Analytica Chimica Acta, accepted, 07/07/2015. This is the accepted version without proofing
corrections. DOI: 10.1016/j.aca.2015.06.011
.
Page 3 of 26
contributions and extract relevant and useful analytical information from complex MDF data
[33]. Broadly speaking, there are three major approaches to MDF data analysis. First, use
multi-way versions of principal component analysis (PCA) to assess how spectra differ.
Second, use supervised regression based methods to correlate the observed spectral variations
with useful variables like concentration. Third, use multi-way decomposition methods like
multivariate curve resolution (MCR) [34-37], parallel factor analysis (PARAFAC) [38-40].
MCR and PARAFAC are used to identify the spectral contributions of individual constituents,
which can then be associated with specific fluorophore emission, and component scores can
be correlated with concentration. However, even with chemometric analysis, the emission is
often not resolvable, so one has to examine the use of other properties to facilitate
fluorophore resolution. Combining anisotropy with MDF [41] enabled the differentiation and
quantification of fluorophores (protein from free amino acids) with similar emission
properties in complex mixtures, based on their rotational speed and hydrodynamic volume,
and thus the molecular size, or for macromolecules the mobility/flexibility of the constituent
fluorophores.
However, in the course of that study, we noted that in multi-fluorophore proteins like
BSA and HSA, anisotropy was not constant across the full emission space. This 4D data
generated by aniso-TSFS (excitation wavelength (λ ex ), wavelength offset (Δλ), intensity (I),
anisotropy (r)) from proteins was not easy to understand or interpret. To develop a robust
analytical methodology for the quantitative analysis of this unusual aniso-TSFS data we took
a well-known macromolecule, HSA, which could be reproducibly varied structurally via
simple thermal and chemical denaturation. Chemometric methods had then to be evaluated
and developed because while there was some available literature on 4-way data analysis [42,
43], none involved fluorescence anisotropy/polarization.
Comprehensive Aniso-TSFS data were collected from HSA solutions, which have
been subjected to both thermal and chemical denaturation. We looked first at the constituent
polarized spectral data using MCR to identify each emitter, and then determined how each
individual constituent emitter (both fluorescent and phosphorescent) behaved in the different
polarization states in terms of spectral emission and contribution. From this, an initial
protein-unfolding model was generated that agreed with the extensive HSA literature.
Second, we validated the unfolding model by calculating the anisotropy values and anisoTSFS plots assessing how each emitter behaved in terms of mobility/accessibility changes. It
was this combination of methods that provided a unique and very comprehensive picture of
how each intrinsic fluorophore was affected by protein structural change. This unique
combination of anisotropy, MDF measurements, and factor-based chemometrics was termed
“anisotropy resolved multidimensional emission spectroscopy” (ARMES).
2. Materials and Methods
2.1 Materials: HSA (>99%, essentially globulin free), guanidine hydrochloride GuHCl
(>99%), sodium sulphite (>98%) and phosphate buffer saline (PBS) tablets were purchased
from Sigma-Aldrich. Potassium iodine was purchased from Riedel-deHaën. All compounds
were used without further purification. PBS HSA stock solutions were prepared using
sterilized high purity water and were membrane filtered (0.22 µm). All stock solutions were
stored at –70 °C and defrosted overnight at 4 °C before aliquoting. For thermal denaturation
corrections. DOI: 10.1016/j.aca.2015.06.011
.
Page 3 of 26
contributions and extract relevant and useful analytical information from complex MDF data
[33]. Broadly speaking, there are three major approaches to MDF data analysis. First, use
multi-way versions of principal component analysis (PCA) to assess how spectra differ.
Second, use supervised regression based methods to correlate the observed spectral variations
with useful variables like concentration. Third, use multi-way decomposition methods like
multivariate curve resolution (MCR) [34-37], parallel factor analysis (PARAFAC) [38-40].
MCR and PARAFAC are used to identify the spectral contributions of individual constituents,
which can then be associated with specific fluorophore emission, and component scores can
be correlated with concentration. However, even with chemometric analysis, the emission is
often not resolvable, so one has to examine the use of other properties to facilitate
fluorophore resolution. Combining anisotropy with MDF [41] enabled the differentiation and
quantification of fluorophores (protein from free amino acids) with similar emission
properties in complex mixtures, based on their rotational speed and hydrodynamic volume,
and thus the molecular size, or for macromolecules the mobility/flexibility of the constituent
fluorophores.
However, in the course of that study, we noted that in multi-fluorophore proteins like
BSA and HSA, anisotropy was not constant across the full emission space. This 4D data
generated by aniso-TSFS (excitation wavelength (λ ex ), wavelength offset (Δλ), intensity (I),
anisotropy (r)) from proteins was not easy to understand or interpret. To develop a robust
analytical methodology for the quantitative analysis of this unusual aniso-TSFS data we took
a well-known macromolecule, HSA, which could be reproducibly varied structurally via
simple thermal and chemical denaturation. Chemometric methods had then to be evaluated
and developed because while there was some available literature on 4-way data analysis [42,
43], none involved fluorescence anisotropy/polarization.
Comprehensive Aniso-TSFS data were collected from HSA solutions, which have
been subjected to both thermal and chemical denaturation. We looked first at the constituent
polarized spectral data using MCR to identify each emitter, and then determined how each
individual constituent emitter (both fluorescent and phosphorescent) behaved in the different
polarization states in terms of spectral emission and contribution. From this, an initial
protein-unfolding model was generated that agreed with the extensive HSA literature.
Second, we validated the unfolding model by calculating the anisotropy values and anisoTSFS plots assessing how each emitter behaved in terms of mobility/accessibility changes. It
was this combination of methods that provided a unique and very comprehensive picture of
how each intrinsic fluorophore was affected by protein structural change. This unique
combination of anisotropy, MDF measurements, and factor-based chemometrics was termed
“anisotropy resolved multidimensional emission spectroscopy” (ARMES).
2. Materials and Methods
2.1 Materials: HSA (>99%, essentially globulin free), guanidine hydrochloride GuHCl
(>99%), sodium sulphite (>98%) and phosphate buffer saline (PBS) tablets were purchased
from Sigma-Aldrich. Potassium iodine was purchased from Riedel-deHaën. All compounds
were used without further purification. PBS HSA stock solutions were prepared using
sterilized high purity water and were membrane filtered (0.22 µm). All stock solutions were
stored at –70 °C and defrosted overnight at 4 °C before aliquoting. For thermal denaturation