A fluorimetry comparison study has been established for the labeling of human serum albumin (HSA) using two boronic acid functionalized squarylium dyes, one is monofunctional boronic acid dye (SQ-BA4) while the other one is bifunctional boronic acid dye (SQ-DBA2). The spectral properties of these two squarylium dyes have been studied under a variety of solutions to explore the optimal conditions for non-covalent protein labeling. Both dyes exhibited very low fluorescence intensity in aqueous solutions presumably due to H-aggregate formation in absence of the protein but showed a significant enhancement in the fluorescence intensity with the addition of HSA in the optimal acidic citrate buffer. Stability constant studies were conducted and SQBA4 dye was shown to have much stronger binding affinity with HSA than SQ-DBA2. Also, the stoichiometries of non-covalent complexes were found to be (1:1) for SQ-BA4 and different (1:2) for SQ-DBA2 with HSA. These studies are claimed to be the optimal for CE-LIF work which will be progressed in our continuing project.

Sebaiy MM, El-Shanawany AA, Baraka MM, Abdel Aziz LM, Colyer CL (2015) Monofunctional and Bifunctional Boronic Acid Functionalized Squarylium Dyes as Non-Covalent Protein Labels: A Fluorimetric Study. J Drug Des Res 2(3): 1019.

•    Fluorimetry
•    Human serum albumin
•    Boronic acid
•    Squarylium dyes
•    Non-covalent
•    Labeling 

The development of novel methods and techniques for protein determination is very important in clinical and analytical applications. Spectral detection methods are widely used due to their high sensitivity and selectivity [1]. Analysis can usually be carried out using absorbance detection. However, laser induced fluorescence (LIF) detection has become a more attractive detection method [2]. This is due to the lower background, increased sensitivity, and increased selectivity that accompanies this method compared to UV-Vis absorbance [2-4]. LIF detection requires analytes (protein) to be made fluorescent by some labeling procedure if they are not already natively fluorescent at the wavelengths associated with the excitation laser sources [2,5]. Analytes can be rendered fluorescent through one of two derivatization reactions: covalent labeling and non-covalent labeling reactions [6]. Non-covalent labeling is an attractive option, because it requires less sample preparation and less analysis time than covalent labeling procedures [7,8]. Two main classes of dyes that demonstrate fluorescence in the NIR region have been explored as non-covalent protein labels: cyanine and squarylium dyes [2,7-10].

Squarylium dyes were first synthesized over fifty years ago through a condensation reaction of squaric acid with two aromatic and/or heterocyclic compounds, which produces the 1,3-disubstituted compound [11]. They are characterized by their ability to exhibit effective light absorption with long excitation and emission wavelengths, high fluorescence quantum yields, and resistance to photodegradation [9]. Structure modifications of the dye by incorporating boronic acid functional groups is claimed to improve the affinity of these dyes towards the target protein analyte as these boronic acid-based dyes had showed excellent selectivity towards monosaccharides [12,13] and gram-positive becteria [14,15] as previously established in our laboratory.

Human serum albumin (HSA) will be studied as a model binding protein. HSA, being the most abundant protein of blood plasma, has many important physiological functions like regulating colloidal osmotic pressure, and transporting numerous endogenous compounds such as fatty acids, hormones, bile acids, amino acids, metals and toxic metabolites [16,17]. Additionally, there is a wide variety of drugs that are delivered to their targeting organs/tissues by binding with HSA and so, it does not only protect the bound drugs against oxidation and influences the in vivo drug distribution, but also alters the pharmacokinetic and pharmacodynamic properties of drugs [18]. Binding properties between HSA and ligands have been studied for several decades using a variety of methods [19–23] but all of them were nonfluorescence based studies. As such, the aim of current work is to compare the optical spectroscopic properties of two related squarylium dyes, an asymmetric monofunctional boronic acidbased dye (SQ-BA4) and a symmetric bifunctional boronic acidbased dye (SQ-DBA2) while both dyes have the same squarine core structure (see Figure 1). The absorbance and emission properties of the dyes in a variety of solvents and aqueous buffers at different pHs were determined prior to titration of fixed quantities of the dyes with HSA. Also, stability constant studies and stoichiometric ratio calculations were applied to give us an overall idea about the nature of such protein-dye interactions. The solution conditions found to yield the most favorable fluorescence emission response in these studies are supposed to be optimal for determination and separation of three model proteins by CE-LIF technique in our continuous project.

Reagents, buffer, and sample preparation

Squarylium dyes, SQ-BA4 and SQ-DBA2 were supplied by Prof. H. Nakazumi (Osaka Prefecture University, Japan) [13,24]. Both stock solutions of 4.00x10-4 M SQ-BA4 and 3.50x10-5 M SQ-DBA2 were prepared by weighing out an appropriate amounts of the solid dyes and dissolving in methanol (GJ Chemicals, Newark, NJ, USA) and dimethylsulfoxide, DMSO (Sigma Aldrich, St. Louis, MO, USA), respectively. These two stock solutions were stored in the dark at 4° C while not in use, under which conditions they are stable for several months or longer. Dilute working stock solutions were prepared just before use from the original stock solutions by dilution with additional methanol or DMSO to a concentration of 1.00x10-5 M, and these solutions were also stored in the dark at 4° C when not in use. Human serum albumin (HSA) was purchased from Sigma Aldrich (St. Louis, MO, USA) and used as received. Stock solution was freshly prepared weekly to a concentration of 1.00x10-5 M in MilliQ distilled, deionized water (Millipore, Bedford, MA, USA), and was stored in the dark at 4° C while not in use. Dilute protein working stock solution was prepared from the original stock solution by dilution with additional MilliQ water to a concentration of 1.00x10-6 M, and was stored in the dark at 4° C when not in use. Four buffers were used in this study to span the pH range from acidic to basic. Appropriate quantities of citric acid (Sigma Aldrich, St. Louis, MO, USA), tris (hydroxymethyl) aminomethane (AMRESCO, Solon, OH, USA), ammonium carbonate (Sigma Aldrich, Buchs, Switzerland), and ammonium dihydrogen phosphate (JMC, Royston, Hertfordshire, UK) were dissolved in MilliQ distilled, deionized water prior to pH adjustment by the addition of 1 M NH4 OH or 1 M HCl (Fisher Scientific, Pittsburgh, PA, USA). All buffers were stored in high density poly ethylene plastic bottles at room temperature for 1-2 weeks, and were filtered through 0.20 μm nylon syringe filters (VWR International, Houston, TX, USA) prior to use. Protein-dye solution mixtures were prepared by adding 5.00 μL of the 1.00x10-5 M solution of SQ-BA4 or SQ-DBA2 to a 500 μL quartz cuvettes. Some volumes of protein working stock solution were then added to this aliquot of dye, along with sufficient buffer to achieve the final desired concentrations of both protein and dye. The final volume for all solutions was 500 μL. The total organic solvent content in aqueous buffer solutions was kept below 2% (v/v).

Synthesis and characterization of squarylium dyes

SQ-BA4 was synthesized according to the method described previously by Ouchi K et al. [13], who also presented a detailed characterization of the dye. Briefly, SQ-BA4 was synthesized by condensation of both 2,3,3-Trimethyl-5- (4,4,5,5-tetramethyl-1,3,2-dioxaborolane-2-yl)-3H-indole and 3-(4-dibutylaminophenyl)-4-hydroxy-3-cyclobuten-1,2-dione, which were synthesized according to the literature [25]. The crude product was purified by two stages of column chromatography followed by recrystallization from CHCl3 /hexane, yielding the pure compound SQ-BA4 as a brown solid in 18% yield. The full synthesis and characterization of the SQ-DBA2 dye is to be detailed in a future publication by T. Maeda et al., but the dye was used here to provide a useful comparison with the behavior of the previously published dye SQ-BA4 towards model proteins.

Instrumentation

Absorbance studies were conducted over a scan range of 400- 800 nm with an integration time of 0.5 sec and an interval of 1 nm using a Hewlett Packard HP8453 UV-Vis spectrometer (Waldbronn, Germany) and 1-cm, semi-micro (total volume 500 µL) quartz cuvettes (Perkin-Elmer, Shelton, CT, USA).

Fluorimetry studies were conducted using a Perkin-Elmer LS50B luminescence spectrometer (Shelton, CT, USA). Excitation and emission slit widths were 10 nm, the excitation wavelength (as specified in Table 1) was selected according to the wavelength of maximum absorbance for the sample or solution under consideration and the emission wavelength range was scanned from 630 nm to 800 nm at a rate of 250 nm/ min. Fluorescence spectra of both free dyes and the protein-dye mixtures were measured using a 1-cm, semi-micro (500 µL) quartz cuvette (Perkin-Elmer, Shelton, CT, USA).

Spectral properties of SQ-BA4 & SQ-DBA2 alone and protein-dye complex

Spectroscopic analysis of boronic acid-based squarylium dyes were conducted under various solution conditions in order to evaluate their suitability as fluorescent probes for proteins. Also, these studies will help us to better understand the hydrophobic interactions, electrostatic interactions, and hydrogen bonding involved in the formation of non-covalent complexes between the dyes and protein targets. The exact nature and combination of these interactions can be difficult to determine, but evidence of interaction is provided by a change in the emission of the protein –dye complex relative to that of the free, uncomplexed dye, and can be easily monitored by fluorimetric studies [24]. Since SQ-BA4 is a monofunctional boronic acid-based dye while SQDBA2 is a bifunctional boronic acid-based dye, it was expected that differences in their complexation with proteins could be attributed to the boronic acid moieties in both structures, thus providing insight into future design and planning for squarylium dye structures tailored to particular analytical needs. Additionally, this understanding proved to aid in the development of a CE-LIF protocol for the separation of various protein–dye complexes, with the promise of extension to protein assays more generally. UV-Vis absorbance and fluorescence emission studies of the dyes (1.00x10-7 M SQ-BA4 & SQ-DBA2) were conducted in acidic, neutral, and basic solution conditions, in the presence and absence of HSA in order to supposedly determine the optimal buffer conditions for further CE-LIF analysis. A summary of absorbance and emission properties of the dyes under various solution conditions, both with and without added protein, is presented in Table 1.

The absorbance of both SQ-BA4 and SQ-DBA2 dyes in aqueous buffer solutions was blue shifted relative to their values in methanol or DMSO, respectively. The blue shift in case of SQ-BA4 was between 18-36 nm while in case of SQ-DBA2, it was between 39-54 nm with a noticeable more shift in case of neutral and acidic buffers for both dyes. Also the band shape was changed where a weak broad absorbance band (around 500 – 730 nm in case of SQ-BA4 and 520 – 800 nm in case of SQ-DBA2) was observed for both dyes in acidic aqueous conditions, which suggests that they all form water-soluble aggregates (Figure 2b, red trace). Two possible types of aggregates-referred to as J and H- are possible. J-aggregates arise from the head-to-tail alignment of monomers, while H-aggregates are aligned in a head-to-head (or parallel) arrangement. Fukuda and Nakahara reported that squarylium dyes showed a strong absorbance band from 630– 640 nm in chloroform with a blue shift for H-aggregates and a red shift for J-aggregates [26]. Figure 2 (b & a, red vs. blue traces) also shows the blue shift in acidic aqueous conditions relative to non-aqueous conditions. Under non-aqueous conditions (methanol or DMSO), it is assumed that the dyes would exist predominantly in their monomeric forms with a relative much stronger and sharper band as seen in Figure 2a (around 500–670 nm in case of SQ-BA4 and 570–720 nm in case of SQ-DBA2). This would indicate that both dyes in this study were most likely forming H-aggregates in aqueous solution, in the absence of protein. Upon the addition of an equimolar amount of our model protein, HSA, λex for each dye was red shifted relative to λex for each dye alone (in the absence of protein) under all buffer conditions as seen in Table 1. Figure 2c (green trace) shows the change in absorbance for each dye upon the addition of HSA under acidic aqueous conditions. These spectral changes are indicative of a non-covalent complex being formed between each dye and the protein. Also, variations in the extent of shifts are indicative of differing strengths of protein– dye interactions under different solution conditions. In addition, these spectral changes imply disaggregation, since the λex of the protein–dye complex closely resembled that of the dye in methanol or DMSO, where it is presumed to exist in its monomeric state. Furthermore, these longer wavelength absorbances can be beneficial because absorbances at longer wavelengths, in general, allow the optical detection to be shifted away from the region of native protein absorbance bands, therefore reducing interference from the bands with the absorbance signal of the protein–dye complex. This reduces background and in turn can be beneficial in improving detection limits.

Regarding the fluorescence studies, the changes in wavelength of maximum fluorescence emission (λem) for dye alone and upon the addition of HSA are also shown in Table 1. SQ-BA4 showed a blue shift in λem while SQ-DBA2 showed a red shift in λem upon the addition of HSA, but to varying extents under different solution conditions. In case of SQ-BA4, a larger blue shift in λem was seen in the acidic buffer (15 nm) which decreases gradually once pH increases in neutral or basic buffers. On the other hand, SQ-DBA2 showed the largest red shift in λem in both acidic (pH 3.50) and basic (pH 9.80) conditions (12 nm), followed by a shift of 7 nm in neutral conditions and a shift of 3 nm in highly basic (pH 11.3) conditions. In both cases, it is noted that upon the addition of HSA to the dyes in aqueous buffered media, there was a shift in the wavelength of maximum emission, which confirms that a non-covalent complex was being formed for both dyes as was seen in the absorbance spectra. A larger shift in wavelength is desired since the wavelength for the complex is then further separated from that of the dye alone, reducing possible spectral interferences and this will ultimately improve fluorescence sensitivity. Determining the optimal buffer (in terms of analytical response) for each dye also required a study of the emission intensity of the protein–dye complexes in order to discover which buffer would provide the greatest sensitivity for determination of the complex.

According to Table 1, acidic buffer was the optimal for maximum fluorescence enhancement in case of both dyes when complexed with HSA. Fluorescence emission increased by a factor of 49.59 and 84.95 (highlighted in bold) in the case of SQ-BA4 and SQ-DBA2, respectively, upon the addition of equimolar amount of the protein under acidic solution conditions. The fluorescence enhancement factor is calculated as the difference between emission intensities in the presence and absence of analyte divided by the emission intensity of dye alone. SQ-DBA2 complex with the protein has a much higher enhancement factor than SQ-BA4 (with a ratio of 1.71 times more sensitive) and this may be attributed to the extra boronic acid moiety in the dye structure. However, both dyes showed a very sharp decrease in the fluorescence enhancement factor upon addition of protein in neutral and basic conditions as mentioned in Table 1 while the lowest enhancement factor was recorded at pH 11.30. Since the isoelectric point (pI) of HSA is 5.30 [27] (which means that HSA at this pH is neutral and carries no net electrical charge and so at a pH below its pI, it carries a net positive charge while above its pI, carries a net negative charge), both neutral and basic conditions cause HSA in addition to the dyes to be negatively charged, leading to electrostatic repulsion which would suggest an unfavorable condition for interaction at these pHs, and so such kind of hydrogen bonding, intercalation or hydrophobic interactions must contribute to the formation of the protein-dye complex. It should be noted also that this explanation is consistent with the results in Table 1 where the fluorescence enhancement factor is getting much lower values once pH increases from neutral to basic to a highly basic for both dyes. The concentration of buffer solution, however, had an impact on fluorescence enhancements where the optimal concentration of ammonium citrate (pH 3.50) was found to be between 10 up to 25 mM while increasing the concentration to 50, 75 or even 100 mM showed a less enhancement of the fluorescence emission for both dyes complexes. Through the spectroscopic analysis of these two dyes, it can be concluded that SQ-BA4 and SQ-DBA2 show an optimal fluorescence with HSA in 25 mM citrate buffer (pH 3.50).

A protein titration (with a constant dye concentration and increasing protein concentrations) was also conducted for both dyes in citrate buffer. Figure 3a shows the titration of SQBA4 (1.00×10-7 M) with HSA from 1:0 (dye with no added HSA) up to a 1:10 dye: protein ratio (beyond this ratio the emission intensity is saturated and a clear emission peak is not detected).

A linear response in emission intensity for SQ-BA4 at 621 nm as a function of added protein concentrations in citrate buffer was seen up to a concentration of 1.00×10-7 M or a 1:1 dye:protein ratio. On the other hand, the protein titration conducted for SQ-DBA2 (Figure 3b) spanned a 1:6 dye: protein ratio and the concentration of the SQ-DBA2 dye used was also 1.00×10-7 M. A linear response in emission intensity for SQ-DBA2 at 657 nm as a function of added protein concentration was observed up to a concentration of 2.00×10-7 M or a 1:2 dye: protein ratio. Table 2 compares the calibration curve results for both dyes titrated with HSA, corresponding to the linear portion of the protein titrations and linearity was high in both cases (R2 > 0.992). The sensitivity of the method is defined as the slope of the line of best fit for each calibration curve. The calibration curve (fluorescence emission as a function of protein concentrations) for SQ-DBA2 displayed steeper slope (3x10+09) and thus more sensitive than SQ-BA4 (2x10+09). A steep slope means that it is easier to see subtle changes in the concentration of analyte since the small changes in concentration lead to large changes in signal (fluorescence emission in this case).

Stability constants for protein-dye complexes

The stability constant (Ks ), also known as formation constant or binding constant, is the equilibrium constant governing the formation of a complex between two or more molecules.

It is a measure of the strength of the interaction between the components of a complex. To quantify the strength of interaction between squarylium dyes and HSA, the stability constant (Ks ) for non-covalent complexes formed between HSA and each dye was determined by performing protein titrations (recall Figure 3) and applying the method established by Patonay and coworkers [28]. According to this method, the interaction between dye (D) and protein (P) can be expressed as:

From the equilibrium expression, the stability constant Ks are defined as:

From the fluorimetric titration data it is possible to calculate Ks according to the following equation:

Where F is the measured fluorescence emission intensity of the complex and k is a constant dependent upon the instrumentation. A saturable non-linear plot of fluorescence versus the protein concentration as in case of SQ-BA4 with HSA was formed due to the dye saturation after 1:1 complex ratio (see the right top inset in Figure 4). As a result, a double reciprocal plot of 1/F versus 1/ [P] will yield a straight line. The intercept divided by the slope of this straight line equals Ks .

Stability constants thus determined varied greatly for both dyes since the boronic acid moieties affected not only their spectroscopic properties but also their affinities towards protein targets. The protein titration data shown in Figure 3 was used to construct double reciprocal plots for each dye, as shown in Figure 4. The interesting results obtained showed that SQ-BA4 formed a more stable non-covalent complex with HSA, with Ks = 3.67×106 M-1 than SQ-DBA2 (Ks = 1.17×105 M-1). These results (specially for SQ-BA4) are shown to be superior to other stability constants for some squarylium dye–protein complexes, having higher concentrations of both the dye and HSA that have been reported elsewhere. For example, samples containing concentrations of 5.00×10−6 M HSA with either 5.00×10−7 M Red-1a, Red-1b, Red-1c or Red-1d in trizma buffer (pH 7.4) were reported by Nakazumi H et al. to have association constants of 8.70×105 M-1, 1.10×106 M-1,8.00×105 M-1 and 1.70×105 M-1, respectively [7] while stability constant on the order of 105 M-1 was obtained by by Welder F et al. through the interactions of 5.25×10−6 M HSA with 5.25×10−7 M SQ-3 in citric acid buffer [24]. These reported values, determined by the same method described by Patonay and coworkers, verifies that the boronic acid group in the dye does, indeed, have an impact on its binding affinity towards protein and increased sensitivity at lower concentrations than other relative squarylium dyes. It is important to note that one limitation of this method used to calculate Ks is that it is only strictly valid for 1:1 stoichiometric complexes of dye and protein (as was the case for SQ-BA4 with HSA). However, it is still also used as a tool to support us with related approximate values for the stability constant with other stoichiometries (as for SQ-DBA2 with HSA). Other methods for measuring the association of dye and protein, which are not dependent upon stoichiometry, include frontal analysis (FA) [8] or non-equilibrium capillary electrophoresis of equilibrium mixtures (NECEEM) [29] for more precise measurements.

Dye: protein stoichiometries

The stoichiometries of both protein–dye complexes were calculated through the method of continuous variation (Job’s plot) [30]. In our fluorescence studies, a plot of fluorescence emission as a function of mole fraction of dye is constructed for various protein-dye mixtures possessing a fixed total concentration of dye plus protein which was 2.00×10-7 M. Two straight lines are formed within the plot and the mole ratio corresponding to the point at which they intersect can be used to determine the predominant stoichiometry of the protein-dye complex.

Under the prescribed optimal acidic buffer condition discussed above, SQ-BA4 and SQ-DBA2 each formed different non-covalent complexes with HSA. For SQ-BA4, the fluorescence maxima in the resulting Job’s plot occurred at a dye mole fraction of 0.50, corresponding to a predominant stoichiometry of 1:1 (dye1–HSA1). On the other hand, the fluorescence maxima in the resulting Job’s plot in case of SQ-DBA2 occurred at a dye mole fraction of 0.30 corresponding to a predominant stoichiometry of 1:2 (dye1–HSA2), as shown in Figure 5. It can be also seen from the figure that when the amount of protein was greater than that of the dye, a relative increase in signal was seen. As soon as more dye than protein was present in the solution, a decrease in signal was observed. This may be attributed to the probable chance of dye aggregation in the aqueous solution with its increasing concentration while the protein concentration is relatively decreasing and thus a tangible quenching in the fluorescence of the complex can be observed. Overall, these findings suggest that the monofunctional or bifunctional boronic acid moieties in the dye structure impact the dye binding with the protein may be through H-bonding with the protein functional groups located on its binding sites. Also, we can’t neglect the importance of hydrophobic interactions between squarylium dyes and proteins which have been previously shown [14] and the present binding studies support this. As such, the combination of H-bonding, electrostatic and hydrophobic interactions renders these dyes suitable for on-column, precolumn, and post-column protein labeling strategies under different solution conditions for subsequent protein analysis by CE-LIF.

Table 1: Spectral properties for SQ-BA4 & SQ-DBA2 with and without added HSA in organic solvents or various aqueous buffer systems (25 mM). Absorbance maxima for protein–dye complex were used as excitation wavelengths for fluorescence emission studies. Dyes and HSA concentrations were 1.00×10-6 M and 1.00×10-7 M for absorbance and fluorescence measurements, respectively.

Table 2: Calibration curves based on lines of best fit from the titration of SQ-BA4 and SQ-DBA2 (1.00×10-7 M) dyes with HSA. Experimental conditions are listed in Figure 3.

Dye	Range of HSA concentration (μM)	Slope	Intercept	R2
SQ-BA4	1.00× – 1.00×	2×	16.23	0.9933
SQ-DBA2	1.00×– 2.00×	3×	-19.47	0.9980

The spectral properties of the two squarylium dyes, monofunctional boronic acid-based SQ-BA4 and bifunctional boronic acid-based SQ-DBA2, were explored under a variety of solutions in order to determine the best conditions for non-covalent protein labeling. These dyes exhibit very low fluorescence intensity in aqueous solutions likely due to H-aggregate formation in absence of the protein. However, both dyes showed a significant enhancement in the fluorescence intensity with the addition of HSA in the optimal citrate buffer (pH 3.50). Stability constant studies were conducted and SQBA4 was shown to have much stronger binding affinity with HSA than SQ-DBA2, although it demonstrated a lower fluorescence enhancement compared to SQ-DBA2 with HSA. However, the stability constant values were shown to be superior to other values of some reported squarylium dye–HSA complexes. Additionally, the stoichiometries of non-covalent complexes were found to be (1:1) for SQ-BA4 and different (1:2) for SQDBA2 with HSA. The variations in stoichiometry indicate the importance of a combination of electrostatic, hydrophobic, and H-bonding interactions between dye and protein, and suggest the suitability of these dyes as candidates for on-column or pre-column protein labeling strategies under different solution conditions for subsequent protein analysis by CE-LIF. However, more studies will need to be conducted to that end. Those upcoming studies should include systematic buffer studies to determine the optimal conditions for protein assays by CE-LIF

The authors hereby acknowledge the original design and first-ever synthesis of SQ-DBA2 by S. Saito and T. Maeda and coworkers (Saitama University and Osaka Prefecture University), and the provision of dye samples by H. Nakazumi (Osaka Prefecture University). The authors also thank Wake Forest University, the Embassy of the Arab Republic of Egypt Cultural & Educational Bureau (Washington, D.C.), and the Egyptian government scholarship #JS2736 (M. Sebaiy) for funding.

1. Sun C, Yang J, Li L, Wu X, Liu Y, Liu S. Advances in the study of luminescence probes for proteins. J Chromatogr B Analyt Technol Biomed Life Sci. 2004; 803: 173-190.

2. Moody ED, Viskari PJ, Colyer CL. Non-covalent labeling of human serum albumin with indocyanine green: a study by capillary electrophoresis with diode laser-induced fluorescence detection. J. Chromatogr. B. 1999; 729: 55–64.

3. Grossman PD. Capillary electrophoresis theory and practice. Academic Press, San Diego, CA., USA. 1992.

4. Banks PR. Fluorescent derivatization for low concentration protein analysis by capillary electrophoresis. Tren. Anal. Chem. 1998; 17: 612–622.

5. Colyer C. Noncovalent labeling of proteins in capillary electrophoresis with laser-induced fluorescence detection. Cell Biochem Biophys. 2000; 33: 323-337.

6. Bardelmeijer HA, Lingeman H, de Ruiter C, Underberg WJ. Derivatization in capillary electrophoresis. J Chromatogr A. 1998; 807: 3-26.

7. Nakazumi H, Colyer CL, Kaihara K, Yagi S, Hyodo Y. Red luminescent squarylium dyes for noncovalent HSA labeling. Chem. Lett. 2003; 32: 804–805.

8. Yan W, Colyer CL. Investigating noncovalent squarylium dye-protein interactions by capillary electrophoresis-frontal analysis. J Chromatogr A. 2006; 1135: 115-121.

9. Yan WY, Sloat AL, Yagi S, Nakazumi H, Colyer CL. Protein labeling with red squarylium dyes for analysis by capillary electrophoresis with laser-induced fluorescence detection. Electrophoresis 2006; 27: 1347–1354.

10. Lin X, Rockett S, Massie T, Turner GB, Maeda M, Nakazumi H, et al. Asymmetric mono- and bis-squarylium dyes as pre-Column and oncolumn labels for protein analysis by capillary electrophoresis with laser induced fluorescence detection. J. Anal. Bioanal. Tech. 2012; 9.

11. Sprenger HE, Ziegenbein W. Condensation products of squaric acid and tertiary aromatic amines. Angew. Chem., Int. Ed. Engl. 1966; 6: 894.

12. Saito S, Massie TL, Maeda T, Nakazumi H, Colyer CL. A long-wavelength fluorescent squarylium cyanine dye possessing boronic acid for sensing monosaccharides and glycoproteins with high enhancement in aqueous solution. Sensors 2012; 12: 5420−5431.

13. Ouchi K, Colyer CL, Sebaiy M, Zhou J, Maeda T, Nakazumi H, Shibukawa M, Saito S. Molecular design of boronic acid-functionalized squarylium cyanine dyes for multiple discriminant analysis of sialic acid in biological samples: Selectivity toward monosaccharides controlled by different alkyl side chain lengths. Anal. Chem. 2015; 87: 1933−1940.

14. Saito S, Massie TL, Maeda T, Nakazumi H, Colyer CL. On-column labeling of gram-positive bacteria with a novel boronic acid functionalized squarylium cyanine dye for analysis by polymer-enhanced capillary transient isotachophoresis. Anal. Chem. 2012; 84: 2452−2458.

15. Saito S, Maeda T, Nakazumi H, Colyer CL. An application of polymer-enhanced capillary transient isotachophoresis with an emissive boronic acid functionalized squarylium dye as an on-capillary labeling agent for gram-positive bacteria. Anal. Sci. 2013; 29: 157-159.

16. Fasano M, Curry S, Terreno E, Galliano M, Fanali G, Narciso P, et al. The extraordinary ligand binding properties of human serum albumin. IUBMB Life. 2005; 57: 787-796.

17. Fanali G, di Masi A, Trezza V, Marino M, Fasano M, Ascenzi P. Human serum albumin: from bench to bedside. Mol Aspects Med. 2012; 33: 209-290.

18. Yamasaki K, Chuang VT, Maruyama T, Otagiri M. Albumin-drug interaction and its clinical implication. Biochim Biophys Acta. 2013; 1830: 5435-5443.

19. Ascoli GA, Bertucci C, Salvadori P. Ligand binding to a human serum albumin stationary phase: Use of same-drug competition to discriminate pharmacologically relevant interactions. Biomed. Chromatogr. 1998; 12: 248–254.

20. Watanabe H, Tanase S, Nakajou K, Maruyama T, Kragh-Hansen U, Otagiri M. Role of arg-410 and tyr-411 in human serum albumin for ligand binding and esterase-like activity. Biochem J. 2000; 349 Pt 3: 813-819.

21. Lucas LH, Price KE, Larive CK. Epitope mapping and competitive binding of HSA drug site II ligands by NMR diffusion measurements. J Am Chem Soc. 2004; 126: 14258-14266.

22. Sinha SS, Mitra RK, Pal SK. Temperature-dependent simultaneous ligand binding in human serum albumin. J Phys Chem B. 2008; 112: 4884-4891.

23. Garg A, Manidhar DM, Gokara M, Malleda C, Suresh Reddy C, Subramanyam R. Elucidation of the binding mechanism of coumarin derivatives with human serum albumin. PLoS One. 2013; 8: 63805.

24. Welder F, Paul B, Nakazumi H, Yagi S, Colyer CL. Symmetric and asymmetric squarylium dyes as noncovalent protein labels: a study by fluorimetry and capillary electrophoresis. J. Chromatogr. B. 2003; 793: 93–105.

25. Arunkumar E, Chithra P, Ajayaghosh A. A controlled supramolecular approach toward cation-specific chemosensors: Alkaline earth metal ion-driven exciton signaling in squaraine tethered podands. J. Am. Chem. Soc. 2004; 126: 6590–6598.

26. Fukuda K, Nakahara H. Spectroscopic studies on aggregation of squarylium dyes and the photopolymerization of a diacetylenic derivative in monolayers at the air/water interface. Colloids Surf., A. 1995; 102: 57-68.

27. Langer K, Balthasar S, Vogel V, Dinauer N, von Briesen H, Schubert D. Optimization of the preparation process for human serum albumin (HSA) nanoparticles. Int J Pharm. 2003; 257: 169-180.

28. Tarazi L, Narayanan N, Sowell J, Patonay G, Strekowski L. Investigation of the spectral properties of a squarylium near-infrared dye and its complexation with Fe(III) and Co(II) ions. Spectrochim Acta A Mol Biomol Spectrosc. 2002; 58: 257-264.

29. Sloat A, Roper MG, Lin XL, Ferrance JP, Landers JP, Colyer CL. Electrophoresis 2008; 29: 3446–3455.

30. Mendham J, Denny R, Barnes J, Thomas K. Vogel qualitative analysis. Pearson Education Limited. 7th ed. 2000.