option and is not fact as many researchers in the field utilize the frequency domain method (including myself). I recommend that the authors reword this as to not offend researchers whom use a different method. Suggested references to add Ranjit S, Malacrida L, Jameson DM, Gratton E. (2018) Fit-free analysis of fluorescence lifetime imaging data using the phasor approach. Nat Protoc. Sep;13(9):1979-2004. this review is relatively recent, but it provides a comprehensive overview of the phasor method as well as its application to the in vivo NADH area. Jameson, DM etr al. 1989 Time-resolved fluorescence studies on NADH bound to mitochondrial malate dehydrogenase.

We agree with the reviewer that the way we put it was misleading and appears ignorant – this was not our intention and we deeply apologize for this accident. Our long-term perspective is to use NAD(P)H-FLIM in vivo in mouse models. In order to reach the region of biological interest e.g. germinal center in lymph node, we need to use a pulsed low-energy excitation (2PM). This is the reason why we characterized the fluorescence lifetime properties of the NAD(P)H-depending enzymes by TCSPC and not by “frequency domain” FLIM and why we called it “the method of choice”. Apart from that, we are also establishing frequency-domain FLIM in an ongoing project, in which high throughput is needed – evidently a major advantage of “frequency-domain” FLIM over “time-domain” FLIM, in general, and TCSPC, in particular. We discuss these aspects in the revised version of the manuscript and included the suggested references accordingly.

Poor English examples:

Line 85: “Already 1992, Lakowicz and his group….”                                                                                     

Line 127: “We verified the performance of the here developed analysis framework…..”                           

Line 226: “The fundament of our approach to interpret NAD(P)H-FLIM data consists of the most probable states we may found NADH and NADPH in cells and tissues.”

Line 335: “…a cell line exhibiting mesenchymal stromal cell-like properties and being able to differentiate towards adipocytes upon stimulation.”

Line 405: “The transformation results in complex number which real and imaginary part…”

We apologize for both style and grammar errors in our manuscript. Dr. Mairi McGrath helped us to rewrite the manuscript accordingly.