Quantitative Time-Lapse Fluorescence Microscopy in Single ...web.mit.edu/biophysics/papers/ARCDB2009.pdf · ANRV389-CB25-13 ARI 4 September 2009 19:45 Quantitative Time-Lapse Fluorescence

ANRV389-CB25-13 ARI 4 September 2009 19:45

Quantitative Time-LapseFluorescence Microscopyin Single CellsDale Muzzey1,2 and Alexander van Oudenaarden1

1Department of Physics, Massachusetts Institute of Technology, Cambridge,Massachusetts 021392Harvard University Graduate Biophysics Program, Harvard Medical School, Boston,Massachusetts 02115; email: [email protected]

Annu. Rev. Cell Dev. Biol. 2009. 25:301–27

First published online as a Review in Advance onJuly 8, 2009

The Annual Review of Cell and DevelopmentalBiology is online at cellbio.annualreviews.org

This article’s doi:10.1146/annurev.cellbio.042308.113408

Copyright c© 2009 by Annual Reviews.All rights reserved

1081-0706/09/1110-0301$20.00

Key Words

cell-to-cell variability, automated image analysis, lineage construction,microfluidic devices, fluorescent proteins

AbstractThe cloning of green fluorescent protein (GFP) 15 years ago revolution-ized cell biology by permitting visualization of a wide range of molecularmechanisms within living cells. Though initially used to make largelyqualitative assessments of protein levels and localizations, fluorescencemicroscopy has since evolved to become highly quantitative and high-throughput. Computational image analysis has catalyzed this evolution,enabling rapid and automated processing of large datasets. Here, we re-view studies that combine time-lapse fluorescence microscopy and au-tomated image analysis to investigate dynamic events at the single-celllevel. We highlight examples where single-cell analysis provides uniquemechanistic insights into cellular processes that cannot be otherwiseresolved in bulk assays. Additionally, we discuss studies where quan-titative microscopy facilitates the assembly of detailed 4D lineages indeveloping organisms. Finally, we describe recent advances in imagingtechnology, focusing especially on platforms that allow the simultaneousperturbation and quantitative monitoring of biological systems.

301

Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


Contents

INTRODUCTION . . . . . . . . . . . . . . . . . . 302INTERCELL COMPARISONS

OF SPATIO-TEMPORALDYNAMICS . . . . . . . . . . . . . . . . . . . . . . 303Temporal Noise. . . . . . . . . . . . . . . . . . . . 303Revealed Pitfalls of Bulk Assays . . . . . 306Subcellular Localization as a Proxy

for PosttranslationalModifications . . . . . . . . . . . . . . . . . . . 306

INTRACELL COMPARISONS OFSPATIOTEMPORALDYNAMICS . . . . . . . . . . . . . . . . . . . . . . 308Timing of Multiple Events . . . . . . . . . 308Quantitative Fluorescence Speckle

Microscopy . . . . . . . . . . . . . . . . . . . . . 311Evidence for Feedback Loops

in Dynamic Measurements . . . . . . 311CONSTRUCTION OF 4D

DEVELOPMENTAL LINEAGESUSING QUANTITATIVEMICROSCOPY . . . . . . . . . . . . . . . . . . . 313Yeast . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313Worms . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314Plants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314Flies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316Fish . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316

MICROFLUIDIC DEVICES. . . . . . . . . 317Specialized Stimuli . . . . . . . . . . . . . . . . . 317Highly Parallel, Long-Term

Imaging . . . . . . . . . . . . . . . . . . . . . . . . 319THE ROAD AHEAD . . . . . . . . . . . . . . . . 319

INTRODUCTION

In many standard biological assays there isa tradeoff among molecular specificity, spa-tial resolution, and temporal sampling. For in-stance, electron micrographs have extremelyhigh spatial resolution, but they generally can-not image specific molecules, and the tempo-ral dimension is lost because sample prepa-ration requires fixation. Southern, Northern,and Western blots are similarly compromised:They have high molecular specificity but lack

spatial resolution because they measure molec-ular abundance at the population level, not thesingle-cell level. Even the polymerase chain re-action, which is highly sequence-specific andcan be performed on single cells (Bengtssonet al. 2005, Peixoto et al. 2004), requires thatcells be lysed, preventing sequential samplingfrom the same cell over time.

Genetically encoded fluorescent proteinsovercome this tradeoff. They enable highmolecular specificity via direct fusion to pro-teins of interest using molecular cloning tech-niques. They afford high spatial resolution: Viatechniques such as flow cytometry and mi-croscopy, they allow detection of fluorescentproteins at the single-cell level and even per-mit spatial localization at effectively arbitraryresolution when sparsely distributed through-out the cell (Kural et al. 2005). Finally, fluores-cent proteins permit temporal sampling over awide range of timescales in live cells, becausefluorescence is generally nontoxic.

Because imaging experiments can involveseveral distinct molecular species (e.g., usingdifferently colored fluorophores), thousands ofindividual cells, and hundreds to thousands oftime points, manual attempts at quantitativeanalysis of time-lapse microscopy data wouldbe tedious, error-prone, and frankly impossiblein many cases. Fortunately, automated imageanalysis can quickly and quantitatively detectdynamic features of interest in an unbiasedmanner. In general, image analysis algorithmssegment an image into regions based on theintensities of adjacent groups of pixels. Theseregions can then be rapidly classified based onmany criteria including their intensity, shape,size, velocity, pixel-to-pixel variability, colocal-ization with regions in other color channels, andcountless more. Regions of interest can then betracked over time, and any number of quantita-tive phenotypes can be extracted.

Though countless research efforts in celland developmental biology have exploited flu-orescent proteins, we restrict our focus here topapers that specifically couple time-lapse flu-orescence microscopy with automated imageanalysis. We first review papers in which the

302 Muzzey · van Oudenaarden

Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


use of quantitative, time-lapse, single-cell imag-ing revealed cell-to-cell variability in the timingof cellular events. Second, we survey papersthat focus less on inter-cell comparisons andmore on intra-cell dynamics such as the co-ordination of multiple events that overlap inspace and time, the behavior of macromolecularstructures in the cytoskeleton, and the variousregulatory roles of feedback loops. Third, wediscuss novel insights gained from genealogi-cal data that involve dynamically tracking sin-gle cells within a growing population. Fourth,we highlight research that couples microfluidicdevices with microscopy, allowing exquisitecontrol over extracellular conditions while si-multaneously monitoring dynamic intracellularresponses. Finally, we feature new develop-ments in single-cell quantitative imaging thatincrease resolution, throughput, and controlover extracellular and intracellular conditions.

INTERCELL COMPARISONS OFSPATIO-TEMPORAL DYNAMICS

It is now well established that a population ofgenetically identical cells can exhibit extensivecell-to-cell variability in the expression levelsof many genes (Elowitz et al. 2002, Newmanet al. 2006, Ozbudak et al. 2002). This vari-ability, termed noise, is reflected in the widthof a population distribution, or histogram, rep-resenting the whole-cell fluorescence valuesfor all single cells; mathematically, it is oftendefined as the ratio between the standard de-viation and the mean expression level of thepopulation. In principle, a histogram can sum-marize how any quantitative trait is distributedacross a population of cells. Thus, in additionto noise in gene expression, noise in many otherphenotypes can also be measured and studied.Because noise in gene expression is nicely re-viewed elsewhere (Kaern et al. 2005, Maheshri& O’Shea 2007, Raj & van Oudenaarden 2008),below we focus on papers that investigate cell-to-cell variability in other quantitative metricsthat rely specifically on the spatial and/or tem-poral information that time-lapse microscopyprovides.

Temporal Noise

To quantify noise in the timing of a seriesof sequential events, histograms reflecting thetime elapsed between pair-wise events in apopulation of single cells must be compiled.The Ramanathan lab recently used quantita-tive microscopy to measure noise in the timingof meiotic stages in budding yeast (Nachmanet al. 2007). Triggered by nutrient deprivation,meiosis in diploid yeast cells consists of a se-ries of sequential events: expression of meioticregulators, a round of DNA replication, andtwo successive nuclear division events (calledMI and MII), ultimately yielding four haploidspores. The researchers could temporally dis-tinguish several of these meiotic stages by track-ing the level and localization of the early mei-otic regulator Dmc1, which they fused to YFP(Dmc1-YFP).

Because Dmc1 binds meiotic chromosomes(Bishop 1994), automated image analysis ofDmc1-YFP clusters in single cells could de-termine whether a cell had undergone zero,one, or two nuclear-division events (leading to1, 2, or 4 clusters of Dmc1-YFP, respectively)(Figure 1a). There was considerable cell-to-cell variability in the timing between nutrientdeprivation and MII, effectively the first andlast meiotic landmarks, respectively (stress-to-end time in Figure 1b). However, because theresearchers had time measurements of multi-ple intermediate steps, they could determineto what extent variability in the entire pro-cess was influenced by variability in the in-termediate processes. They found that totalmeiotic timing noise was dominated specifi-cally by variability in the time between nutrientdeprivation and MI; furthermore, they deter-mined that this variability was largely accountedfor by variability in the time between nutri-ent deprivation and the onset of expression ofmeiotic regulators (Figure 1b). They also usedtheir rich single-cell dataset to demonstrate thatthe duration of successive stages was largely un-correlated with the duration of the time re-quired to induce the meiotic regulators. Thisobservation is noteworthy because correlation

www.annualreviews.org • Quantitative Time-Lapse Fluorescence Microscopy in Single Cells 303

Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


Entry MI MII EndStress

Time

Num

ber

of c

ells

Stress-to-endtime

Stress-to-endtime

Stress-to-endtime

1

2

3

4

5

6

7

Cell

#

Time

Entry MI MII EndStress1

2

3

4

5

6

7

Cell

#

Anticorrelated CorrelatedEntry MI MII EndStress

1

2

3

4

5

6

7Ce

ll #

Time

Num

ber

of c

ells

Time

Num

ber

of c

ells

Time Time

a b

c

EndMII

Entry

MI

Figure 1Cell-to-cell variability in the timing of meiotic events. (a) Meiotic regulator Dmc1 was fused to YFP andvisualized in live budding yeast cells. Because Dmc1 associates with chromosomes, the stage of meioticprogression (e.g., entry, MI, MII, end) could be tracked using image-analysis software that effectively detectsthe number of Dmc1-YFP foci (e.g., entry has one focus, MI has two, and MII has four). (b) Top:Hypothetical data for seven single cells based on the results in Nachman et al. (2007), where the length ofthe shaded bars corresponds to the amount of time spent in each stage. Nutritional stress was applied to allcells at the same time—hence their alignment at the stress marker on the left. Bottom: Hypotheticalhistograms reflecting the time spent in each stage across many cells. Note that variability in the total time(i.e., stress-to-end time) is dominated by the stress-to-entry time and not by variability in the duration ofother stages, as observed by Nachman et al. (c) Left: Hypothetical single-cell profiles (top) and histograms(bottom) illustrating how anti-correlation between the stress-to-entry time and the MII-to-end time leads to areduction in stress-to-end variability as compared to (b). Right: Correlation between stress-to-entry andMII-to-end times leads to more variability in the stress-to-end time than in (b). Panel (a) adapted fromNachman et al. 2007 with permission of Cell.

(anti-correlation) among the durations of inter-mediate events would increase (decrease) noiseof the whole event (Figure 1c). In sum, thiswork by Nachman et al. (2007) highlights and

critically relies upon several key advantagesof quantitative time-lapse imaging and auto-mated analysis: Their phenotype of interestwas defined by the subcellular localization of a


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


fluorescent marker and required that single cellsbe tracked over long time periods.

The Siggia and Cross labs also elegantly in-vestigated temporal noise in the budding yeastcell cycle, focusing on the Start checkpoint,which marks the beginning of the G1/S tran-sition and commitment to cell division. Startinvolves the coordinated activation of manyevents, including transcriptional activation ofa battery of genes and bud emergence. To mea-sure temporal coordination between two ofthese events (gene activation and bud emer-gence) in single cells, the researchers expressedGFP under the control of the promoter forthe cyclin CLN2 gene (CLN2pr-GFP), which isone of the genes upregulated at Start. Time-lapse fluorescence microscopy coupled withautomated image analysis quantified cell-to-cell variability in the time between bud emer-gence and peak CLN2pr-GFP expression (bud-to-peak time in Figure 2a). Interestingly, bycomparing histograms compiled in wild-typeand mutant cells, the authors subsequently

−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−→Figure 2Variability in the timing of mitotic cell-cycle eventsand positive feedback revealed by single-cell traces.(a) Top: Schematic of single-cell events tracked viaimage analysis software in investigation ofbudding-yeast Start checkpoint. The timing ofcytokinesis was determined by the rapiddisappearance of the Cdc10-GFP fusion protein(red ) from the bud neck. The green signalcorresponds to expression of GFP under the controlof the CLN2 promoter (CLN2pr-GFP). Bottom:Hypothetical histograms based on findings in whichswi4� cells exhibit considerably higher variability inthe timing both between bud emergence and peakCLN2pr-GFP signal and between successivecytokinesis events. (b) Hypothetical single-cell data( faded lines) illustrating how bulk measurements(dotted opaque lines) mask the observation thatCLN2pr-GFP is expressed earlier in wild-type cellsthan in cln1�cln2� cells, as shown in Skotheim et al.(2008). The average traces overlap for much of thetrajectory and give no indication that wild-type cellsexpress CLN2pr-GFP earlier than mutant cells.However, the average time at which single-cellCLN2pr-GFP levels cross an arbitrary threshold(dashed line) is notably different between the celllines (see dots and error boundaries above the plot).

identified Swi4 as a protein required for lowcell-to-cell variability in the bud-to-peak time.This increase in bud-to-peak timing variabilityin the absence of Swi4 is noteworthy, as it coin-cided with considerably higher temporal vari-ability in cell-division times, which they trackedvia a fluorescent fusion protein that marks theseptin ring and fades rapidly upon cytokinesis(Figure 2a). This study is significant in thatit identifies a common molecular determi-nant of two distinct temporal noise phenotypes

Cytokinesis CytokinesisBud

emergencePeakGFP

Time

Wild-type

swi4∆swi4∆

Nu

mb

er

of

Ce

lls

Wild-type

Bud-to-peaktime

Cytokinesis-to-cytokinesis time

Time

Time

Mo

ck fl

uo

resc

en

ce (

arb

. un

its)

Time needed for single cellsto cross dotted threshold

(average +/– std. dev.)

Threshold

Wild-type

cln1∆ cln2∆

a

b


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


(in bud-to-peak timing and cell-cycle duration)that can only be measured with quantitative mi-croscopy. In a follow-up investigation (Di Taliaet al. 2007), this group further partitioned theG1 stage of the cell cycle into two phases dis-tinguished by the determinants of their tempo-ral noise properties. Differences in cell volumeand Cln3 levels largely accounted for variabilityin the first phase, whereas temporal variabilityin the second phase was volume-independentand arose mainly from differential expression ofCLN2.

Revealed Pitfalls of Bulk Assays

A recent paper from the Cross and Siggialabs nicely illustrated the perils of bulk mea-surements that single-cell analysis avoids(Skotheim et al. 2008). Previous bulk mea-surements showed that activation of CLN2expression occurred with similar kinetics inwild-type cells and in cln1� cln2� cells (Stuart& Wittenberg 1995), leading to the conclu-sion that Cln1 and Cln2 do not positivelyautoregulate their own expression. However,single-cell analysis of CLN2pr-GFP withtime-lapse microscopy revealed several criticalexpression differences between wild-type andcln1� cln2� cells (Skotheim et al. 2008). Inparticular, CLN2pr-GFP expression occurredmore rapidly in wild-type cells than in mutantcells, indicative of positive feedback by Cln1and Cln2. At the same time, though, CLN2pr-GFP expression was higher and had largertemporal variability in mutant cells than inwild-type cells. Together, these three effects(i.e., changes in magnitude, timing, and timingnoise) mask the evidence for positive feedbackwhen averaged over many cells (Figure 2b).In a population average, the few early andhighly expressing mutant cells compensate forthe late and lowly expressing cells to yield ameasurement nearly identical to that of thewild-type population. Thus, Skotheim et al.used quantitative single-cell microscopy toreveal a critical positive feedback loop thatthey later showed is required for robust budemergence and coherence (i.e., low temporal

noise) in the transcriptional activation of G1/Sgenes.

Single-cell measurements have also demon-strated inaccuracies in our understanding of thebehavior of the tumor suppressor p53 in re-sponse to DNA damage (Lahav et al. 2004).Bulk Western blot analysis of p53 showed thatits levels undergo damped oscillations in re-sponse to DNA damage (Lev Bar-Or et al.2000). However, damped oscillations in bulkcan result from several different types of single-cell behavior (Figure 3). When Lahav andcolleagues fused CFP to p53 and assayed thelevel of p53-CFP in single human cells in re-sponse to DNA damage, they found that thedamping observed in bulk was not due to auniform decrease in the magnitude of succes-sive pulses (Figure 3a) or decaying synchrony(Figure 3b). Instead, each cell had a discretenumber of pulses of roughly equivalent mag-nitude (Figure 3c), and the number of cellswith n pulses decreases as n increases. IncreasedDNA damage was found to increase the aver-age number of pulses in each cell, rather thanmodulating the frequency or amplitude of eachpulse. The difference between damped oscilla-tions (Figure 3a) and a variable-length seriesof roughly identical pulses has major conse-quences on the modeling of single-cell behav-ior. Thus, when possible, the basis of modelsmeant to describe single-cell dynamics shouldbe the results of single-cell assays.

Subcellular Localization as a Proxyfor Posttranslational Modifications

Another powerful application of single-cell mi-croscopy and analysis is the study of a pro-tein’s posttranslational modifications. Althougha fluorescent tag allows easy assessment oftotal protein levels (Wu & Pollard 2005), a pro-tein’s posttranslational modifications are muchharder to detect, because total fluorescenceremains unchanged upon modification. Pro-tein modifications that cause major confor-mational changes can potentially be measuredusing intramolecular resonance-energy trans-fer (Charest et al. 2005, Lohse et al. 2007),


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


Moc

k si

gna

l (ar

b. u

nits

)

Moc

k si

gna

l (ar

b. u

nits

)

Moc

k si

gna

l (ar

b. u

nits

)

a Damped, synchronized b Undamped, decaying synchrony c Undamped, variable number ofpulses, otherwise sychronized

Time Time Time

Figure 3Damped oscillations observed in bulk measurements can result from fundamentally different behaviors at the single-cell level. In (a),(b), and (c), all data are hypothetical; the mean is shown in purple, and single-cell traces are light green. (a) Single cells behave like themean and undergo synchronized, damped oscillations. (b) Single cells are initially synchronized, but their synchrony decays with time.(c) Cells exhibit a discrete number of undamped and synchronized signal pulses, but the number of cells pulsing diminishes with time,consistent with the observations in Lahav et al. (2004) of p53 expression. A baseline increase to the mock signal is added in (c) such thatthe mean trace resembles those in (a) and (b).

but this technique requires the fusion of twofluorescent molecules and serendipity in theirconformation-dependent orientations. Alter-natively, modifications for which specific anti-bodies have been developed can be measuredvia immunofluorescence, though cells must befixed, thereby eliminating the possibility ofdetecting correlated features within individ-ual cells across multiple time points. However,in special cases, a protein’s subcellular local-ization changes upon being posttranslationallymodified. Fluorescent fusions to such proteins,coupled with time-lapse microscopy and auto-mated image analysis, provide a powerful toolto study dynamics in protein activity in singlecells.

In elegant work from the Elowitz lab, theactivation status of the yeast calcium-responseregulator Crz1 was monitored in single cellsvia the subcellular localization of Crz1-GFP(Cai et al. 2008). Normally sequestered in thecytoplasm in a phosphorylated state, Crz1 isdephosphorylated upon the introduction of ex-tracellular calcium and rapidly translocates tothe nucleus, where it directly regulates the ex-pression of more than 100 genes (Yoshimotoet al. 2002). The authors used the ratio be-tween the cell’s brightest pixels and its mean

intensity as a proxy for the activation statusof Crz1. Remarkably, they found that calciumstress elicited one synchronized burst of Crz1activity followed by an asynchronous series ofbursts among individual cells (Figure 4). Av-eraging this response over many cells yieldedan activation profile consistent with microarrayresults of the Crz1-regulated genes (Yoshimotoet al. 2002). However, it is noteworthy that at-tempts to deduce single-cell behavior from themicroarray results would have been highly mis-representative (Figure 4). The authors showedthat more severe calcium stress changed the fre-quency of bursts but not their amplitude or du-ration; these data suggest that burst frequencyis a particularly important parameter in co-ordinating the downstream response. Indeed,using a simplified model and subsequent exper-iments, the authors argue that the frequency-modulated activation of Crz1 is optimal forthe coordinate regulation of genes with diversepromoter architectures over a range of stressmagnitudes. Because they also found that pul-satile activation of transcription factors is notunique to Crz1 (e.g., the activity of generalstress-response factor Msn2 also pulses), it willbe fascinating to see how this behavior affectsother cellular processes.


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


Synchronizedfirst pulse

Asynchronouslater pulses

Moc

k nu

clea

r loc

aliz

atio

n

Time

Figure 4One synchronized pulse followed by a series of unsynchronized pulses yields anaverage trajectory (dotted red line) that fails to represent single-cell activity (palecolored lines). Mock data are shown only for four hypothetical single cells, butthe average trace was calculated from a simulation of 100 single cells. Thisschematic resembles the data of Cai et al. (2008), who quantified activity ofCrz1-GFP by using image-analysis software to measure its nuclear enrichmentin single cells.

Other key studies have exploited the subcel-lular localization of fusion proteins to detecttranscription-factor activation (Hersen et al.2008, Mettetal et al. 2008); these works are re-viewed in the microfluidic section below.

INTRACELL COMPARISONS OFSPATIOTEMPORAL DYNAMICS

Although quantitative single-cell microscopyhas provided extensive evidence for the preva-lence and importance of cell-to-cell variabil-ity, the high spatiotemporal resolution of thesesame assays also provides information aboutthe dynamics of processes within a single cell,irrespective of cell-to-cell variability. In thefirst section below, we review studies thatprobe the dynamics of multiple and often over-lapping events within single cells. Next, we

describe quantitative fluorescent speckle mi-croscopy, a technique that monitors the dynam-ics of macromolecular structures, such as thecytoskeleton, in living cells. Finally, we discussrecent works that have identified and/or char-acterized feedback loops that are illuminated bysensitive dynamic measurements, thereby high-lighting the fact that quantitative, time-lapsesingle-cell analysis can reveal key insights aboutnetwork structure.

Timing of Multiple Events

The development of spectrally unique, genet-ically encoded GFP variants (Ai et al. 2007;Shaner et al. 2004, 2005) has helped elucidatemechanisms that occur in the same subcellularlocation and/or at the same time. For instance,two noteworthy independent investigationsused GFP and red fluorescent protein (RFP)fusion proteins to resolve the long-standingquestion regarding the dynamic composition ofproteins in the cisternae of the Golgi apparatus(Losev et al. 2006, Matsuura-Tokita et al. 2006).Prior to this research, it was unclear whether thecis, medial, and trans Golgi stacks maintainedfixed protein composition over long time peri-ods, or whether a single cisterna progressivelytransitioned from cis to medial to trans by dy-namically adjusting its constituent proteins. Inthese two studies in yeast, cis- and trans-specificproteins were labeled with differently coloredfluorophores in the same cell. An individualcisterna first fluoresces owing to the cis-specific protein for ∼150 seconds (Figure 5).As the cis-specific signal falls, the trans-specificsignal in that same cisterna rises and persistsfor another ∼200 seconds. Thus, rather thanretain exclusively cis- or trans-specific proteinsfor a long time, cisternae rapidly mature,with secretory-processing factors progres-sively shuttling through them. Because thesemeasurements were gathered in living cellsover time, Losev et al. (2006) went furtherto demonstrate that the timescale of proteinsecretion (∼6 minutes) was roughly equivalentto that of the observed Golgi maturationprocess (Losev et al. 2006). These data suggest


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


that much of the trafficking among Golgistacks consists of Golgi-specific processingfactors and not the secreted proteins. Together,these studies elegantly demonstrate the powerof quantitative time-lapse microscopy to dis-tinguish among possible mechanisms, extractimportant information about their kineticproperties, and generate new hypotheses.

Deducing the order of molecular events incomplicated pathways is frequently challeng-ing. Where distinct phenotypes exist, epista-sis experiments can sometimes resolve whetherone factor acts before another. Alternatively,biochemical time-series experiments can beperformed on synchronized samples to iden-tify which events precede others; for instance,chromatin immunoprecipitation was used toidentify the order of transcription-factor re-cruitment to the yeast HO promoter (Cosmaet al. 1999). For events that cannot be resolvedwith epistasis or that occur on timescales fasterthan bulk biochemistry experiments can probe,time-lapse fluorescence microscopy can shedlight on the ordering of events.

For instance, Dultz et al. (2008) monitoredthe dynamic assembly and disassembly of therat nuclear pore complex (NPC), which con-sists of approximately 30 nucleoporin proteins(Nups) and is disassembled during mitosis. Theauthors created multiple cell lines, each bear-ing a GFP-fused Nup and DiHcRed fused to aprotein (here simply called RFP) that leaves thenucleus upon breakdown of the nuclear enve-lope and re-enters via the NPC upon restora-tion of the nuclear envelope at the end ofmitosis. The position of chromosomes was alsomeasured simultaneously via DNA staining.For each GFP-Nup fusion, the time betweenGFP-Nup/DNA colocalization (a proxy for in-corporation into the NPC) and RFP nuclear ac-cumulation (a proxy for NPC competence) wasmeasured. Because the RFP construct and itsnuclear entry and exit dynamics were commonto all cell lines, the relative GFP-Nup/DNAcolocalization times could be determined byreference to the RFP signal. Thus, for 11 Nupsrepresenting eight major subcomplexes of theNPC, nuclear incorporation could be detected

a

b

7:27 9:00 9:45 9:57 10:21 12:00 12:54

7:27 9:00 9:45 9:57 10:21 12:00 12:54

100

75

50

25

0400 500 600

Time (s)

Flu

ores

cenc

e in

tens

ity

(% m

ax)

700 800

Figure 5The composition of processing factors in single cisternae of the Golgiapparatus changes rapidly. (a) The early processing factor Vrg4 was tagged withGFP, and the late processing factor Sec7 was tagged with DsRed; thelocalization of each signal was measured dynamically (Losev et al. 2006).Image-analysis software was used to track individual cisternae. Top: Onecisterna, marked by the white arrow, is followed over time as it transitions fromgreen to red. Times shown at the bottom-left of each image are in mm:ss(minutes:seconds). Bottom: Same as top, except fluorescence channels are onlyshown for the cisterna marked with an arrow in top. (b) Quantification of thesignal in (a), indicating that the early processing factor vacates the cisternaabout when the late processing factor enters. Note that the time between entryof early factors and exit of late factors is ∼7 minutes, close to the time neededfor secreted proteins to undergo processing in the Golgi. Figure redrawn fromLosev et al. 2006 with permission of Nature.

with high precision at subminute timescales us-ing clever quantitative analysis of three sig-nals in parallel cell lines (Figure 6). The studyshowed that the order of disassembly is not sim-ply the reverse of the assembly ordering; fur-thermore, the nuclear pore is partially com-petent for protein transport even when someNups are not yet incorporated. These con-clusions highlight the promise of time-lapsemicroscopy and automated image analysis for


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


–2 0

0.2

0.4

0.6

0.8

1.0

1.2

2 4 6 8–10 –8–14 –12 –6 –4

Anaphaseonset t½ (import)

Time (min)

Nor

mal

ized

fluo

resc

ence

inte

nsit

y

Nup133

Nup153

Nup50

POM121

Nup98

Nup93

Nup58

Nup214

IBB

Figure 6Sequential assembly of components into the nuclear-pore complex (NPC) tracked with high-temporalresolution. The thin black curve represents the nuclear localization of IBB-DiHcRed, a fusion proteinimported following cell division via the NPC; the time 0 min [i.e., t1/2(import)] is set when the black curvereaches half its maximal level. The relative rates of nuclear localization of GFP-tagged nucleoporins (Nups)and IBB-DiHcRed in strains bearing only one GFP-tagged Nup were used to determine the relativeincorporation times of 11 different Nups. Data reproduced from Dultz et al. 2008, Rockefeller UniversityPress.

deconstructing cellular pathways, even thosewith fast dynamics.

An important feature of dynamic cellularmeasurements is that knowledge of an event’stimescale can suggest which factors may regu-late the event. This virtue is nicely illustratedby another paper from the Ellenberg lab thatuses multiple fluorophores to elucidate the or-dering of a pathway (Schuh & Ellenberg 2007).The authors used quantitative time-lapse mi-croscopy to investigate the dynamics of spindleassembly in mouse oocytes. Spindle assemblyin these cells is remarkable because it pro-ceeds without the centrosomes that typicallycoordinate spindle assembly in other cells. Bytracking the subcellular localizations of bothchromosomes and microtubules, Schuh and

Ellenberg found that more than 80 microtubuleorganizing centers (MTOCs) self-assemble toform a bipolar spindle-like structure that subse-quently aligns the chromosomes. In the processof formulating this interesting event-orderingresult, they used automated image-analysis totrack the 3D positions of individual cytoplasmicMTOCs over time. They noted that the calcu-lated MTOC velocities, as well as their ballistic(as opposed to random) trajectory toward thenucleus, is consistent with MTOC transport bymicrotubule motors. Though this hypothesisof motor-protein-assisted transport may not beentirely surprising, it underscores how knowingthe quantitative dynamics of a particular pro-cess via single-cell microscopic measurementscan distinguish between alternate mechanisms.


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


Quantitative Fluorescence SpeckleMicroscopy

Quantitative fluorescence speckle microscopy(qFSM) combines live-cell imaging and sophis-ticated analysis algorithms to probe the dy-namics of macromolecular structures (reviewedin Danuser & Waterman-Storer 2006). qFSMrelies on fluorescently tagging only a smallpercentage (typically 0.5 to 2%) of a struc-ture’s subunits. For instance, microtubules con-tain roughly 1625 tubulin dimers per 1 μm(Danuser & Waterman-Storer 2006); in aqFSM experiment, GFP-tagged tubulin wouldbe carefully expressed at a level such that only∼8–32 molecules are fluorescent per 1 μm.This low-level expression of fusion proteinsleads not to uniform fluorescence across the mi-crotubule, but rather to the presence of distinctand randomly distributed fluorescent foci (i.e.,speckles). Each speckle contains roughly threeto eight fluorophores (Danuser & Waterman-Storer 2003) that were incorporated in closeproximity during microtubule assembly. Bymonitoring the movement of speckles, as wellas their appearance and disappearance, re-searchers can evaluate the motion of structuresin the cytoskeleton and the kinetics of their as-sembly and disassembly. However, such evalu-ation can require the tracking of tens of thou-sands of speckles over time, underscoring theneed for automated imaging to extract quanti-tative and mechanistic features from the data.

Several groups have used qFSM to investi-gate actin dynamics in migrating cells. Cell mi-gration involves a series of steps: cell membraneprotrusion, adhesion of the protruded mem-brane to the extracellular matrix, cytoskeleton-mediated pulling of the cell body against theadhesion molecules, and the severing of ad-hesion molecules from the rear of the cell.Monitoring nearly a thousand speckles of fluo-rescently tagged actin monomers in membraneprotrusions (lamellipodia) revealed that actinpolymerization is enhanced within 1 μm of thelamellipodium edge (Watanabe & Mitchison2002). A subsequent study tracked more than10,000 actin speckles over time, determining

their trajectories and lifetimes (Ponti et al.2004). Interestingly, this analysis identified par-tially overlapping regions of the lamellipodiumin which quantitatively different actin dynam-ics occurred. Specifically, short-lived speckleswith high velocity were enriched only near theedge; however, long-lived speckles with low ve-locity were found throughout the entire lamel-lipodium and adjacent lamella. Exploiting thesequantitative phenotypes and a range of small-molecule inhibitors, the authors showed thatspecific actin-associated molecules regulate thedifferent speckle behaviors. This research isparticularly noteworthy because of the tem-poral basis for distinguishing phenotypes: Allspeckles were the same color, they occupied thesame subcellular localization, and their inten-sities were roughly identical. Thus, their life-times and velocities—two quantitative featuresextracted via automated image analysis—werethe distinguishing features.

Evidence for Feedback Loopsin Dynamic Measurements

Analysis of large-scale gene-expression datasetsand protein-protein interaction networks in-dicates that positive and negative feed-back loops are highly overrepresented motifs(Yeger-Lotem et al. 2004). They serve a rangeof functions in controlling the dynamics ofcellular processes (Brandman & Meyer 2008).For instance, negative feedback can give rise tooscillations and stability of a particular molec-ular state, whereas positive feedback frequentlyleads to all-or-none, switch-like behavior. Theconverse is also noteworthy: Specifically, ob-servation of oscillations often implies negativefeedback, and switch-like behavior frequentlysuggests the existence of positive feedback.Because bulk measurements can disguiseswitch-like responses (Ferrell & Machleder1998), time-lapse single-cell measurements areuniquely poised to reveal or verify the existenceof important positive-feedback mechanisms, asa recent report from Holt and colleagues beau-tifully illustrates (Holt et al. 2008). In light ofthe fact that segregation among many discrete


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


chromosomes during anaphase is both rapidand nearly simultaneous (i.e., switch-like), theauthors hypothesized that the anaphase regu-latory mechanism may contain a critical posi-tive feedback loop. They performed biochemi-cal analysis of the yeast protein securin, whichinhibits anaphase by repressing separase, itselfthe factor responsible for cleaving the cohesinmolecules holding sister chromatids togetherbefore separation at anaphase. They found thatthe phosphatase Cdc14 promotes degradationof securin. Because separase was earlier shownto activate Cdc14, the positive feedback loopconsisting of one positive and two negative con-nections became clear: Securin inhibits sepa-rase, which activates Cdc14, which effectivelyinhibits securin.

To explore the role of this putative feed-back loop in living cells, the authors labeled twoindividual chromosomes using GFP-fused Tetand Lac repressors, which bound to arrays ofgenomically incorporated operator sites onchromosomes IV and V. Chromosomes werevisualized as spots in the cell, and spot track-ing via automated image analysis software couldidentify the time at which sister chromatidssegregated (i.e., one spot became two distinctspots). In wild-type cells, 90 seconds on aver-age elapsed between the chromatid segregationof the respective chromosomes. In a cell linewhere positive feedback was impaired, however,

the difference in segregation times nearly dou-bled to 170 seconds on average. In sum, thesingle-cell analysis performed by Holt et al.(2008) demonstrates the existence of a positivefeedback loop that has a dramatic effect on thesynchrony of events in anaphase.

Similar analysis of spatiotemporal data per-formed by Ozbudak et al. (2005) supports theexistence of a negative feedback loop that reg-ulates bud formation in yeast. The authorstracked the membrane-specific localization ofa fusion protein that binds activated Cdc42,which in yeast marks the site of bud for-mation by becoming localized at a particularpatch of the cell periphery. This aggregationof the Cdc42 signal (the Cdc24 polar cap)had been previously shown to result from apositive feedback loop (Irazoqui et al. 2003,Wedlich-Soldner et al. 2003). A positive feed-back loop alone, however, would predict thatas soon as a Cdc42 polar cap nucleates, its in-tensity could increase as a function of time butits location should not change. Remarkably, incells where the mechanisms that typically pro-gram the yeast budding pattern were disabled,researchers found that the Cdc42 polar capfreely traveled around the periphery of the cell(Figure 7), and these spatial deviations contin-ued up to ∼15 minutes before bud emergence.A concise quantitative model was developed toanalyze the cap dynamics, and it was shown

Wild-type

rsr1Δ

0 3 6 8 11 18 24 28

0 1 2 6 8 11 13 15

Figure 7In the absence of bud-site landmarks, the polar cap of active Cdc42 wanders around the cell periphery. The CRIB domain of Gic2,which binds specifically to active Cdc42, was tagged with GFP and monitored in wild-type cells and in mutant cells lacking Rsr1 (a.k.a.Bud1). Rsr1 marks the site of budding, and active Cdc42 recruits factors that mediate the budding process. In rsr1� cells, budformation still occurs but at a random location. These time-lapse images (where the number at the upper left is the number of minutes)indicate that the polar cap of active Cdc42 rapidly traverses the cell periphery instead of randomly picking one location and remainingfixed there. Figure adapted from Ozbudak et al. 2005 with permission of Developmental Cell.


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


that positive feedback alone was insufficient toexplain the system behavior, but the additionof a negative feedback loop could capture theobserved cap migration. Thus, a phenomenononly detectable via time-lapse imaging at thesingle-cell level revealed the likely existence ofa novel network feature whose molecular basiscan be subsequently investigated.

CONSTRUCTION OF 4DDEVELOPMENTAL LINEAGESUSING QUANTITATIVEMICROSCOPY

The decade-long and Nobel Prize–winningeffort of John Sulston and his colleaguesdetermined the entire embryonic lineage ofCaenorhabditis elegans (Sulston et al. 1983).Since this tremendous achievement, C. eleganshas become a key model organism for the studyof development. In recent years, it has becomepossible to determine the lineage of a range oforganisms with considerably less effort by ex-ploiting the power of fluorescence time-lapsemicroscopy coupled with automated image-analysis techniques. Below we discuss studiesthat span a wide range of organismal complex-ity, each of which offers exciting insights intodevelopmental processes.

Yeast

Though yeasts are unicellular organismsand thus not generally considered a modelorganism for development, Kaufmann, Yang,and colleagues showed that tracing the pedi-gree of a growing colony can reveal importantinsight into the molecular memory passed downthrough generations (Kaufmann et al. 2007).Budding yeast cells expressing a fluorescent re-porter of Gal-network activation were trackedvia microscopy for up to 12 hours, leading to ex-pansion of a colony from one cell to ∼20 cells.The authors used a strain in which the switch-like activity of the Gal network was mutatedsuch that activation (switching on) occurredstochastically with a probability of ∼10% pergeneration (Acar et al. 2005); this allowed for

00

40

80

120

0

40

80

120

150 300 450

Time (min)

Flu

ores

cenc

e (a

rb. u

nits

)

600 750 900

Cell 1born

Mother(1)

Daughter(1-2)

τ1

τ1|1-21|1-2

τ1-21-2

τ1

τ1|1-2

τ1-2

Cell 1-2born

t = 30 min t = 570 min t = 600 min t = 690 min

1 1

1-1-1-11-1-1

1-2

1

1-1-1-11-1-1

1-2

11-1

1-1-1-11-1-1

1-2

a

b

Figure 8Correlated switching times among closely related single cells revealed viasimultaneous monitoring of lineage and gene expression. (a) YFP expressedunder control of the GAL1 promoter (GAL1-YFP) is shown in purple.Genealogy is indicated by the numbering scheme in which hyphens separategenerations (e.g., 1–1–1 is the daughter of 1–1 and the granddaughter of 1), andthe number indicates siblings (e.g., 1–2 is the sister of 1–1 and the seconddaughter of 1). Although all cells in the rightmost panel are close relatives ofcell 1, only a subset expresses GAL1-YFP. (b) Quantification of GAL1-YFPexpression in a mother and daughter shows that both switch to an expressingstate nearly simultaneously. It was shown that related cells separated by up tofour generations tended to activate GAL1-YFP expression at a similar time.Figure adapted from Kaufmann et al. 2007.

switching events to occur within the 12-hourobservation period (Figure 8a). By simulta-neously tracking reporter gene expression anda cell’s ancestry, it was shown that mothersand daughters maintain a level of synchrony intheir switching times over multiple generations(Figure 8b). As expected, this memory de-cayed as a function of time, but remarkably, thecorrelation persisted for four doublings. Thislong-term memory was consistent with a model


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


in which Gal-regulatory factors are expressedin large but infrequent bursts. The frequencyand amplitude of gene-expression bursts arewell-characterized parameters of gene expres-sion noise (Maheshri & O’Shea 2007, Raj &van Oudenaarden 2008). It will be interestingto see the extent to which noise regulation in thedevelopment of multicellular organisms affectssynchronized gene expression programs acrossthe cells in a pedigree.

Worms

In their automated-lineage analyses of devel-oping C. elegans embryos (Bao et al. 2006, Zhaoet al. 2008), the Waterston lab provides an il-lustrative example of the comparisons that canbe made using rich phenotypes. Their generalapproach was to label histone proteins with flu-orescent proteins (Murray et al. 2006). Becausehistones are ubiquitously associated with DNA,the nucleus of every cell in the organism up tothe ∼350-cell stage is resolvable via microscopy.Furthermore, because nuclei are approximatelyspherical and the fluorescent-histone signal fillsthe nucleus, resolving individual cells via image-analysis software is relatively straightforward ascompared to the difficulty involved in segmen-tation using cell boundaries rather than nucleifor cells with a range of shapes. Owing to therelatively large size of the embryo, a z-stack of2D images must be captured in rapid succes-sion in order for sophisticated image-analysisroutines (Murray et al. 2006) to assemble a 3Ddepiction of nuclei positions. The results of thismethod are quite striking: An entire lineage canbe constructed in just six hours of image acquisi-tion and two to four hours of image analysis. Ofcourse, the lineage has amazingly high tempo-ral precision with full-embryo sampling everyminute. However, it also contains tremendousspatial information: The dynamic position ofevery cell is known throughout development.

Like the work in yeast performed byKauffman et al., Waterston and his colleaguesmapped gene expression data onto the lin-eage by driving histone-RFP expression with

developmentally relevant promoters (a histone-GFP fusion under control of the endogenoushistone promoter was still present to trace theentire lineage) (Murray et al. 2008). Amongother findings, they showed that neural devel-opment regulator cnd-1 was expressed earlierand in a much broader range of cells than pre-viously characterized. A separate study fromthe Waterston group highlighted the possibil-ity of quantitatively comparing lineage charac-teristics across species (Zhao et al. 2008). Theauthors constructed lineages for embryos ofC. elegans and the related worm Caenorhabdi-tis briggsae. Remarkably, despite extensive ge-nomic differences between the species, theirdevelopment appeared to be quite similar(Figure 9). There was a one-to-one correspon-dence in the lineages between the species up toat least the 350-cell stage. The timing of di-visions was also highly conserved among mostbranches of the lineage, but the assay was pre-cise enough to identify a particular subset ofcells with a different cell-division rate. Just asinterspecies comparisons of genome sequencehave enabled extensive evolutionary analyses,large-scale comparative lineage analyses havethe potential to broadly impact the study of de-velopment and evolution.

Plants

Automated lineage analysis has also beenperformed in developing Arabidopsis thalianameristems (Reddy et al. 2004, 2007). In a recentstudy, a technique similar to the one used forlineage analysis was applied to explore the re-lationship between microtubule alignment andmorphological stress across cell walls in devel-oping plant meristems (Hamant et al. 2008). InA. thaliana, cortical microtubules form a stri-ated bundle in each cell with a clear orienta-tion (Figure 10a). Others had proposed thatanisotropic stress across the cell wall exerted byneighboring cells played a role in microtubuleorientation, so the authors investigated thispossible mechanism via time-lapse microscopyin a developing shoot apical meristem. They


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


a

b

Figure 9Lineage comparison reveals developmentalsimilarities between (a) C. elegans and (b) C. briggsaeembryos. Spots represent nuclei, which werevisualized via GFP-labeled histones and trackedcomputationally in 3D over time. Color-codingindicates descendants of a common precursor cell.Figure reproduced from Zhao et al. 2008 withpermission of Developmental Biology.

fused GFP to a microtubule-associated proteinand tracked the microtubule orientation overtime in the developing plant via analysis soft-ware. They found that microtubule-orientationfluctuations changed over time. Specifically, atthe tip of the meristem, microtubule orienta-tions shifted frequently in the first ∼12 hoursof a cell cycle but stabilized (i.e., stopped ro-tating) in the ∼4 hours preceding cell division.The stabilized orientation was consistent withthe predictions of a model the authors proposedto deduce cell-wall stress from the 3D shape ofthe developing meristem.

To confirm this suggested link between ori-entation and stress, the authors performed two

a b

20 μm

Figure 10Orientation of microtubule bundles in A. thaliana corresponds highly withpredictions from a model of cell-wall stress. (a) The microtubule-bindingdomain of a microtubule-associated protein was tagged with GFP. GFP signalabove a high threshold signified cell boundaries (pseudocolored red), and signalbelow the threshold indicated cell-traversing microtubules (pseudocoloredgreen). The letter P (red ) indicates the position of a developing primordium, aposition characterized by rapid cell growth and a lack of alignment inmicrotubule orientation. To the upper left of the primordium is a region wheremicrotubules are tightly aligned in the southwest-to-northeast direction.(b) A model that predicts stress across the cell wall can also accurately predictmicrotubule orientations (red ). Inputs to the model include the 3D tissue shapeand cell boundaries from (a). Note the consistency between the model’spredictions and actual microtubule orientations from (a) in the region to theupper left of P. Figure reproduced from Hamant et al. 2008 with permission ofScience.

key experiments, one using laser ablation andthe other involving application of direct com-pression force. In the laser-ablation experiment,an individual cell was rapidly killed such thatcell-wall stress on the adjacent cells was di-minished; within six hours, the authors ob-served reorientation of microtubules in nearbycells, indicative of adjustment to a new cell-wallstress landscape. In the direct-compression ex-periment, the plant was pinched between twoteflon blades, dramatically altering the direc-tion of stress across cell walls. Shortly after thestress began, microtubule orientation destabi-lized and within six hours, most cells in prox-imity to the blades had reoriented their micro-tubules such that they were in alignment withthe imposed cell-wall stress. Together, theseexperiments demonstrate how quantitative lin-eage analysis, coupled with the ability to makeperturbations and sensitively measure theireffects, can significantly advance our under-standing of development.


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


Flies

Tracking the dynamics of development in liv-ing organisms via fluorescently labeled histoneproteins was recently performed in Drosophilaembryos. Stathopoulos and colleagues focusedon the creation of the mesoderm duringgastrulation (McMahon et al. 2008). The meso-derm is formed from a tube-shaped invagina-tion of the ectoderm that subsequently flattensand spreads along the ectoderm to formtwo distinct but adjacent cell layers. Thoughthis sequence of events is evident from anti-body staining of fixed embryos, dynamic datain a living embryo provided much highertemporal and spatial resolution, thereby eluci-dating novel features of the development path-way. Individual nuclei were tracked for morethan two hours and sampled every ∼50 sec-onds; approximately 100 mesoderm and 1500ectoderm nuclei were followed in each embryo.Interestingly, because all cells bore the samefluorescent histone marker, mesoderm and ec-toderm cells could only be distinguished laterby tracing their trajectories back to their initialpositions in the embryo (i.e., before the flat-tening of the mesoderm); this underscores theneed for sophisticated image analysis and a longsampling duration. The authors quantitativelyverified a previous qualitative observation sug-gesting that movement of ectoderm and meso-derm cells was coupled. To reach this conclu-sion, they examined the correlation between thevelocity of a mesoderm cell and its six nearestectoderm neighbors. Remarkably, of the threevelocity components (i.e., anterior to posterior,left to right, and center to ectoderm), only theanterior-to-posterior velocity displayed a cou-pling between mesoderm and ectoderm move-ment. This highly specific finding suggests acommon mechanism for this directed motionbut also the existence of independent mecha-nisms regulating the other motion components.The single-cell traces along the left-to-rightaxis suggest that some controlling mechanismis at work—as opposed to simple diffusion—because the relative left-to-right positions ofcells were largely maintained through two

cell divisions and the formation of a flattenedmesoderm.

Finally, in another beautiful quantitativeanalysis, the authors showed via lineage tracingthat anomalous mesoderm development in flieswith defective fibroblast-growth-factor signal-ing is due to a very specific cell-migration prob-lem. Cells initially in the lower furrow (i.e.,the ectoderm-proximal lower half of the tube-shaped, preflattened mesoderm structure), andthus close to the ectoderm, underwent normalmigration; however, those in the upper furrowfar from the ectoderm had trajectories dissimi-lar to wild-type cells, leading to a very differentflattened mesoderm pattern. In all, the workof McMahon et al. very nicely illustrates howthe rich phenotypes measured via quantitativelineage analysis can refine our understandingof developmental mechanisms and suggest thepresence of novel forms of regulation.

Fish

An exciting recent paper about zebrafish devel-opment (Keller et al. 2008) sheds light on howfar lineage analysis can take us in unraveling de-velopmental mechanisms of higher organisms.Keller et al. developed a novel optical techniquethat permitted high-resolution 3D scanning ofthe entire zebrafish embryo every 90 seconds.Again exploiting the fluorescent labeling ofhistones to find nuclei, the authors tracked ev-ery nucleus position from shortly after fertiliza-tion until the embryo contained ∼16,000 cells(Figure 11). The enormity of data generatedand analyzed cannot be understated, as fourhigh-resolution images were captured everysecond for 24 hours. After image analysis andcomputational 3D reconstruction of the em-bryo, a breathtaking movie of the develop-ing embryo was constructed (SupplementalMovie 1; follow the Supplemental Mate-rial link from the Annual Reviews home pageat http://www.annualreviews.org). Aestheticsaside, it also contains an inordinate amount ofinformation that can be analyzed for years tocome. The authors performed a few analyses intheir initial report such as identifying the time


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


of symmetry breaking and exploring the specificevents in a mutant zebrafish that lead to defec-tive mesendoderm formation. In the future, itwill be fascinating to use these data and tech-niques to assess the robustness of development,discover the effects of a range of mutations,and perform interspecies comparisons amonghigher vertebrates.

MICROFLUIDIC DEVICES

The ability to perturb a cell’s surroundingsdynamically can be critical for unravelingregulatory mechanisms. Additionally, whereasmonitoring cells on short timescales is vitallyimportant, so too is the capability to measurecellular properties over very long timescales,potentially days or weeks, using microscopy.Described below are advances in both extracel-lular manipulation and long-term microscopymade possible by utilizing microfluidic devices.

Specialized Stimuli

Several recent studies have powerfully exploitedmicrofluidic devices to measure living cells’dynamic responses to an exquisitely controlledextracellular environment. Though two workswe mention utilize the device to alter the sur-roundings temporally, we also review a studythat uses a device to control the spatial distri-bution of a signaling input.

−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−→Figure 11Development of the zebrafish embryo imaged usingGFP-labeled histones. (a) Positions and velocities ofnuclei represented at three time points. At each timepoint, a 3D image is generated by acquiring a stackof 2D images (x and y dimensions) across a range ofz positions. In the black-and-white images at left,the maximum intensity across the whole z-stack forevery pixel in x,y space is plotted. Shading in thecolored images at right indicates the velocities(cyan = slow, orange = fast) of single nuclei asdetermined by their relative position in adjacentframes. (b) Individual nuclei from a 280-minutesample in (a) can be imaged at very high resolution.Figure adapted from Keller et al. 2008 withpermission from Science.

a

b

100 min(64-cell stage)

280 min(early blastula)

b

550 min

Blastomere nuclei

100 μm

10 μm


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


Using microfluidics for temporal control ofa stimulatory signal, Mettetal et al. (2008) andHersen et al. (2008) both investigated dynamicsof the Hog1 MAP kinase cascade that controlsthe hyperosmotic shock response in budding

Time

1.0

0.5

0.0Osm

osh

ock

mag

nitu

de

Output

Time

Moc

k H

og1

nucl

eare

nric

hm

ent

Osm

osh

ock

mag

nitu

de

Osm

osh

ock

mag

nitu

de

Moc

k H

og1

nucl

ear e

nric

hm

ent

Input

Time

1.0

0.5

0.0Input

Effectiveinput

Time

1.0

0.5

0.0

Output

Time

a Below bandwidth frequency

b Above bandwidth frequency

Figure 12Illustration of bandwidth as a measure of pathway responsiveness. (a) In bothHersen et al. (2008) and Mettetal et al. (2008), a microfluidic device deliveredpulses of media with differing osmolyte concentrations (input panel) to yeastcells, causing the activation and nuclear enrichment of the fluorescently taggedMAP kinase Hog1 (output panel). Below the bandwidth frequency, the outputtracks the input signal. (b) For input frequencies above the bandwidth, however,the cells cannot reliably process the input, leading to an effective input shownschematically in blue, which is approximately the integral of the input. Theoutput very poorly represents the fluctuations in the input signal.

yeast. In its inactive state, Hog1 is primarilycytoplasmic, but it translocates to the nucleusrapidly upon hyperosmotic shock. In both stud-ies, the authors fused a fluorescent protein toHog1 and used custom image-analysis softwareto measure the nuclear accumulation of Hog1as a proxy for its activation state in living cells.Additionally, in both studies yeast cells were af-fixed to a coverslip and placed in a microfluidicdevice that could rapidly switch between mediaof different salt concentrations. Hersen et al.focused their analysis on the system’s band-width, which corresponds to the upper limitof input frequencies (i.e., oscillating pulses ofsalt) below which the cells can accurately trackthe input. For example, for low-frequency sig-nals below the bandwidth, Hog1 activity os-cillates with a similar frequency as the input(Figure 12a). However, for high-frequency sig-nals above the bandwidth, the Hog1 osmotic-stress response is not fast enough to reflectthe input fluctuations and instead responds tothe average of the quickly fluctuating signal(Figure 12b). Such analysis is important be-cause it reflects the responsiveness constraintsof the system and contains important mechanis-tic information as well. For instance, all steps inthe response must be faster than the bandwidth.To identify the rate-limiting step, the authorsadjusted the duration of high-salt and low-saltperiods. They found that the system’s slowestreaction, which is primarily responsible for set-ting the bandwidth frequency, regulates deacti-vation of the Hog1 response. It is important tonote that the approach of Hersen et al. is verygeneral: It can be applied to any system wherethe input is well known and the output can bemeasured at a rate faster than the frequency offluctuations in the input signal.

Recent work by Mettetal et al. (2008) usedengineering principles to gain insight into thenetwork topology of the Hog1 system. Theyexposed cells to a range of different frequen-cies of salt stress and measured the amplitudeand phase shift of the corresponding oscilla-tions in Hog1 activity. Similar to the band-width analysis of Hersen et al. (2008), thefrequencies at which dramatic changes in the


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


output response occur correspond to impor-tant reaction rates that dominate the entire sys-tem’s dynamics. The authors found two suchdramatic changes, implying that the networkdynamics could be modeled with only two dif-ferential equations (as opposed to tens of differ-ential equations, one for every reaction in thenetwork). The equations deduced from theirfrequency analysis suggested a network struc-ture highly similar to the known architecture ofthis well-characterized system; thus, frequencyanalysis could prove very enlightening for sys-tems with unknown architecture but that oth-erwise have well-known and measurable in-puts and outputs. Together, the experiments ofHersen et al. and Mettetal et al. demonstratethat microfluidic devices coupled with quanti-tative time-lapse microscopy can yield impor-tant insights into the dynamics and topology ofcellular networks.

Aside from temporal-signal manipulation,microfluidic devices can also create a veryprecise spatial gradient. Hao et al. (2008)used a microfluidic device to create a lin-ear concentration gradient of yeast matingpheromone. Yeast growing in the device ex-hibited three distinct states of growth: Veg-etative growth occurred in the region withlow pheromone, chemotropic growth at inter-mediate pheromone, and shmoo formation athigh pheromone. In wild-type cells undergo-ing chemotropic growth, cell division occursin the direction of higher pheromone. Theauthors measured the angle of budding rela-tive to the gradient, which was made possiblebecause the gradient orientation was very pre-cisely controlled and cells could be imaged asthey grew. Using this quantitative phenotype,they found that only one of the two pheromone-responsive MAP kinases could yield directedgrowth toward high pheromone. The work ofHao et al. is particularly interesting becausethey observed several distinct cellular phe-notypes (e.g., vegetative growth, chemotropicgrowth, and shmoo formation) as a function ofconcentration in just one experiment. This ap-proach could be extended to perform a rangeof titration experiments using a microscope:

Researchers could scan the device for the phe-notype of interest and then determine the pre-cise extracellular conditions causing the pheno-type based on the position within the device.

Highly Parallel, Long-Term Imaging

Microfluidics are also powerful because theyprevent stress and permit long-term imaging(Melin & Quake 2007). Quake and colleaguesdemonstrated that Escherichia coli could be cul-tured for up to 200 hours in a device that per-mitted simultaneous imaging by microscopy(Balagadde et al. 2005). Their device is partic-ularly noteworthy because E. coli typically formbiofilms when cultured to high density, therebycomplicating more traditional long-term mi-croscopy methods. The authors engineered amicrofluidic device where cells grow in a loopedchamber with 16 segments. At specified times,one segment can be cordoned off, washed withlysis buffer to remove biofilm-nucleating cells,rinsed of lysis buffer, and then reincorporatedinto the loop of flowing media. In a subse-quent study, the authors showed that 96 inde-pendently controllable chambers on one micro-scope slide could be used to culture mammaliancells for more than a week while imaging si-multaneously (Gomez-Sjoberg et al. 2007). Weexpect the use of microfluidics to grow rapidlyas an experimental platform that allows highlyparallel, long-term imaging of cells in an envi-ronment that can be very tightly controlled.

THE ROAD AHEAD

In this review, we have discussed pioneer-ing studies that specifically couple quantita-tive, time-lapse fluorescence microscopy in liv-ing single cells with automated image-analysistechniques to explore the temporal and spatialproperties of biological systems. In the nearfuture, we anticipate tremendous progress inthis field on multiple fronts: enhanced abil-ity to perturb the cells’ interior and exterior, awider range of fluorophores and imaging tech-nologies, higher spatial resolution, greater par-allelization, and more refined image-analysis


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


software. Below we highlight examples of suchprogress.

The ability to perturb a system is criti-cal for distinguishing correlation from causa-tion. As described earlier, microfluidic devicesand flow chambers permit precise control ofthe extracellular environment, but other newtechnologies are needed to affect intracellularprocesses on short timescales. One such ex-ample is chromophore-assisted laser inactiva-tion (CALI) (Bulina et al. 2006a,b; Liao et al.1994; Tanabe et al. 2005; Wang et al. 2008).The basic principle of CALI is that excitationof specific fluorophores (e.g., KillerRed, a vari-ant RFP molecule) fused to a protein of interestcreates free radicals that inactivate the proteinof interest within minutes. Though CALI canobviously be used to assess the rapid knockdownof a protein, it could also inactivate a repres-sor, thereby leading to sharp accumulation of aprotein of interest. Because the microscope it-self perturbs the system, data can be gathered atall time points before and after the stress; thus,even the short-timescale function of a proteincan be detected.

The refinement and development of newfluorophore technologies have been rampantin recent years (Wang et al. 2008), and we ex-pect progress to persist. Quantum dots (Dahanet al. 2003, Giepmans et al. 2006) and theresonance-energy transfer technologies (e.g.,FRET, BRET, BiFC)—largely excluded fromthis review because they have been nicelyreviewed elsewhere ( Joo et al. 2008)—areincreasingly being used in live-cell assays,especially those classifying protein-protein in-teractions in vivo (Kerppola 2008). The devel-opment of spectrally distinct, genetically en-coded fluorophores (Shaner et al. 2005) has alsosignificantly advanced the field, since more fac-tors can be monitored in the same cell concur-rently. Particularly interesting from the stand-point of tracking dynamic events in live cells arephotoswitchable fluorophores (Patterson 2008)and so-called fluorescence timers (Subach et al.2009). The former can be converted among twocolors (or light and dark states) via UV light ex-posure either reversibly (Ando et al. 2004) or

irreversibly (Chudakov et al. 2004, Patterson &Lippincott-Schwartz 2002); the latter convertfrom one color to another color, without in-tervention, over a range of timescales. Thesetechnologies will allow researchers to effec-tively timestamp particular molecules in eithera switch-like or graded manner.

Though imaging with light microscopy re-veals a wealth of dynamic information withinsingle cells, its spatial resolution has tradi-tionally been capped by the diffraction limit,which is roughly half the fluorophore’s excita-tion wavelength (typically 200 nm at best). Sev-eral revolutionary technologies have shatteredthe diffraction limit, yet they impressively stilluse far-field light microscopy (Hell 2007, 2009).Stimulated emission depletion (STED) (Heinet al. 2008, Klar et al. 2000, Nagerl et al. 2008),stochastic optical reconstruction microscopy(STORM) (Bates et al. 2007, Huang et al. 2008,Rust et al. 2006), and the related photoacti-vated localization microscopy (PALM) (Betziget al. 2006, Shroff et al. 2008) effectively allowimaging at an arbitrary resolution, and already-published reports have resolved features sepa-rated by <20 nm (Betzig et al. 2006, Eggelinget al. 2008, Rust et al. 2006). The compar-ison between STED- and STORM-acquiredimages and those from confocal microscopyare striking (Figure 13). In addition to highresolution in the x,y plane, subdiffraction-limitimaging in the z direction permits 3D imag-ing using both STED (Nagerl et al. 2008) andSTORM (Huang et al. 2008). Multiple colorscan be detected with both techniques, but thusfar only STED can use genetically encoded flu-orophores (Willig et al. 2006); STORM andPALM both require special fluorophores thatcan be cycled between light and dark statesrepeatedly. Also, though STORM and PALMcan capture only a few high-resolution imagesper minute at best, the temporal sampling ofSTED is remarkable (e.g., 28 frames per sec-ond in Westphal et al. 2008). The major draw-back of all of these technologies is that they re-quire very high-power illumination sources thatare potentially damaging to the cells over longimaging periods. Critically, however, they can


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


a Confocal STED

b Confocal

STORM

500 μm

1 μm

500 μm3 μm

500 μm500 μm3 μm

Figure 13Fluorescence imaging below the diffraction limit highlights features occluded even using confocalmicroscopy. (a) A yellow fluorescent protein fused to a sequence that targeted it to the endoplasmicreticulum was imaged using confocal microscopy (left) and STED (right). Arrowheads indicate positionswhere STED detects a ring in the ER structure not visible by confocal imaging. Scale bar = 1 μm.(b) Comparison of immunofluorescence staining of microtubules using confocal microscopy (top) andSTORM (bottom). The middle and rightmost panels are zoomed portrayals of the colored boxes in the leftpanels. The pixelation of these zoomed panels is apparent in the confocal images but not STORM, becauseSTORM imaging can identify fluorophore positioning with considerably higher spatial resolution thanconfocal imaging. (a) Adapted from Hein et al. 2008 with permission from the National Academy ofSciences, U.S.A. and (b) modified from Bates et al. 2007 with permission of Science.


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


all image events in live cells, setting them apartfrom other impressive subdiffraction-limit ap-proaches, such as structured-illumination mi-croscopy (Schermelleh et al. 2008), that cur-rently require cell fixation.

We expect the high temporal and spatial res-olution of these techniques to lead to signif-icant advances in quantitative systems model-ing. For instance, even with standard far-fieldimaging, parameters measured from live-cell,microscopy-based assays have been instrumen-tal in quantitative modeling (Marco et al. 2007).Higher temporal and spatial resolution will onlyenable better parameter estimation, therebyfacilitating dynamic modeling efforts. STED,STORM, PALM, and their variants should al-low very sensitive measurement of a range ofreaction rates in vivo, effectively permitting invivo biochemistry.

Intercell differences are easily detectedwith current microscopic methods, but high-throughput comparison of multiple samples(e.g., different mutants, species, drug treat-ments) is less straightforward. As discussed ear-lier, microfluidic devices are one tool that canmake automated microscopy more parallel, andtheir use will inevitably become more commonin the future. Tissue arrays are another platformthat shows promise (Sauter et al. 2003). Imagingadherent cells within 384-well plates have beenused to measure phenotypic profiles of a rangeof drugs (Perlman et al. 2004), and microar-rays with thousands of tissue samples spottedon a microscope slide allow for massively paral-lel microscopy-based phenotyping (Sauter et al.2003). Improvements will be needed, however,

before tissue microarrays can be used for live-cell imaging.

Finally, we expect the maturation of image-analysis software to continue. Many softwarepackages are already available that range in bothprice and complexity. Matlab and the freewareImageJ are two popular software suites for im-age analysis, but many other commercial pack-ages are available as well. Because no programcould possibly perform every quantitative anal-ysis one would want, it is important for image-analysis software to have scripting capabilitiesso the users can define custom metrics. We areeager for speed increases in automated imageprocessing as well as better integration betweenanalysis and acquisition software. Progress onthese two fronts could potentially allow on-the-fly analysis of images that then informs theacquisition process. For instance, because flu-orophore bleaching after many exposures is acommon problem with time-lapse microscopy,image-processing software could partially alle-viate this problem by dynamically adjusting theimage-acquisition rate based on how quicklythe parameter of interest is changing. A sim-ilar process is used to minimize processingtime when numerically integrating differentialequations. Image-processing software couldsimilarly be programmed to intervene (e.g.,switching the extracellular media, deactivatingproteins via CALI, switching the color of pho-toswitchable dyes, etc.) only when a particularevent is detected. Thus, in silico feedback onthe in vivo system will allow the profiling of alarge variety of phenotypes in response to com-plex system inputs.

SUMMARY POINTS

1. Automated image-processing software is an invaluable and flexible tool for extractingquantitative phenotypes from microscopy datasets.

2. Gene expression noise is just one example of widespread variability among geneti-cally identical cells in a population. Any quantitative trait—such as event timing ortranscription-factor enrichment in the nucleus—can have noise properties that reflectunderlying regulatory mechanisms and potentially refute the results of bulk assays.


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


3. Measuring dynamic processes at high temporal and spatial resolution in a single livingcell can resolve the ordering of events in complicated pathways and identify importantnetwork properties such as feedback loops.

4. Quantitative microscopy can track spatiotemporal and ancestral trajectories at the single-cell level in a broad range of developing organisms.

5. Microfluidic devices allow concurrent imaging of cells and manipulation of the extra-cellular environment. Microscopy experiments using microfluidics can be highly paralleland long-lasting (e.g., more than one week).

FUTURE ISSUES

1. Fluorophore development will continue its rapid growth, providing molecules that rangein color, brightness, maturation time, photostability, switching ability, and other qualitieswe cannot even anticipate.

2. Subdiffracion-limit imaging techniques will become more accessible.

3. Repositories for raw microscopy data will be established, providing a resource for cellbiologists similar to the protein data bank for structural biologists.

4. Cell and developmental biologists, physicists, engineers, chemists, computer scientists,statisticians, and applied mathematicians will increasingly work together to extend thefrontiers of quantitative imaging in biology.

DISCLOSURE STATEMENT

The authors are not aware of any biases that might be perceived as affecting the objectivity of thisreview.

ACKNOWLEDGMENTS

We apologize to all researchers whose relevant studies could not be included here owing to spaceconstraints. We are grateful to Suzanne Komili, Arjun Raj, Jeff Gore, Hyun Youk, and John Tsangfor their helpful comments on the manuscript. This work was supported by an NSF GraduateResearch Fellowship to D.M. and NIH grant R01-GM068957.

LITERATURE CITED

Acar M, Becskei A, van Oudenaarden A. 2005. Enhancement of cellular memory by reducing stochastictransitions. Nature 435:228–32

Ai HW, Shaner NC, Cheng Z, Tsien RY, Campbell RE. 2007. Exploration of new chromophore structuresleads to the identification of improved blue fluorescent proteins. Biochemistry 46:5904–10

Ando R, Mizuno H, Miyawaki A. 2004. Regulated fast nucleocytoplasmic shuttling observed by reversibleprotein highlighting. Science 306:1370–73

Balagadde FK, You L, Hansen CL, Arnold FH, Quake SR. 2005. Long-term monitoring of bacteria undergoingprogrammed population control in a microchemostat. Science 309:137–40


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


Bao Z, Murray JI, Boyle T, Ooi SL, Sandel MJ, Waterston RH. 2006. Automated cell lineage tracing inCaenorhabditis elegans. Proc. Natl. Acad. Sci. USA 103:2707–12

Bates M, Huang B, Dempsey GT, Zhuang X. 2007. Multicolor super-resolution imaging with photo-switchable fluorescent probes. Science 317:1749–53

Bean JM, Siggia ED, Cross FR. 2006. Coherence and timing of cell cycle start examined at single-cellresolution. Mol. Cell 21:3–14

Bengtsson M, Stahlberg A, Rorsman P, Kubista M. 2005. Gene expression profiling in single cells from thepancreatic islets of Langerhans reveals lognormal distribution of mRNA levels. Genome Res. 15:1388–92

Betzig E, Patterson GH, Sougrat R, Lindwasser OW, Olenych S, et al. 2006. Imaging intracellular fluorescentproteins at nanometer resolution. Science 313:1642–45

Bishop DK. 1994. RecA homologs Dmc1 and Rad51 interact to form multiple nuclear complexes prior tomeiotic chromosome synapsis. Cell 79:1081–92

Brandman O, Meyer T. 2008. Feedback loops shape cellular signals in space and time. Science 322:390–95Bulina ME, Chudakov DM, Britanova OV, Yanushevich YG, Staroverov DB, et al. 2006a. A genetically

encoded photosensitizer. Nat. Biotechnol. 24:95–99Bulina ME, Lukyanov KA, Britanova OV, Onichtchouk D, Lukyanov S, Chudakov DM. 2006b.

Chromophore-assisted light inactivation (CALI) using the phototoxic fluorescent protein KillerRed.Nat. Protoc. 1:947–53

Cai L, Dalal CK, Elowitz MB. 2008. Frequency-modulated nuclear localization bursts coordinate generegulation. Nature 455:485–90

Charest PG, Terrillon S, Bouvier M. 2005. Monitoring agonist-promoted conformational changes ofbeta-arrestin in living cells by intramolecular BRET. EMBO Rep. 6:334–40

Chudakov DM, Verkhusha VV, Staroverov DB, Souslova EA, Lukyanov S, Lukyanov KA. 2004. Photoswitch-able cyan fluorescent protein for protein tracking. Nat. Biotechnol. 22:1435–39

Cosma MP, Tanaka T, Nasmyth K. 1999. Ordered recruitment of transcription and chromatin remodelingfactors to a cell cycle- and developmentally regulated promoter. Cell 97:299–311

Dahan M, Levi S, Luccardini C, Rostaing P, Riveau B, Triller A. 2003. Diffusion dynamics of glycine receptorsrevealed by single-quantum dot tracking. Science 302:442–45

Danuser G, Waterman-Storer CM. 2003. Quantitative fluorescent speckle microscopy: where it came fromand where it is going. J. Microsc. 211:191–207

Danuser G, Waterman-Storer CM. 2006. Quantitative fluorescent speckle microscopy of cytoskeletondynamics. Annu. Rev. Biophys. Biomol. Struct. 35:361–87

Di Talia S, Skotheim JM, Bean JM, Siggia ED, Cross FR. 2007. The effects of molecular noise and size controlon variability in the budding yeast cell cycle. Nature 448:947–51

Dultz E, Zanin E, Wurzenberger C, Braun M, Rabut G, et al. 2008. Systematic kinetic analysis of mitoticdis- and reassembly of the nuclear pore in living cells. J. Cell Biol. 180:857–65

Eggeling C, Ringemann C, Medda R, Schwarzmann G, Sandhoff K, et al. 2009. Direct observation of thenanoscale dynamics of membrane lipids in a living cell. Nature 457:1159–62

Elowitz MB, Levine AJ, Siggia ED, Swain PS. 2002. Stochastic gene expression in a single cell. Science297:1183–86

Ferrell JE, Machleder EM. 1998. The biochemical basis of an all-or-none cell fate switch in Xenopus oocytes.Science 280:895–98

Giepmans BN, Adams SR, Ellisman MH, Tsien RY. 2006. The fluorescent toolbox for assessing proteinlocation and function. Science 312:217–24

Gomez-Sjoberg R, Leyrat AA, Pirone DM, Chen CS, Quake SR. 2007. Versatile, fully automated, microfluidiccell culture system. Anal. Chem. 79:8557–63

Hamant O, Heisler MG, Jonsson H, Krupinski P, Uyttewaal M, et al. 2008. Developmental patterning bymechanical signals in Arabidopsis. Science 322:1650–55

Hao N, Nayak S, Behar M, Shanks RH, Nagiec MJ, et al. 2008. Regulation of cell signaling dynamics by theprotein kinase-scaffold Ste5. Mol. Cell 30:649–56

Hein B, Willig KI, Hell SW. 2008. Stimulated emission depletion (STED) nanoscopy of a fluorescent protein-labeled organelle inside a living cell. Proc. Natl. Acad. Sci. USA 105:14271–76


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


Hell SW. 2007. Far-field optical nanoscopy. Science 316:1153–58Hell SW. 2009. Microscopy and its focal switch. Nat. Methods 6:24–32Hersen P, McClean MN, Mahadevan L, Ramanathan S. 2008. Signal processing by the HOG MAP kinase

pathway. Proc. Natl. Acad. Sci. USA 105:7165–70Holt LJ, Krutchinsky AN, Morgan DO. 2008. Positive feedback sharpens the anaphase switch. Nature 454:353–

57Huang B, Wang W, Bates M, Zhuang X. 2008. Three-dimensional super-resolution imaging by stochastic

optical reconstruction microscopy. Science 319:810–13Irazoqui JE, Gladfelter AS, Lew DJ. 2003. Scaffold-mediated symmetry breaking by Cdc42p. Nat. Cell Biol.

5:1062–70Joo C, Balci H, Ishitsuka Y, Buranachai C, Ha T. 2008. Advances in single-molecule fluorescence methods

for molecular biology. Annu. Rev. Biochem. 77:51–76Kaern M, Elston TC, Blake WJ, Collins JJ. 2005. Stochasticity in gene expression: from theories to phenotypes.

Nat. Rev. Genet. 6:451–64Kaufmann BB, Yang Q, Mettetal JT, van Oudenaarden A. 2007. Heritable stochastic switching revealed by

single-cell genealogy. PLoS Biol. 5:e239Keller PJ, Schmidt AD, Wittbrodt J, Stelzer EH. 2008. Reconstruction of zebrafish early embryonic develop-

ment by scanned light sheet microscopy. Science 322:1065–69Kerppola TK. 2008. Bimolecular fluorescence complementation (BiFC) analysis as a probe of protein inter-

actions in living cells. Annu. Rev. Biophys. 37:465–87Klar TA, Jakobs S, Dyba M, Egner A, Hell SW. 2000. Fluorescence microscopy with diffraction resolution

barrier broken by stimulated emission. Proc. Natl. Acad. Sci. USA 97:8206–10Kural C, Kim H, Syed S, Goshima G, Gelfand VI, Selvin PR. 2005. Kinesin and dynein move a peroxisome

in vivo: a tug-of-war or coordinated movement? Science 308:1469–72Lahav G, Rosenfeld N, Sigal A, Geva-Zatorsky N, Levine AJ, et al. 2004. Dynamics of the p53-Mdm2 feedback

loop in individual cells. Nat. Genet. 36:147–50Lev Bar-Or R, Maya R, Segel LA, Alon U, Levine AJ, Oren M. 2000. Generation of oscillations by the

p53-Mdm2 feedback loop: a theoretical and experimental study. Proc. Natl. Acad. Sci. USA 97:11250–55Liao JC, Roider J, Jay DG. 1994. Chromophore-assisted laser inactivation of proteins is mediated by the

photogeneration of free radicals. Proc. Natl. Acad. Sci. USA 91:2659–63Lohse MJ, Bunemann M, Hoffmann C, Vilardaga JP, Nikolaev VO. 2007. Monitoring receptor signaling by

intramolecular FRET. Curr. Opin. Pharmacol. 7:547–53Losev E, Reinke CA, Jellen J, Strongin DE, Bevis BJ, Glick BS. 2006. Golgi maturation visualized in living

yeast. Nature 441:1002–6Maheshri N, O’Shea EK. 2007. Living with noisy genes: how cells function reliably with inherent variability

in gene expression. Annu. Rev. Biophys. Biomol. Struct. 36:413–34Marco E, Wedlich-Soldner R, Li R, Altschuler SJ, Wu LF. 2007. Endocytosis optimizes the dynamic local-

ization of membrane proteins that regulate cortical polarity. Cell 129:411–22Matsuura-Tokita K, Takeuchi M, Ichihara A, Mikuriya K, Nakano A. 2006. Live imaging of yeast Golgi

cisternal maturation. Nature 441:1007–10McMahon A, Supatto W, Fraser SE, Stathopoulos A. 2008. Dynamic analyses of Drosophila gastrulation provide

insights into collective cell migration. Science 322:1546–50Melin J, Quake SR. 2007. Microfluidic large-scale integration: the evolution of design rules for biological

automation. Annu. Rev. Biophys. Biomol. Struct. 36:213–31Mettetal JT, Muzzey D, Gomez-Uribe C, van Oudenaarden A. 2008. The frequency dependence of osmo-

adaptation in Saccharomyces cerevisiae. Science 319:482–84Murray JI, Bao Z, Boyle TJ, Boeck ME, Mericle BL, et al. 2008. Automated analysis of embryonic gene

expression with cellular resolution in C. elegans. Nat. Methods 5:703–9Murray JI, Bao Z, Boyle TJ, Waterston RH. 2006. The lineaging of fluorescently-labeled Caenorhabditis elegans

embryos with StarryNite and AceTree. Nat. Protoc. 1:1468–76Nachman I, Regev A, Ramanathan S. 2007. Dissecting timing variability in yeast meiosis. Cell 131:544–56Nagerl UV, Willig KI, Hein B, Hell SW, Bonhoeffer T. 2008. Live-cell imaging of dendritic spines by STED

microscopy. Proc. Natl. Acad. Sci. USA 105:18982–87


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


Newman JR, Ghaemmaghami S, Ihmels J, Breslow DK, Noble M, et al. 2006. Single-cell proteomic analysisof S. cerevisiae reveals the architecture of biological noise. Nature 441:840–46

Ozbudak EM, Becskei A, van Oudenaarden A. 2005. A system of counteracting feedback loops regulatesCdc42p activity during spontaneous cell polarization. Dev. Cell 9:565–71

Ozbudak EM, Thattai M, Kurtser I, Grossman AD, van Oudenaarden A. 2002. Regulation of noise in theexpression of a single gene. Nat. Genet. 31:69–73

Patterson GH. 2008. Photoactivation and imaging of photoactivatable fluorescent proteins. Curr. Protoc. CellBiol. 38:21.6.1–10

Patterson GH, Lippincott-Schwartz J. 2002. A photoactivatable GFP for selective photolabeling of proteinsand cells. Science 297:1873–77

Peixoto A, Monteiro M, Rocha B, Veiga-Fernandes H. 2004. Quantification of multiple gene expression inindividual cells. Genome Res. 14:1938–47

Perlman ZE, Slack MD, Feng Y, Mitchison TJ, Wu LF, Altschuler SJ. 2004. Multidimensional drug profilingby automated microscopy. Science 306:1194–98

Ponti A, Machacek M, Gupton SL, Waterman-Storer CM, Danuser G. 2004. Two distinct actin networksdrive the protrusion of migrating cells. Science 305:1782–86

Raj A, van Oudenaarden A. 2008. Nature, nurture, or chance: stochastic gene expression and its consequences.Cell 135:216–26

Reddy GV, Gordon SP, Meyerowitz EM. 2007. Unraveling developmental dynamics: transient interventionand live imaging in plants. Nat. Rev. Mol. Cell Biol. 8:491–501

Reddy GV, Heisler MG, Ehrhardt DW, Meyerowitz EM. 2004. Real-time lineage analysis reveals oriented celldivisions associated with morphogenesis at the shoot apex of Arabidopsis thaliana. Development 131:4225–37

Rust MJ, Bates M, Zhuang X. 2006. Sub-diffraction-limit imaging by stochastic optical reconstruction mi-croscopy (STORM). Nat. Methods 3:793–95

Sauter G, Simon R, Hillan K. 2003. Tissue microarrays in drug discovery. Nat. Rev. Drug Discov. 2:962–72Schermelleh L, Carlton PM, Haase S, Shao L, Winoto L, et al. 2008. Subdiffraction multicolor imaging of

the nuclear periphery with 3D structured illumination microscopy. Science 320:1332–36Schuh M, Ellenberg J. 2007. Self-organization of MTOCs replaces centrosome function during acentrosomal

spindle assembly in live mouse oocytes. Cell 130:484–98Shaner NC, Campbell RE, Steinbach PA, Giepmans BN, Palmer AE, Tsien RY. 2004. Improved monomeric

red, orange and yellow fluorescent proteins derived from Discosoma sp. red fluorescent protein. Nat.Biotechnol. 22:1567–72

Shaner NC, Steinbach PA, Tsien RY. 2005. A guide to choosing fluorescent proteins. Nat. Methods 2:905–9Shroff H, Galbraith CG, Galbraith JA, Betzig E. 2008. Live-cell photoactivated localization microscopy of

nanoscale adhesion dynamics. Nat. Methods 5:417–23Skotheim JM, Di Talia S, Siggia ED, Cross FR. 2008. Positive feedback of G1 cyclins ensures coherent cell

cycle entry. Nature 454:291–96Stuart D, Wittenberg C. 1995. CLN3, not positive feedback, determines the timing of CLN2 transcription

in cycling cells. Genes Dev. 9:2780–94Subach FV, Subach OM, Gundorov IS, Morozova KS, Piatkevich KD, et al. 2009. Monomeric fluorescent

timers that change color from blue to red report on cellular trafficking. Nat. Chem. Biol. 5:118–26Sulston JE, Schierenberg E, White JG, Thomson JN. 1983. The embryonic cell lineage of the nematode

Caenorhabditis elegans. Dev. Biol. 100:64–119Tanabe T, Oyamada M, Fujita K, Dai P, Tanaka H, Takamatsu T. 2005. Multiphoton excitation-evoked

chromophore-assisted laser inactivation using green fluorescent protein. Nat. Methods 2:503–5Wang Y, Shyy JY, Chien S. 2008. Fluorescence proteins, live-cell imaging, and mechanobiology: Seeing is

believing. Annu. Rev. Biomed. Eng. 10:1–38Watanabe N, Mitchison TJ. 2002. Single-molecule speckle analysis of actin filament turnover in lamellipodia.

Science 295:1083–86Wedlich-Soldner R, Altschuler S, Wu L, Li R. 2003. Spontaneous cell polarization through actomyosin-based

delivery of the Cdc42 GTPase. Science 299:1231–35Westphal V, Rizzoli SO, Lauterbach MA, Kamin D, Jahn R, Hell SW. 2008. Video-rate far-field optical

nanoscopy dissects synaptic vesicle movement. Science 320:246–49


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.


Willig KI, Kellner RR, Medda R, Hein B, Jakobs S, Hell SW. 2006. Nanoscale resolution in GFP-basedmicroscopy. Nat. Methods 3:721–23

Wu JQ, Pollard TD. 2005. Counting cytokinesis proteins globally and locally in fission yeast. Science 310:310–14

Yeger-Lotem E, Sattath S, Kashtan N, Itzkovitz S, Milo R, et al. 2004. Network motifs in integrated cellularnetworks of transcription-regulation and protein-protein interaction. Proc. Natl. Acad. Sci. USA 101:5934–39

Yoshimoto H, Saltsman K, Gasch AP, Li HX, Ogawa N, et al. 2002. Genome-wide analysis of gene expressionregulated by the calcineurin/Crz1p signaling pathway in Saccharomyces cerevisiae. J. Biol. Chem. 277:31079–88

Zhao Z, Boyle TJ, Bao Z, Murray JI, Mericle B, Waterston RH. 2008. Comparative analysis of embryoniccell lineage between Caenorhabditis briggsae and Caenorhabditis elegans. Dev. Biol. 314:93–99


Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.

AR389-FM ARI 14 September 2009 14:58

Annual Reviewof Cell andDevelopmentalBiology

Volume 25, 2009

ContentsChromosome Odds and Ends

Joseph G. Gall � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 1

Small RNAs and Their Roles in Plant DevelopmentXuemei Chen � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �21

From Progenitors to Differentiated Cells in the Vertebrate RetinaMichalis Agathocleous and William A. Harris � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �45

Mechanisms of Lipid Transport Involved in Organelle Biogenesisin Plant CellsChristoph Benning � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �71

Innovations in Teaching Undergraduate Biologyand Why We Need ThemWilliam B. Wood � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �93

Membrane Traffic within the Golgi ApparatusBenjamin S. Glick and Akihiko Nakano � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 113

Molecular Circuitry of Endocytosis at Nerve TerminalsJeremy Dittman and Timothy A. Ryan � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 133

Many Paths to Synaptic SpecificityJoshua R. Sanes and Masahito Yamagata � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 161

Mechanisms of Growth and Homeostasis in the Drosophila WingRicardo M. Neto-Silva, Brent S. Wells, and Laura A. Johnston � � � � � � � � � � � � � � � � � � � � � � � � � 197

Vertebrate Endoderm Development and Organ FormationAaron M. Zorn and James M. Wells � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 221

Signaling in Adult NeurogenesisHoonkyo Suh, Wei Deng, and Fred H. Gage � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 253

Vernalization: Winter and the Timing of Flowering in PlantsDong-Hwan Kim, Mark R. Doyle, Sibum Sung, and Richard M. Amasino � � � � � � � � � � � � 277

Quantitative Time-Lapse Fluorescence Microscopy in Single CellsDale Muzzey and Alexander van Oudenaarden � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 301

Mechanisms Shaping the Membranes of Cellular OrganellesYoko Shibata, Junjie Hu, Michael M. Kozlov, and Tom A. Rapoport � � � � � � � � � � � � � � � � � � � � 329

The Biogenesis and Function of PIWI Proteins and piRNAs: Progressand ProspectTravis Thomson and Haifan Lin � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 355

vii

Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.

AR389-FM ARI 14 September 2009 14:58

Mechanisms of Stem Cell Self-RenewalShenghui He, Daisuke Nakada, and Sean J. Morrison � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 377

Collective Cell MigrationPernille Rørth � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 407

Hox Genes and Segmentation of the Hindbrain and Axial SkeletonTara Alexander, Christof Nolte, and Robb Krumlauf � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 431

Gonad Morphogenesis in Vertebrates: Divergent Means to aConvergent EndTony DeFalco and Blanche Capel � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 457

From Mouse Egg to Mouse Embryo: Polarities, Axes, and TissuesMartin H. Johnson � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 483

Conflicting Views on the Membrane Fusion Machinery and the FusionPoreJakob B. Sørensen � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 513

Coordination of Lipid Metabolism in Membrane BiogenesisAxel Nohturfft and Shao Chong Zhang � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 539

Navigating ECM Barriers at the Invasive Front: The CancerCell–Stroma InterfaceR. Grant Rowe and Stephen J. Weiss � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 567

The Molecular Basis of Organ Formation: Insights from theC. elegans ForegutSusan E. Mango � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 597

Genetic Control of Bone FormationGerard Karsenty, Henry M. Kronenberg, and Carmine Settembre � � � � � � � � � � � � � � � � � � � � � � 629

Listeria monocytogenes Membrane Trafficking and Lifestyle:The Exception or the Rule?Javier Pizarro-Cerda and Pascale Cossart � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 649

Asymmetric Cell Divisions and Asymmetric Cell FatesShahragim Tajbakhsh, Pierre Rocheteau, and Isabelle Le Roux � � � � � � � � � � � � � � � � � � � � � � � � � � � 671

Indexes

Cumulative Index of Contributing Authors, Volumes 21–25 � � � � � � � � � � � � � � � � � � � � � � � � � � � 701

Cumulative Index of Chapter Titles, Volumes 21–25 � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 704

Errata

An online log of corrections to Annual Review of Cell and Developmental Biology articlesmay be found at http://cellbio.annualreviews.org/errata.shtml

viii Contents

Ann

u. R

ev. C

ell D

ev. B

iol.

2009

.25:

301-

327.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

MA

SSA

CH

USE

TT

S IN

STIT

UT

E O

F T

EC

HN

OL

OG

Y o

n 10

/12/

09. F

or p

erso

nal u

se o

nly.

Documents

Quantitative Time-Lapse Fluorescence Microscopy in Single ...web.mit.edu/biophysics/papers/ARCDB2009.pdf · ANRV389-CB25-13 ARI 4 September 2009 19:45 Quantitative Time-Lapse Fluorescence