391
13 Vibrational Spectroscopy: Disease Diagnostics and Beyond
An established mechanism for the transport of nanoparticles in cells is via endosomes and later lysosomes [170]. A number of studies have also demonstrated that,
in the later stages of trafficking, individually endocytosed nanoparticles accumulate
in larger multivescular bodies [165] and they have also been seen to be localized
in the endoplasmic reticulum and golgi apparatus [171]. Notably, in Fig. 13.19 Ia,
the majority of the nanoparticles, identified in green as cluster 5, are surrounded
by cytoplasmic regions of cluster 1, which is relatively spatially extensive, particularly compared to typical sizes of lysosomes (1–2 µm). In the white light image of
Fig. 13.19 Ib, this region can be identified as the Endoplasmic Reticulum (ER),
which may be expected to be rich in lipidic compounds, and a PCA of the different
regions of the cytoplasm confirms this. Cluster 4, on the other hand is relatively rich
in nucleic acids, which may be due to the presence of mitochondria. Co-staining
the cells to highlight the ER and examination using confocal fluorescence microscopy indicates a strong correlation between the ER stain and the fluorescence of the
nanoparticle label, and any locally concentrated fluorescence which may be indicative of encapsulated nanoparticles is contained within the ER [151]. the analysis
supports the attribution of the subcellular environment of K-means cluster 1 as endoplasmic reticulum.
the study represents an initial proof of concept, to demonstrate the potential use
of Raman spectroscopy to identify and locate nanoparticles in cells and to probe
their subcellular environment. No attempt has, as yet, been made to quantify the
amount of nanoparticles located or the sensitivity and detection limit of the technique. Instrumentally, the spatial resolution can be improved by moving to lower
wavelength sources, although the excitation of any fluorescence from the nanoparticles and the cell constituents should be avoided. however, as it scales linearly with
wavelength, a move from 785 to 473 nm should result in an improvement in lateral
resolution by a factor of ~ 1.7. ultimately, the confocal mode of operation can provide 3d localisation of the nanoparticles at submicron resolution.
other multivariate analytical approaches have also been applied in the field of
Raman microspectroscopy of cells. In addition to KmCA, other clustering methods
such as Fuzzy C means clustering (FCm) and hierarchal cluster analysis (hCA)
have been used to separate the cellular Raman data into clusters and subsequently
reshape the data into images [172, 173]. however, as highlighted by headegaad et
al., these approaches have their own limitations. In particular, boundaries between
sub-cellular features can often result in the addition of extra clusters with mixed
spectral signatures. this can be overcome by increasing the number of clusters;
however, this in turn can result in added complexity to interpretation and inaccuracies in regional separation. Additionally, the reproducibility of these methods can
also be questioned as the starting point for the centroid based KmCA and FCm is
subjective [172].
PCA and vertex component analysis (vCA) have also been used to separate out
distinct regions of the cell. With regards to PCA, separation is based on the variances between the spectra in the data set, the majority of the variance being described
by the first three principal components [172]. thus, the score values can be used to
construct a composite image of the cell in which the biochemical contributions of
13 Vibrational Spectroscopy: Disease Diagnostics and Beyond
An established mechanism for the transport of nanoparticles in cells is via endosomes and later lysosomes [170]. A number of studies have also demonstrated that,
in the later stages of trafficking, individually endocytosed nanoparticles accumulate
in larger multivescular bodies [165] and they have also been seen to be localized
in the endoplasmic reticulum and golgi apparatus [171]. Notably, in Fig. 13.19 Ia,
the majority of the nanoparticles, identified in green as cluster 5, are surrounded
by cytoplasmic regions of cluster 1, which is relatively spatially extensive, particularly compared to typical sizes of lysosomes (1–2 µm). In the white light image of
Fig. 13.19 Ib, this region can be identified as the Endoplasmic Reticulum (ER),
which may be expected to be rich in lipidic compounds, and a PCA of the different
regions of the cytoplasm confirms this. Cluster 4, on the other hand is relatively rich
in nucleic acids, which may be due to the presence of mitochondria. Co-staining
the cells to highlight the ER and examination using confocal fluorescence microscopy indicates a strong correlation between the ER stain and the fluorescence of the
nanoparticle label, and any locally concentrated fluorescence which may be indicative of encapsulated nanoparticles is contained within the ER [151]. the analysis
supports the attribution of the subcellular environment of K-means cluster 1 as endoplasmic reticulum.
the study represents an initial proof of concept, to demonstrate the potential use
of Raman spectroscopy to identify and locate nanoparticles in cells and to probe
their subcellular environment. No attempt has, as yet, been made to quantify the
amount of nanoparticles located or the sensitivity and detection limit of the technique. Instrumentally, the spatial resolution can be improved by moving to lower
wavelength sources, although the excitation of any fluorescence from the nanoparticles and the cell constituents should be avoided. however, as it scales linearly with
wavelength, a move from 785 to 473 nm should result in an improvement in lateral
resolution by a factor of ~ 1.7. ultimately, the confocal mode of operation can provide 3d localisation of the nanoparticles at submicron resolution.
other multivariate analytical approaches have also been applied in the field of
Raman microspectroscopy of cells. In addition to KmCA, other clustering methods
such as Fuzzy C means clustering (FCm) and hierarchal cluster analysis (hCA)
have been used to separate the cellular Raman data into clusters and subsequently
reshape the data into images [172, 173]. however, as highlighted by headegaad et
al., these approaches have their own limitations. In particular, boundaries between
sub-cellular features can often result in the addition of extra clusters with mixed
spectral signatures. this can be overcome by increasing the number of clusters;
however, this in turn can result in added complexity to interpretation and inaccuracies in regional separation. Additionally, the reproducibility of these methods can
also be questioned as the starting point for the centroid based KmCA and FCm is
subjective [172].
PCA and vertex component analysis (vCA) have also been used to separate out
distinct regions of the cell. With regards to PCA, separation is based on the variances between the spectra in the data set, the majority of the variance being described
by the first three principal components [172]. thus, the score values can be used to
construct a composite image of the cell in which the biochemical contributions of