Available online at www.sciencedirect.com

www.elsevier.com/locate/optcom

Optics Communications 281 (2008) 888–894

In vivo monitoring of multiple circulating cell populationsusing two-photon flow cytometry

Eric R. Tkaczyk a,b,*, Cheng Frank Zhong a,b, Jing Yong Ye a,b, Andrzej Myc b,Thommey Thomas b, Zhengyi Cao b, Raimon Duran-Struuck c, Kathryn E. Luker d,e,

Gary D. Luker d,e, Theodore B. Norris a,b, James R. Baker Jr. b

a Center for Ultrafast Optical Science, EECS Department, University of Michigan, 2200 Bonisteel Blvd., Ann Arbor, MI 48109-2099, United Statesb Michigan Nanotechnology Institute for Medicine and the Biological Sciences, University of Michigan, Ann Arbor, MI 48109-0648, United States

c Unit for Laboratory Animal Medicine and Department of Internal Medicine, Ann Arbor, MI 48109-0614, United Statesd Department of Radiology, University of Michigan Medical School, Ann Arbor, MI 48109-0648, United States

e Department of Microbiology and Immunology, University of Michigan Medical School, Ann Arbor, MI 48109-0648, United States

Received 16 April 2007; received in revised form 5 October 2007; accepted 9 October 2007

Abstract

To detect and quantify multiple distinct populations of cells circulating simultaneously in the blood of living animals, we developed anovel optical system for two-channel, two-photon flow cytometry in vivo. We used this system to investigate the circulation dynamics inlive animals of breast cancer cells with low (MCF-7) and high (MDA-MB-435) metastatic potential, showing for the first time that twodifferent populations of circulating cells can be quantified simultaneously in the vasculature of a single live mouse. We also non-invasivelymonitored a population of labeled, circulating red blood cells for more than two weeks, demonstrating that this technique can also quan-tify the dynamics of abundant cells in the vascular system for prolonged periods of time. These data are the first in vivo application ofmultichannel flow cytometry utilizing two-photon excitation, which will greatly enhance our capability to study circulating cells in cancerand other disease processes.� 2007 Elsevier B.V. All rights reserved.

1. Introduction

There is intense interest in analyzing circulating cells inthe blood stream, including tumor cells and endothelialcells, to study cancer biology and response to therapy.The number of circulating tumor cells is an independentpredictor of survival in patients with metastatic breast can-cer [1]. The genotype of circulating tumor cells appears tobe representative of cells in tumor tissue, suggesting thatmonitoring cancer cells in the blood may reveal genetic

0030-4018/$ - see front matter � 2007 Elsevier B.V. All rights reserved.

doi:10.1016/j.optcom.2007.10.106

* Corresponding author. Address: Center for Ultrafast Optical Science,EECS Department, University of Michigan, 2200 Bonisteel Blvd., AnnArbor, MI 48109-2099, United States. Tel.: +1 734 763 0209; fax: +1 734763 4876.

E-mail address: etkaczyk@umich.edu (E.R. Tkaczyk).

changes that occur over the course of disease [2]. Breastcancer cells also can be detected in the circulation ofpatients without clinical evidence of metastatic disease,although the biologic significance of these cells remainsto be established [3]. Furthermore, changes in numbers ofcirculating endothelial cells have been used to monitorresponses to therapy [4]. These data emphasize the signifi-cance of circulating cells for cancer biology and treatmentand emphasize the need to define molecular features thatregulate dynamics of circulating cells in living animalsand patients.

One key obstacle to studies of circulating cells in mousemodels of cancer and other diseases is the challenge of opti-cally interrogating cells in the vascular system of livingmice. Two approaches to so-called in vivo flow cytometryhave been demonstrated in the vasculature of living mice

E.R. Tkaczyk et al. / Optics Communications 281 (2008) 888–894 889

[5,6]; while these initial studies showed promise, they havebeen limited to detecting single color biomarkers. How-ever, numbers of circulating cells enumerated with single-color detection may vary in response to changes in ananimal’s physiology, such as vasoconstriction or fluctua-tions in heart rate, in addition to intrinsic changes in circu-lating cells. With single color detection of biomarkers,errors caused by different physiological conditions cannotbe compensated. Thus, accurate assessments of circulatingcells in vivo are difficult to obtain.

To simultaneously enumerate different populations ofcirculating cells under varying physiological conditionsover time in an animal, we developed a novel two-photonflow cytometry system for two-channel detection of fluores-cent cells in vivo. Although detection of circulating cellsin vivo has been achieved in a confocal geometry using sin-gle-photon excitation [7], that technique has been limited todetection of a single fluorescent species. In contrast, ourtwo-photon system has the important advantage of simul-taneously detecting multiple cell populations because a sin-gle femtosecond near infrared (NIR) laser can be used toexcite multiple fluorescent dyes via two-photon transitions.The large separation between NIR excitation and emissionwavelengths attenuates scattered excitation light while col-lecting the entire fluorescence spectrum with high effi-

Fig. 1. Two-channel, two-photon flow c

ciency, thereby decreasing detection background [8].Further, multiphoton excitation brings to in vivo flowcytometry all the advantages of multiphoton microscopy,such as reduced photodestruction outside the interrogatedregion [9]. We used this innovative system to quantify fluo-rescently labeled red blood cells in vivo for periods of morethan two weeks and to monitor two populations of breastcancer cells in the same mouse. This research establishestwo-photon flow cytometry for real-time studies of multi-ple circulating cell populations in vivo, which will advanceour ability to probe the biology of circulating cells duringdisease progression and response to therapy.

2. Materials and methods

2.1. Two-photon flow cytometer

The two-photon flow cytometry apparatus (Fig. 1) useda Ti:Sapphire laser (Coherent Mira), which generates 50-fspulses at 800 nm with a repetition rate of 76 MHz. Thefemtosecond NIR laser beam was focused with a long-working distance objective (Olympus 40·) into a blood ves-sel of a mouse ear. A dichroic mirror directed the laser intothe microscope objective while transmitting the fluores-cence collected through the same objective. A secondary

ytometer for in vivo measurements.

890 E.R. Tkaczyk et al. / Optics Communications 281 (2008) 888–894

dichroic mirror and bandpass filters were used to separatethe fluorescence from different labeling dyes into two chan-nels, which were detected with two distinct photomultipliertubes (PMT Hamamatsu HC7421-40) and recorded withtwo multichannel photon counting scalers (MCS StanfordResearch SR430) which are capable of time resolution aslow as 41 ls. The signals (number of photons countedper each 1.3 ms bin of sampling interval) were sent to acomputer for data analysis.

For in vivo measurements, a custom-made heated stagewas used to restrain anesthetized mice. An 8-mm diameterhole was drilled at the center of the stage, on top of which aglass slide was placed. The vasculature and blood flow ofthe mouse ear were visualized on a CCD camera via trans-mitted light from a fiber optic illuminator (EW-41500-50,Cole-Parmer).

A Matlab program was used to extract fluorescent peaksabove the background noise level from the multichannelphoton counting scaler (MCS) trace signals. The particledetection threshold in a channel was set above the maxi-mum signal level from control traces. The programscanned the trace signals for fluorescent peaks above thethreshold. Once a peak was located, the peak characteris-tics, such as height (maximum fluorescent signal withinthe peak), width (number of consecutive bins above thebackground threshold) and location (the index of the max-imum bin) were stored. For two-channel measurements,the data from each channel were analyzed separately. Adouble-peak event was treated as a single event if the fluo-rescent signal between the two peaks did not fall below thebackground threshold value.

2.2. The in vivo flow cytometry procedure

Five to six weeks old, specific-pathogen-free femaleimmunocompromised nude (NU/NU CD-1 or NU/JFoxn1nu) or immunocompetent (CD-1) mice were pur-chased from Charles River Laboratories (Portage, Michi-gan) and housed in a specific pathogen-free animal facilityat the University of Michigan Medical Center in accordancewith the regulations of the University’s Committee on theUse and Care of Animals as well as with federal guidelines,including the principles of Laboratory Animal Care.

Prior to in vivo flow cytometry experiments, mice wereinjected with 0.5 ml of sterile saline heated to 37 �C tomaintain hydration. Mice were anesthetized with inhala-tion of isoflurane (4%) and then maintained on 1–2% iso-flurane throughout experiments. For optical detection,mice were placed on a heated stage with the left ear onthe glass slide window. Drops of glycerol were applied tothe mice eyes to prevent drying. The femtosecond NIRlaser beam was focused into the mouse ear with the objec-tive from below. Microscope immersion oil was placedbetween the mouse ear and the glass slide to maintain theposition of the ear during the cytometry measurement.When the short pass filter in the fluorescence collectionoptical path was switched off, the back-scattered light from

the femtosecond NIR laser beam could be visualized on theCCD camera to align the excitation beam with the bloodvessel. An � 5 lm diameter arteriole flow was selected forcytometry measurements. The location of this arteriolewas marked to allow subsequent measurements to beobtained from the same blood vessel.

Before injection of the solution of fluorescent-dye-labeled particles or cells, the background signals in bothshort wavelength channel (S-channel) and long wavelengthchannels (L-channel) were recorded as a control. Solutionscontaining the fluorescence-labeled particles were injectedthrough the tail vein. In vivo flow cytometry typically beganwithin 5 min of injection. The laser power used was below20 mW at the focus, and no photodamage to the ear of themice was observed after the experiments. During analysis,the thresholds were set such that no peaks existed in thecontrol traces.

2.3. In vivo flow cytometry of fluorescent microspheres and

DeepRed-labeled splenocytes

Two micron, yellow–green fluorescent (Ex505/Em515)microspheres (Molecular Probes, Inc. Eugene, OR) werewashed, resuspended in 200 ll PBS (total number of beads2.3 · 109), and injected via tail vein to CD-1 mice. Spleenswere collected from euthanized animals and were disruptedin PBS to obtain a single cell suspension, after which thecells were washed in PBS and the red blood cells were lysedusing ammonium chloride. To stain the splenocytes, thecells were treated with MitotrackerTM DeepRed 633 (Molec-ular Probes, Inc., Eugene, OR), incubated for 1 h at 37 �C,trypsinized, rinsed and resuspended in PBS, pH 8. A totalof 3.6 · 106 DeepRed-labeled splenocytes were injectedinto a female NU/NU CD-1 mouse (chosen to preventimmune system recognition of the injected splenocytes).The two-channel fluorescence signals were recorded imme-diately after each injection for 10 or 20 consecutive minutesand repeated at the same vessel location roughly 2 h andalso 1 day after the initial injection.

2.4. In vivo flow cytometry of DiD labeled red blood cells

Blood (25–50 ll) was collected from mice by retro-orbi-tal puncture, and 12 ll were centrifuged to remove serumand the buffy coat as described above. Cells were washedonce with PBS and then stained with 10 ll 1,1 0-Dioctade-cyl-3,3,3 0,3 0-tetramethyllindodicarbocyanine perchlorate(DiD, a lipophilic dye that labels membrane bilayers)(Vybrant DiD, Labeling Solution, Invitrogen) in 100 llPBS for 30 min at 37 �C. Cells were washed once in PBSto remove unincorporated DiD and then resuspended in100 ll sterile 0.9% NaCl. Samples (containing approxi-mately 1 · 107 stained RBCs) were injected intravenouslyinto a female NU/J Foxn1nu mouse via a tail vein forin vivo flow cytometry experiments. The long channel fluo-rescence signal was recorded at roughly the same vessellocation several times over the next several weeks (0 min,

Fig. 2. Two-channel raw data from the two-photon flow cytometer. Shortwavelength channel (‘‘S-channel,’’ green line) and long wavelengthchannel (‘‘L-channel,’’ red line) traces in a mouse injected with Dee-pRed-labeled splenocytes at time zero (dashed line). The control trace isshown to the left of the dashed line. The data is essentially a histogramshowing the number of photons counted by the multichannel scaler ineach 1.31 ms bin on the x-axis. Detail of individual fluorescent peaks areshown in the inset. (For the interpretation of color in this figure legend, thereader is referred to the Web version of this article.)

Fig. 3. Depletion dynamics of injected cells. Number of detected events(log scale) are plotted over 20 min after the injection of 3.6 · 106

DeepRed-labeled splenocytes at time 0.

E.R. Tkaczyk et al. / Optics Communications 281 (2008) 888–894 891

20 min, 5 h, 24 h, 48 h, 4 days, and 17 days after injection).The frequency was calculated as the number of peaks in theL-channel within 214 s.

For measurements beginning on day 4, the location inthe vessel was optimized manually in real time to maximizethe amplitude and frequency of the detected events. Thiswas possible since the large number of events enabled a fre-quency count every few seconds between manual adjust-ments of the objective. It is noteworthy that thisimprovement is only possible when the number of detectedevents is at least several per second, and thus cannot beused in one-color analysis of metastatic cancer cells orother rare populations of cells in the circulation.

2.5. Conventional flow cytometry of red blood cells

Blood samples (50–100 ll) were collected from mice viaretro-orbital puncture using heparinized capillary tubes.Blood was transferred to heparinized microfuge tubes andcentrifuged for 1 min at 10,000 RPM. Serum and the buffycoat were removed, and the cell pellet was resuspended in750 ll FACS buffer (PBS with 0.1% sodium azide and 1%heat-inactivated fetal calf serum). Flow cytometry was per-formed on a FACS Calibur System (BD Biosciences), anddata were analyzed with CellQuest software.

2.6. In vivo flow cytometry of two cell lines labeled with two

color quantum dots

To simultaneously monitor circulating MCF-7 andMDA-MB-435 in the same mouse in vivo, cells were labeledwith 15 nM or 7.5 nM of Qtracker quantum dots (Invitro-gen) that emit at 585 nm (Qdot585) or 655 nm (Qdot655),respectively. Quantum dots were mixed with an equal vol-ume of transfection reagent provided with the Qtracker kitaccording to the manufacturer’s protocol. Breast cancercells were seeded at 4 · 106 cells/60 mm dish (MDA-MB-435) or 3 · 106 cells per dish (MCF-7) 6 h prior to labelingwith quantum dots mixed with 200 ll of cell culture med-ium. Cells were rocked every 15 min for 1 h to distributethe quantum dots, and cell culture medium then was addedto a final volume of 4 ml to continue labeling overnight.Cells were treated with 2 mM EDTA to release them fromthe culture dish, washed with PBS, and resuspended in ster-ile 0.9% saline for injection. Because initial studies withthese cells suggested that fewer MDA-MB-435 cells couldbe detected immediately after injection, we injected twiceas many of these cells: 1 · 106 435 cells were co-injectedwith 0.5 · 106 MCF-7 cells in 125 ll PBS. The two-channelfluorescence signals were recorded immediately after eachinjection and repeated at the same vessel location roughly2 h and 1 day after the initial injection. Another set of micewere injected with tumor cells labeled with the reverse pat-tern of quantum dots (MCF-7 labeled with Qdot585 andMDA-MB-435 labeled with Qdot655) to confirm that theparticular fluorophore assignment does not influence thedetectability of these cell lines in vivo.

3. Results

3.1. Monitoring circulating fluorescent particles in vivo

As initial validation of the two-photon technique forin vivo flow cytometry, we labeled splenocytes ex vivo withthe fluorescent dye DeepRed and injected these cells intra-venously into mice. Within 30 s after the start of detection,we detected fluorescent peaks from DeepRed-labeled cellsin the long wavelength detection channel (Fig. 2). Circulat-ing splenocytes were monitored continuously over 20 min,at which point events dropped to less than 10% of their ini-tial frequency (Fig. 3). Very low numbers of circulating

892 E.R. Tkaczyk et al. / Optics Communications 281 (2008) 888–894

splenocytes could be detected approximately 24 h afterinjection (data not shown). These data were confirmed bystandard ex vivo flow cytometry that showed rapid deple-tion of splenocytes from the circulation of mice.

We used fluorescent microspheres to verify that the sys-tem could detect signals in the short wavelength channelin vivo. Following intravenous injection of microspheresinto CD-1 mice, large fluorescence peaks were detectedonly in the short wavelength channel. As expected, thesesynthetic microspheres cleared rapidly from the circulationand were not detectable approximately 24 h after injection(data not shown). Overall, these initial studies establish thefeasibility of using the two-photon flow cytometer to detectfluorescence in two different detection channels.

3.2. Monitoring of circulating red blood cells in vivo for

extended periods of time

To establish that the two-photon flow cytometer couldmonitor an abundant population of cells for prolongedperiods of time, we interrogated red blood cells (RBC) inthe circulation of living mice over 17 days. RBC werelabeled with the membrane dye 1,1 0-dioctadecyl-3,3,3 0,3 0-tetramethyllindodicarbocyanine perchlorate (DiD) andinjected intravenously into nude mice. Numbers of RBCin repeated measurements were highly variable, despitefocusing the laser at the same artery in the ear (Fig. 4). Itwas observed that the operator could significantly increasethe number of detected events (and therefore sampled vol-ume) by manually adjusting the laser focus during thecourse of data acquisition, presumably to the maximumflow position in the artery. In vivo data for circulatingRBC were validated with conventional flow cytometryex vivo on blood drawn on day 18. According to conven-

Fig. 4. Monitoring for extended times in vivo. Number of detectedcirculating DiD-labeled RBCs are plotted after injection of approximately1 · 107 labeled RBCs at time 0. Focal point in the mouse ear was manuallyoptimized for maximal counts in real time for measurements beginning at100 h. Both axes are log scale.

tional flow, 4.40% of 1.5 million blood cells counted werelabeled with DiD. No labeled cells were detected by eitherconventional or two-photon flow cytometry 1.5 monthsafter injection, which could be the result of clearance ofcells from the circulation or instability of the dye for thisperiod of time in blood.

3.3. Simultaneous monitoring of two populations in the

circulation of a mouse using two-color two-photon cytometry

To establish the capability of two-photon flow cytome-try to enumerate two distinct, rare populations of circulat-ing cells, we measured dynamics of two different breastcancer cell lines in the circulation of living mice. HumanMDA-MB-435 cells are known to spontaneously formmetastases in mice [10], while MCF-7 cells have very lowmetastatic potential [11]. Thus, we hypothesized that thesecells would show differences in their circulation kineticsbased on differences in ability to metastasize.

We enumerated circulating MDA-MB-435 and MCF-7cells with in vivo two-photon flow cytometry at approxi-mately 10, 100, and 1000 min after injection. As shown inFig. 5, the dynamics of each cell line in the circulation weremarkedly different. Ten minutes after injection, we detectedapproximately five hundred MCF-7 cells during a 430-speriod of data acquisition. Numbers of MCF-7 cells at thistime were more than 10-fold greater than MDA-MB-435cells, despite the small difference in total numbers of cellsinjected. These data are similar to those reported by Georg-akoudi et al., who showed using single channel in vivo flowcytometry that exit of prostate cancer cells from thecirculation correlated with relative differences in metastatic

Fig. 5. Simultaneous monitoring of two populations of circulating cancercells. Detected events are shown after the injection of 5 · 105 Qdot585stained (low metastatic potential) MCF-7 cells simultaneously with 1 · 106

Qdot665 stained (high metastatic potential) 435 cells into the same mouseat time 0. The frequency was calculated as the number of events counted inthe short wavelength S-channel (open markers) or long wavelength L-channel (closed markers) within 430 s. The experiment was repeated inthree different mice, and the average results are shown, with error barsindicating the variation over the three animals. Both axes are log scale.

E.R. Tkaczyk et al. / Optics Communications 281 (2008) 888–894 893

potential [12]. Interestingly, MCF-7 cells cleared almostcompletely from the circulation by 1000 min after injection,while MDA-MB-435 cells remained detectable throughoutthe entire 1000 min of the experiment. Thus, the two celllines had very different circulation kinetics. Both cell linesretained their detectability level in vivo when the quantumdot assignments were switched, indicating that the mea-sured kinetics of each cell line were not biased by the emis-sion wavelength of quantum dots. These data areconsistent with prior work showing that behavior of quan-tum dot-labeled tumor cells and unlabeled cells in vivo areindistinguishable following tail vein injection [13].

Studies in vitro showed that the stability of labeling iscomparable between 435 and MCF-7 cells, and bothremain labeled adequately for detection for over 5–6 days.Microscopic analysis of excised organs further confirmedthat cells retained quantum dot labels in vivo. Fig. 6 showsrepresentative fluorescence microscopy images of each cellline with quantum dots 48 h after labeling. Conventionalflow cytometry showed essentially 100% labeling efficiencywith quantum dots in both cell lines and confirmed micro-scopic images for comparable persistence of labeling (datanot shown). These data establish that differences between

Fig. 6. Representative fluorescence images 48 h after labeling of (a) MCF-7 cells with 585 QDots and (b) MDA-MB-435 with 655 QDots.

circulating MCF-7 and MDA-MB-435 cells are not dueto differences in labeling between the two cell lines.

To determine the relationship between circulating tumorcells and cells in tissues, we quantified numbers of quantumdot-labeled MDA-MB-435 and MCF-7 cells in excisedlungs and liver 48 h after injection. We found approxi-mately 3-fold greater MDA-MB-435 cells in these organsthan MCF-7. These data, in combination with the resultsof two-photon in vivo flow cytometry, suggest that greaternumbers of 435 cells arrest in the lungs and liver initiallyand persist in these organs relative to MCF-7 cells. It islikely that fewer MCF-7 cells are detected by 48 h becausethese cells are reported to undergo apoptosis in secondaryorgans, particularly in the absence of estrogen supplemen-tation that these cells require for proliferation in mice.These data are consistent with previous studies showingthat MCF-7 cells do not form metastases after tail veininjection [14].

4. Discussion

Recent studies have focused attention on detecting andquantifying absolute numbers and dynamics of circulatingcells in cancer and inflammation to determine disease prog-nosis, identify potential therapeutic targets, and monitorthe response to therapy (reviewed in [15,16]). To betterunderstand normal and pathologic regulation of circulat-ing cells, it is necessary to develop techniques that allowsensitive, long-term real-time monitoring of multiple popu-lations of cells in vivo. The present study describes an inno-vative two-color two-photon flow cytometer thataccomplishes these experimental objectives. Using this sys-tem, we were able to perform simultaneous two-colordetection of multiple circulating cell lines in vivo, whichhas not been accomplished with previous techniques forin vivo flow cytometry.

Many unique studies are possible with the novel two-photon approach to in vivo flow cytometry. Investigationswith RBCs show that an abundant population of cellscan be monitored for periods of greater than two weeks,thereby establishing that the system can detect single eventsat high counting rates over extended periods of times. Thismay be enabled by the limitation of multiphoton absorp-tion to the interrogated region, reducing potential collat-eral damage to tissues outside of the vessel lumen.Monitoring circulating cells over the course of weeks hasnot been demonstrated previously with other in vivo flowcytometers. Interrogating populations of hematopoieticcells, such as red blood cells (RBCs) or leukocytes, couldprovide new insights into normal physiology or diseaseprocesses, as well as serving as an internal standard forhemodynamics and perfusion. For example, if the dynam-ics of a cell population, such as RBCs, are well known, thisprovides a statistically reliable internal control of the inter-rogated blood volume per unit time, if more than onedetection channel is available (as in our unique multipho-ton system).

894 E.R. Tkaczyk et al. / Optics Communications 281 (2008) 888–894

Variations in measurements during the RBC experimentemphasize the importance of multiple channel detection.While single channel in vivo flow cytometry is sufficientfor a qualitative analysis of circulating cell dynamics, it isgenerally limited by variations in event counts in repeatedexperiments. Between separate times of detection, bloodflow velocity at the measurement site may vary. This occurson a short time-scale due to pulsatile flow, but also on alonger time scale due to drift in the positioning of the laserfocus. This can be caused when the anesthetized mousechanges position slightly, which results in the laser focusingon a site closer to the wall of the vessel, where blood flowvelocity is lower. Further, physiological conditions can behighly variable between experiments, whether examiningthe same or a different mouse. These factors are com-pounded by statistical fluctuations when small sample setsare acquired for studies of rare circulating cells. Without ameans to determine the volume and total events sampledover a given time interval, these issues will complicate theanalyses of populations of cells in mice using single colorflow cytometry.

With multicolor, two-photon in vivo flow cytometry, dif-ferent populations of cells can be compared simultaneouslyin a single animal with a single laser source, overcomingmany variables intrinsic to in vivo experiments. In particu-lar, we used this technique to quantify two populations ofbreast cancer cells in the same mouse and demonstratedintrinsic differences in the dynamics of these cells in the cir-culation. Because we can monitor multiple cell populationsin the same animal, two-photon flow cytometry greatlyreduces experimental variability related to animal-to-ani-mal differences in perfusion. We expect that further studieswill allow us to interrogate functions of specific moleculesimplicated in metastatic breast cancer, such as the chemo-kine CXCL12 and its receptor CXCR4 [17]. These types ofstudies may improve our understanding of circulating can-cer cells and the progression to metastatic disease. Thesewould also facilitate the in vivo testing of new compoundstargeted against specific molecules or pathways.

One challenge to in vivo cytometry is the small numberof labeled cells passing through the blood vessel in thedetection region. Using the simultaneous injection oflabeled RBCs as a reference control for focal point optimi-zation will enable far more reliable experiments on rareevent circulation dynamics. One solution to increase thestatistical power of the measurements would be to accesslarger blood vessels with higher flow rates, thereby increas-ing the number of detected cells; this should be enabled bythe ability of the femtosecond NIR laser source to pene-trate much deeper through biological tissue than a contin-uous wave laser in visible region for one-photon excitation[18], and to potentially reduced photo damage to surround-ing tissue for extended monitoring.

We anticipate that two-photon flow cytometry willgreatly enhance our ability to interrogate circulating cellsin pre-clinical models of cancer and other diseases. Forexample, we envision that this technology can be used to

investigate (1) changes in gene expression in circulatingcells through imaging probes injected into mice or geneti-cally-engineered cell lines with fluorescent reporters; and(2) analyze effects of specific adhesion molecules in traffick-ing of cancer or immune cells using genetically-engineeredmice. These studies will be aided by technical advancessuch as the real-time imaging of blood vessels to monitorand control the position of the laser beam throughoutthe experiment and the development of approaches to labelspecific cells in vivo, rather than ex vivo.

Acknowledgements

This project has been funded in whole or in part withFederal funds from the NASA Ames Research Center, Na-tional Aeronautics and Space Administration, under Con-tract No. NAS2-02069. Funding also is provided by theSusan Komen Foundation, Sidney Kimmel Foundation,JTM Research, and NIH P50 CA93990. E.T. was sup-ported by an NSF Graduate Research Fellowship.

References

[1] M. Cristofanilli, G.T. Budd, M.J. Ellis, A. Stopeck, J. Matera, M.C.Miller, J.M. Reuben, G.V. Doyle, W.J. Allard, L.W. Terstappen,D.F. Hayes, N. Engl. J. Med. 351 (2004) 781.

[2] S. Meng, D. Tripathy, S. Shete, R. Ashfaq, H. Saboorian, B. Haley,E. Frenkel, D. Euhus, M. Leitch, C. Osborne, E. Clifford, S. Perkins,P. Beitsch, A. Khan, L. Morrison, D. Herlyn, L. Terstappen, N.Lane, J. Wang, J. Uhr, PNAS 103 (2006) 17361.

[3] S. Meng, D. Tripathy, E.P. Frenkel, S. Shete, E.Z. Naftalis, J.F.Huth, P.D. Beitsch, M. Leitch, S. Hoover, D. Euhus, B. Haley, L.Morrison, T.P. Fleming, D. Herlyn, L.W. Terstappen, T. Fehm, T.F.Tucker, N. Lane, J. Wang, J.W. Uhr, Clin. Cancer Res. 10 (2004)8152.

[4] P. Mancuso, M. Colleoni, A. Calleri, L. Orlando, P. Maisonneuve, G.Pruneri, A. Agliano, A. Goldhirsch, Y. Shaked, R.S. Kerbel, F.Bertolini, Blood 108 (2006) 452.

[5] J. Novak, I. Georgakoudi, X. Wei, A. Prossin, C.P. Lin, Opt. Lett. 29(2004) 77.

[6] V. Zharov, E. Galanzha, V. Tuchin, J. Biomed. Opt. 10 (2005)051502-1.

[7] H. Lee, C. Alt, C.M. Pitsillides, M. Puoris’haag, C.P. Lin, Opt.Express 14 (2006) 7789.

[8] K. Konig, J. Microsc. 200 (2000) 83.[9] W. Denk, K. Svoboda, Neuron 18 (1997) 351.

[10] J.E. Price, A. Polyzos, R.D. Zhang, L.M. Daniels, Cancer Res. 50(1990) 717.

[11] S.W. McLeskey, J. Kurebayashi, S.F. Honig, J. Zwiebel, M.E.Lippman, R.B. Dickson, F.G. Kern, Cancer Res. 53 (1993) 2168.

[12] I. Georgakoudi, N. Solban, J. Novak, W. Rice, X. Wei, T. Hasan, C.Lin, Cancer Res. 64 (2004) 5044.

[13] E.B. Voura, J.K. Jaiswal, H. Mattoussi, S.M. Simon, Nature Med. 10(2004) 993.

[14] C. Alix-Panabieres, J.P. Brouillet, M. Fabbro, H. Yssel, T. Rousset,T. Maudelonde, G. Choquet-Kastylevsky, J.P. Vendrell, J. Immunol.Methods 299 (2005) 177.

[15] A. Luster, R. Alon, U. von Andrian, Nature Immunol. 6 (2005) 1182.[16] P.K. Goon, C.J. Boos, P.S. Stonelake, G.Y. Lip, Future Oncol. 1

(2005) 813.[17] K. Luker, G. Luker, Cancer Lett. 238 (2006) 30.[18] M. Gu, X. Gan, A. Kisteman, M.G. Xu, Appl. Phys. Lett. 77 (2000)

1551.