11.07.2015 Views

PixFRET, an ImageJ plug-in for FRET calculation ... - ResearchGate

PixFRET, an ImageJ plug-in for FRET calculation ... - ResearchGate

PixFRET, an ImageJ plug-in for FRET calculation ... - ResearchGate

SHOW MORE
SHOW LESS
  • No tags were found...

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

54 J.N. FEIGE ET AL.Fig. 3. The ECFP SBT ratio c<strong>an</strong> be fitted as a function of ECFP<strong>in</strong>tensity. The data from Figure 1 were fitted with a l<strong>in</strong>ear (A) or <strong>an</strong>exponential (B) model.st<strong>an</strong>t, irrespective of the ga<strong>in</strong> applied to the PMT(Fig. 2B). This result suggests that the variations observedwith the set-up with two PMTs, or with the arrayof PMTs, reflect a differential response of the PMTs.Improv<strong>in</strong>g <strong>FRET</strong> Calculation by Model<strong>in</strong>g SBTRatios as a Function of Fluorophore IntensityTo circumvent the problem of SBT variation, us<strong>in</strong>g aunique PMT both <strong>for</strong> the detection of the donor <strong>an</strong>d ofthe acceptor seems there<strong>for</strong>e to be the best solution.However, this imposes to acquire the three ch<strong>an</strong>nelssequentially because of mech<strong>an</strong>ical movement of thePMT, a time-consum<strong>in</strong>g process that is not compatiblewith the <strong>FRET</strong> <strong>an</strong>alysis of rapidly diffus<strong>in</strong>g complexes.El<strong>an</strong>gov<strong>an</strong> et al. (2003) have already reported thevariation of SBT ratios as a function of fluorophore<strong>in</strong>tensity <strong>an</strong>d proposed <strong>an</strong> eleg<strong>an</strong>t but only commerciallyavailable algorithm to tackle this problem. Alternately,we propose <strong>an</strong> easily applicable method thatconsists <strong>in</strong> model<strong>in</strong>g SBT ratios as a function of fluorophore<strong>in</strong>tensity. For the problem reported here<strong>in</strong>, theCFP SBT ratio c<strong>an</strong> be fitted as a function of CFP <strong>in</strong>tensityeither with a l<strong>in</strong>ear or <strong>an</strong> exponential model(Fig. 3). These models are then used to estimate theSBT ratio correspond<strong>in</strong>g to each donor <strong>in</strong>tensity <strong>an</strong>dc<strong>an</strong> be implemented <strong>in</strong> the N<strong>FRET</strong> <strong>for</strong>mula describedby Xia <strong>an</strong>d Liu (2001) as follows:where F, CFP, <strong>an</strong>d YFP are the <strong>in</strong>tensities measuredwith the <strong>FRET</strong>, CFP, <strong>an</strong>d YFP sett<strong>in</strong>gs (correspond toF, D, <strong>an</strong>d A <strong>in</strong> the nomenclature proposed by Gordonet al. (1998)), a <strong>an</strong>d b are the average donor <strong>an</strong>dacceptor SBT ratios, <strong>an</strong>d c, d, e, f, <strong>an</strong>d g are the const<strong>an</strong>tsdeterm<strong>in</strong>ed by the fitt<strong>in</strong>g of the SBT ratio,accord<strong>in</strong>g to Figure 3.To validate this approach, the three <strong>FRET</strong> <strong>calculation</strong>methods have then been compared with the sameset of data (Fig. 4). Cells were tr<strong>an</strong>sfected with expressionvectors <strong>for</strong> ECFP <strong>an</strong>d EYFP (negative control),ECFP fused to EYFP (positive control), or PPAR-ECFP<strong>an</strong>d RXR-EYFP. The three <strong>for</strong>mulas give similar results<strong>for</strong> both the positive <strong>an</strong>d negative <strong>FRET</strong> controls as,under these experimental conditions, variations betweencells are small <strong>an</strong>d all fluorescence <strong>in</strong>tensities are close tothe average of the population used to calculate SBT ratios(Fig. 4A). The l<strong>in</strong>N<strong>FRET</strong> values are slightly higherbecause of the underestimation of the ECFP SBT <strong>for</strong> bothlow <strong>an</strong>d high ECFP <strong>in</strong>tensities (see Fig. 3A). The mostdramatic effect is seen when N<strong>FRET</strong> is calculated <strong>for</strong> the<strong>in</strong>teraction between the two nuclear receptors. Indeed,the more accurate l<strong>in</strong>N<strong>FRET</strong> <strong>an</strong>d expN<strong>FRET</strong> <strong>calculation</strong>ssignific<strong>an</strong>tly reduce st<strong>an</strong>dard deviation, with amore pronounced effect <strong>for</strong> expN<strong>FRET</strong>. To better characterizethe reduction of the variability of the measurementswhen model<strong>in</strong>g SBT ratios, we plotted the differencebetween expN<strong>FRET</strong> or l<strong>in</strong>N<strong>FRET</strong> <strong>an</strong>d N<strong>FRET</strong> as afunction of fluorophore <strong>in</strong>tensity (Fig. 4B). In both cases,the highest variation was observed <strong>for</strong> extreme ECFPvalues <strong>for</strong> which us<strong>in</strong>g <strong>an</strong> average SBT ratio leads to agreat under- or overestimation of the correction factor(Figs. 4B <strong>an</strong>d 4C). The results obta<strong>in</strong>ed with l<strong>in</strong>N<strong>FRET</strong><strong>an</strong>d expN<strong>FRET</strong> are hence not only more precise, but alsoimprove the statistical properties of the data, allow<strong>in</strong>gthe user to compare more accurately mild ch<strong>an</strong>ges. Wealso tried to fit the donor SBT with a second order polynomialmodel rather th<strong>an</strong> <strong>an</strong> exponential growth, but thismodel greatly overestimates SBT values when donor<strong>in</strong>tensities are very low or very high (data not shown).Pixel-by-Pixel <strong>FRET</strong> Analysis <strong>an</strong>d VisualizationCells are highly org<strong>an</strong>ized <strong>an</strong>d prote<strong>in</strong> distributionas well as <strong>in</strong>teractions are often limited to specific compartments.It is hence crucial <strong>for</strong> biologists to be able tomap <strong>in</strong>teractions with precision with<strong>in</strong> a cell or a populationof cells to better underst<strong>an</strong>d the spatial org<strong>an</strong>izationof cellular activities. We there<strong>for</strong>e developed <strong>an</strong><strong>ImageJ</strong> <strong>plug</strong>-<strong>in</strong> called ‘‘<strong>Pix<strong>FRET</strong></strong>’’ that allows one togenerate normalized <strong>FRET</strong> images, by comput<strong>in</strong>g pixelN<strong>FRET</strong> ¼ F CFPþYFP CFP CFPþYFP 3 a YFP CFPþYFP 3 bpffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiCFP CFPþYFP 3 YFP CFPþYFPð1Þl<strong>in</strong>N<strong>FRET</strong> ¼ F CFPþYFP CFP CFPþYFP 3 ðc CFP 3 CFP CFPþYFP þ d CFP Þ YFP CFPþYFP 3 bpffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiCFP CFPþYFP 3 YFP CFPþYFPð2ÞexpN<strong>FRET</strong> ¼ F CFPþYFP CFP CFPþYFP 3 ðe CFP 3 expðCFP CFPþYFP 3 f CFP Þ þ g CFP Þ YFP CFPþYFP 3 bpffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiCFP CFPþYFP 3 YFP CFPþYFPð3Þ

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!