mommammmem
`
`fiytonjratry a.-.4
`
`Mlfiowcyt
`
`William Violet Floorophorea: A New titans oi
`Ultrahright Fluorescent finrnpounria tor
`
`lmmunotlnoreseanea Experiments
`
`Pratip K. C‘ria‘ttoparitlftyay‘,l ’l Brent Gaylnml,2 Adrian Painter,2 Nan lining,3
`Mary A. Rayenf Geoff Let/via,4 Morgan A. Renter," A.K.M. Nuruur llahrnan,5
`David. A. Prince,6 Michael. R. Betta?” Mario Roederer1
`
`‘lrrrrrrunnTachnoiogy Section, Vaccine
`Research (teeter, i‘llAiD, NEH. Bethesda.
`Maryland
`
`2Sirigen. Ringo/00d, Hampshire,
`United Kingdom
`
`2BioLegenzl, San Diego, California
`‘Neurnsnience Fiesearnli institute and
`Department of Molecular Cellular,
`Developmental Biology, University at
`California, Santa Barbara, California
`
`EDepartment of Microbiology, University of
`Pennsylvania, Philadelphia, Pennsylvania
`
`6institute of intention and lrrrrrrunity,
`{lardiff University School of Medicine,
`Cardifi, Whales, United Kingdom
`
`Received 28 November 2011; Revision
`Fiat
`“ti i2 February 2m; Accepted 28
`
`February rim
`Grant sponsors: intramural Research
`Program of the l‘iiAlD, hill},
`
`Additional Supporting Information may be
`tmmc‘ in the nrilinn version nithis article.
`
`*Carresponclenca in: Preiip
`Channpadhyay, immunoTeehnn-lntgy
`Seerion.\iaccine Research (trainer,
`NlAED, trill-l, Bethesda, MD 20392, HERA
`
`Email: pcha'ttnptllmaiiningmr
`
`Published trnlirre 6 April 2Gi2 in Wiley
`Dnlina Library iwaleynnlinelibrarycom)
`DDl: 10.100219th 3.22043
`
`Published 201?. Wiley Periodicals, lE’iC.
`lTliin article is 3 US government work
`and, as such, is in the nniilin domain in
`the United grates of America
`
` International
`
`
`
`Cytoroetiy Part A to 81A: AEBASB 2012
`
`8 Ahetraet
`The Nobel Prize in Chemistry was awarded in 2000 for the discovery of conductive
`organic polymers, which have subsequently been adapted for applications in ultrasensi-
`tive biological detection. Here, we report the first use of this new class of fluorescent:
`probes in a diverse range of cytometric and imaging applications. We demonstrate that
`these “Brilliant Violet” reporters are dramatically brighter than other UV—violet excita~
`ble dyes, and are of similar utility to phycoerythrin (PE) and allophycocyanin (AFC).
`They are thus ideally suited for cytornetrir assays requiring high sensitivity. such as
`l‘vll-lC-tnultirner staining or detection of intracellular antigens. Furthermore,
`these
`reporters are sensitive and spectrally distinct options for fluorescence imaging, two—
`photon microscopy and imaging cytornetry. These ultra—bright materials provide the
`first new high-sensitivity fluorescence probes in over 25 years and will have a dramatic
`impact on the design and implementation of multicolor panels for high—sensitivity
`irnrnunotluorescence assays. Published 2012 Wiley Periodicals, Inc.l
`
`*6 Key terms
`brilliant Violet;
`polymers
`
`ilnorochrorncs; dyes;
`
`imrnunofluorcscencc; Violet
`
`laser; conductive
`
`THE power oi" flow cytorneti'y, as a tool to analyze complex biological systems. is still
`limited by the variety of fluorochrorues currently available (i). Existing fluorescent
`probes fall into three classes: large protein—based molecules leg, phycobiliproteins),
`inorganic fluorescent nanocrystals (e.g., quantum dots), and small orgruiic dyes ( erg,
`fluorescein), All have been incorporated into standard immunoiluorescence staining
`technology, but many exhibit undesirable qualities in terms ofbrightness, stability, or
`applicability to different techniques (lilte intracellular staining) (2,3). in this contri-
`bution, i 'e examine a fourth class of fluorescent materials that otter distinct advan-
`tages in performance and versatility.
`ln principle, the sensitivity of fluorescent probes in cell staining is limited by
`three factors. First, and most importantly, is the intrinsic brightness of the probe,
`quantified generally by its alisorhance cross-section and quantum efficiency. Second
`is the degree to which in‘trirrSic background (auto fluorescence) of the sample overlaps
`with the dye’s emission. And third, for multipararneter applications,
`is how broad
`the excitation and erriission spectra are, leading to “spillover” or overlap of multiple
`probe emissions. This requires mathematical correction i known as compensation).
`For
`imaging applications, a fourth factor becomes important:
`resistance to
`photohleaching. The properties of phycoerythrin (PE) and ( to a lesser extent)
`allophycocyunin (AFC), two phycohiliproteins in use for the last 25 years, make them
`the brightest probes currently used in most irnrnunoilnorescence experimentst Both
`
`1
`
`BD00000091
`
`TFS1 057
`
`TFS1057
`
`1
`
`

`

`have
`
`consid Erable extinction coefficients
`
`(l,960,000 and
`
`700,000 cm "I Mr"1 respectively) and high quantum etficien-
`cies {0.82 and 0.68 respectively).
`Beyond PE and AFC, other fluorescent: probes are now
`available, but most have important limitations that complicate
`the design of multiparameter flow cytometry or multilabel
`imaging experiments (4,5). Between 1995 and 2004, polychro~
`matic (>5 color) flow cytometry grew rapidly, driven by an
`interest in studying complex biological systems. the develop
`ment of new hardware, and the introduction of energy transfer
`tandems of PE and AFC. These tandem dyes. now standard
`components in multiparameter staining, use covalently bound
`acceptor dyes to achieve progressively redder emissions (6).
`Thus, this family of dyes contributes a wide variety of “colors”:
`however, the dyes exhibit significant spectral overlap, vary in
`donor:acceptor ratios from, l,ot-to--lot (complicating compen-
`sation), can be unstable, and can have relatively low intrinsic
`brightness (making them far less sensitive than PE or AFLL).
`Notably, similar dyes are limited for UV-violet excitation, leav--
`ing a spectral gap fo r additional multicolor probes.
` on of UV—yiolet excitable
`More recently,
`the introdi
`quantum dots (QD) probes, emitting light at various wave—
`lengths between 525 nm and 800 nm, provided a dramatic
`increase in the number of parameters that could be measured
`simultaneously. ultimately leading to ldcolor cytometry {7).
`QB fluorescence is read in regions of the spectrum with low
`autofluorescence (an important advantage of these dyes);
`however, they stiffer from dramatic spillovers (from rnultilaser
`excitation). Also, QDs have been somewhat difficult to reliably
`conjugate and thus the commercial catalogue of reagents
`remains limited (2). Finally, the purification of free fluroro-
`chrome from conjugaes is also complicated and inefficient,
`further limiting their use for intracellular staining.
`The third class of
`tluorochronies consists of small
`
`molecule organic dyes. These dyes have well-defined, wave-
`length-tunable st uctures, and are available in a Wide range of
`colors. However, they lack brightness relative to QDs and phy-
`cobiliproteins because of their inherently low molar extinction
`
`: which
`coefficients. This is particularly true for co umarine d y
`remain the primary reporters used for UV—violet excitation
`and emission below 525 mn. Thus, despite the introduction of
`dozens of new fluorescent probes over the past decades, none
`rival PE or APC for sensitivity—putting a significant limita—
`tion on the design of multicolor staining panels requiring
`multiple high --sensitivity detections.
`Similarly, multilabel fluorescence microscopy has been
`limited by the availability of suitable, UV—violet excitable dyes.
`Currently-available dyes bleach rapidly in fluorescence and
`confocal microscopy applications (e.g AlexadOfi, AMCA‘) or
`are poor matches for con‘unon epifluorescence and n‘iultiplio--
`ton filter sets (e.g. lily/light 405‘). Although many microscopes
`are equipped with a Violet light source and filters, the imaging
`is frequently limited to DAPI (Al’,6~diamidino~1—phenylindole)
`and Hoechst 33342 stains of nuclei.
`
`Given these limitations, we have been developing a new
`family of fluorescent molecules for highly sensitive chemical
`and biological detection over the past decade (8,9). These
`
`Cytornetiy Part A to 81A: AEFASB 2012
`
`g m iNfitfi. &§§"§‘§ .fifi
`
`ltnown as ids-conjugated polymers, found initial
`molecules,
`utility in the development of polymer-based organic electron--
`ics (Olil‘iDs, photovoltaics, printable circuitry, etc.) (10). More
`recently, such materials were synthetically adapted for aqueo us
`solubility to take advantage of their unique optical properties
`in various biosensor schemes (11,12).
`Like other organic dyes,
`'Ir—conjugated polymers have a
`synthetically tunable networl< of ;r~orbi‘tals that allows for elec—
`tron delocali'zation, absorption from the UV to near lll region
`of the spectrum, and efficient fluorescence. However, unlike
`organic dye molecules, for which the u—network is confined to
`a discrete set of atoms, the backbone structure of conjugated
`polymers allows for delocalization to occur over many repeat
`units in a polymer chain. In this way, the polymers act as large,
`extended collections of optical subunits, which are able to
`respond cooperatively in various energy transfer and/ or
`quenching processes (10,l3). This collective behavior leads
`to extraordinarily high molecular extinction coefficients
`
`(>105 M "l cm ’7 X) that rival phycobiliproteins and QDS. Their
`unique optical properties can also be tuned synthetically to
`produce materials with intense photoluminescence at a wide
`range of wavelengths which. like organic dyes, retain discreet
`excitation pro tiles (14). These properties are not characteristic
`of any lluorochrome produced to date;
`thus, these polymers
`represent a new, fourth category of fluorescent materials.
`Our focus has been to create well—defined conjugated
`polymer structures that allow for facile bioconjugation to a
`variety of biological probes (antibodies, streptavidin, nucleic
`acids, etc.). In this way, the polymers are utilized as discrete
`
`(single chain) tluor
`ent reporters wh ere the optical, proper--
`ties are controlled primarily by the chemical structure rather
`
`than particle size or aggrega ‘ be avior (ll 77713). This depar-
`ture front previous methodologies affords more specific
`control of biological recognition events using conventional
`detection strategies. Furthermore, the materials can be man u--
`factured reproducibly, consistently, and on a large scale. The
`first of these probes, Brilliant Violet
`(BX/421‘),
`is excitable at
`405 nm, demonstrates high intrinsic brightness, and has little
`
`nonsr
`'fic binding. lmportantly, it is reliably conjugated to
`antibodies through known, well—defined functional sites (both
`in terms of number and location). in addition, we generated
`reproducible polymerrdye tandem structures by controlled
`incorporation of orthogonal conjugation sites. These unique
`tandem dyes benefit from the collective energy transfer beha—
`vior of the underlying polymer structure esulting in amplified
`signal response from the acceptor (15).
`Thus, this article reports the production and performance
`of various antibody conjugates of RVLlZl and demonstrates
`their utility in multicolor flow cytometry and fluores :ence mi-
`croscopy. we show that BV/lll is as bright, and often signifi-
`cantly brighter, than currently available dyes and is suitable for
`intracellular staining. In addition, we describe the first in a se-
`ries of novel tandem reporters produced for violet laser excita~
`tion, BV57C. Vv’e characterize the performance of these dyes in
`a variety of cytometric and imaging applications. lmportantly,
`BV421 is the first new fluorescent probe rivaling PE and APC
`in useful brightness, and represents a new class of reporters
`
`457
`
`BD00000092
`
`2
`
`

`

`EN §MA£A§€¥§
`
`ifi
`
`(developed from a novel chemistry), providing an important
`new tool in multiparameter analysis.
`
`METHODS
`
`Flow Cytomo‘trv
`antibody
`fluorochronte—conjugated
`The
`following
`reagents were obtained from Biolegend (San Diego. CA;
`clones indicated in parentheses): mouse anti—human CD3
`BV‘lZl, PB, and PE (l7Al): CD8 BVélZl, 3‘7570, BVfilZ, PE,
`and PE (RPA—TS); CCR'," BV421 and PB (41312); CD95 PE
`(DXE); CD127 BV421 and PB (servos); IFNy BV421, PB,
`and EVE/'0 (4333); and CD154 BV421 (24—31). Additionally,
`the following. antibody conjugates i 'ere obtained from various
`vendors, or conjugated. iii-house, as indicated: 11.2 Alexa/'00"
`APC (11D Biosciences, San lose, CA), IFNy FlTC (13D), TNF
`Cy7PE (BD), CCil7 Alexafittf) (in-house), CD/tSRA QDBSS
`(in-house),
`(117127 CySPii
`(Beckman Coulter, Miami,
`171.),
`CDFJ QD800 (in-house), (ID/15110 C)y7APC (in-house), CD4
`FlTC (in-house), CD14 QD545 (in-house),
`(11319 QD545
`(inrhouse), Aqua live/dead fixable dye (1nvitrogen), and CD25
`APC (BD). Peptide~lVlHC Class I (le—lCl) multitners were
`produced by
`combining
`st ‘eptavidinvAPC (1nvitrogen),
`~QD655 (lnvitrogen), or ~BV421 (BioLegend) with biotinyl—
`ated HLA—AZ molecules loaded with a peptide derived from
`the pp65 protein of CMV (NLVPi‘x/l‘iUtTV), as described pre—
`viously (7). Samples were analyzed on a twenty parameter
`LSR—ll equipped as described previously (7); relevant filter
`combinations are described. in the text and Tables 1 and 3.
`
`Sample Preparation for imaging Experiments
`Amnis ImageStream. (:r'yopreserved human peripheral blood
`mononuclear cells (PBMCs) from healthy donors were thawed
`and rested overnight at 370C in complete medium (RPMl--
`1640, 10% fetal bovine serum (PBS), l% penicillin/streptomy--
`cin and 1% L—glutaniine). Cells were washed with phosphate—
`buffered saline (PBS) and imrnunostained for anti—EDS,
`la—
`1 eled with either BV/lll or PacBlue (Biolegend).
`in FACE-
`buffer (PBS, 1% BSA and 0.1% sodium azide) for 50 minutes
`at room temperature. Cells were washed with FACS buffer and
`fixed in 1% parafornialdehyde. Immunostained cells were ana—
`lyzed on an Aninis lrnageStreamX imaging flow cytometer
`configured for lZ—channel acquisition.
`
`imaging. Tissue was obtained from Z--month--old
`(Ionfocal
`mice that had been deeply anesthetized with 120 mg/kg
`sodium pentobarbital (i.p.) and perfused intracardially with
`0.9% saline followed by 4% paratormaldehyde in 0.l M
`sodium phosphate buffer (pl—i 7.2 at 200C). The procedures
`conforn‘ted to the NIH Guide for the Care and Use of labora-
`
`tory Animals and were conducted under authorization by the
`Institutional Animal Care and Use Committee at UCSB.
`
`Brains were dissected from the skull following a 15—min post—
`fixation step, and were prepared for irnniunolluorescence by
`100 um vibratonie sectioning. Tissues were protein blocked
`with a solution of 5% normal goat serum in PBS pH 7.2 and
`1% Triton—)4 for 3 h and then rinsed with PBS. A mouse
`
`458
`
`monoclonal antibody to beta. [ll-tubulin (1:200; No. T4026;
`Sigma, St. Louis, MO) or mouse monoclonal antibody to neu--
`rofilaments NF70 (1:1,000; #‘NJABlGlS; Millipore, Billerica,
`MA) were used to detect beta lIl-tubulin or neurolilaments,
`respectively. Primary antibodies were detected using BV421 or
`Dvliighttt05 coniugated goat anti-mouse lgGs (laclzson La is,
`‘West Grove, PA). Some sections were then labeled for Glial
`Fibrillar'y Acidic Protein (GPAP) with a GUS—conjugated
`mouse monoclonal antibody (H.000; No.
`(19205; Sigma)
`and/or F—actin detected with Oregon Green 488 phalloidin
`(stock diluted 1:200; No. 07466; lnvitrogen, Carlsbad, CA).
`For multichannel confocal images nuclei were counterstained
`with ToPro—3 ( 1:1,000; No. T3605 lnvitrogen) and then rinsed
`with PBS. Sections were mounted on slides in phosphate
`buffer and covered with No. 1.5 thickness cover glass.
`
`(127,16 weeks of age) male
`imaging. Adult
`Three—color
`()57Bl.6] mice were killed with (303, and the eyes were enu—
`cleated and placed in 61% pau‘afortnaldel'iyde in sodium phos—
`phate buffer (0.1 M; pH 7.4). After overnight fixation,
`the
`cornea, and lens were removed from the eye at which time the
`retinas were gently peeled from the underlying pigment
`epithelium in preparation for the imniunohistochemistry, as
`follows. The retinas, processed as flatmo unts, were first rinsed
`in PBS for a minimum of l h and subsequently incubated in
`blocking serum (normal donltey serum 1:20 in PBS, 0.5% bo-
`vine serum albumin [BSAla 0.1% Triton X-—100, and 0.1% az-
`ide [PB'l‘Ali overnight on a rotator at 40C. Primary antibodies
`in P BTA were added the following day and placed on a rotator
`overnight at 43C. Probes to the following proteins were used:
`anti-glial fibrillary acidic protein (GFAP) anti-neurofilament,
`and biotinylated—lsolectin B4. The following day, primary
`antibodies were rinsed in PETA and corresponding secondary
`antibodies were added and incubated overnight on a rotator at
`4°C. Atiti—GFAP (green) labels retinal astrocytes, anti—neurod
`lament (red) labels ganglion cell axons and the Isolectin B4
`(blue)
`labels blood vessels. On the last day, sections were
`rinsed in PETA and mounted in 5% n—propyl gallate in glyc—
`erol on glass slides and sealed with a coverslip.
`
`Depth penetration. A 1.5"an cube of brain tissue incubated
`in primary antibody and for 3 days, rinsed and incubated in
`the BV/lll conjugate for 24 11 showed that the BV421 pene~
`trated deep into the fixed tissue. The tissue was subsequently
`sectioned co ronally by first removing 500 [on from the stained
`surface and then sectioning Million sections.
`
`imaging
`The Amnis ImageStream X was configured for IZ-Chan-
`nels of simultaneous fluorescence detection and contained
`
`four excitation lasers: 405 mu, 488 nm, 658 nm, and 785 nm.
`Without BV/lZl, the 405 nm laser was under utilized clue to a
`lack of ivailable fluorophores, as Pacific Blue. Pacific Orange,
`and Quantum dots are either inefficiently detected or have sig~
`nificant overlap with other channels. As such, the 405 nni laser
`was typically used only for nuclear stains such as DAPI. Data
`were collected with all channels open to detect any crossover
`
`Brit/[ant Violet Dyes for Fluorescence Applications:
`
`BD00000093
`
`3
`
`

`

`g or
`
`iNfi’afi. &§§"§‘§ .fifi
`
`Table 1. Spectral properties or” BV421 compared to P8
`
`(opens)
`yam-m (onsss‘
`rooms)
`was-510 (onsoo;
`
`:2 ic-
`BV/tZl
`4
`0.11
`0
`0
`0
`PB
`0.3
`0.3
`0
`0
`
`L2)n—
`
`ts.
`
`s{
`
`
`
`PB
`
`BV42 l
`PB
`
`0
`OJ,
`
`(1600-2
`
`
`
`BV42 l
`
`
`
`Values represerttthe percent spillover for each fluorochrome (BV42’l or PE) into spectral regions assigned for detection of otherfluor-
`ochromes (e.g., 00545). These other fluorochromes are arranged first by the laser used for excitation (violet (V), blue (E3), green (G), or red
`(Hi), and then by the optimal wavelengths (hm) for detection. Thus, 2% or" the BV42‘! signal can be detected in the 533577557 hm region o
`the spectrum used for detection ofODSdSflthe ”compensation” for BV421 into the QD545 channel is 2%.
`
`between lasers or channels. with channel 7 being the primary
`B‘s/‘42} and Pacific Blue detection channel. Confocal images
`were obtained using an Olympus Fluoview 500 Confocal
`Microscope equipped with a 40X oil
`immersion objective
`(NA. 1.30) and four confocal lasers (405, 488, 543 and 633
`nm). Multichannel confocal images were collected with a line
`sequential scan for maximal spectral separation. Comparisons
`of photoblcaching were made using the same laser dwell—time
`and using the 405 nm diode laser and. i% laser transmittance.
`The photomultiplier detector sensitivity t'as decreased from
`651 for DyLithOS to 581 for the BV421.
`images were col-
`lected as 512 X 51?. pixels sampled at 8.0 its/pixel in 12 bit
`gray. Multiphoton images were collected with an {:Dlympus
`Fluoview 1000i [PE equipped with a 25X (NA. 105) water
`immersion-dipping objective and tunable Mai Tai I-{P feinto-
`second pulsed laser. The laser was tuned between 700 nm and
`1.000 nm for imaging and the Violet/Green filter for the exter~
`nal detector was employed (Dix/i485, BAAlIZtl—ZLEO, BA495~
`SAlOHQ‘). Images we e collected as 512 X 512 pixels sampled at
`8.0 LiS/pixel in 12 bit gray. Three—color imaging was performed
`with a laser scanning confocal microscope (Firoiew 50");
`Olympus, Center Valley, PA). Each image represents a projec—
`tion of four z—planes collected at 0.5 ,um intervals. During the
`image acquisition time, which lased approximately 1 min, no
`photobl caching was observed with any of the tluorochromes.
`
`Resorts
`
`Physical and Spectral Properties of
`Brilliant Violet Dyes
`An advantage of fluorescent polymer probes is that both
`their physical and optical properties can be modified in a well—
`defined way, often by altering the structure of the monomer
`subunits before polymerization. For application in flow cyto—
`metry, high solubility and low nonspecific binding are critical
`factors. to address these criteria, a range of pendant side
`
`Cytornetiy Part A 9 81A: AEFASB 2012
`
`chains were screened. beginning with various combinations of
`cationic and/or anionic groups. to evaluate the properties of
`the 'esulting polymer. However, many of these structures \ 'ere
`prone to aggregation, which ultimately reduced fluorescence
`and resulted in unacceptable levels of nonspecific binding to
`mammalian cells. Through further synthetic manipulation we
`arrived at a neutral polymer structure with water solubility in
`excess of 50 mg/mL, which exhibited no appreciable nonspeci—
`fic interaction with cultured lurltat cells in initial screening
`assays. The resulting material, BV421, is intensely photolumi--
`escent with a quantum efficiency ot‘rj9% .-_t-_ 3% in a range of
`typical flow buffers (PBS, DPBS. etc.) and has an exceedingly
`short fluorescence lifetime (<1 ns‘). Through control o‘fthe po--
`lymeriza‘tion process and subsequent purification, we deter--
`mined a molecular weight range that afforded a very large
`extinction coefficient (2.5 X 106 MW] ciif 1) and exhibited (ac--
`ceptable reproducibility in optical performance. For co mpari—
`son, Pacific. Blue (l19 B), a single molecule dye with near maxi—
`mum absorption at 405 rim. has an extinction coefficient ofrlb
`/ 104 MW?" cm "1, a quantum efficiency of 78%, and a longer
`lifetime of 9 ns. Practically, the brightness ofa fluorochrome is
`proportional to the extinction coefficient and quantum effi—
`ciency, and inversely related to the fluorescence lifetime.
`The structure of }3V42,J, also specifically controls for the
`number and position of functional sites available for covalent
`attachment to antibodies. This numerical and spatial consis-
`tency resembles that found in classic organic dyes and is in
`stark contrast to other macromolecular fluorochromes such as
`
`PE or QDs, whose properties complicate conjugation and
`quality processes. importantly, these functionalities are come
`patible with a variety of procedures routinely used for the con-
`jugation of iluorochromes to antibodies, or for conjugations
`to other bio—molecules like streptavidin or oligonucleotides.
`Figure 1A provides the absorption and emission spectra
`of the BV421 polymer. showing peak absorption at 407 nm
`and pealc emission at 421 nm. 'l‘hese spectra align well with
`
`459
`
`BD00000094
`
`4
`
`

`

`RI
`
`INAL ARTI HE
`
`A
`
`hue
`
`.3
`
`Ame
`
`ho a
`
`250
`
`450
`350
`Wavelength (nun)
`
`550
`
`650
`
`.1l
`
`600
`500
`ml]
`Wavelength (nm)
`
`700
`
`samples with serial dilutions of each antibody. Figure 2A
`shows that CD8—BV421 exhibits a remarkably bright signal,
`well separated from background, compared to PB. For come
`parison, cells were also stained with CD8 conjugated to PE,
`the brightest fluorochrome currently available.
`Figure 2B plots the median fluorescence intensity of the
`CD84r and CD87 populations for each of the conjugates, along
`with the staining index (SI). The latter metric (l6) compares
`detection sensitivity among fluorochromes, which may be
`impacted differently by non—specific binding or cellular auto—
`fluorescence (i.e. “background”), despite similar strength of
`positive signal. For display, reagent dilutions were normalized
`to the calculated binding constant (KA) to overlay by the effec—
`tive antibody concentration. This allows for direct comparison
`of the reagents. When stained with anti—CD8 BV421, CD84r
`cells are more than 50X brighter than those stained with the
`PB conjugate (left panel), while CD87 cells exhibit similar
`background staining (middle panel). As a result,
`the SI of
`BV421 is much higher than PB (right panel), and comparable
`to PE (Fig. 2B). As shown in Supporting Information Figure 2,
`similar
`results were obtained by detailed comparison of
`anti—CD3 conjugates (without normalization).
`We ensured the specificity of the BV421 conjugate by
`co—staining cells with BV421—CD3 and PE—CD3. We observe
`complete correlated corstaining (not shown), indicating that
`there was no substantial non—specific binding, and all antigen—
`bearing cells are identified by the reagent.
`To directly test the sensitivity of BV421, we conjugated it
`to antibodies against cellular proteins that stain dimly or are
`characterized by a continuum of expression (from dim to
`bright). Figure 3A shows that the higher staining index of
`BV421 allows for far better resolution of CD127 expression
`compared to the PB conjugate of the same antibody (Fig. 3B).
`Similarly, when CCR7 expression is measured using a BV421
`conjugate, central memory T—cells
`(which express CCR7
`dimly) can be easily resolved fiom CCR7’ memory/effector
`T—cells (Fig. 3C). Like PB, spillover of BV421 into the channels
`used to detect other fluorochromes is minimal; the maximum
`compensation value is 2% into the channel we use to detect
`QD545 (Table I). These properties illustrate the high sensitiv—
`ity achieved with BV421. In fact, when the staining indices of
`various fluorochromes are compared (Table 2), EV421 ranks
`second only to PE, with a staining index comparable to APC.
`
`Evaluation of BV570: Comparison to QD585 and PE
`To examine the fluorescence characteristics of BV570, we
`conjugated the tandem polymer to a mouse anti—human CD8
`antibody and stained PBMC for analysis on a flow cytometer
`(equipped with a 50 mW Violet laser, a 570 nm long—pass filter,
`and a 585/42 bandpass filter). Supporting Information Figure
`3A compares this reagent to QD585 and PE conjugates of the
`same antibody. The staining pattern of the BV570 tandem is
`similar to the QD585 reagent, but with a mildly higher SI at
`high antibody concentrations (Supp. Info. Fig. 3B). Notably,
`residual signal from the donor dye (BV421) is low, and typical
`of that observed for other tandem dyes.
`
`Brilliant Violet Dyes for Fluorescence Applications
`
`BD00000095
`
`B
`
`hue
`
`Eta/V
`
`
`
`SignalIntensity,arbitraryunits
`
`
`
`
`
`hu e
`
`300
`
`Figure 1. Excitation and emission spectra for BV dyes. (A) Sche—
`matic illustrating polymer structure of BV421, as conjugated to
`antibody. Energy from a violet laser (hva) excites electrons in the
`polymer and blue light (hve) is emitted. Excitation (purple) and
`emission spectra (blue) for BV421 are also shown. (B) Schematic
`illustrating structure of BV57O tandem dye, as conjugated to
`antibody. The BV421 donor is represented by the repeating poly—
`mer, while the Cy3 acceptor is represented by the orange unit.
`Energy from a violet laser (hva) excites electrons in the polymer
`(donor) and energy is transferred Cy3 (the acceptor); Cy3 then
`emits yellow/orange light (hve). Excitation (purple) and emission
`spectra (orange) for BV57O are also shown.
`
`standard flow cytometer instrument configurations commonly
`used for Cascade Blue and Pacific Blue.
`
`By introducing primary amine groups into the polymer
`structure and covalently attaching organic fluorochromes, we
`also generated a family of Brilliant Violet tandem dyes. Sup—
`porting Information Figure I shows the emission of various
`BV tandems (at different stages of development) which span
`the Visible spectrum and well into the near IR. The resulting
`tandems, of which BV570 is the first example (Fig. 1B), have
`important advantages over conventional phycobilliprotein
`tandems. First, the number of sites for acceptor dye attach—
`ment
`is controlled synthetically. Therefore, BV tandems
`exhibit far less lot—to—lot variation in the ratio of donor to
`
`acceptor dye, and thus less spectral variability, than PE and
`similar protein based tandems. Because the number of dye
`attachment sites is fixed, repeated optimization of dye stoichir
`ometry is not necessary.
`
`Evaluation of BV421: Comparison to PB and PE
`To compare the staining characteristics of BV421, PB,
`and PE, we conjugated a mouse anti—human CD8 monoclonal
`antibody to each of these dyes, and stained parallel PBMC
`
`460
`
`5
`
`

`

`CD8 BV421
`
`CDB PacBlue
`
`QRIQ I NAL ARTIQLE
`
`A
`
`1.3E07
`Antibody Dilution Factor —)
`
`200 5,5EDG
`
`200
`
`12800
`
`2.5
`
`
`
`
`
`
`B
`
`MFI Positive
`
`MFI Negative
`
`105
`
`10‘
`
`101
`
`1o=
`
`105
`
`10‘
`
`10°
`
`z
`10‘
`
`BV421
`PacBlue
`PE
`
`10'3
`
`10"
`
`1D"
`
`10“
`
`10'
`
`102
`
`10'3
`
`10'2
`
`10"
`
`10“
`
`10‘
`
`10}:
`
`102:
`
`10';
`
`100‘E
`
`10-"
`
`SI
`
`10'1
`
`10”
`
`10"
`
`10“
`
`10‘
`
`102
`
`NwmalizedAnnbodyOonmflafion
`[uL antibody! 100m. calm/K
`
`3
`
`Figure 2. Comparison of BV421, PB, and PE signal. (A) Staining patterns for BV421 (left), PB (middle), and PE (right) conjugates of anti-
`human CD8. (B) Median fluorescence intensity (MFI) of CD8+ population (left panel) and CD8’ population (middle panel). Comparison of
`SI for BV421, PB, and PE conjugates (right panel). The X—axis shows normalized antibody concentrations, relative to the binding constant
`(KA) calculated for each reagent.
`
`Table 3 compares the spillover characteristics for the
`BV570 tandem polymer and QD585. Residual signal from
`BV421 (in the Violet 450 channel) is minimal (7%, Table 3).
`The emission spectrum of BV570 is broader than that of
`QD585, leading to more overlap into the neighboring QD565
`detector. This has little practical consequence, however, since
`QD staining panels also do not typically include both QD565
`and QD585 at the same time. In addition, BV570 has more
`fluorescence in the red and farrred regions of the spectrum
`than QD585. Again, this probably has little consequence, since
`QD655, QD705, and QD800 are quite bright and therefore
`less susceptible to “spreading error.“
`
`Application of BV Fluorochromes to Common
`Flow Cytometry Assays
`We next explored various applications for BV reporters,
`which might offer new staining options beyond what is cur—
`rently possible. For example,
`intracellular cytokine staining
`assays cannot typically take advantage of the unique fluores—
`cent properties of QDs. This is because cytokine reagents, con—
`jugated to QDs, are not available commercially and in—house
`conjugates exhibit unacceptable background staining (often
`because free QD remaining from the conjugation cannot be
`efficiently removed from the conjugate). We postulated that
`
`Cytometry PartA 0 81A: 4567466, 2012
`
`BV fluors were less likely to exhibit these problems, and tested
`this by staining SEE—stimulated cells with anti—human IFNy
`BV421 and BV570. Indeed, Supporting Information Figure 3C
`demonstrates strong antibody signals from both the BV421
`and the related BV570 tandem.
`
`In addition, we tested the stability of BV421 in the setting
`of co—culture assays, which are necessary to optimally detect
`antigen—specific CD44r and CD84r T—cells that express CD154
`and CD107a expression (17
`19). In our previous study, we
`noted that only PE and APC conjugates of CD154 remained
`fluorescent after remaining overnight with stimulated T—cells
`in a PBMC culture; fluorescence was abrogated for CD154
`conjugates of Alexa dyes, QDs, and FITC (18). Supporting
`Information Figure 4 demonstrates the strong signal achieved
`with BV421 CD154 in this assay, which suggests that BV
`reporters are resistant to endosomal degradation (the primary
`means by which surface CD154 is recycled). The co—expression
`of various cytokines with CD154 is consistent with the litera
`ture describing CD154 as a global marker of CD44r T—cell acti—
`vation, and demonstrates the specificity of the BV421—labeled
`antibody in the co—culture setting.
`Finally, we reveal the value of BV421 as the fluorochrome
`in peptide MHC—class I (pMHC—I) multimers, used for direct,
`ex vivo detection of antigen—specific CD8+ T—cells. Figure 4A
`
`461
`
`BD00000096
`
`6
`
`

`

`RI
`
`INAL ARTI
`
`lE
`
`CD127 BV421
`
`CD127 Pacific Blue
`
`
`CCR7
`
`
`121028
`
`1:80
`
`1:10
`
`1:1028
`
`1:80
`
`1:10
`
`COR? BV421
`Naive
`
`1:1028
`
`1:80
`
`1:40
`
`Antibody Dilution
`
`Figure 3. BV421 conjugates allow better resolution of dim markers. (Al Anti-human CD127 BV421 exhibits a bright signal allowing clear
`discrimination of positive and negative populations. (B) Anti-human CD127 PB exhibits nearly one log less fluorescence than BV421 conju-
`gate. (C) Bright, dim, and negative levels of CCR7 expression can be distinguished clearly with the BV421 conjugate, allowing resolution of
`naive, central memory, and other memory cell subsets.
`(D) CCR7 expression on T—cell subsets. Note that central memory (CD45RA’
`CCR7+) cells express lower levels of CCR7 than naive (CD45RA+ CCFi7+) cells; thus, efforts to identify T—cell subsets with CCR7 require very
`bright fluorochromes.
`
`shows the staining patterns for pMHCI multimers (directed
`against CD8+ T—cells specific for the NLV epitope of CMV)
`produced in parallel from SAV—BV421, SAV—APC, and SAV—
`QD605. pMHCI multime