PBXL-1: a New Fluorochrome Applied to Detection of Proteins on Membranes John P

PBXL-1: A New Fluorochrome Applied to Detection of Proteins on Membranes John P. Morseman, Mark W. Moss, Steven J. Zoha and F.C.Thomas Allnutt Martek Biosciences, Columbia, MD, USA

BioTechniques 26:559-563 (March 1999)

ABSTRACT paramount importance. PBXL-1 has a large complex mass (ca. 15 MDa) that protects many chromophores (1400) in a An easy, sensitive and direct fluorescent immunodetection protein matrix. This large number of fluorophores physically method for proteins is described using the new fluorochrome amplifies signal (i.e., a large and constant number of fluo- PBXLÔ-1 imaged with the FMBIOÒII Laser Scanning Imaging rophores with each binding event). This is in contrast to enzy- System. PBXL-1 is derived from a protein supra-molecular complex matic fluorometric and colorimetric methods, where amplifi- that contains a large number of chromophores. This complex, the cation depends on enzyme activity, substrate availability and phycobilisome, is extracted from a red alga then chemically stabi- incubation time for amplification. lized to allow its use in specific binding assays. PBXL-1 was cross- The phycobilisome is exploited to facilitate direct fluores- linked to goat anti-rabbit IgG or streptavidin with heterobifunction- cent detection of molecules that have sufficient access for this al cross-linkers. The detection limit of PBXL-1 was determined by large label. In our laboratory and in those of collaborators, the applying it on nitrocellulose membranes then imaging the membrane utility of these large fluors for detection of target molecules using an ytterbium aluminum garnet (YAG) laser. Evaluation of bound to cell surfaces (e.g., flow cytometry) (8) and in mi- PBXL-1 sensitivity in a specific binding assay was tested on strepta- croplate immunoassays (e.g., thyroid-stimulating hormone) vidin/biotin and an antibody system. PBXL-1 provides high sensitiv- (9) has been demonstrated. The large complex mass does not ity in direct fluorescent applications due to a physical amplification hamper these detection formats. As discussed here, it seemed of signal (i.e., a large number of fluorophores per binding event). obvious that proteins exposed on membranes would also be PBXL-1 provides a linear response over two orders of magnitude readily detected. Recently, another PBXL dye, PBXL-3, has while providing sub-amol sensitivity, indicating broad applicability been qualified for western blotting on the STORMÒ system for detection of a variety of targets. To our knowledge, this is the (Molecular Dymamics, Sunnyvale, CA, USA) utilizing a dif- most sensitive direct fluorescent detection method available. ferent laser than used for this study (1). PBXL-1 contains two major sections: the core and periph- eral rods. The core contains allophycocyanin (APC) (6). The INTRODUCTION rods that surround this core contain R-phycocyanin (R-PC) proximal to the APC core and B-phycoerythrin (B-PE) on the PBXLÔ-1 is a new fluorescent label composed of chemi- distal end. The phycobiliproteins are assembled in the alga to cally stabilized phycobilisomes, the light-harvesting antennae maximize efficient energy transfer from B-PE to the terminal complex, of the red alga Porphyridium cruentum (3). PBXL-1 emitter (a lower-energy form of APC), providing emission at provides high-sensitivity direct fluorescent detection, en- 666 nm. PBXL-1 absorbs light maximally at 545 nm and abling the use of direct fluorescence, where sensitivity is of emits at 666 nm, with a Stokes shift over 100 nm. However, it has a broad absorption spectrum that could be further utilized Chemical, Rockford, IL, USA) at a ratio of 6 streptavidin per for optimal excitation (Figure 1). PBXL-1 is functionally wa- PBXL-1 (3). CF-4 was conjugated to streptavidin using ter-soluble and excitable with the 488 nm argon and green SATA/SMCC chemistry at a ratio of 2.5 CF-4 dyes per strep- HeNe lasers. Note:PBXL-1 is compatible with 543-nm heli- tavidin. SMCC, SATA and bovine serum albumin (BSA)-bi- um/neon, 520 and 531-nm krypton ion and 537-nm otin were obtained from Pierce Chemical. Affinity-purified helium/cadmium lasers. The water solubility is referred to as polyclonal goat anti-rabbit (GAR) IgG was purchased from functional solubility since these are very large complexes that OEM Concepts (Toms River, NJ, USA). Streptavidin (SA) can be removed by centrifugation yet act as if they are water- was purchased from Prozyme (Palo Alto, CA, USA). Phos- soluble under assay conditions. This dye family holds the po- phate-buffered saline (PBS) was obtained from Boehringer tential to enable the use of direct fluorescent detection in Mannheim (Indianapolis, IN, USA), however, it was modified many applications now restricted to enzymatic or radioiso- to include additional sodium phosphate to 100 mM (PBS100) topic methods. or with 100 mM phosphate plus 1% BSA (PBS100B). For In this report, the PBXL-1 label was tested in two model washes, PBS100 was modified by the addition of 0.1% systems to examine its utility in direct fluorescent detection of TweenÒ20 detergent (PBS100T). The blocking buffer used membrane-immobilized proteins. Detection was with the (MBB) was 1.5% BSA, 1% casein, 0.5% gelatin and 0.1% FMBIOÒII Laser Scanning Imager. PBXL-1 has nearly opti- Tween 20 in PBS100 (4). All other reagents were from Sigma mal excitation at the 532-nm ytterbium aluminum garnet (St. Louis, MO, USA). The membrane used for blots was (YAG) laser line used by the FMBIO II. Both the detection HybondÔ-ECL from Amersham Pharmacia Biotech (Piscat- limit of PBXL-1 on the FMBIO II and its sensitivity in specif- away, NJ, USA); other membranes tested were WestranÒ ic binding assay formats were evaluated. Direct fluorescent Hydrophobic PVDF (Schleicher & Schuell, Keene, NH, detection offers advantages in ease of use; however, it is un- USA) and positively charged nylon (Boehringer Mannheim). derutilized due to a lack of sensitivity vs. radioisotopic and enzymatic methods. This study demonstrates sensitive and Laser Scanning Imaging quantitative protein detection on membranes with the new PBXL-1 fluorescent label on the FMBIO II. PBXL-1 and its conjugates were imaged on membranes with a FMBIO II Fluorescence Imaging System (Hitachi Software Engineering America, South San Francisco, CA, MATERIALS AND METHODS USA). Blots were placed face down on optically pure glass and wetted thoroughly with PBS100, then all of the bubbles Reagents and Chemicals were removed. The software provided by Hitachi for the FM- BIO II was used to analyze data. Depending on the applica- PBXL-1, CryptoFluorÒ-4 (CF-4) and their conjugates tion, adjustments were made to the software parameters, in- were produced by Martek Biosciences (Columbia, MD, cluding: changing (i)the sensitivity of the photomultiplier USA). CF-4 is a smaller molecular weight phycobiliprotein tube, (ii)stringency of the scanning area, (iii)emission filters, (about 35000 Da) from cryptomonad algae with optimal exci- (iv)focusing depth of the laser scanning area and (v)physical tation at 555 nm and emission at 580 nm (5). Streptavidin properties of the membrane to decrease fluorescent back- labeled with PBXL-1 (SA*PBXL-1) was made using two het- ground. Membranes were read wet to eliminate the back- erobifunctional cross-linkers, N-succinimidyl S-acetylthioac- ground caused by the interface between the membranes and etate (SATA) and sulfosuccinimidyl 4-(N-maleimidometh- glass and to maintain optimal PBXL-1 fluorescence. yl)cyclohexane-1-carboxylate (SMCC) (both from Pierce Detection of Unconjugated of PBXL-1 on Nitrocellulose Membrane

A dilution series of PBXL-1 (0.38 pg–10 ng) was immobilized on nitrocellulose to determine the optimal molar sensitivity of PBXL-1. PBXL-1 was diluted in PBS100B, then 0.5- mL aliquots were manually applied to nitrocellulose as previously described (2). Samples were allowed to dry for 10 min to assure adequate binding to the membrane. Blots were immersed in PBS100, then imaged on the FMBIO II using a 650 (±15)-nm filter. Data were analyzed as described above.

Detection of BSA-Biotin with PBXL-1 Streptavidin Conjugates

BSA-biotin was manually applied to nitrocellulose in 0.5- µL vol from 0.1 pg to 100 ng in PBS100B. Membranes were air-dried for 5 min then blocked in MBB for 1 h at room tem- perature (RT). SA*PBXL-1 at 10 mg/mL in MBB was reacted with the membranes for 30 min at RT with shaking. These de- Figure 1. Excitation and emission spectra of PBXL-1 in PBS100.Excita- tion of the emission spectrum was at 545 and 666 nm emission was moni- tection times, conjugate concentrations and MBB composi- tored. Spectra were normalized and are uncorrected. tion were experimentally determined. Membranes were washed 3´with PBS100T at RT. Developed membranes were with CF-4-labeled streptavidin, while the rest of the proteins then scanned on the FMBIO II using a 650-nm (±15-nm band on the blot were detected with GAR*PBXL-1 to evaluate pass) filter and analyzed as described above. PBXL-1 in a two-color optical separation. The emission max- imum of CF-4 is at 580 nm and that of PBXL-1 is 666 nm, Detection of Rabbit IgG with Goat Anti-Rabbit Labeled such that two-color imaging can easily be done with the with PBXL-1 FMBIO II. Since the filters were more restrictive, some loss of sensitivity occurs when run in the two-color mode as com- Rabbit IgG was applied to Hybond-ECL from 0.1–100 pg pared to single-color detection. in PBS. After application, membranes were air-dried for 5 min then blocked in MBB for 1 h at RT. Goat anti-rabbit IgG labeled with PBXL-1 (GAR*PBXL-1) was added to the RESULTS AND DISCUSSION membrane at 10 mg/mL in PBS100B then incubated for 45 min at RT. Membranes were then washed with PBS100T for several minutes at RT before being scanned on the FMBIO II. Sensitivity of PBXL-1 Direct Detection Additional washes were done as required. PBXL-1 was serially diluted in PBS100B, spotted onto ni- Immunoblotting of Rabbit IgG trocellulose then imaged with the FMBIO II to determine the detection sensitivity and linear dynamic range. The Hybond- Western transfers were carried out to evaluate PBXL-1 ECL (nitrocellulose) membrane provided low-background conjugates in detection of proteins electrophoretically trans- fluorescence emission with the 660-nm filter. Several other ferred to nitrocellulose. Rabbit IgG was electrophoresed in a membranes were evaluated before choosing the Hybond- 15% sodium dodecyl sulfate polyacrylamide gel electro- ECL. These had either higher levels of background (Westran phoresis (SDS-PAGE) gel using standard methods (7) then Hydrophobic PVDF) or were not compatible with the 532-nm transferred to nitrocellulose membrane using a HoefferÒ laser (positively charged nylon). The lower limit of detection Western Blotting Apparatus (Hoeffer Pharmacia Biotech, San for PBXL-1, using a 650 (±15)-nm filter, was 0.78 pg; this Francisco, CA, USA) at 200 mA for 1 h. The membrane was equals 2.60 ´10-20mol or 31300 molecules of PBXL-1 (Fig- removed, washed in PBS100, then blocked in MBB for 1 h. ure 2). The dynamic linear range was two orders of magnitude GAR*PBXL-1 was used to develop the membrane at a con- (on a molar basis) or three orders of magnitude (on a pg ba- centration of 10 mg/mL in PBS100B. The membrane was de- sis). The linear portion of the PBXL-1 dilution curve (0.1–10 veloped for 45 min at RT, washed in PBS100T, then imaged. amol) had a correlation coefficient (r2) value of 0.991 (Figure Biotinylated low-molecular-weight standards (Bio-Rad, Her- 2). One possible limiting factor to the sensitivity achieved was cules, CA, USA) were also run with the streptavidin/biotin that the 650 (±15)-nm filter captured only about 30% of the model system. In most cases, these standards were developed total emission of PBXL-1. To further increase sensitivity, a with SA*PBXL-1. In one case, the standards were developed 630-nm-long bandpass filter was used to increase the number

Figure 2. Fluorescent imaging of PBXL-1 directly spotted onto Hybond- Figure 3. Fluorescent imaging of PBXL-1 streptavidin conjugate bound ECL.(A) A graphical representation of the image in Panel B. PBXL-1 was se- to biotinylated BSA directly spotted onto Hybond-ECL.(A) A graphical rially diluted from the right of the membrane in Panel B, starting at 10 ng and representation of the image in Panel B. The spotted BSA-biotin was a 1/10 diluted 10-fold for the next 2 dilutions (10–0.1 ng), dilutions were then by half dilution of the far right spot for three spots (10–0.1 ng) then a 1/2 dilution relative to the prior spot (50–0.38 pg). The 0.38 pg spot was not observable. from then on (50–0.78 ng). of photons captured from PBXL-1 emission. The 630-long- Direct Detection of Immobilized Rabbit IgG with Goat bandpass filter increased the lower limit of detection signifi- Anti-Rabbit IgG Labeled with PBXL-1 cantly, to 0.195 pg, 1.3 ´10-20mol or 7830 molecules. How- ever, the 630-nm-long bandpass filter had an adverse effect on For direct detection of rabbit IgG, dilutions of rabbit IgG the linear dynamic range, as light leakage and membrane were applied in 0.5-mL aliquots on nitrocellulose membranes, background fluorescence increased dramatically. A more opti- air-dried, blocked with MBB, then detected with mal filter at 666 nm with a 20-nm bandpass should provide GAR*PBXL-1. While there was some noise in these data, the higher sensitivity and will be applied as the method is further response was linear over several orders of magnitude (r2= optimized. Without further optimization, PBXL-1 can be de- 0.9647). The lower level of detection of 0.39 ng of rabbit IgG tected at sub-amol levels with the FMBIO II. corresponds to 2.4 ´10-15mol of rabbit IgG (Figure 4). This low-affinity, antibody-driven detection using GAR*PBXL-1 provides fmol detection of rabbit IgG. Higher affinity anti- Direct Detection of Biotinylated BSA with Streptavidin- Labeled PBXL-1 bodies and better filter matching to the PBXL-1 excitation should provide increased sensitivity. BSA-biotin was applied in 0.5 mL vol on nitrocellulose, air-dried, blocked with MBB then detected with SA*PBXL- 1. SA*PBXL-1 concentrations vs. various blocking buffers were run to optimize signal-to-noise ratios (data not shown). Once optimal concentrations of detection reagent and noise reduction were achieved, replicate experiments were run to provide representative data for detection of BSA-biotin with SA*PBXL-1. The lowest BSA-biotin detected was 0.78 pg or 1.51 ´10-16mol (Figure 3). The linear dynamic range was two orders of magnitude with an r2value of 0.9985. It may be possible to achieve detection at lower levels of BSA-biotin with more optimal filters and dilutions below 0.78 pg. How- ever, with the current data, BSA-biotin can be detected with SA*PBXL-1 at the 10 amol level.

Figure 4. Direct fluorescent detection of rabbit IgG immobilized to nitrocellulose using GAR*PBXL-1.(A) A graphical representation of the image in Panel B. The dilution series on the membrane in Panel B is rabbit IgG diluted from 100 to 0.39 ng in half-step dilutions from right to left.

Figure 5. Western blot of rabbit IgG detected with GAR*PBXL-1.Lane 1 had biotinylated standards, which were developed with streptavidin labeled with PBXL-1. Lanes 2–4 had rabbit IgG at 25, 10 and 5 ng, respectively. Detection of Electroblotted Rabbit IgG with Goat Anti- REFERENCES Rabbit IgG Labeled with PBXL-1 1.Anonymous.1998. What’s new: PBXL-3 labelled antibodies. Life Sci- Western blots were performed to show the applicability of ence News 23:44. PBXL-1 dyes and to allow comparison to existing methods. 2.Batteiger, B.1988. Blocking of immunoblots, p. 145-150. InO. Bjerrum and N. Heegaard (Eds.), Handbook of Immunoblotting of Proteins. CRC The background of the PBXL-1-labeled, anti-rabbit IgG Press, Boca Raton. (GAR*PBXL-1) was explored further to decrease back- 3.Cubicciotti, R., inventor, assignee. 1997. Phycobilisomes, derivatives, ground problems seen in some of the dot blots (e.g., Figure 4). and uses therefor. US Patent 5,695,990. 1997 Dec. 7. Substantial sensitivity was achieved with minimal back- 4.Gordon, J. and P. Billing.1988. Dot immunoblotting - general principles and procedures, p. 27-30. InO. Bjerrum and N. Heegaard (Eds.), Hand- ground in direct fluorescent detection of rabbit IgG (Figure book of Immunoblotting of Proteins. CRC Press, Boca Raton. 5). The nanogram sensitivity obtained in the western blots 5.Hill, D.R. and K.S. Rowan.1989. The biliproteins of the Cryptophyceae. compares well with existing nonradioactive detection meth- Phycologia 28:455-463. ods. Additional optimization of this method can be achieved, 6.MacColl, R. and D. Guard-Friar.1987. Phycobiliproteins, p. 218. CRC so these results should be considered a conservative estimate Press, Boca Raton. 7.Sambrook, J., E. Fritsch and T. Maniatis. 1989. Molecular Cloning: A of the potential of this method. Laboratory Manual, 2nd ed. CSH Laboratory Press, Cold Spring Harbor, Western blots with rabbit IgG and biotinylated low-molec- NY. ular-weight protein standards were carried out to evaluate the 8.Zoha, S. 1998. PBXLÔfluorescent dyes for ultrasensitive direct fluores- potential of PBXL-1 as a visualization reagent for protein de- cent detection. InAdvances in Molecular Labels, Signaling and Detec- tion. Cambridge Healthtech Institute, San Diego. tection using multiple labels. Streptavidin labeled with 9.Zoha, S., S. Ramnarain and F. Allnutt.1998. Ultrasensitive direct fluo- CryptoFluor-4 (SA*CF-4) and GAR*PBXL-1 were used si- rescence immmunoassay for thyroid stimulating hormone. Clin. Chem. multaneously to develop the western blot. The membrane was 44:2045-2046. scanned once with the FMBIO II and detected on two chan- nels (585- and 660-nm filters; data not shown). Rabbit IgG Address correspondence to Dr. F.C. Thomas Allnutt, was detectable to 10 ng. Additional steps should be taken to Martek Biosciences Corporation, Fluorescent Products, 6480 decrease the background fluorescence of PBXL-1, but, Dobbin Road, Columbia, MD 21045-5825, USA. Internet: nonetheless, utility in two-color format was demonstrated. [email protected] The SA*CF-4-detected biotinylated standards came out very well at the micromolar range used. The concentration of the standards was 1000 times higher than the rabbit IgG. This could explain some of the background seen with PBXL-1 as the software was manipulated to balance the two fluors being used. The lower detection levels seen with single-color detection (Figure 5) were probably the result of software compensation to allow imaging of both fluors simultaneously. It is possible that better compensation combined with better filter sets for the two dyes could provide lower detection levels.

CONCLUSIONS

The recently introduced PBXL dyes are demonstrating broad application in high-sensitivity direct fluorescent detection. This study details PBXL-1 applicability for sensitive immunodetection of proteins imaged on the FMBIO II. The physical amplification achieved by PBXL-1’s large size and the large number of fluors per binding event provide amol sensitivity in direct detection. The large mass of PBXL-1 was not a problem in western blotting or detection on membranes, where targets were presented on the surface. For western blotting, superior sensitivity vs. existing direct fluorescent detection systems was demonstrated in both streptavidin:biotin and antibody model systems. Additionally, PBXL-1 sensitivity compares favorably with enzymatic methods while having the advantage of requiring fewer steps with fewer reagents (no enzyme or substrate).

ACKNOWLEDGMENTS

The authors thank Dr. Sheila Colby, Ms. Deborah Smead and Hitachi Software Engineering America Ltd. for allowing the use of their instrument and providing useful technical in- formation for the completion of this study.