AmCyan was originally derived from the fluorescent protein amFP486. AmFP486 was cloned from the sea anemone Anemonia majano (GenBank accession number AF168421), and belongs to the family of fluorescent proteins (FPs) isolated from coral reef organisms (1); green fluorescent protein (GFP), derived from the jellyfish Aequorea victoria, is its most famous representative. These GFP-like proteins have different colors, depending on their absorbance and fluorescence emission spectral profiles, which make them excellent tools to simultaneously track multiple proteins in live cells and to study protein–protein interactions. AmCyan displays about 29% identity and 40% similarity in amino acid sequence with GFP. Both proteins have similar chromophore precursors: S-Y-G in GFP and M-Y-G in AmCyan. During post-translational modifications, the precursor undergoes spontaneous internal cyclization and dehydration of the central tyrosine residue and the mature protein becomes fluorescent. Molecular oxygen is required to complete the M-Y-G cyclization.
Both amFP486 and AmCyan have a unique barrel-like structure which consists of 11 β-strands (2). The dimensions of the barrel are 24 × 42 Å. The entire structure is stabilized by hydrogen bonds, which makes it resistant to proteolysis and denaturation. The fluorophore is positioned inside the barrel and surrounded with the hydrophobic residues. This fluorophore-containing cavity protects the fluorophore from quenching by water.
AmCyan is one of the existing violet fluorochromes for flow cytometry analysis of human, mouse, and nonhuman primate leukocytes. In 2004, BD introduced several AmCyan violet reagents for cell analysis. AmCyan has a broad excitation spectrum with a peak at 458 nm that can be excited by the violet laser at 405 nm and, partially, by the blue laser at 488 nm. AmCyan has an emission peak of 489 nm, which is close to the emission peak of FITC at 519 nm. Emitted fluorescence at 489 nm from AmCyan after blue laser excitation is recognized as FITC fluorescence in the FITC channel, which hence requires significant compensation. This leads to limitations in the use, since its use can result in decreased resolution of dim signals in the FITC detector.
Based on these spillover and resolution issues, our key objective was to change the excitation profile of AmCyan to significantly decrease spillover into the FITC detector. To achieve this, we decided to narrow the excitation spectrum of AmCyan and/or shift it to the violet spectral region.
MATERIALS AND METHODS
All monoclonal antibodies were from BD Biosciences, CA. The following conjugates were used: CD3 (clone SK7) labeled with APC and with BD Horizon V450 (V450); CD4 (clone SK3) labeled with FITC, V450, BD Horizon V500 (V500), PerCP Cy5.5 and AmCyan; CD8 (SK1 clone) labeled with V500 and AmCyan; CD27 (clone L128) labeled with APC; CD45 (clone 2D1) labeled with PerCP Cy5.5; CD45RA (clone L48) labeled with FITC; TNF-α (clone 6401.1111) labeled with FITC; IL-2 (clone 5344.111) labeled with PE; IFN-γ (clone B27) labeled with V450.
AmCyan, amplified from the vector pVEXHN-AmCyan-K6 (Nature Technology, Lincoln, NE), was used as the template for mutagenesis. QuikChange II site-directed mutagenesis kit (Stratagene, La Jolla, CA) was used to introduce site-directed point mutations.
GeneMorph II random mutagenesis kit (Stratagene, La Jolla, CA) was used for random mutagenesis in conditions optimal for six mutations per 1,000 bp. The library of 700,000 AmCyan mutants was created and used to transform E. coli competent ElectroMAX DH5α-E™ cells (Life Technologies, Carlsbad, CA). Transformed cells were aliquoted, frozen in liquid nitrogen and stored at −80°C upon screening.
DNA sequencing was provided by Sequetech (Mountain View, CA). DNA oligos were synthesized by Integrated DNA Technologies (Coralville, IA).
Screening of Mutants
Aliquots of frozen cells from the libraries of random mutants were thawed and plated on LB/agar Petri dishes supplemented with 50 μg/ml carbenicillin. A 400-nm LED flash light was used as an excitation light source. Colonies with more intense fluorescence were picked and cultured in 96-well plates in 250 μl per well of LB medium containing 50 μg/ml carbenicillin at room temperature (RT). Protein expression was started by a 0.5 mM IPTG. After incubation at RT, the full emission spectra were measured with SpectraMax M5 plate reader (Molecular Devices, Sunnyvale, CA) after excitation at 405 nm.
Protein expression was induced in E. coli BL21 cells by adding a 0.5 mM IPTG to a bacterial culture grown at 37°C in LB medium to an optical density of OD600∼0.8. After overnight incubation at RT, the bacteria were harvested and re-suspended in 150 mM NaCl, 1 mM β-mercaptoethanol, 50 mM Na phosphate buffer (pH 7.4). Cells were disrupted ultrasonically followed by a 30-min centrifugation at 40,000g. AmCyan or its mutants were purified on a Ni-NTA agarose (QIAGEN GmbH, Hilden, Germany) and eluted with 250 mM imidazole. The proteins were further purified by size exclusion chromatography on Superdex 200 10/30 (GE Healthcare Bio-Sciences AB, Uppsala, Sweden) column in a 150 mM NaCl/ 50 mM Na phosphate buffer (pH 7.4).
Fluorescence quantum yields were determined as described earlier (3) with quinine sulfate (Sigma-Aldrich, St. Louis, MO) as a reference standard, and literature values (0.51 in 50 mM sulfuric acid) were used for quantum efficiency (4). All readings were taken at ambient temperature. AmCyan fluorescent protein and its mutants were serially diluted in 25 mM Na phosphate buffer/ 150 mM NaCl with A375 ≤ 0.05, and quinine sulfate was serially diluted (with A375 ≤ 0.1) in a 50 mM sulfuric acid. Light absorbance at defined wavelengths and fluorescence emission spectra were registered in 96-well plates on a SpectraMax M5 microplate reader (Molecular Devices, Sunnyvale, CA). Data were plotted as integrated fluorescence intensity versus correspondent absorbencies and analyzed using Microcal Origin 7.5 software (OriginLab, Northampton, MA).
Absorbance spectra were recorded in 1 cm cuvettes on an Agilent 8453 UV–Visible spectrophotometer (Agilent Technologies, Santa Clara, CA). Molar extinction coefficients of the chromophores were calculated based on the theoretical protein sequences from the corresponding molar extinction coefficients at 280 nm using ExPASy ProtParam tool (5). Fluorescence emission spectra were recorded at ambient temperature on a Fluorolog-3 spectrofluorometer (HORIBA Jobin Yvon, Edison, NJ) in 1 × 1 cm2 cuvettes.
Preparation of AmCyan100-mAb Conjugates
Monoclonal CD4 or CD8 antibody was buffer exchanged into labeling buffer on spin columns (Bio-Spin 30, Bio-Rad Laboratories, Hercules, CA) and concentrated to 2–10 mg/ml, if necessary. SMCC modified AmCyan or AmCyan100 mutants were incubated with antibodies. Labeling was done for 60 min at ambient temperature. A reaction mixture was separated on a Bio-Spin column and analyzed for dye-to-protein ratio (D/P) and antibody concentration spectrophotometrically.
CD4 (clone SK3) and CD8 (clone SK1) AmCyan100-mAb conjugates were tested by flow cytometry. Both antibodies were titrated for surface and intracellular staining by staining cells with serial concentrations ranging from 7.5 ng to 1 μg/100 μl of cell suspension.
For surface staining, cells were stained following a lyse-wash protocol. Briefly, 100 μl of whole blood were stained in 5 ml tubes for 25 min at RT in the dark and then treated with 1 ml of BD FacsLysing solution (BD Biosciences, San Jose, CA) for 10 min in the dark. Cells were then washed using PBS/0.5% BSA/0.1% NaN3 (for now on referred as wash buffer). For intracellular staining, peripheral blood mononuclear cells (PBMCs) were isolated from BD Vacutainer CPT cell preparation tube with sodium heparin (BD, Franklin Lakes, NJ) and washed with wash buffer. Cells were then treated with 1 ml of 1x BD FACS Lysing solution for 10 min, washed, and then treated with 1 ml of 1x BD FACS Permeabilizing Solution 2 (BD Biosciences, San Jose, CA) for 10 min. Once cells were permeabilized and washed with wash buffer, mAbs were added to the cell suspension and incubated for 25 min at RT. Finally, cells were washed with wash buffer, resuspended in wash buffer or 1% paraformaldehyde (PFA), and then analyzed on a standard BD LSRII, BD Canto II or BD FACSVerse flow cytometers. The BD LSRII and Canto II were fitted with a 488nm Argon laser, Red HeNe 633 nm diode and violet diode 405 nm with BD FACSDiva Software v.6.1.3. The BD FACSVerse was fitted with a 488 nm Argon laser, Red HeNe 640 nm diode and violet diode 405 nm with BD FACSuite Software v.1.01 The filters used to detect AmCyan100 were the same across the three instruments: a 500-nm long pass dichroic mirror, and a 528/45 nm bandpass filter.
Instrument setup was performed on the standard LSRII and Canto II by optimizing the PMT voltages for each detector so that the measured standard deviation of unstained cells had less than a 10% contribution of electronic noise and the positively stained populations were within the linear range of the detector. Both of the standard deviation and the maximum linearity of the detector were assessed by using cytometer setup and tracking (CS&T) fluorescent beads (BD Biosciences, San Jose, CA) and the CS&T software module available on all digital flow cytometers running BD FACSDiva software. Instrument setup on the BD FACSVerse was performed by running CS&T beads in BD FACSuite software, and using the default lyse-wash reference settings for AmCyan.
The conjugates were tested individually and in different multicolor flow cytometry assays. Analysis was done using BD FACSDiva software or BD FACSuite software and stain indices (SIs), a measure of resolution of the fluorescence conjugates, were calculated by applying the formula: SI = D/W, where D = MFI positive population – MFI negative population, and W = 2 × rSD (robust standard deviation negative population) (6).
RESULTS AND DISCUSSION
Quantum Yield of AmCyan
This parameter describes the capacity of AmCyan fluorescent protein and its mutants to capture and subsequently release light. The quantum yield (QY) of AmCyan is fairly comparable with quantum yields of other fluorescent proteins. We determined that the QY of AmCyan, the N34S/K68M mutant of amFP486, is equal to 0.42, which agrees well with data from Lukyanov's group for the K68M mutant of amFP486 (7).
AmCyan Mutant as a Starting Point for Random Mutagenesis
AmCyan Y69 mutants
Our key objective was to narrow the excitation spectrum of AmCyan and to shift this spectrum towards the ultraviolet. We considered two point mutations in fluorescent proteins that could potentially change AmCyan's photochemical properties. We first replaced a tyrosine residue in the fluorophore-forming tripeptide X-Y-G with a tryptophan or histidine residue, which is known to cause a blue shift in GFP (8). Two Y69 mutants of the amFP486 were reported earlier (9); the mutant Y69W had excitation maxima at 412 nm and a comparatively small shift of emission maximum up to 475 nm, the other mutant, Y69H, was reported as not fluorescent at all. In order to prove this concept, we made and characterized three point mutants of AmCyan: Y69F, Y69H, and Y69W. Only the Y69W mutant had a remarkable fluorescence. The other two mutants appeared to be dim. Absorbance maxima of all mutants changed their position to 354 nm for Y69F, 378 nm for Y69H, and 413 nm for Y69W (Fig. 1A). Their emission peaks were blue-shifted too, with maxima at 448 nm for Y69F, 458 nm for Y69H, and 472 nm for Y69W (Fig. 1A).
In addition to the Y69 mutation in the fluorophore, we introduced another point mutation—a substitution of W145 in AmCyan with phenylalanine. It is known that Y145F substitution may increase the fluorescence and accelerate chromophore formation in a blue-fluorescent variant of GFP (8). Indeed, the double mutant Y69W/W145F had higher light absorbance (21.3 mM−1 cm−1 vs. 19.9 mM−1 cm−1 per monomer) and a better A405/A488 ratio than Y69W single mutant. However, a distinctive property of all analyzed Y69 AmCyan mutants was low quantum yields (not more than 0.05) and, as a result, low brightness.
AmCyan S148 mutants
An alternative approach to shifting the excitation peak of AmCyan to the violet spectral region involves changing the protonation state of the phenolate group, which determines the spectral properties of the fluorophore center. Analysis of the amFP486 chromophore environment leads to a conclusion that S148 mutation may affect an anionic character of the phenolate (2, 10) and change its fluorescent properties. In this case, the Y69 remains unchanged. To prove this hypothesis, we made a set of S148 mutants, where the serine residue was replaced either with histidine, phenylalanine or tryptophane. Those mutants were expressed in E.Coli and purified. Light absorbance and fluorescence emission spectra of S148H, S148F, and S148W are summarized in Figure 1B. S148H and S148W mutants had absorbance spectra shifted from 458 to 392 nm and 385 nm, respectively (Fig. 1B). These mutants had a remarkable fluorescence and relatively broad Stokes shifts—99 and 121 nm. Light absorbance of the S148F mutant was very low and it appeared to be dim. The S148W mutant had two absorbance peaks—one with a maximum at 385 nm, and another with a maximum at 475 nm. The clone exhibiting the most promise for cell analysis applications had an insert of seven amino acid residues (W148-INS) and had an absorbance maximum at 393 nm and an emission maximum at 495 nm (Fig. 1B).
We found out that most of the tested AmCyan W148 mutants had good expression levels in E.Coli and most were soluble and stable in solution. The S148W mutant with the insert W148-INS yielded the highest quantum yield of 0.25. Therefore, we choose it as a template for a new library of random mutants.
Random Mutagenesis—AmCyan100 Variant
In order to increase the brightness of the new blue-shifted S148W mutant with the insert, we utilized the random mutagenesis approach. Error-prone PCR was used to generate a library of genetic variants. This library was used to transform E. coli competent ElectroMAX DH5α-E cells. We screened and picked the colonies that yielded the highest fluorescent signals when illuminated at 400 nm and cultured them in 96 well plates. Protein expression was then initiated with IPTG. After incubation at room temperature, the full emission spectra were measured with a SpectraMax M5 plate reader and analyzed. Finally, the brightest mutant, AmCyan100, was selected. Besides the insert and S148W mutation, it has four other mutations. It appeared that the AmCyan100 mutant has a Stokes shift much longer than the original AmCyan—about 105 versus 31 nm (Fig. 2). The AmCyan100 fluorescence quantum yield is lower than that of AmCyan (0.25 vs. 0.42, respectively). Nevertheless, taking into account that brightness is a function of fluorescence quantum yield and molar extinction coefficient, when excited at 405 nm AmCyan100 is brighter than AmCyan (6.0 vs. 5.6 per monomer, respectively). Importantly, it has a very low fluorescence when excited at 488 nm. The latter characteristic makes it the best-suited label for multicolor applications (i.e., it can be excited by the violet laser and has almost no spillover into the FITC channel as a result of blue laser excitation).
AmCyan exists in solutions as a tetramer protein. It appeared that altering the amino acid residues around S148 changed the oligomeric state of the protein. Indeed, using size exclusion chromatography, we determined that AmCyan100 has a dimeric structure. Mutagenesis and insertion of extra amino acid residues most likely modifies a region on the surface of the protein molecule responsible for the interaction between two AmCyan dimers, thus preventing formation of tetramers.
Performance of AmCyan100-mAb Conjugates in Flow Cytometry
In order to evaluate flow cytometry performance, we first titered the CD4 and CD8 AmCyan100 mAbs for both surface and intracellular staining. For surface staining an optimal titer, determined by highest SI, of 125 ng (CD4) and 500 ng (CD8) per 100 μl of cell suspension, was established. Identical titers were found when performing intracellular staining.
Next, we compared the brightness of the AmCyan100 conjugates to the AmCyan or V500 equivalent (same clone) conjugates. As shown in Figure 3A, the SI of AmCyan100 CD4 and CD8 mAbs was similar to the one of the AmCyan and V500 reagents, indicating comparable sensitivity resolution/brightness. In addition, identical percentages of positive cells (data not shown) were detected across the three different CD4 or CD8 conjugates.
The use of beads (BD CompBeads or equivalent) as opposed to cells is common practice in multicolor flow cytometry as it provides multiple advantages: it allows for the use of the same antibody conjugates as in the experimental samples, creates bright and uniform positive fluorescence peaks, and avoids using sometimes limited biological material. We wanted then to verify that the spillover values obtained with AmCyan100 conjugates bound to BD CompBeads were an acceptable surrogate for equivalently stained cells. Almost identical values were obtained under both conditions (data not shown) indicating that AmCyan100 reagent compensation can be satisfactorily calculated using beads.
CD4 and CD8 AmCyan100 reagents were found to have minimum spillover into other detectors and the % of spillover calculated was consistent across the two reagents that were tested. The highest spillover detected was into the V450 detector (12–13%) and was less than for AmCyan (about 19%, see Table 1). Most importantly, and as expected based on their excitation profile, AmCyan100 conjugates exhibited low spillover into the FITC detector (4.8–5.0%, Table 1) that was eight to nine times lower than the spillover obtained for the equivalent AmCyan conjugates (Table 1) and two times lower than the spillover of the V500 conjugates into the FITC detector (Table 1). These differences were easily visualized when examining single color stained cells and comparing the spillover of CD4 AmCyan, CD4 AmCyan100 and V500 into V450, FITC and PE (Fig. 3B); clearly the AmCyan100 conjugate was the one that exhibited the least spillover into any of the other detectors.
Table 1. Comparison of AmCyan, AmCyan100, V500 % SOV into other detectors
Since fixation with PFA is commonly used to preserve samples to be analyzed by flow cytometry, we tested the effect of fixation with 1% PFA in the performance of AmCyan100 reagents. Noteworthy, and in contrast to AmCyan conjugates, AmCyan100 conjugates were stable 24 h after fixation, exhibiting similar resolution and spillover values into other detectors compared to nonfixed preparations (Table 2).
Table 2. Effect of fixation of AmCyan100 % SOV into other detectors
SOV time 0
SOV 18 h post-fixation PFA 1%
SOV time 0
SOV 18 h post-fixation PFA 1%
SOV time 0
SOV 18 h post-fixation PFA 1%
Finally, we compared the performance of AmCyan100 reagents to that of AmCyan and V500 reagents in multicolor flow applications. A six-color panel aimed at detecting CD4 and CD8 T-cells producing cytokines (IL-2, IFN-γ, and TNF-α) upon antigen specific stimulation was tested. In this assay, and after stimulation, fixation, and permeabilization as described in the Materials and Methods section, human PBMCs were stained with the following mAb cocktail: TNF-α-FITC/IL-2-PE/CD4-PerCP Cy5.5/CD3-APC/IFN-γ-V450/CD8-AmCyan100 or CD8-AmCyan or CD8-V500. This cocktail was designed so that the effect of the spillover of a bright marker conjugated with the AmCyan100 or an equivalent fluorochrome into the FITC and V450 detectors could be easily assessed. The gating strategy used to evaluate the cytokine response is presented in Figure 4A. As shown in Figure 4B, in the CD4 versus CD8 plots in the top row, the CD8-AmCyan100 antibody provides acceptable intra-cellular staining, better than that seen with CD8-AmCyan, and much better than that seen with CD8-V500. Most importantly, the spillover of CD8-AmCyan100 into the FITC and V450 detectors does not degrade the resolution of the population distributions in the CD8+ and CD4+ gated V450 versus FITC plots, as compared to the visually obvious spread and reduced separation of the populations and seen with the cocktails including CD8-AmCyan or CD8-V500. Similar observations were made when running a five-color panel aimed at discriminating different subsets of memory/effector CD4+ T-cells. For this assay, all the staining was done as surface and the following combination was used: CD45RA-FITC/CD45-PerCP Cy5.5/CD27-APC/CD3-V450/CD4-AmCyan. As shown in Figure 4C, the resolution of the CD45RA-FITC population distribution of the CD4+ T-cells is better when using CD4-AmCyan100 compared to CD4-AmCyan.
We introduce a new variant of AmCyan fluorescent protein that has been purposely designed for multicolor cell analysis. AmCyan100, the new fluorescent protein that we developed using both site directed and random mutagenesis approaches, has an emission profile similar to AmCyan with an emission maximum at 500 nm, and an excitation maximum shifted to 395 nm, which coincides more closely with the violet laser line at 405 nm compared with the original AmCyan's excitation maximum of 458 nm. This new protein has a Stokes shift of more than 100 nm, compared to a Stokes shift of 31 nm in its precursor. The brightness of AmCyan100 CD4 and CD8 conjugates appear to be similar to that of AmCyan CD4 and CD8 conjugates and, more importantly, they do not have significant fluorescence emission as a result of blue laser excitation and do demonstrate greatly reduced spillover into the FITC channel. Spillover values into FITC and other channels remain practically unaffected after cell staining followed by fixation in 1% PFA. Our findings demonstrate that AmCyan100-mAb conjugates can potentially replace the existing AmCyan-mAb conjugates in multicolor flow cytometry without any changes in instrument setup and existing reagent panel design while at the same time enhancing overall data resolution due to reduced spillover into spectrally adjacent detectors.
The authors thank Dr. N. Gadol for her help in reviewing and editing the manuscript.