QD-Antibody Conjugates via Carbodiimide-Mediated Coupling: A

ARTICLE pubs.acs.org/Langmuir

QD-Antibody Conjugates via Carbodiimide-Mediated Coupling: A Detailed Study of the Variables Involved and a Possible New Mechanism for the Coupling Reaction under Basic Aqueous Conditions Daniel A. East,† Daniel P. Mulvihill,† Michael Todd,§ and Ian J. Bruce*,† †

Nanobiotechnology Research Group, School of Biosciences, University of Kent, Giles Lane, Canterbury, Kent, CT2 7NJ, United Kingdom § School of Science, University of Greenwich, Central Avenue, Chatham Maritime, Kent, ME4 4TB, United Kingdom

bS Supporting Information ABSTRACT: A detailed study into the optimization of carbodiimide-mediated coupling of antibodies (Ab) and quantum dots (QD) for use in cellular imaging has been undertaken. This involved the grafting of commercially available carboxyl-modified QDs (Evident Technologies “Lake Placid Blue” Evitag and eBioscience’s eflour nanocrystals) with anti-Cdc8 Abs to produce conjugates with specific affinity for fission yeast tropomyosin Cdc8 protein. The water-soluble carbodiimide 1-ethyl3-(3-dimethylaminopropyl)carbodiimide (EDC) was used to activate the QDs prior to their incubation with antibody, and a range of QD-carboxyl/EDC/Ab mole ratios were used in the experiments in attempts to optimize fluorescence and bioaffinity of the conjugate products (EDC to QD-carboxyl-600 nmol/15pmol to 0.12 nmol/15 pmol and QD to Ab 120 pmol/24 pmol to 120 pmol/1.2 pmol). It was observed that a specific “optimum” ratio of the three reactants was required to produce the most fluorescent and biologically active product and that it was generated at alkaline pH 10.8. Increasing the ratio of Ab to QD produced conjugate which was less fluorescent while reducing the ratio of EDC to QD in the activation step led to increased fluorescence of product. Conjugates were tested for their possession of antibody by measurement of their absorption at OD280 nm and for their fluorescence by assay λmaxem at 495 nm. A quantitative assay of the bioactivity of the conjugates was developed whereby a standardized amount of Cdc8 antigen was spotted onto nylon membranes and reacted with products from conjugation reactions in a sandwich-type colormetric assay The “best” conjugate was used in intracellular imaging of yeast Cdc8 protein and produced brighter, higher definition images of fixed yeast cell actin structure than a fluorescein Ab conjugate routinely produced in our laboratory. The QD Ab conjugate was also significantly more resistant to photobleaching than the fluorescein Ab conjugate. Results from other experiments involving EDC, the water-soluble carbodiimide 1-cyclohexyl-3-(2-morpholinoethyl)carbodiimide metho-p-toluenesulphonate (CMC), and EDC.HCl have suggested a new reaction mechanism for EDC coupling under basic aqueous conditions. In summary, a robust understanding of commercial QD-COOH surface chemistry and the variables involved in the materials’ efficient conjugation with a bioligand using carbidiimide has been obtained along with an optimized approach for Ab QD conjugate production. A novel assay has been developed for bioassay of QD Ab conjugates and a new mechanism for EDC coupling under basic aqueous conditions is proposed.

’ INTRODUCTION QDs are small, nanometric sized crystal clusters of semiconductor materials, e.g., CdSe 10 to 20 nm diameter, which have the capability (once excited) to fluoresce in a way that is related to the material from which they are made and their sizes and shapes. When QDs are photoexcited, an electron hole pair is created which, when recombined, leads to the release of a photon. The photon released is lower in energy than that absorbed and is related to the size of the QD, i.e., the degree of quantum confinement imposed. The energy of the emitted photon is directly related to its wavelength and therefore the r 2011 American Chemical Society

color observed. “Tuning” of QDs by control of their sizes can be used to cause them to emit light of different wavelengths, i.e., light of different colors. In practice, this means that the smaller the QD the shorter the wavelength of light observed after their excitation (i.e., the “bluer” it is), while larger quantum dots emit light at longer wavelengths, toward the red end of the spectrum. For a full review of quantum dot structure and function, see ref 1. Received: August 21, 2011 Revised: September 17, 2011 Published: October 04, 2011 13888

dx.doi.org/10.1021/la203273p | Langmuir 2011, 27, 13888–13896

Langmuir The surfaces of commercially available QDs are functionalized with chemical groups that permit them to be conjugated to ligands such as proteins and nucleic acids (for use in biological applications). Very commonly, the group is streptavidin which can be linked to biotinylated biological ligands.2 4 Unfortunately, these materials are relatively expensive, and given the high affinity of the two groups, conjugation reactions are difficult to influence in terms of final surface density of bioligand. Other types of surface functional groups can also be found on QDs, e.g., carboxyl and amine groups, but are employed much less frequently by biologists in making biomolecule QD conjugates. Examples of commercially available quantum dots and a representation of their structures can be found in ref 5. In the context of biomedicine and biological research, organic fluorophores have long been used either nonspecifically for general imaging of cells and tissues or as conjugates with specific biologands for bioaffinity based imaging—most particularly when they are conjugated to antibodies or single-stranded nucleic acids. However, these types of materials often suffer from a lack of sensitivity, are relatively unstable with short half-lives, and need complex imaging systems when used in combination. QDs possess several advantages over organic fluorophores in these contexts in that they are significantly more photostable, possess greater extinction coefficients, and are (relatively) insensitive to pH.6 8 They also possess longer fluorescence lifetimes and relatively large Stokes shifts with emission spectra that are narrower and more symmetrical, as well as larger quantum yields. In addition, QDs can also be excited over a very wide range of wavelengths which means that special filter pairs currently used for work with organic fluorophores are unnecessary and that a range of different QDs can be excited simultaneously using a common light source. Effectively, QDs therefore yield brighter signals for longer periods of time with respect to organic fluorophores. In a practical sense, these are also important attributes, as when imaging fluorescently labeled cells, the slide on which they are fixed must be scanned for a single cell or group of cells of sufficient quality to use for experimental purposes. Organic fluorophores fade quickly during this process, and well labeled cells must be located and imaged quickly to avoid losing image quality. QD Ab conjugates have the potential to overcome this problem and allow slides to be scanned for long periods, without any noticeable decrease in fluorescence intensity. QDs also potentially allow the possibility of storing “stained” fixed cells for reimaging for longer periods than cells labeled with classical fluorophores. Consequently, QD conjugates with bioligands, e.g., Abs or oligonucleotides, have been used for diagnostic and research purposes for in vivo and in vitro imaging of disease and cellular processes and structures. For a comprehensive list of references relating to QD conjugates, preparation, and applications in biological systems, see ref 9 from which it is also clear that no applications have used carbodiimide carboxyl amine coupling chemistry in the production of conjugates. Rather streptavidin biotin interaction has been employed. Given the above, it is surprising that no detailed investigation into the variables involved in the formation of the QD Ab conjugates using classical carbodiimide amine conjugation chemistry has been undertaken so far. Such conjugation is cheap (relatively when compared with biotin streptavidin) and facilitates the density of surface-conjugated groups to be modulated so as to optimize the materials in bioapplications. A probable reason for this (from personal experience within the School of Biosciences at the University of Kent) is that many biologists fail

ARTICLE

to successfully conjugate amines with carboxyls using carbodiimides because they lack a complete understanding of the chemical mechanism involved and what to do when it fails. EDC mediated couplings are common and taken for granted, but even so, most chemists also appreciate that the coupling reaction does not always work in the way either expected or desired. Finally, QDs are expensive and bioresearchers generally do not want to expend effort in optimizing a chemical method simply to produce something that can be tried as an alternative to classical fluorophore cellular imaging. This study has attempted to aid in that process and details an investigation utilizing COOH surface functionalized Evitag QDs and eBioscience’s eflour nanocrystals with an antifission yeast tropomyosin Ab to produce a conjugate with usefulness in intracellular imaging. A comparison of the “best” QD Ab conjugation product with an existing Ab fluorescein conjugate routinely used in our laboratory was performed, and results indicated that robust, reproducible QD Ab conjugates could be made which were capable of better (longer and higher intensity) imaging of cells than the organic fluorophore conjugate alternative. In addition, a novel bioassay for QD Ab conjugates was developed, and a new reaction mechanism for basic aqueous carbodiimide-mediated conjugation of carboxyl and amine groups is proposed.

’ MATERIALS AND METHODS Antibody Production. See Supporting Information. Quantum Dot Antibody Conjugation (General Protocol). Carboxyl-modified quantum dots (QDs) (Evident technologies’ “Lake Placid Blue” Evitag, Evident Technologies, New York, USA; or eBioscience’s eflour nanocrystals - Carboxyl-Functionalized eFluor 490NC, eBioscience Inc., San Diego, USA—see later) were activated by adding EDC (free base Sigma-Aldrich product number 39391) at the mole ratios shown in Table 1 in a total volume of 11 μL. The mixtures were incubated for 5 min at ambient temperature after which affinity-purified antibody (referred to as anti-Cdc 8 or Ab—see Supporting Information and ref 10) was added followed by 10 μL of 10 phosphate buffered saline (PBS) (Sigma-Aldrich, Poole, UK) and deionized water to a volume of 100 μL. The reaction was then incubated at ambient temperature for 1.5 h on a rotating wheel in the dark. This protocol is based on that described by the manufacturers, and both Evident Technologies and eBioscience indicate that there is an average of eight COOH groups per QD. All QDs were supplied as buffered suspensions/solutions. Conjugate Separation. Ultrafiltration. Vivaspin 500 filter cartridges (Sartorius, Gottingen, Germany) were pre-rinsed 3 times with 200 μL 0.1 mM KOH, by centrifugation at 6000 g, for 5 min to remove glycerine and sodium azide and then washed 3 times with 200 μL deionized water in the same manner. QD Ab conjugation reactions were then added to the filter cartridges which were centrifuged at 6000 g for 5 min to produce a retentate volume (sample above the filter) of approximately 25 μL. The retentate was washed 3 times with 200 μL PBS to remove unconjugated Ab and QDs by centrifugation at 6000 g, for 5 min, and subsequently removed and made up to a final volume of 100 μL in PBS. Conjugate Spectroscopy (Final Protocol). OD280 nm. 100 μL conjugate/filter flowthrough was assayed at OD280 nm using a UV spectrophotometer (Varian Carey 50 Bio, Palo Alto, USA) in microvolume cuvettes (BrandTech cat no. 7592 00, Essex, USA) with a path length of 1 cm. Fluorescence. 100 μL of conjugate/fractions were assayed for fluorescence in white 384 well microtiter plates (Corning Costar, Corning, USA) using a fluorescence spectrophotometer (Varian Carey Eclipse, 13889

dx.doi.org/10.1021/la203273p |Langmuir 2011, 27, 13888–13896

Langmuir

ARTICLE

Agilent Technologies, Santa Clara, CA, USA), at 600 v at λmax excitation 468 nm and λmax emission 495 nm. Western Blot Assay. 100 μL of standards, QD Ab conjugates, and flowthroughs from QD Ab separations were applied directly to a methanol-activated PVDF membrane (Millipore, Billerica, MA, USA) and dried at 37 °C for 4 h. Dried membranes were then incubated at room temperature with antirabbit alkaline phosphotase secondary antibody (1 in 10 000 dilution, Sigma-Aldrich, Poole, UK) in PBS and 3% w/v skimmed milk powder for 1 h with gentle agitation followed by 3 washes in PBST, 10 min each. Washed membranes were then incubated with development buffer (0.1 M NaCl, 0.1 M Tris, pH 9.6, 5 Mm MgCl2) at ambient temperature for 1 min, the excess buffer discarded, 3 mL of BCIP/NBT-Purple Liquid Substrate (Sigma-Aldrich, Poole, UK) added, and the reaction terminated by washing the membranes in deionized water. The membranes were dried and scanned using an Epson Perfection 1640SU scanner (Tokyo, Japan) and data analyzed using ImageJ software (http://rsb.info.nih.gov/ij/). A standard curve constructed using a known antibody concentration range was used for comparison in each experiment: standards of affinity purified anti-Cdc8 antibody were prepared at concentrations of 1.23 10 2, 6.15 10 3, 2.46 10 3, 1.23 10 3, and 6.15 10 4 nanomol, spotted onto membrane, and treated in an identical manner to that described above for QD Ab conjugates. Values plotted represent an average of three separate results. Cellular Imaging. Fixed cell samples on glass slides were visualized using an Olympus IX71 microscope (Olympus Corp., Tokyo, Japan)

Table 1. Ratios of Reaction Components Employed in Conjugations ratio

mole ratios of EDC to QD

mole ratios of EDC to COOH

40000:1 5000:1

4800 nmol:120 pmol 600 nmol:120 pmol

600 nmol:15 pmol 75 nmol:15 pmol

1000:1

120 nmol:120 pmol

15 nmol:15 pmol

500:1

60 nmol:120 pmol

7.5 nmol:15 pmol

250:1

30 nmol:120 pmol

3.75 nmol:15 pmol

8:1

0.96 nmol:120 pmol

0.12 nmol:15 pmol

fitted with a PlanApo 100 OTIRFM-SP 1.45 NA lens mounted on a PIFOC z-axis focus drive (Physik Instrumente, Karlsruhe, Germany) and illuminated using an automated 300 W xenon light source (Sutter, Novato, CA) with ET-sedat filters (Chroma, Bellows Falls, VT). Images were recorded using a QuantEM CCD camera (Photometrics, Tucson, AZ) controlled by Metamorph software (Molecular Devices, Sunnyvale, CA). Digital deconvolution was performed using MediaCybernetics Autoquant X software (Maryland, USA).

’ RESULTS AND DISCUSSION QD Ab Conjugate Production. Commercially available QDs with surface COOH functional groups were obtained from Evident Technologies and anti-Cdc8 antibodies were produced by us. These were used in all experiments unless otherwise indicated. In all QD Ab conjugation reactions, a surface QD carboxyl group density of 8 per QD was assumed (information provided by Evident Technologies). It was also assumed that the reaction stoichiometry of the carbodiimidemediated coupling was 1:1:1 ( COOH/EDC/Ab) and reactions were always carried out as a single-step procedure where all reactants were present as a mixture. Ab was produced as described in the Supporting Information where a gel showing purified Ab protein along with a description of its concentration estimation is also presented. In our experiments, nanomole ratios of 4800, 600, 120, 60, 30, and 0.96:120 pmol EDC to QD were employed, equating to mole ratios of 600 nmol:15 pmol, 75 nmol:15 pmol, 15 nmol:15 pmol, 7.5 nmol:15 pmol, 3.75 nmol:15 pmol, and 0.12 nmol:15 pmol with respect to QD surface carboxyl groups. Picomole ratios of 120:24, 120:12, 120:6, 120:1.2, 12:24, 12:12, and 12:6 EDC activated QD to Ab were employed (data not shown for the last three ratios). Control reactions consisted of QD reacted with EDC without Ab and QDs reacted with Ab but without EDC, and all reactions were performed in PBS pH7.0, at ambient temperature for 90 min with end-over-end mixing as described by the Evident Technologies literature.

Table 2. Fluorescence and Absorbance of Conjugation Reaction and Control Retentates and Flowthroughsa

Conjugate Ab/IgG content as assayed using the immunoassay approach developed in this work. Values represent the mean arbitrary fluorescence and OD280 nm values (as measured fluorimetrically or spectrophotometrically after zeroing against buffer) with corresponding range in brackets and the amount of Ab in the conjugate as determined by immunoassay (see later) in nanomoles, which is also expressed as a percentage of total Ab added to the reaction. “Optimum” conditions are highlighted in green. Control reactions are highlighted in grey. N = at least 3 for each value. The reaction EDC:QD 0:120 pmol, QD:Ab 120 pmol:0 (yellow highlight) was not subjected to filtration. (Wash fluorescence and OD280 nm values are not displayed, as they were equal to or less than the blanks used for zeroing the fluorimeter and spectrophotometer.) a

13890


Langmuir Initially, several unsuccessful attempts at size exclusion chromatography (as recommended by Evident Technologies) for the separation of conjugates from excess reactants were made (see Supporting Information). While some groups have successfully used this approach for QD Ab conjugate separation from reactants,11 others have specifically highlighted a problem when attempting to elute QD fractions from gel filtration columns.12 In our case, this was likely to have resulted from nonspecific adsorption of QDs to the column matrix material and QDs were also observed to bind nonspecifically to various other substrates throughout this work. This observation is mirrored by other reports where surface-functionalized QDs have been observed to bind nonspecifically to proteins and cell membranes.13,14 While attempting to develop a microtiter plate based assay for conjugate Ab content determination, it was observed that QD nonspecific binding to plate wells coated with Cdc8 protein could be abolished by the addition of a blocking agent, bovine serum albumin (see Supporting Information). Consequently, separation of the conjugation reactions by ultrafiltration (as used by others7,15,16) was employed in all experiments (see Materials and Methods) and the retentate and flowthrough fractions (as well as washings) from all conjugations were assayed for their fluorescence and antibody content by absorbance at λ280 nm (see Materials and Methods). Table 2 presents results from these experiments. It is apparent from Table 2 that the sum of fluorescence values from the retentate, flowthrough, and washes for QDs alone (row 15) did not sum to the value observed for QD fluorescence (row 16). In fact, in no case did the sum of fluorescence values for retentates, flowthroughs, and washes reach that value. This suggested that either (i) some of the QDs or QD Ab conjugates were retained by the filter membrane or (ii) QD fluorescence was reduced by the other components present in the reactions. In support of (i) was the fact that when observed closely the membranes used in row 16 experiments appeared slightly yellow in color, similar to that of the QD stock solution, which even extensive washing could not remove. This nonspecific adsorption of QDs to the filter membrane could have been responsible for reducing conjugate yield and have contributed to the degree of variation observed in terms of fluorescence between batches of conjugate produced. The second hypothesis is supported by the fact that fluorescence values observed for control reactions where QDs were present with either Ab or EDC were lower than for QD alone. Interestingly, almost all fluorescence measured postseparation for all reactions was present in the retentate indicating that the QD/EDC, QD/Ab mixtures, and QD Ab conjugates did not pass through the filter. No fluorescence was observed in either retentate or flowthrough for control reactions where only Ab was present. A number of trends were apparent from the results, which can be summarized as (i) lowering EDC concentration increased retentate fluorescence at constant Ab concentration, (ii) increasing Ab concentration decreased retentate fluorescence, (iii) lowering EDC concentration increased retentate OD280 nm at constant Ab concentration, (iv) increasing Ab concentration decreased retentated OD280 nm, (v) lowering EDC concentration decreased flowthrough fluorescence at constant Ab concentration, and (vi) lowering EDC concentration decreased flowthrough OD280 nm irrespective of Ab concentration. From Table 2, it can be observed that the greatest fluorescence was observed in those conjugation reactions where EDC and Ab concentration were lowest in terms of ratio to QD. In later

ARTICLE

experiments, conjugate Ab content was also assayed in terms of its biological activity. The pH of all conjugation reactions was measured, and this is reported in the final column of Table 2. Carbodiimide coupling procedures are typically carried out at a pH range 3.0 5.0, and it is believed necessary to keep accurate control over the pH during the coupling reaction. This is because, on one hand, the carbodiimide needs to be activated by protonation of an imide group nitrogen and, on the other hand, the carboxyl group should ideally remain mostly in its deprotonated form (RCOO ) so as to act as a nucleophile in the reaction. In no case did the pH of the conjugation reactions approach those values (see later). In fact, a specific mole ratio of EDC to QD carboxyl groups was required for most efficient conjugation regardless of antibody concentration. Mole ratios of EDC to QD higher or lower than this optimum resulted in conjugates possessing little to no Ab. Overall the efficiency of conjugation was low with the percentage of Ab incorporated into the “best” conjugate at only 11% of the total added to the reaction. It has been reported that EDC reacts with carboxyl groups most efficiently between pH 3.5 and 4.5 and that amide formation occurs optimally at pH 5.0 (see later), but interestingly, our best results were produced at a pH of 10.8. Theoretically, all reactions were buffered to pH 7.5. Another series of experiments was conducted in order to investigate the possible contributions of EDC and Ab to fluorescence and OD280 nm values of conjugates. First, fluorescence and OD280 nm values for the three different EDC concentrations used in conjugation reactions were measured (Table 3A), and it was observed that fluorescence in all cases was close to zero, but OD280 nm of the solutions was EDC concentration dependent. Increasing EDC concentration led to higher OD280 nm values. Second, the fluorescence and OD280 nm values for the three EDC concentrations and a single QD concentration were measured (see Table 3B). It this case, it was observed that the presence of QDs increased the OD280 nm values for each EDC concentration but that fluorescence of the three mixtures was more or less equal. Third fluorescence and OD280 nm values for mixtures of the three EDC concentrations with QD and Ab were measured before (i) and after (ii) separation by ultrafiltration. See Table 3C(i) and (ii). In the case of (i), it can be observed that the addition of Ab reduced fluorescence at the highest EDC concentration but that at the two lower concentrations fluorescence was almost equal. The addition of Ab produced no significant difference in OD280 nm compared to when only EDC and QDs were present. After separation by ultrafiltration (see Material and Methods), retentate and flowthrough fractions were assayed for fluorescence and OD280 nm. Table 3C(ii) illustrates results from these experiments where it can be seen that lowering EDC concentration resulted in increased retentate fluorescence and OD280 nm. As EDC concentration was reduced, flowthrough fluorescence and OD280 nm decreased. These observations are consistent with the trends observed in Table 2. The observation that Ab could quench QD fluorescence in a concentration-dependent manner has also been reported recently by others who have observed QD quenching when they and fluorescent proteins have been used in FRET (F€orster resonance energy transfer) applications.17 In fact, up to 90% quenching of QD fluorescence was observed by Hering et al.18 Fluorescence quenching of QDs by carbon nanotubes,19 metal nanoparticles,20,21 and by modified oligonucleotides22 has also been reported. Given these observations and the observations of others (noted above), it was effectively impossible to quantify the amount of 13891


Langmuir

ARTICLE

Table 3. Fluorescence and OD280 nm Absorbance Values for EDC and EDC, QD, and Ab Mixturesa EDC

Fluor

OD280

4800 nmol

5.28 ( 0.76

0.68 ( 0.001

600 nmol

4.10 ( 0.65

0.17 ( 0.007

120 nmol

1.73 ( 0.55

0.02 ( 0.006

Fluor

OD280

EDC:QD 4800 nmol:120 pmol

717.83 ( 12.30

1.11 ( 0.004

600 nmol:120 pmol

810.15 ( 5.87

0.49 ( 0.004

120 nmol:120 pmol

700.18 ( 3.29

0.35 ( 0.001

(i) Total EDC:QD:Ab

Fluor

OD280

4800 nmol:120 pmol:12 pmol

345.73 ( 8.51

1.213 ( 0.046


713.33 ( 63.55

0.471 ( 0.009


762.41 ( 8.04

0.401 ( 0.040

(ii) Separated EDC:QD:Ab

R Fluor

R OD280

FT Fluor

FT OD280


14.6 ( 1.76

0.015 ( 0.007

38.84 ( 3.65

0.214 ( 0.012


181.73 ( 34.76

0.214 ( 0.020

10.40 ( 4.39

0.041 ( 0.007


214.20 ( 1.54

0.215 ( 0.010

7.130.49

0.023 ( 0.012

a

Values refer to ratios of components with respect to COOH groups and relate to those used in conjugation reactions. EDC (A), EDC and QDs (B), and EDC, QDs with Ab before (C i) and after separation by ultrafiltration (C ii). R = retentate, FT = flowthrough.

Ab QD conjugate which was being formed in the reactions. In addition, the conjugations represented very small amounts of expensive material, rendering alternative physical or chemical analysis impossible. It was therefore decided that the best way to analyze the conjugates was to measure their biological activity. Antibody Content and Biological Activity of QD-Ab Conjugates. After several different attempts to measure the biological activity of the conjugates, a sandwich-type immunoassay was developed to measure conjugate ability to react with antigen (see Materials and Methods and Supporting Information). In this approach, a known volume of conjugate was spotted and dried onto a PVDF membrane and then reacted with an alkaline phosphatase coupled secondary antibody. A soluble, uncolored substrate was then added which could be converted to an insoluble blue product by the enzyme. Blue color development depended on the amount of enzyme present and time of incubation. (For a full description, see Materials and Methods and Supporting Information.) Conjugates produced under various conditions (4800 nmol:120 pmol, 600 nmol: 120 pmol, 120 nmol: 120 pmol EDC:QD, and 120 pmol:12 pmol, 120 pmol:6 pmol, 120 pmol:1.2 pmol QD:Ab) were then subjected to the assay, and one example of a “developed membrane” is presented in Figure 1. The standard curve used for comparison in this experiment is also presented and the results are given in Table 2. Details of the development of the assay are given in the Supporting Information. From these results, it was clear that a ratio of 600 nmol:120 pmol EDC:QD appeared to produce best conjugates in terms of antibody amount and activity regardless of antibody concentration. Ratios of 4800 nmol:120 pmol and 120 nmol:120 pmol always resulted in conjugates possessing little to no Ab. Using a ratio of 120 pmol:6 pmol QD to Ab resulted in a conjugate containing the highest percentage (11.83%) of Ab when compared to the total amount used in the reaction and the greatest

amount, 3.88 10 4 nmol (as calculated by comparison with the standard curve). Lower ratios of QD to Ab resulted in a greater percentage of Ab detectable in flowthrough fractions. Interestingly, the pH of the reaction producing the “best” conjugate in terms of this assay was 10.8, which was unusually alkaline for a watersoluble carbodiimide mediated coupling. QD Ab Conjugate and a FITC Ab Conjugate Photobleaching Characteristics. In order to compare the relative resistance to photobleaching of QD Ab conjugates against that of classical fluorophores, a bleaching experiment was performed. Equivalent samples of a fluorescein isothiocyanate (FITC) conjugated antirabbit antibody (Sigma Aldrich, Poole, UK; produced by diluting the commercial fluorophore Ab conjugate until it possessed the same fluorescence as the QD Ab) and a QD Ab conjugate (produced using optimum conditions—see Table 2 highlighted green) were illuminated with a white light source in a fluorescence microscope (Materials and Methods) and fluorescence intensity for the two materials recorded over time. The values were normalized by expressing each intensity value as a decimal of the maximum observed intensity, and plotted as a graph (Figure 2). While the FITC conjugate had “bleached” to background level within 200 s, the QD conjugate still remained fluorescent after 600 s. Cellular Imaging Using Quantum Dot Antibody Conjugates. Immunofluorescence microscopy was used to assess the performance of the QD Ab conjugate produced in intracellular imaging. These experiments were performed with conjugate produced using eBioscience’s eflour nanocrystals (CarboxylFunctionalized eFluor 490NC, eBioscience Inc., San Diego, CA, USA) which were recommended by Evident Technologies as the equivalent alternative as a consequence of the company ceasing business. This also provided the opportunity to test another carboxylated QD in the optimized coupling reaction. 13892


Langmuir

ARTICLE

Figure 3. Comparison of quantum dot and FITC immunofluorescence of fixed yeats cell actin structure. Immunofluorescence microscopy was performed against Cdc8 in WT S. pombe cells using a QD-labeled primary antibody (direct) and a FITC-labeled secondary antibody (indirect). The top two panels show a maximum projection of 21 sections taken at separate focal planes, using identical exposure settings. The CAR is clearly visible (arrow heads). The bottom two panels show the same cells after deconvolution by imaging software. Cdc8 decorated actin filaments can be seen (fine arrows). An unknown structure can be observed in the images produced with QDs (thick arrows). Scale bar equal to 5 μm.

Table 4. EDC, CMC, and EDC.HCl Solution pHs in Water and PBSa EDC

Figure 1. Membrane immunoassay. (A) One example of a “developed” PVDF membrane spotted with known concentrations of anti-Cdc8 antibodies and the retentate and flowthrough of 14 samples (see table). (B) Standard curve constructed from a range of anti-Cdc8 antibody (Ab) concentrations with plotted values representing the average of 3 values for each antibody concentration. The amount of antibody in each conjugate was calculated using the gradient of the trend line.

a

Figure 2. Effects of photobleaching on QD Ab and FITC Ab conjugate. Fluorescent intensity of a QD Ab conjugate and a FITC Ab conjugate over time. Values for intensity have been normalized against the maximum for each fluorophore.

Wild-type S. pombe cells were grown to mid log phase, fixed, permeablized, and incubated with conjugate produced using the

CMC

EDC.HCl

pH in

pH in

pH in

pH in

pH in

pH in

Amount

H2O

PBS

H2O

PBS

H2O

PBS

4800 nmol

11.66

11.57

6.40

7.29

6.10

7.30

600 nmol

11.36

11.05

6.22

7.31

6.40

7.34

120 nmol

10.69

9.48

6.50

7.32

6.04

7.35

60 nmol

10.26

7.89

6.52

7.34

6.54

7.32

30 nmol

9.90

7.48

6.20

7.31

6.32

7.33

0.96 nmol

8.19

7.30

6.40

7.32

6.50

7.33

Controls: pH H2O = 8.10; pH PBS = 7.35.

reaction conditions highlighted green in Table 1 (see Materials and Methods) with the exception that an additional step to quench EDC activated, but non-antibody-conjugated QD surface groups was included (following the manufacturer’s instructions). Such groups can serve as sites for nonspecific interaction of the materials if left unquenched and the process involved incubating the QD Ab conjugate with 50 mM Tris, pH 7.0, for 15 min after ultrafiltration and before conjugate washing. Cells which were probed with nonquenched QD Ab conjugates demonstrated increased levels of nonspecific background fluorescence (data not shown). Images were obtained from cells probed with a QD Ab conjugate and compared with those from cells probed with an unlabeled primary Cdc8 antibody followed by FITC labeled secondary antibody23 and control experiments involved probing cells with EDC-activated, but non-Ab-conjugated QDs. In this 13893


Langmuir

ARTICLE

Figure 4. Reaction mechanism for EDC-mediated carboxyl-amine coupling under basic aqueous conditions. Stage 1 represents formation of isoacylurea, while stage 2 represents conjugation and the reaction likely to occur in excess base, i.e., above pH 10.8. 13894


Langmuir latter case, no defined intracellular fluorescent labeling was observed, but rather a low level of background nonspecific fluorescence (data not shown). Results from the QD Ab and primary/FITC-labeled secondary antibody experiments are shown in Figure 3, the top 2 panels of which show cells imaged with FITC and QDs where the contractile actomyosin ring (CAR) was clearly visible. These figures are composites composed of multiple images taken at different focal planes throughout the cell, which are then flattened into a two-dimensional image (maximum projection). All images were captured under identical exposure conditions and are directly comparable. It is immediately evident that the image acquired using QD Ab conjugate is significantly brighter than when FITC was employed. The bottom panels show the digitally deconvolved version of the image (see Materials and Methods.). Digital deconvolution is a process that removes background fluorescence and sharpens the image to produce the best possible picture. In this process, contrast and brightness is automatically adjusted to an optimum, and it is therefore no longer possible to directly compare fluorescence intensity between the two approaches. However, the image produced using QD Ab conjugate is at least as good as that obtained when using FITC, and the signal appears sharper and more detailed in the case of the QD conjugate. Cdc8 protein localized to actin filaments around the cell periphery can also be observed in both images, but appears more discrete in the image produced using QD Ab conjugates. Cdc8 decorated actin cables can also be observed in the QD Ab composite image, but in the FITC image, they are indistinguishable from the background. Another distinct structure can also be observed in the image produced using the QD Ab conjugate. The structure resembles a cortical ring, located close to the cell tip (Figure 3 thick arrows). The only structure like this previously reported is the acto-myosin ring at the medial plane of cells undergoing mitosis.24 The function of this structure is as yet unknown, but one possibility is that this structure represents a “scar” of CAR proteins left from a previous cell division. This structure has never been observed when imaging Cdc8 with classical fluorophores. No fluorescent signals were observed from control experiments. pH and Its Effect on the QD Ab Conjugation Reaction. As noted previously, the best QD Ab coupling was observed to occur at very alkaline pH with respect to those noted by others.25 From the results presented in Table 2, it seemed likely that it was the EDC that was affecting reaction pH, and consequently, it was decided to dissolve EDC into water and phosphate buffer (at the concentrations used in conjugation reactions) so as to measure the resulting solutions pHs. The results are described in Table 3 and clearly show that increasing EDC concentration leads to an increasingly alkaline solution even in phosphate buffer. Our reactions were all buffered at pH 7.35 according to the QD manufacturer’s protocols in 0.01 M phosphate buffer with 0.154 M NaCl, and although a wide range of carbodimide concentrations were used in these experiments, they are representative of values reported by others in the literature (e.g., see refs 26, 27, 28, and 29 [for EDC-N-hydroxysuccinimide activation]). It is clear from Tables 2 and 4 that at any mole ratio of EDC to carboxyl group greater than 15 nmol the pH of the reaction was above 7.35 and the buffering capacity of the system had been exceeded. Interestingly, the reaction chosen as producing the best conjugate occurred at pH 10.8, and it seems very unlikely that the mechanism proposed by Nakajima and Ikada25 for carbodiimide-mediated carboxyl amine coupling

ARTICLE

reactions under acidic aqueous conditions was functioning, i.e., that the reaction was driven by the dissociation of the carboxylic acid to its conjugate base (carboxylate ion) and acid (H+) along with subsequent protonation of the carbodiimide and attack of the positively charged carbon center by the carboxylate ion. We suggest that the reaction scheme given above could be a possible alternative for carbodiimide coupling under basic conditions to that proposed by Nakajima and Ikada for coupling under acidic conditions with EDC.HCl. In our proposal, the EDC dimethylamino group is protonated by water leading to structure A (Figure 4). This would account for the increase in pH observed in conjugation reactions and in aqueous solutions of the EDC as a consequence of the accumulation of OH ions in solution. In the first stage of the reaction, coupling (Figure 4) OH ions would then cause the deprotonation of the carboxylic acid and the carboxylate (structure B, Figure 4) generated would be available to attack the carbodiimide carbon. The strained NdCdN group would react with the carboxylate ion as shown causing the formation of the isoacyl urea intermediate (Structure C, Figure 4) which would subsequently react with the primary amine as described above in stage 1 in an identical manner to that proposed by Nakajima and Ikada to yield structures D and E (Figure 4). We also postulate that the reason that this reaction is terminated at very high pH is because a dynamic equilibrium exists between the carboxylic acid and OH species such that when OH predominates it outcompetes the carboxylate and reacts with the isoacyl urea to form structures F and G (urea and the carboxylic acid; Figure 4). The proposal of Nakajima and Ikada that formation of the acid anhydride may also promote conjugation is unlikely to be the case in our system, as any anhydride that may form would be rapidly hydrolyzed and removed from the reaction under basic aqueous conditions. In a brief attempt to confirm the above hypothesis, an experiment was conducted in which the water-soluble carbodiimide, 1-cyclohexyl-3-(2-morpholinoethyl)carbodiimide toluene metho-p-sulfonate (CMC), was dissolved at the same concentrations as those tested for EDC in water and phosphate buffer. CMC has no dimethylamino group but does possess a quaternary ammonium salt group in its side chain. The resulting solution pHs were measured, and in this case, no change in pH was observed regardless of CMC concentration. See Table 4 for results. Interestingly, when we repeated the same experiment with EDC.HCl, the commonly used salt of EDC also used in coupling reactions, this also had no effect on solution pH either in water or in phosphate buffer. These results strongly confirm the hypothesis above. An analogy can be drawn between this proposal and what happens in aqueous solutions of amino-functionalized organosilanes such as aminopropyltriethoxy silane. Such solutions rapidly become strongly alkaline as a consequence of protonation of the silane amine group and accumulation of OH ions in solution. These ions then drive the hydrolysis of the silane alkoxy groups promoting silane polymerization and condensation.30 Such a situation does not occur in the case of tetra alkoxysilanes where no change in their aqueous solution pH is noted.

’ CONCLUSIONS This work has shown that it is possible to reproducibly synthesis QD-Ab conjugates for intracellular imaging of cells which perform better than those currently available where organic fluorophores are present via a water-soluble carbodiimide mediated approach for carboxyl amine coupling. The work has extensively investigated 13895


Langmuir the effects of reaction variables on the formation of conjugate and has established a new method for conjugate bioactivity assay. Optimum conditions for the conjugation reaction in phosphate buffer represented a mole ratio of EDC:QD carboxyl group/Ab of 75 nmol:15 pmol:6 pmol. Images obtained when using the QD Ab conjugates in immunofluoresence of yeast cellular actin structure were observed to be significantly brighter than those obtained when using organic fluorophores and displayed a greater degree of structural detail. The QD Ab conjugates were also significantly more resistant to photobleaching than a comparable organic fluorophore Ab conjugate. The optimum pH for the coupling reaction was observed to be 10.8, and a new mechanism has been proposed for water-soluble carbodiimide-mediated amine carboxyl coupling under basic aqueous conditions.

’ ASSOCIATED CONTENT

bS

Supporting Information. Additional results from experiments relating to Ab production and purification as well as separation and quantification of QD-Ab conjugates. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION

ARTICLE

(16) Yu, H.; Lee, J.; Kim, S.; Nguyen, G.; Kim, I. Anal. Bioanal. Chem. 2009, 394, 2173. (17) Dennis, A. M.; Bao, G. Nano Lett. 2008, 8, 1439–1445. (18) Hering, V. R.; Faulin, T. E.; Triboni, E. R.; Rodriguez, S. D.; Bernik, D. L.; Schumacher, R. I.; Mammana, V. P; Faljoni-Alario, A.; Abdalla, D. S.; Gibson, G.; Politi, M. J. Bioconjugate Chem. 2009, 20, 1237. (19) Biju, V.; Itoh, T.; Baba, Y.; Ishikawa, M. J. Phys. Chem. B 2006, 110, 26068–26074. (20) Dyadyusha, L.; Yin, H.; Jaiswal, S.; Brown, T.; Baumberg, J. J.; Booy, F. P.; Melvin, T. Chem. Commun. (Camb.) 2005, 3201. (21) Zhang, J.; Badugu, R.; Lakowicz, J. R. Plasmonics 2008, 3, 3. (22) Lee, J.; Choi, Y.; Kim, J.; Park, E.; Song, R. ChemPhysChem 2009, 10, 806. (23) Martin-Garcia, R.; Mulvihill, D. P. J. Cell Sci. 2009, 122, 3862. (24) Noguchi, T.; Arai, R.; Motegi, F.; Nakano, K.; Mabuchi, I. Cell Struct. Funct. 2001, 26, 545. (25) Nakajima, N.; Ikada, Y. Bioconjugate Chem. 1995, 6, 123–130. (26) http://www.siercheng.com/UploadFile/200961194140360.pdf (27) http://www.appliedcytometry.com/Sample_Protocols/MagPlex_ beads/Magnetic_Oligo_Coupling.pdf (28) http://probes.invitrogen.com/media/pis/mp19020.pdf (29) Sam, S.; Touahir, L.; Andresa, J. S.; Allongue, P.; Chazalviel, J.-N.; Gouget-Laemmel, A. C.; Henry de Villeneuve, C.; Moraillon, A.; Ozanam, F.; Gabouze, N.; Djebbar, S. Langmuir 2010, 26, 809. (30) Bruce, I. J.; Sen, T. Langmuir 2005, 21, 7029.

Corresponding Author

*E-mail: [email protected].

’ ACKNOWLEDGMENT We thank the European Union FP6 NACBO project, contract number 500804 for financial support, and Dr. Dag Rother for the reaction graphic. ’ REFERENCES (1) Alivisatos, A. P. Science 1996, 271, 933. (2) Giepmans, B. N.; Deerinck, T. J.; Smarr, B. L.; Jones, Y.; Ellisman, M. Nat. Methods 2005, 2, 743. (3) Kingeter, L. M.; Schaefer, B. C. BMC Biotechnol. 2009, 9, 49. (4) Li, R.; Dai, H.; Wheeler, T. M.; Sayeeduddin, M.; Scardino, P. T.; Frolov, A.; Ayala, G. E. Clin. Cancer Res. 2009, 15, 3568. (5) http://www.ebioscience.com and http://www.invitrogen.com (6) Alivisatos, A. P.; Gu, W.; Larabell, C. Annu. Rev. Biomed. Eng. 2005, 7, 55. (7) Bruchez, M., Jr.; Moronne, M.; Gin, P.; Weiss, S.; Alivisatos, A. P. Science 1998, 281, 2013. (8) Nirmal, M.; Brus, L. Acc. Chem. Res. 1999, 32, 407. (9) http://www.invitrogen.com/site/us/en/home/brands/MolecularProbes/Key-Molecular-Probes-Products/Qdot/Qdot_Citations.html and others at http://www.invitrogen.com (10) Skoumpla, K.; Coulton, A. T.; Lehman, W.; Geeves, M. A.; Mulvihill, D. P. J. Cell Sci. 2007, 120, 1635. (11) Hua, X.; Liu, T.; Cao, Y.; Liu, B.; Wang, H.; Wang, J.; Huang, Z.; Zhao, Y. Anal. Bioanal. Chem. 2006, 386, 1665. (12) Fernandez-Arguelles, M. T.; Costa-Fernandez, J. M.; Pereiro, R.; Sanz-Medel, A. Analyst 2008, 133, 444. (13) Bentzen, E. L.; Tomlinson, I. D.; Mason, J.; Gresch, P.; Warnement, M. R.; Wright, D.; Sanders-Bush, E.; Blakely, R.; Rosenthal, S. J. Bioconjugate Chem. 2005, 16, 1488. (14) Kairdolf, B. A.; Mancini, M. C.; Smith, A. M.; Nie, S. Anal. Chem. 2008, 80, 3029. (15) Kaul, Z.; Yaguchi, T.; Harada, J. I.; Ikeda, Y.; Hirano, T.; Chiura, H. X.; Kaul, S. C.; Wadhwa, R. Biochem. Cell Biol. 2007, 85, 133. 13896


QD-Antibody Conjugates via Carbodiimide-Mediated Coupling: A

Recommend Documents