TECHNICAL INNOVATION
Novel Use of the BD FACS
TM
SPA to Automate
Custom Monoclonal Antibody Panel
Preparations for Immunophenotyping
Abigail S. Kelliher,1* David W. Parent,1
David C. Anderson,1 Michelle E. Dorn,1 Jessica L. Hahn,1 Sara Eapen,2and Frederic I. Preffer1
1
Department of Pathology, Massachusetts General Hospital, Boston, Massachusetts 2
Dana Farber Cancer Institute, Boston, Massachusetts
Background: The effective and accurate diagnosis of hematologic malignancies relies on flow cytomet-ric immunophenotyping. Selected combinations of monoclonal antibodies (mAbs) arranged in multicolor panels allow for the accurate definition of normal and abnormal hematologic cell populations. The most time-consuming and crucial step in the staining process involves dispensing combinations of multiple mAbs into their appropriate staining tubes. This step is prone to error, requires concentration and accu-racy, and is dependent on technologist experience.
The Becton Dickinson BioScience (BD) FACSTMSample Prep Assistant (SPA) is touted as a breakthrough in automated in vitro diagnostic sample preparation. The SPA is designed to automate BD MultiTESTTM and BD TriTest lyse/no-wash assays. However, because most cases in our laboratory require tedious appli-cation of unique four-color mAb cocktails for leukemia and lymphoma testing, we wondered whether the SPA would be helpful in accurately dispensing these mixtures.
Methods: The mAb panels were prepared by the SPA in two separate timed runs and on separate days. Eleven specimens (nine from patients and two from normal volunteers) were split and stained with four-color cocktails created by the SPA or manually. The percentage of positive (%P) cells and mean fluores-cent intensity for each mAb pair were determined. These values were plotted against each other and cor-relation values were calculated. To quantitate timesaving in the laboratory, two technologists prepared individually the same mAb panels and were timed.
Results: The correlation between the two methods was high; r2was 0.988 for 158 %P antigen pairs; no bias between the manual and robotic methods was detected with the Wilcoxon rank test. Bland-Altman analysis indicated no obvious relation between the difference and the mean of %P cells, suggesting that the SPA successfully dispensed antibodies for leukemia/lymphoma panels. The two methods may be inter-changeable, although the limited sample size prohibits this conclusion from Bland-Altman statistics alone. In addition, one possible error was detected in the SPA-prepared panels. The SPA averaged 65 min/run, the experienced technologist 12.95 min/run, and the inexperienced technologist 54.9 min/run.
Conclusions: SPA dispensing time was twice the average manual dispensing time; however, SPA use was completely automated and freed the technologist to perform other tasks. SPA use permitted preemp-tive preparation of mAb panels and thus streamlined processing; however, the cost of the assay and the amount of reagent waste increased. It is certain that software modifications by BD could decrease the SPA reagent dispense time and decrease the cost associated with reagent waste when the SPA is used in this novel fashion. q2005 Wiley-Liss, Inc.
Key terms: flow cytometry; immunophenotyping; robotics; staining techniques; SPA; automation
Becton Dickinson BioSciences (BD; San Jose, CA, USA) released the BD FACSTM Sample Prep Assistant (SPA) in 2002. The SPA, a class I medical device appropriate for in vitro diagnostic use in the preparation of samples for further acquisition and analysis by flow cytometry, is touted as a breakthrough in automated sample process-ing. The SPA is freestanding and independent from a
*Correspondence to: Abigail S. Kelliher, B.A., Warren Building Room 112, Massachusetts General Hospital, Boston, MA 02114.
E-mail: [email protected]
Received 18 March 2004; Accepted 11 March 2005
Published online 25 May 2005 Wiley InterScience (www. interscience.wiley.com).
DOI: 10.1002/cyto.b.20055
flow cytometer and only prepares specimens for subse-quent analysis. Designed to work in tandem with the BD MultiTestTM and TriTestTM lyse/no wash assays and to automate T-cell and lymphocyte subset enumerations, the SPA provides a robotic preparatory mechanism that dispenses patient blood, monoclonal antibodies (mAbs), and lyse reagents without manual intervention.
For processing specimens such as T-cell subsets, patient specimen VacutainerTM tubes are placed in the SPA specimen tube rack and automatically mixed. A syringe-driven needle pierces the VacutainerTMseal and a preset volume of whole blood is aspirated and trans-ferred to secondary sample test tubes. Subsequent to robotic addition of mAbs, the secondary test tubes are mixed and incubated for a user-defined length of time. BD FACSTMlysing solution is added; tubes are mixed and again incubated. The resulting stained and fixed cells are then ready for analysis. The SPA dovetails well in the lab-oratory using the BD MultiTestTM lyse/no-wash assay to determine lymphocyte subsets because the assay is then almost completely ‘‘hands-off’’ and frees the technologist for other tasks.
In addition to the SPA, we are aware of two other robotic instruments claiming similar function, the Coulter Prep PlusTM and the Partec Robby Automat1. This laboratory had access only to the SPA and was inter-ested in determining whether its value in the laboratory could be expanded beyond its intended purpose in the preparation of four-color immunophenotyping leukemia/ lymphoma panels.
Immunophenotyping for leukemia and lymphoma diag-noses is a critical responsibility for the clinical flow cytometry laboratory (1–7). In this laboratory, this testing requires selected combinations of four-color conjugated mAbs organized into panels designed to identify T, B, nat-ural killer, and myeloid cell populations. These panels are predominantly laboratory defined and different panels consist of various numbers of four-color staining tubes, depending on the clinical question and the complexity of the test ordered. For example, ‘‘leukemia’’ panels (which generally require identification of lymphoid and myeloid populations) require an eight-tube panel. A ‘‘lymphoma’’ panel generally requires identification only of lympho-cytic populations and requires only a four-tube staining panel. The ‘‘unknown’’ lymphoma panel is used for bone marrow specimens when a diagnosis of lymphoma is requested. Bone marrow specimens require the identifica-tion of immature myeloid and erythroid populaidentifica-tions and lymphocyte populations, resulting in a panel of five tubes. A crucial, tedious, and time-consuming step of the staining protocol is aliquoting each of four fluoro-chrome-conjugated mAbs for every tube. In a four-tube/ patient lymphoma panel, this translates into 15 individ-ual aliquots/patient (the kappa and lambda antibodies are packaged by the manufacturer as a single-dispense cocktail). A leukemia panel consists of eight tubes, necessitating 31 individual aliquots. In a busy clinical lab-oratory, the technologist often prepares ‘‘batches’’ of specimens, each requiring its own panel. A batch can
routinely consist of seven or eight patient specimens, each with different test requests. On average, a single batch could conceivably require 161 individual mAb ali-quots. This is a tedious process requiring the full con-centration of the technologist. Automating this step would free up a large portion of a technologist’s time and decrease errors. Because the SPA can add any num-ber of reagents to particular tubes, we wondered whether we could use this feature of the SPA to dis-pense consistently and accurately single mAb reagents in our test-panel combinations, thus freeing the technolo-gist for performing other tasks.
MATERIALS AND METHODS Manual Staining
The manual staining process for these samples is straightforward; whole blood and bone marrow speci-mens are washed before staining to remove endogenous patient serum immunoglobulin. Patient cell suspensions obtained by fine-needle aspirate or minced solid tissue samples are concentrated by centrifugation to obtain the maximum number of cells for analysis. In our laboratory, the technologist determines the appropriate staining panel and then aliquots each antibody into tubes by standard techniques (1). Briefly, cell suspensions are added, mixed, and incubated for 15 min at room temper-ature. A lysing reagent is added to remove mature red blood cells, and the staining tubes are mixed and incu-bated for a second time. After this incubation, the tubes are centrifuged to remove the red cell debris and washed a final time with phosphate buffered saline to remove residual debris. Cells are resuspended in a solu-tion of 1% formaldehyde/phosphate buffered saline.
SPA Setup
The SPA platform consists of a rotating specimen rack that holds 13- 75-mm or 16- 100-mm VacutainerTM patient sample tubes. Beside the rack is a removable reagent block, which can be chilled to retard the warm-ing of the monoclonal antibody reagent bottles. There is a carousel in which the individual staining tubes are placed. The entire unit is housed in a smoked plastic hood to prevent light from entering and to protect the technologist from accident. A separate reagent ‘‘caddy’’ holds reservoirs for sheath fluid, cleaning reagents, lysing reagents, and deionized water. The SPA is configured with a personal computer using Windows that runs its operational software. The dispensing of reagents is per-formed with a syringe-based system. A single needle is used to dispense specimen, antibodies, and lysing reagents. This needle is washed and rinsed thoroughly between each function to avoid carryover.
The SPA is controlled by software in which an operator with administrator privileges can define reagents, panels, and tubes, combinations of which make up a staining protocol. The ‘‘reagent rack editor’’ identifies sets of anti-body reagents. The editor contains lists of reagent sets defined by BD or the operator. Each set or rack can
con-sist of any combination of single mAbs or multiple mAb cocktails. Each reagent block is capable of housing 24 BD-sized antibody vials. Although the SPA reagent block was designed to physically hold BD antibody vials, we have circumvented this limitation by recycling empty, washed BD vials with whatever mAb is needed, regard-less of the vendor. Each mAb vial is assigned a location in the reagent block; the location of the reagent is then crit-ical for appropriate reagent distribution.
After a reagent rack is defined with the location of mAb vials identified, panels and tube settings are defined using the ‘‘panel editor.’’ Within this editor, manufac-turer-defined panels are listed and operator panels are defined. For each panel, reagent incubation and red cell lysing times are defined. Because the SPA is being used in this case only to dispense mAbs and not to perform a complete staining procedure, the reagent incubation time and the lyse incubation time were set to 0 min.
The operator adds tubes to each panel through the ‘‘tube editor.’’ This editor defines the specimen volume, the reagents to be added, and their respective volumes. The SPA is currently limited because the reagent vol-umes are restricted to 20, 10, and 5 ml. Due to this restriction, the reagent volumes were re-titered to fit this template. The lyse reagent volume is also defined in this editor; for this application the amount was set to 0ml.
Sample Testing
To compare the manual setup of mAb testing panels versus that of the SPA, the SPA was programmed to ali-quot several antibody staining panels (leukemia, lym-phoma, and unknown lymphoma; Table 1). Patient speci-mens were split and stained in parallel with SPA-dis-pensed mAb panels or manually disSPA-dis-pensed panels. This process was performed over a consecutive 2-day period when the SPA dispensed antibody panels in two separate runs. In total, 11 specimens were processed: three peripheral bloods, three bone marrow aspirates, two fine-needle aspirates, one lymph node, one bronchial lav-age, and one commercially available stabilized blood product (BD Multi-Check). Each pair of specimens (man-ually and SPA-prepared antibody panels) was analyzed in the clinical setting to examine the white blood cells for possible hematopoietic malignancy. In all cases, the lym-phocytes and blasts were initially gated using CD45 and side light scatter in conjunction with forward and side light scatter gates. All percentage of positive (%P) popu-lations and mean fluorescence intensities obtained were from populations gated on forward/side scatter. These patient specimens and their resulting clinical diagnoses are summarized in Table 2.
To determine laboratory timesaving, each SPA run was timed to determine the amount of time required to ali-quot the antibody staining panels. Two technologists (one with 10 years of flow cytometric experience and one with 3 months of flow cytometric experience) were also timed when preparing identical antibody panels.
RESULTS
To determine whether antibody panels created by the SPA produced identical results as those panels prepared manually, comparisons between %P of 158 pairs of anti-gens were made and the results were plotted on a scat-tergram (Fig. 1; r2 ¼ 0.988). Eleven antigen pairs were eliminated from analysis where antibody was not added or the incorrect antibody was added or the data quality was unduly affected due to poor specimen quality such that accurate interpretation of the data was impossible. In addition, the mean fluorescence intensity of each anti-gen was compared (Fig. 2; r2 ¼ 0.958). Four additional antigen pairs were eliminated from analysis due to unin-terpretable data because of specimen quality. The corre-lation between the two methods was excellent and there was no obvious relation between the difference and the mean of %P cells, suggesting that the SPA successfully dispensed antibodies for leukemia/lymphoma panels.
For more in depth analysis, each antigen was examined separately and %P cells were determined. The degree of agreement between the two methods was evaluated using Bland-Altman plots (Fig. 3). Paired difference analysis of %P cells was also carried out using the Wilcoxon sign rank test. In addition, the values were plotted against each other and correlation values were calculated.
Each of the 11 samples had measurements on 14 markers and each marker was analyzed separately. There were nine extra markers that were not included in the analysis because very few samples had measurements made on them. Missing measurements were treated as random and not included in the analysis. There were two manual errors; either the wrong antibody was added or an antibody was omitted and one robotic error was
Table 1
Clinical Leukemia/Lymphoma Panels Used for Testing* Panel/tube FITC PE PerCP APC Leukemia 1 CD13 CD33 CD45 CD14 2 CD7 CD8 CD3 CD4 3 CD10 CD23 CD20 CD5 4 Kappa Lambda CD20 CD19 5 CD43 CD10 CD19 CD33 6 CD71 HLA-DR CD45 CD33 7 TdTa CD10 CD20 CD19 8 MPOa CD117 CD45 CD33 Lymphoma 1 CD43 CD19 CD45 CD14 2 CD7 CD8 CD3 CD4 3 CD10 CD23 CD20 CD5 4 Kappa Lambda CD20 CD19 Unknown lymphoma 1 CD71 CD33 CD45 CD14 2 CD7 CD8 CD3 CD4 3 CD10 CD23 CD20 CD5 4 CD43 CD10 CD19 CD33 5 Kappa Lambda CD20 CD19
*APC, allophycocyanin; FITC, fluorescein isothiocyanate; MPO, myeloperoxidase; PE, phycoerythrin; PerCP, peridinin chlorophyll protein; TdT, terminal deoxynucleotidyl transferase.
aThese are antibodies to intracellular antigens and are added
deduced from an aberrant pattern of observed CD45 expression.
The Bland-Altman plots for each of the 14 markers showed no relation between difference and average %P cells, i.e., there was no evidence for an increasing trend in variance of the difference as the average increased (data not shown). The degree of agreement between the two methods was summarized by the mean difference and standard deviation of the differences. Table 3 presents results for 14 markers representing the T, B, natural killer, and myeloid cell populations. Column 2 in Table 3 presents the average of the manual results for each marker. Column 3 in Table 3 lists the mean of dif-ferences (robot vs. manual subset percentages) for 14 markers. Column 4 presents the Gaussian 95% confi-dence intervals for the mean difference plus or minus twice the estimated standard deviation of the difference. These limits provide an indication of the extent of the differences and whether the two methods are inter-changeable, assuming the distribution of the differences were Gaussian. Column 5 presents the P values from a two-sided Wilcoxon sign-rank test, testing that the median difference was 0, which corresponds to no bias.
Only three of the means (CD19 peridinin chlorophyll protein, kappa and lambda) were negative; all the rest
were positive, suggesting that the robot method had larger measurements than the manual method. The means ranged from 0.67 to 2.54, with a mean of 0.81 and standard deviation of 1.06. All values were within two standard deviations of the mean. There was one sig-nificant test comparing the measurements on CD4 allo-phycocyanin (P ¼0.004) and one marginally significant test comparing CD8 phycoerythrin (P¼0.05). Using the Bland-Altman statistic, there was decreased power to detect significant bias due to the small sample.
In the laboratory timesaving evaluation, the experi-enced technologist dispensed the antibodies in the panel sets in 12.95 min. The inexperienced technologist pensed the antibodies in 54.95 min, and the SPA dis-pensed identical patterns in 65 min.
DISCUSSION
Although the SPA aliquoted antibodies slower than manual aliquoting, utilization of the SPA allowed the technologist additional time for performing other tasks and created a predictable time span for other necessary activities. In an economy where labor freezes are com-monplace but the workload continues to increase, auto-mation is one key to maintaining service by enabling two (or more) tasks to be completed at the same time, thus contributing to the overall efficiency of the labora-tory. As noted in the manual staining timing tests, there is a wide variation in setup time from person to person. Using the SPA to prepare the panels decreases the detri-mental effect inexperience can contribute to laboratory workflow because accurate panel preparation is one of the more difficult and crucial tasks for new technologists to learn and do efficiently. This laboratory currently uses the SPA to prepare panels to be stored for later use to streamline the staining process; thus there is no delay while waiting for antibody panels to be prepared. We have estimated that the SPA frees up anywhere from 45 min to 3 h of time per day depending on the experi-ence and speed of the technologist.
With a few software modifications by the manufac-turer, the dispense time for the SPA could be improved.
Table 2
Diagnosis of Patient Specimens Used for Comparison Testing
Tissue Diagnosis
Lymph node Reactive follicular and paracortical hyperplasia with focal non-necrotizing granulomas Bronchoalveolar lavage Negative for malignant cells
Thyroid FNAa Prominent and reactive lymphoid population consistent with Hashimoto’s thyroiditis
Peripheral blood There is no evidence for a monoclonal B- or an unusual T-cell population
Bone marrow aspirate Normocellular marrow with trilineage hematopoiesis; there is no evidence of lymphoma
Peripheral blood 2% Circulating myeloid blasts in peripheral blood Lymph node Metastatic poorly differentiated adenocarcinoma Bone marrow aspirate Trilineage hematopoiesis; there is no
evidence of lymphoma
Bone marrow aspirate Bone marrow aspirate with trilineage hematopoiesis
aFine-needle aspirate.
FIG. 1. Plot of %P populations of patient and normal specimens using standard lymphoma/leukemia panels dispensed manually versus those dispensed by the SPA (r2¼0.988, n¼158).
In the conventional use of the SPA, the instrument mixes each specimen tube and then aliquots the specimen to the appropriate tube. This step can take up to 10 min. In this novel application, specimen is added manually at a later time and thus this initial specimen-dispensing step is a wasted one. Currently, this step can be elimi-nated when the SPA is operated in its ‘‘service mode,’’ when certain procedures can be disabled for trouble-shooting purposes. When used with the specimen-dis-pensing procedure disabled, an overall decrease in total time is noted. Although it is possible to run in service mode indefinitely, it is not recommended for daily use because user tracking logs would be obviated and secur-ity features would be disabled, thus allowing unauthor-ized access to protocol manipulation. Software adapta-tions that would allow the user to enable or disable this first step within particular defined protocols would make the SPA a more versatile tool.
Dispensing of antibodies is the most critical step of the staining process and the one most likely to contain errors due to the complexity and the number of panels being prepared and the enormous concentration required by the technologist. Errors occurring at this step are irrecoverable, i.e., restaining is the only option. An error of this magnitude could double the staining time and the associated cost for a particular run. Decreasing the error rate could vastly improve the turn-around time on specimens and keep overall laboratory costs down. Using the SPA did not eliminate errors: there was one possible error in the SPA-prepared panels where there was an unusual staining pattern of CD45. It is unknown whether this was an instrument error or a sampling/mixing error occurring later on in the staining process. Although no other similar errors were detected, it can be safely hypothesized that comparable failures may happen in the future. These types of failures can be as costly as when manual errors occur. At this point in time, an insufficient amount of data is available to ascer-tain the failure rate, but it is being closely monitored.
This laboratory used the same quality control meth-ods on the SPA-prepared specimens as on the manually prepared specimens to ascertain whether or not errors
were present. We employ a semiquantitative method that relies on the design of our mAb panels. Each tube was monitored for expected staining patterns, percen-tages of cells (B cells, T cells, etc.) were monitored for tube to tube consistency, and each SPA-prepared speci-men was compared with its manually prepared coun-terpart. It is advisable for all laboratories to employ their own quality control procedure for all staining methods.
There are disadvantages to using the SPA. Because the software allows for only a limited number of antibody volumes, the amount of antibody normally used in man-ual pipetting must be increased. For some antibodies this increased from 8 to 10 ml and in others from 3 to 5ml. This increase in volume represents a 15% increase in cost per tube. Further software adaptations that allow for any number of reagent volumes would eliminate this cost increase. The ability to decrease specimen volumes from the set volumes of 50 or 100ml could significantly decrease the assay cost due to a smaller staining volume. Significant hardware and software adaptations enabling the use of microtiter plates would contribute to decreas-ing the cost of the assay significantly.
In addition to increasing the amount of mAb used in the panels, there is also an increase in wasted antibody. Because panels are pre-prepared before any specimen is received, entire panels could be discarded before use due to their limited shelf life of 3 days. After this time point, evaporation begins and the stability of the fluoro-chromes become questionable (unpublished observa-tions). At the end of a workweek, panels are inevitably discarded and the laboratory must absorb the cost of these wasted reagents.
Ideally, one would use the SPA for the entire staining process including dispensing specimen, reagent, and lys-ing reagent and monitorlys-ing incubation times. Although this is possible with peripheral blood specimens, the nature of bone marrow and solid tissue specimens deserves caution. Specimen is dispensed by using a thin needle. Bone marrows can be more viscous with solid
FIG. 3. Bland-Altman plot depicting the difference in %P versus average %P. The center line indicates the mean and the two horizontal lines are plus or minus two standard deviations from the mean. FIG. 2. Plot of mean fluorescent intensities for antigens on patient
and normal specimens stained with standard lymphoma/leukemia pan-els dispensed manually versus those dispensed by the SPA (r2 ¼
particles (i.e. bone, clots) and solid tissue specimens, even after filtering, often reaggregate, thus increasing the danger of needle clogging. The hypocellular nature of fine-needle aspirates and cerebrospinal and vitreous fluids preclude their use with the SPA. Because the reagent/specimen needle does not descend to the abso-lute bottom of the tube, a minimum specimen volume is necessary for accurate specimen dispensing. Diluting these specimens to this minimum volume would decrease the total number of cells available for staining and acquisition. In addition, common staining protocols involve pelleting and washing cells. The SPA is not intended for these activities; instead BD offers another instrument, the BD FACSTM Lyse Wash Assistant. Ideally the functions of these two instruments will be integrated.
In conclusion, clinical laboratories could routinely use the SPA for automated dispensing of individual antibody reagents for leukemia/lymphoma panels with little effect on data collected and no discernible effect on clinical diagnoses. The instrument liberates a technologist’s time and prepares panels accurately and consistently, allowing for better workload estimates. There is a modest increase in reagent cost associated with SPA use; however, minor
software adaptations by the manufacturer would make the SPA more cost efficient. Although the instrument’s intended use is for staining lymphocytes and T-cell sub-sets, using the instrument in this novel way increases the benefit to the laboratory in cost and effort.
LITERATURE CITED
1. Preffer FI. Flow cytometry. In: Colvin RB, Bhan AK, McCluskey RT, editors. Diagnostic immunopathology. 2nd ed. New York: Raven Press; 1993. p 725–749.
2. Andreef M. Clinical cytometry. Ann NY Acad Sci 1986;468:1–408. 3. Jennings CD, Foon KA. Recent advances in flow cytometry:
applica-tion to the diagnosis of hematologic malignancy. Blood 1997;90: 2863–2892.
4. Martinez A, Aymerich M, Castillo M, Colomer D, Bellosillo B, Campo E, Villamor N. Routine use of immunophenotype by flow cytometry in tis-sues with suspected hematological malignancies. Cytometry 2003; 56B:8–15.
5. McKenna RW, Asplund SL, Kroft SH. Immunophenotypic analysis of hematogones (B-lymphocyte precursors) and neoplastic lymphoblasts by 4-color flow cytometry [review]. Leuk Lymphoma 2004; 45:277–285.
6. Pagnucco G, Vanelli L, Gervasi F. Multidimensional flow cytometry immunophenotyping of hematologic malignancy. Ann NY Acad Sci 2002;963:313–321.
7. Braylan RC, Orfao A, Borowitz MJ, Davis BH. Optimal number of reagents required to evaluate hematolymphoid neoplasias: results of an international consensus meeting [review]. Cytometry 2001;46: 23–27 (erratum Cytometry 2001;46:119).
Table 3
Manual Versus Robot Results for Common Leukocyte Antigens Markera
Mean manual result
Mean of differences
(robot vs. manual) Mean62 SD
Pfrom the Wilcoxon sign-rank test CD45 73.2 0.9 3.565.2 0.40 CD19 PerCP 17.2 0.7 2.961.6 0.25 CD43 FITC 69.1 2.2 5.6610.0 0.19 CD7 FITC 14.8 1.5 8.6611.6 0.56 CD8 PE 18.9 1.7 3.166.6 0.05 CD3 PerCP 47.5 2.5 7.1612.2 0.17 CD4 APC 27.8 2.4 4.268.9 0.004 CD5 APC 48.5 0.3 2.763.2 0.67 CD20 PerCP 13.6 0.4 5.165.8 0.78 CD23 PE 8.4 0.1 2.362.5 1 CD10 FITC 3.6 0.1 0.660.8 1 Kappa FITC 10.1 0.2 3.062.6 0.62 Lambda PE 7.6 0.2 2.762.3 1 CD19 APC 18 0.3 2.262.8 0.62 a
APC, allophycocyanin; FITC, fluorescein isothiocyanate; PE, phycoerythrin; PerCP, peridinin chlorophyll protein.