This P o isional PDF co esponds o he a icle as i appea ed upon accep ance. Fully o ma ed
PDF and ull ex (HTML) e sions will be made a ailable soon.
AccPbFRET: an ImageJ plugin o semi-au oma ic, ully co ec ed analysis o
accep o pho obleaching FRET images
BMC Bioin o ma ics 2008, 9:346 doi:10.1186/1471-2105-9-346
Janos Roszik ([email p o ec ed])
Janos Szollosi ([email p o ec ed])
Gyo gy Ve eb ([email p o ec ed])
ISSN 1471-2105
A icle ype So wa e
Submission da e 23 Ap il 2008
Accep ance da e 19 Augus 2008
Publica ion da e 19 Augus 2008
A icle URL h p://www.biomedcen al.com/1471-2105/9/346
Like all a icles in BMC jou nals, his pee - e iewed a icle was published immedia ely upon
accep ance. I can be downloaded, p in ed and dis ibu ed eely o any pu poses (see copy igh
no ice below).
A icles in BMC jou nals a e lis ed in PubMed and a chi ed a PubMed Cen al.
Fo in o ma ion abou publishing you esea ch in BMC jou nals o any BioMed Cen al jou nal, go o
h p://www.biomedcen al.com/in o/au ho s/
BMC Bioin o ma ics
© 2008 Roszik e al. , licensee BioMed Cen al L d.
This is an open access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion License (h p://c ea i ecommons.o g/licenses/by/2.0),
which pe mi s un es ic ed use, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly ci ed.
- 1 -
AccPbFRET: an ImageJ plugin o semi-au oma ic,
ully co ec ed analysis o accep o pho obleaching
FRET images
János Roszik, János Szöllısi and Gyö gy Ve eb§
Depa men o Biophysics and Cell Biology, Resea ch Cen e o Molecula
Medicine, Uni e si y o Deb ecen, Deb ecen, Hunga y
§Co esponding au ho
Email add esses:
JR: [email p o ec ed]
JS: szo[email p o ec ed]
GV: e [email protected]
- 2 -
Abs ac
Backg ound
The accep o pho obleaching luo escence esonance ene gy ans e (FRET) me hod
is widely used o moni o ing molecula in e ac ions in cells. This me hod o FRET,
while among hose wi h he simples ma hema ics, is obus , sel -con olled and
independen o luo opho e amoun s and a ios.
Resul s
AccPbFRET is a use - iendly, e icien ImageJ plugin which allows ully co ec ed,
pixel-wise calcula ion and de ailed, ROI ( egion o in e es )-based analysis o FRET
e iciencies in mic oscopic images. Fu he mo e, au oma ic egis a ion and semi-
au oma ic analysis o la ge image se s is p o ided, which a e no a ailable in any
exis ing FRET e alua ion so wa e.
Conclusions
Despi e o he widesp ead applicabili y o he accep o pho obleaching FRET
echnique, his is he i s pape whe e all possible sou ces o majo e o s o he
measu emen and analysis a e conside ed, and AccPbFRET is he only p og am which
p o ides he comple e sui e o co ec ions - o egis e ing image pai s, o unwan ed
pho obleaching o he dono , o c oss- alk o he accep o and/o i s pho op oduc o
he dono channel and o pa ial pho obleaching o he accep o . The p og am
e icien ly speeds up he analysis o la ge image se s e en o no ice use s and is
eely a ailable.
- 3 -
Backg ound
Fluo escence esonance ene gy ans e (FRET) is a powe ul echnique ha can
be applied o s udy nanoscale in a- and in e molecula e en s and in e ac ions o
molecules in si u in biological sys ems [1]. In assessing FRET, luo escence o a
spec ally ma ched dono and accep o dye pai can be measu ed o e eal he
adia ionless ans e o exci a ion ene gy om he dono o he accep o , in he case
ha hei dipoles a e p ope ly o ien ed and he wo a e in spa ial p oximi y (usually a
a dis ance o 1-10 nm) [2]. This la e phenomenon is he basis o he popula i y o FRET
in biology: The dis ance o e which FRET occu s is small enough o cha ac e ize he
p oximi y o possibly in e ac ing molecules, unde special ci cums ances i e en p o ides
quan i a i e da a on exac dis ances, and, addi ionally, in o ma ion on he spa ial o ien a ion
o molecules o hei domains. Hence he e y ap opos e m om S ye , who equaled FRET
o a “spec oscopic ule ” [3]. FRET can be measu ed bo h in mic oscopic imaging and in
low cy ome y. While low cy ome ic FRET (FCET) ca ies he ad an age o examining
la ge cell popula ions in a sho ime, mic oscopic app oaches ha e he abili y o p o ide
subcellula de ail and he possibili y o co ela e FRET alues wi h o he biological
in o ma ion gained om luo escen labeling, on a pixel by pixel basis [4]. In a e iew abou
FRET imaging, Ja es-E ijman and Jo in classi ied 22 di e en app oaches o quan i ying
FRET in a sys ema ic way. The echniques all in wo majo g oups: mos o hem a e based
on dono quenching and/o accep o sensi iza ion, and a ew on measu ing emission
aniso opy o ei he he dono o he accep o [5]. In he p ac ice o cell biology, o dina y
con ocal mic oscopy is now b oadly a ailable, and b ings h ee quan i a i e FRET
app oaches wi hin close each. These a e he a ious a iome ic app oaches, dono
pho obleaching FRET and accep o pho obleaching FRET [4, 6]. Some o he app oaches
based on aniso opy [7, 8], luo escence li e ime [9, 10], imaging spec oscopy [11], o
li e ime imaging spec oscopy [12] equi e mo e specialized equipmen , while ye o he s lack
- 4 -
he quan i a i e measu emen o FRET e iciency and ely on a ious FRET pa ame e s ha
a e usually made un eliable by he a ying amoun s and a ios o dono and accep o in each
examined pixel [13].
Dono pho obleaching FRET, exploi ing he dec ease o exci ed s a e li e ime and
consequen ial p o ec ion om pho odes uc ion in he p esence o FRET was he i s
quan i a i e app oach applied o mic oscopy [14, 15, 20] and ca ies he ad an age o
being ela i ely simple o implemen and a he sensi i e, howe e , he need o ex e nal
con ols and he local a ia ions in empe a u e and oxygena ion can cause p oblems. The
a iome ic app oach based on cohe en conside a ion o dono quenching, sensi ized emission
and c oss- alk be ween channels was i s applied in low cy ome y [l] and hen adap ed o
mic oscopy [16]. While i yields i sel eadily o ime-dependen measu emen s, he a he
in ol ed ma hema ics usually sca es biologis s away who hen su ice wi h calcula ing
dubious FRET a ios. A obus , easy o use, sel -con olled FRET me hod, independen
o dono and accep o concen a ion and s oichiome y, is accep o pho obleaching
FRET, which equi es only simple image ma hema ics [4, 17, 18, 19]. The de-
quenching o he dono upon pho odes uc ing he accep o esul s in an inc ease o
he dono luo escence, which is p opo ional o he FRET e iciency E:
( _ _ )
( )
1
Dono quenched by accpe o
Dono de quenched
F
EF−
= −
A measu emen ha exploi s his p opo ionali y is acilely implemen ed in con ocal
mic oscopy, hus p o iding he op ion o dis inguishing a ious molecula associa ion
s a es e en a he subcellula le el. The me hod is also applicable o he e e -
sp eading amily o g een luo escen p o ein (GFP) de i a i es [20].
Image manipula ion and analysis in biological esea ch a e o en pe o med wi h he
ee ImageJ package [21]. In spi e o he nume ous plugins a ailable, he e a e only
h ee ools o help he e alua ion o FRET. Two o hem aid he assessmen o

- 5 -
a iome ic FRET images [22, 23], while ha p o ided by D. S epensky has been he
only eely a ailable ool o accep o pho obleaching FRET [24]. This plugin allows
he calcula ion o FRET e iciency based on a e age luo escence signals (i.e. no on
a pixel-by-pixel basis) om pixels abo e a p e-de ined h eshold in accep o
pho obleaching images. I does no p o ide co ec ion possibili ies, egis a ion, and i
allows selec ion o ec angula shaped ROIs only. As he need o e alua ing la ge
da a se s o molecula in e ac ions inc eases, we ha e unde aken o de elop a
p og am ha add esses all he abo e de iciencies, and is also capable o quick semi-
au oma ic p ocessing o se ial measu emen s.
Implemen a ion
The plugin was w i en in Ja a 1.6, and es ed wi h ImageJ e sion 1.38x.
FRET e iciency E(i,j) is ob ained pixel-by-pixel acco ding o
( ) ( )
1( , ) 1( , )
( , )
2( , ) 1( , ) 1( , ) 1( , )
(1 )( )
1( (1 ) )
D i j A i j
i j
D i j A i j D i j A i j
F F
EF F F F
δ
γ ε
− α −
= − − αδ + − α − α − δ
whe e FD1(i,j) and FD2(i,j) a e he dono luo escence alues o he pixel (i,j) be o e (1)
and a e (2) pho obleaching he accep o , and FA1(i,j) he accep o luo escence o he
same pixel be o e pho obleaching. All F alues a e backg ound co ec ed h oughou .
α, γ, δ and ε a e co ec ion ac o s ha a e desc ibed below.
In some cases, pho obleaching o he accep o is no comple e. As shown by an
Muns e e al. [25], he a e age FRET e iciency is di ec ly p opo ional o he
amoun o a ailable accep o molecules assuming ha pho obleaching occu s
indisc imina ely o all accep o molecules, and he e is no mo e han one accep o pe
dono molecule p esen . To co ec o incomple e accep o bleaching, he co ec ion
ac o α is calcula ed as
- 6 -
2( , ) 1( , )
A i j A i j
F Fα =
whe e FA2(i,j) and FA1(i,j) a e in ensi ies in he accep o channel in pixels abo e
h eshold o he dono and accep o labeled sample, be o e (1) and a e (2)
pho obleaching.
The co ec ion ac o γ o unwan ed pho obleaching o he dono du ing he image
acquisi ion p ocedu e [4, 26] can be calcula ed ei he as
1( , ) 2( , )
Dd i j Dd i j
F F
γ
=
o
1( , ) 2( , )
Dd i j Dd i j
F F
γ
=
whe e FDd1(i,j) and FDd2(i,j) a e dono luo escence in ensi ies o dono only (Dd)
samples in pixels abo e h eshold be o e (1) and a e (2) pho obleaching he
accep o , and he signs deno e mean alue. Since FRET p o ec s he dono om
pho obleaching [14], his ac o calcula ed based on a sample labeled wi h dono only
is no exac ly accu a e. Howe e , he di e ence in p ac ice is 10-20 % o 1-2%, which
may no cause g ea e o s in de e mining FRET. Ne e heless, cau ion needs o be
aken o minimize pho obleaching o he dono du ing he measu emen .
The p og am o e s (also as o he co ec ion ac o s δ and ε), he possibili y o
calcula e he ac o on a pixel-by-pixel basis and hen a e age pixels abo e h eshold
o aw da a, o , al e na i ely, o a e age he aw da a o hese pixels and hen
calcula e an a e age co ec ion ac o .
In he case ha he accep o dye also luo esces in he dono channel, FRET would be
unde es ima ed wi hou co ec ing o his c oss- alk [26]. The app op ia e co ec ion
ac o δ is calcula ed as
1( , ) 1( , )
Da i j Aa i j
F F
δ
=
- 7 -
o
1( , ) 1( , )
Da i j Aa i j
F F
δ
=
whe e FDa1(i,j) and FAa1(i,j) a e signals in he dono and accep o channels in pixels
abo e h eshold o an accep o only labeled sample, be o e (1) pho obleaching he
accep o .
In some cases, pho obleaching he accep o can yield a pho op oduc wi h dis inc
abso p ion and emission p ope ies, which can con ibu e o he pos -bleach dono
signal, esul ing in he o e es ima ion o FRET e iciency [4]. The co ec ion ac o ε
o such accep o -pho op oduc is calcula ed as
2( , ) 1( , )
Da i j Aa i j
F F
ε
=
o
2( , ) 1( , )
Da i j Aa i j
F F
ε
=
whe e FDa2(i,j) and FAa1(i,j) a e in ensi ies in he dono and accep o channels in pixels
abo e h eshold o an accep o only labeled sample, be o e (1) and a e (2)
pho obleaching.
The calcula ion o he cons an s γ, δ and ε equi es aking images wi h he same
pho obleaching p o ocol on samples labeled wi h dono only and accep o only, which
usually need o be aken anyway. To co ec o shi s in he x-y plane, he images a e
egis e ed using he Fas Ha ley T ans o m algo i hm [27] implemen ed in he
ImageJ package. All co ec ions a e op ional and can be ac i a ed / inac i a ed in he
“Co ec ions” menu o he plugin.
Resul s
Wi h ou p og am AccPbFRET, which can be ound in he addi ional ile [see
Addi ional ile 1] o can be downloaded om i s homepage [28], FRET e iciencies
- 8 -
a e calcula ed pixel-by-pixel, and hei dis ibu ion is de e mined o any use de ined
ec angula , polygonal, o eehand ype ROI o subcellula loca ion. Accu a e
selec ion o he examined cellula componen s is u he ed by he p o ision o
in e ac i ely se h eshold alues o dono and, op ionally, accep o luo escence
in ensi ies, and o also ga e using images wi h ele an independen luo escen labels
in he same sample. In addi ion, ou plugin p o ides au oma ic egis a ion o he
images, an absolu e necessi y o pe ec alignmen o dono images aken be o e and
a e pho obleaching. An example analysis wi h and wi hou egis a ion (along wi h
o he examples) can be ound in he addi ional ile [see Addi ional ile 1]. We
compa ed he esul s o he same images ob ained wi h FRETcalc [24] and
AccPbFRET, and we ob ained simila FRET e iciencies, 15.1% and 14.7%,
espec i ely. Howe e , when we used images ha needed egis a ion because o a
ew pixels shi , he esul s changed o 5.7% e sus 14.6%.
O he impo an issues wi h accep o pho obleaching FRET is co ec ing o bleaching
o he dono , o he c oss- alk o he accep o and/o i s p o isional pho op oduc o
he dono channel, and pa ial pho obleaching o he accep o , which a e also sol ed
by AccPbFRET.
The s eps o c ea ing he FRET image a e enume a ed in he main p og am window,
so he use only needs o ollow he ins uc ions (see image o main window on he
igh side o Figu e 1 o de ails). Supplemen a y in o ma ion appea s as a ool ip
when he mouse poin e is ho e ed o e an op ion o bu on ha migh need u he
explana ion. Thus e en no ice use s, hose un amilia wi h FRET and/o wi h Ja a
and ImageJ can quickly go h ough he analysis p ocedu e, wi hou he dange o
commi ing he usual e o s. On a e age, he comple e analysis om loading he
images o a i ing a eliable esul s akes less han a minu e.
- 15 -
Figu es
Figu e 1 – Sc eensho o an example analysis wi h he AccPbFRET plugin
The ImageJ (le op) and AccPbFRET ( igh ) dialog windows a e displayed, oge he
wi h dono channel sou ce images aken by a con ocal mic oscope (labeled
app op ia ely as ‘Dono be o e bleaching’, ‘Dono a e bleaching’), as well as he
calcula ed, co ec ed FRET image (‘T ans e image’), on which ROIs can be selec ed
and s a is ics calcula ed. These s a is ics can be seen in a sepa a e ‘Resul s’ window,
and a his og am o he FRET dis ibu ion is also p esen ed.
Figu e 2 – The analysis p ocess
In his igu e, he changes o dono and accep o images du ing he s eps o he
analysis p ocess a e shown. DB, DA: dono images be o e and a e bleaching he
accep o ; AB AA: accep o images be o e and a e bleaching (same images as in
Figu e 1, c opped o i he page). STEP1: o iginal images; STEP2: images a e
egis a ion (no e he disappea ance o he op lines om DA and AA); STEP3: a e
backg ound sub ac ion; STEP4: a e Gaussian il e ing; STEP 5: h esholded
images. The co ec ed FRET/ ans e image and his og am de i ed om i a e also
displayed.
Addi ional iles
Addi ional ile 1 – Sou ce code and example images
File o ma : comp essed zip ile (.zip)
Ti le: AccPbFRET_ 2_0.zip

- 16 -
Desc ip ion: This ile con ains he AccPbFRET.ja a sou ce code and some example
LSM and TIFF image iles oge he wi h explana ions.
Figu e 1
Figu e 2
Addi ional iles p o ided wi h his submission:
Addi ional ile 1: accpb e _ 2_0.zip, 10687K
h p://www.biomedcen al.com/imedia/6103352862081725/supp1.zip