Research Article pubs.acs.org/chemneuro
Cite This: ACS Chem. Neurosci. XXXX, XXX, XXX−XXX
Raman Imaging of Individual Membrane Lipids and Deoxynucleoside Triphosphates in Living Neuronal Cells during Neurite Outgrowth Giuseppe Pezzotti,*,†,‡,§,∥ Satoshi Horiguchi,∥,⊥ Francesco Boschetto,†,∥ Tetsuya Adachi,∥,⊥ Elia Marin,†,⊥ Wenliang Zhu,† Toshiro Yamamoto,⊥ Narisato Kanamura,⊥ Eriko Ohgitani,∥ and Osam Mazda*,∥ Downloaded via DURHAM UNIV on July 21, 2018 at 01:11:51 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
†
Ceramic Physics Laboratory, Kyoto Institute of Technology, Sakyo-ku, Matsugasaki, 606-8585 Kyoto, Japan Department of Orthopedic Surgery, Tokyo Medical University,6-7-1 Nishi-Shinjuku, Shinjuku-ku, 160-0023 Tokyo, Japan § The Center for Advanced Medical Engineering and Informatics, Osaka University, Yamadaoka, Suita, 565-0871 Osaka, Japan ∥ Department of Immunology, Graduate School of Medical Science, Kyoto Prefectural University of Medicine Kamigyo-ku, 465 Kajii-cho, Kawaramachi dori 602-0841 Kyoto, Japan ⊥ Department of Dental Medicine, Graduate School of Medical Science, Kyoto Prefectural University of Medicine, Kamigyo-ku, Kyoto 602-8566, Japan ‡
ABSTRACT: Recent developments in Raman imaging at the microscopic scale were exploited here with the specific purpose of locating spectral fingerprints of individual membrane lipids and deoxynucleoside triphosphates during neuronal cell networking and separation. After carefully screening the Raman spectra of isolated lipid components, we located an in situ mapped specific Raman fingerprints from individual phospholipids at the micrometric level in comparison with the total lipid distribution within single living cells. We concurrently examined silent zones of lipid emissions and exploited those peculiar spectral ranges for mapping both abundance and localization of individual DNA nucleoside triphosphates. This work represents a first step toward label-free/molecular-selective Raman patterning with high spectral resolution of the relevant chemical species involved with the functionality of neuronal cells.
KEYWORDS: Neurite outgrowth, Raman spectroscopy, phospholipids, nucleotide triphosphates
1. INTRODUCTION Molecular patterning in neuronal cells during differentiation and separation is a topic of great interest in modern medicine because it drives the macroscopic assembly of neurochemical (metabolic) cells that directly correlates with cognitive abilities.1 At the fundamental level in multiscale development of self-organized cell assemblies, specific molecules regulate the behavior of neuronal cells. The strongly nonlinear generation and successive migration of such molecules then lead to the highly hierarchical setup of synaptic connections in neural cell assemblies. The detection of elementary molecules both in membrane lipids and DNA is the first step toward understanding both genetic rules of differentiation and assembly mechanisms in neuronal cells. For this purpose, a number of genetic tags were made available, which include both intrinsically fluorescent proteins and proteins that fluoresces when bound to endogenous or exogenous fluorophores.2 Fluorescent tags, specifically tailored for nucleic acid sequences, and a number of reporter molecules are also available to visualize specific biomolecules in cells.3−5 However, the attachment of fluorescent moieties might impair the function of the targeted © XXXX American Chemical Society
proteins or lipids and, consequently, adversely affect cellular functions.6 Moreover, several tags are usually required for the concurrent visualization of different molecular species. Possible effects on subcellular localization of fusion species as well as their cellular function cannot be ruled out. Consequently, timelapse experiments using fluorescent tags might become problematic and are seldom found in published literature. In situ Raman microprobe spectroscopy offers the advantage of a label-free, concurrent detection of different molecular species. When combined with an optical microprobe, Raman spectroscopy has proven capable of reveal molecular distributions in single living and apoptotic cells.7,8 Highresolution Raman visualization has also been achieved for cell state transitions.9 Exploitation of fast Raman imaging techniques (e.g., slit confocal10 and nonlinear scattering11 microscopy) has enabled time-resolved imaging of the Received: May 14, 2018 Accepted: July 5, 2018
A
DOI: 10.1021/acschemneuro.8b00235 ACS Chem. Neurosci. XXXX, XXX, XXX−XXX
Research Article
ACS Chemical Neuroscience
Figure 1. Raman spectra of (a) PC, (b) PS, (c) PE, and (d) PI in the frequency interval 475−1775 cm−1. Schematic drafts of the respective molecular structures of the lipids are drawn on the right side to each spectrum. Silent zones for lipid emission are emphasized with broken lines.
elementary molecular constituents of cell lipids,12 proteins,13 and nucleic acid.14 This study focuses on in situ Raman assessments of living neuronal cells during division and differentiation. Although several in vivo Raman studies of nerve injuries and brain diseases have previously been published,15−18 in vitro Raman studies of the metabolic response of neuronal cells are yet seldom found in the published literature. Accordingly, the main goal of this investigation is to present a label-free spectral imaging procedure that could enable assessing abundance and localization of individual phospholipids and deoxynucleoside triphosphates (dNTP) during neuronal cell networking and separation. In this study, we discovered several Raman fingerprints that could be utilized as endogenous probes for monitoring and visualizing peculiar aspects of neuronal cell metabolism in health and disease.
phosphatidylinositol (PI) are the main neutral components. During physiological activities of neuronal cells such as cell separation, differentiation, neurite sprouting, and transformation from normal brain tissue to tumors, composition and concentration of lipids take peculiar patterns, which could be used in basic research and diagnostics as well. Raman spectra of elementary lipid components have been precisely collected and partially labeled by Krafft et al.19 The availability of such basic data offers us a chance to further these notions into physiological analyses of neuronal cells through identifying and mapping in situ the spectral fingerprints of specific lipid components. Parts a−d of Figure 1 represent the Raman spectra of PC, PS, PE, and PI, respectively, in the frequency interval 475−1775 cm−1. Schematic drafts of the molecular structure of those lipids are given side by side to each spectrum. Criteria of structural similarity and symmetry could guide us in analyzing vibrational features while searching for fingerprints of specific lipids in the Raman spectra of Figure 1. Different phospholipids share common features in their structures, namely the two fatty acid tails and the phosphate group head connected to each other through a glycerol molecule.
2. RESULTS AND DISCUSSION 2.1. Raman Spectra of Individual Lipids and Nucleoside Triphosphates. Phospholipids are major constituents of neuronal cells and, among them, phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylserine (PS), and B
DOI: 10.1021/acschemneuro.8b00235 ACS Chem. Neurosci. XXXX, XXX, XXX−XXX
Research Article
ACS Chemical Neuroscience
Figure 2. Raman spectra of (a) dATP, (b) dGTP, (c) dCTP, and (d) dTTP in the frequency interval 475−1775 cm−1. Schematic drafts of the respective molecular structures of the nucleoside triphosphates are drawn on the right side to each spectrum. Silent zones for lipid emission are replotted from Figure 1 and emphasized with broken lines.
they belong to their head groups and greatly differentiate according to their different structural details. The symmetric and antisymmetric stretch vibrations of the choline (head) group N+(CH3)3 in PC emit Raman signals at 719 (C−N symmetric stretching) and 875 cm−1 (C−N antisymmetric stretching), respectively.19 The Raman band located at 733 cm−1 is peculiar to PS because it represents a fingerprint emission for the serine residue (C−N stretching). PE instead shows an isolated band at 759 cm−1, which is related to ethanolamine. Peculiar Raman bands for PI are located at 576 and 601 cm−1, which arise from the inositol residue (both are from C−OH bending). Additional emissions at 519 and 777 cm−1 are instead related to ring vibrations and to inositol ring breathing, respectively. Stretching of CO bonds commonly display Raman bands at frequencies in the region 1716−1744 cm−1 for all types of lipid. However, there are important peculiarities in this highfrequency region too, which could be summarized as follows: (i) PS is the only phospholipid displaying two different types
Accordingly, a number of vibrational emissions from those common bonds will display conspicuously the same, although shifts in frequencies at maximum and differences in relative intensities might be found among different phospholipids. This is, for example, the case of a strong activity in the frequency interval 1420−1452 cm−1, which is dominated by C−H bond scissoring.20,21 Being common to all lipids and also shared with proteins, such emissions can hardly serve to single out specific lipid structures. Similar considerations could be made for frequencies at around 1278 cm−1 (overlap of CC groups in unsaturated fatty acids and amide III in proteins)19,22 and 1298 cm−1 (CH2 deformations common to all lipids).23 C−C skeletal stretching (at 1060−1095 and 1128 cm−1 in the backbone of lipids)24,25 and PO2− symmetric stretching (1070−1090 cm−1 in DNA)24,26 are also regions considered not suitable for Raman differentiation among lipids and between lipids and proteins as well. On the other hand, bands in the lower wavenumber region shown in Figure 1 (475−900 cm−1) might allow distinction among phospholipids because C
DOI: 10.1021/acschemneuro.8b00235 ACS Chem. Neurosci. XXXX, XXX, XXX−XXX
Research Article
ACS Chemical Neuroscience
the other relatively strong C−N emission in dATP at 1481 cm−1. A real fingerprint match in the Raman spectrum of dTTP comes from its strongest band located at 1651 cm−1 (cf. Figure 2d), which arises from CC stretching. This band is free from overlap with NH2 scissoring emissions because the NH2 bond is missing in the aromatic ring of dTTP. On the other hand, CO stretching emissions appear in the highest frequency regions reported in Figure 2. In dGTP, they appear at relatively high frequencies (∼1702 cm−1) as compared to other nucleoside triphosphates (cf. Figure 2b). This is due to the fact that CO stretching occurs in the fused pyrimidineimidazole ring structure of guanine. Note also that dATP experiences a silent zone at >1620 cm−1 because no CO bond is present in its structure (cf. Figure 2a). The relatively strong band located at 1675 cm−1 in the dCTP spectrum is assigned to a cumulative emission of CC and CO stretching vibrations. Carbon−hydrogen bond vibrations typically contribute Raman emissions at intermediate frequencies. In dATP, C− H bending and CH2 twisting are found at 1130 and 1249 cm−1, respectively. On the other hand, the C−H bending mode appears as a doublet in dCTP at 1102 and 1163 cm−1 for C−H bonds belonging to the pyrimidine and imidazole sides, respectively. The strong band observed at 1252 cm−1 in the spectrum of dCTP (Figure 2c), above-reported as C−N stretching, partly overlaps with emissions from the CH2 twisting mode. NH2/CH2 scissoring vibrations also contribute the C−N stretching band at 1498 cm−1. Unlike dATP and dCTP, the Raman bands arising from carbon−hydrogen bond vibrations are quite pronounced in the spectrum of dTTP (Figure 2d). They mainly appear at 1182 cm−1 (C−H bending) and 1368 cm−1 (C−H deformation and CH3 bending). Similar to the case of dCTP, also in dTTP C−N stretching bands at intermediate frequencies strongly overlap with CH2 twisting (1235 cm−1) and with CH2 scissoring (1476 cm−1). Specifically regarding the low-frequency silent zone between 600−700 cm−1, which is free from lipid signals, the mildly intense band at 648 cm−1 in dATP and the strong band at 650 cm−1 in dGTP correspond to ring deformation of adenine and guanine groups, respectively. Although peculiar to these specific nucleosides, their almost complete overlap does not allow one to discern between the two. On the other hand, the relatively strong bands at 600 and 623 cm−1 in dCTP and dTTP, respectively, both stem from CO bending. The difference in frequency, which is large enough to deconvolute the two components in a mixed nucleoside environment, stems as already explained above from structural differences. The above analyses of band origin in different lipid and nucleoside components enable one to single out Raman fingerprint of specific molecules in the spectrum of neuronal cells, which usually appear as a mix of all the above components. Figures 3 and 4 summarize the vibrational sensors used in this study for mapping both abundance and localization of individual phospholipids and dNTPs, respectively. 2.2. Conventional Fluorescence and Raman Tags for Lipids and DNA. In this section, we apply and briefly discuss fluorescence3 and Raman18 tags as established by previous researchers to map lipids and DNA in living cells. Parts a−c of Figure 5 show fluorescence images of PC12 neuronal cells after 5 days of culturing in NDF. The confocal laser scanning
of CO stretching, namely the CO bonds in the glycerol units and those in the serine group, and (ii) shifted CO frequencies can be detected for different phospholipids due to structural differences in their functional groups. According to (i), PS is the only phospholipid that displays a CO doublet at 1732 and 1716 cm−1 for glycerol and serine locations, respectively. Moreover, PE displays vibrational stretching of CO bonds clearly shifted toward higher frequencies (∼ or >1744 cm−1) according to arguments in (ii). The main reason for this latter effect is that, unlike the fatty acid chains R in other phospholipids, the C14H31 tails in PE do not contain C C but only C−C single bonds. Low-frequency (related to head-residue) and high-frequency (related to CO stretching) fingerprint bands are labeled in red ink in Figure 1 for easier visualization. Their usefulness in Raman mapping of individual lipids in living neuronal cells shall now depend on the possibility that such peculiar Raman bands could actually remain resolvable in overlap with the Raman emissions from DNA, which are discussed hereafter. Similar to the case of lipids, we also examined Raman spectra collected from individual dNTPs.27,28 The preliminary analysis described hereafter was made with the purpose of locating fingerprint bands, which not only differentiated individual nucleosides among different triphosphates but also did not overlap with bands from lipids. In responding to the latter criterion, the two spectral areas at 600−700 and 1450− 1650 cm−1 were considered as the most suitable because they were silent with respect to emissions from lipids (cf. broken lines in Figure 1). Parts a−d of Figure 2 represent Raman spectra of deoxyadenosine triphosphate (dATP), deoxyguanosine triphosphate (dGTP), deoxycytidine triphosphate (dCTP), and deoxythymidine triphosphate (dTTP), respectively. Schematic drafts of the respective structures are given at the right side of each Raman spectrum. Common to all the four triphosphates is the Raman activity related to C−N vibrations in their nitrogenous bases, while CN stretching vibrations are not found in dTTP because this bond is absent in its structure (cf. Figure 2d). However, the frequencies at which these vibrations appear in the spectra might display shifted records according to the different bond details found in the structures. With reference to Figure 2, one can find the C−N bonds emitting at 1302, 1336, 1370, 1412, 1481, and 1500 cm−1 in dATP, at 1075 (imidazole side), 1170 (pyrimidine side), 1308, 1348, 1382, and 1530 cm−1 in dGTP, at 1222, 1252, 1299, 1378, and 1498 cm−1 in dCTP, and at 1235 and 1476 cm−1 in dTTP. On the other hand, CN bond vibrations appear at 1574 and 1598 cm−1 in dATP, with the former frequency conspicuously coinciding with the 1575 and 1579 cm−1 frequencies experienced by the CN bond in dGTP and dCTP, respectively. For the main purpose of this study, it is important to note that CN bond stretching gives rise to an additional strong emission in dCTP located at 1551 cm−1. This emission is quite isolated from other bands, does not significantly overlap emissions from other nucleosides and, more importantly, lies within the high-frequency silent zone located in the spectra of lipids (cf. Figure 2b). Also the band at 1336 cm−1 is a quite strong feature in the Raman spectrum of dATP and lies in a relatively less crowded zone of lipid emission (cf. Figures 1 and 2a). However, the best frequency candidate for Raman detection of dATP is believed to be at ∼1500 cm−1 because this frequency is bound at the low-frequency side by D
DOI: 10.1021/acschemneuro.8b00235 ACS Chem. Neurosci. XXXX, XXX, XXX−XXX
Research Article
ACS Chemical Neuroscience
Figure 5. Fluorescence images of PC12 neuronal cells after 5 days of culturing in NDF: (a) benzimidazole trihydrochloride trihydrate stain to visualize cell nuclei; (b) OPPC to visualize translocation sites across the membrane; (c) a fluorescent azide attached to the alkyne group of 5-ethynyl-2′-deoxyuridine (EdU), in turn covalently incorporated into DNA as a mimic of thymidine, to visualizes cell nuclei; and (d) an explanatory draft the stain method in (c), as redrawn from ref 3.
Figure 3. Vibrational sensors (and the related frequencies) used in this study for mapping individual phospholipids: (a) PC, (b) PE, (c) PS, and (d) PI.
bind to adenine-thymine regions of DNA, (b) oligopeptide transport system permease protein (OPPC), which is a transport system responsible for translocation of the substrate across the membrane, and (c) a fluorescent azide attached to the alkyne group of 5-ethynyl-2′-deoxyuridine (EdU), which is covalently incorporated into DNA as a mimic of thymidine and visualizes cell nuclei. An explanatory draft of this latter stain method is given in Figure 5d as redrawn from ref 3. The blue fluorescence visualized in Figure 5a is produced by a cell-permeable DNA stain that preferentially binds to adenine−thymine regions of the DNA and thus visualizes cell nuclei. No fixation was required with the selected stain, but its visualization required an ultraviolet light source, which could potentially harm the cells. The fluorescent image of the stained membrane in Figure 5b appears extremely bright and detailed. OPPC is indeed known for completely substituting for the membrane-associate protein OPP and effectively visualizing any location of the membrane. However, unlike OPP, OPPC is incapable of transporting tripeptides. In other words, once stained, the cell metabolism that depends on membrane permeation becomes completely inhibited. In Figure 5c, we mapped the EdU alkynyl group applied after cell fixation in combination with a fluorescent azide in the major groove of the double helix (cf. Figure 5d). This approach leads to bright imaging of DNA definitely better resolved in space as compared to the blue stain in Figure 5a. However, cell fixation is incompatible with life. Note that even when cell fixation is not required, a common issue in fluorescent probes is that they usually experience a heavy molecular weight, which is at least comparable to or even larger than the parent small molecules. Such a circumstance could severely alter biological activity, cellular localization, and dynamics of the parent small molecules.29
Figure 4. Vibrational sensors (and the related frequencies) used in this study for mapping dNTPs: (a) dATP, (b) dGTP, (c) dCTP, and (d) dTTP.
micrographs locate: (a) benzimidazole trihydrochloride trihydrate molecules that visualize cell nuclei and preferentially E
DOI: 10.1021/acschemneuro.8b00235 ACS Chem. Neurosci. XXXX, XXX, XXX−XXX
Research Article
ACS Chemical Neuroscience Raman monitoring of the EdU tag18 was designed to overcome the above shortcomings of fluorescent molecules. Figure 6a shows an optical micrograph of differentiated PC12
Figure 6. (a) Optical micrograph of differentiated PC12 neuronal cells and their linked neurites, and (b) a Raman hyperspectral map of DNA at the location in (a) visualized through Raman emissions from the Edu tag.
neuronal cells, which built up a network of linked neurites. In Figure 6b, a Raman hyperspectral map is shown of the same location after introducing Edu tags in the culture to visualize DNA. EdU alkynes experience a strong Raman emission at ∼2120 cm−1, namely in a conspicuously silent zone for Raman emissions from cells (i.e., 1800−2800 cm−1). The strong Raman emission of the alkyne tag arises from the −CC− triple bond stretching vibrations. Moreover, the alkyne molecules being small and quite mobile, they have been reported to allow cells retaining their biological activity.30 The Raman map in Figure 6b reveals a wealth of details about the location and concentration of DNA species in networking neurons. However, the EdU Raman method does not allow distinguishing among and mapping individual dNTPs. Note also that, while thymidine tags obviously locate cell nuclei,18 DNA nucleosides becomes quite mobile in differentiating neuronal cells and hardly display as a compact nucleic region. 2.3. Raman Mapping of Individual Lipids and Nucleoside Triphosphates. The chance to differentiate among individual phospholipids and dNTPs in living neuronal cells is challenged in this section. A completely label-free Raman approach brings no damages or alterations to cell metabolism. However, due to overlapping emissions from biological molecules, this approach necessarily requires the collection of highly spectral resolved Raman images. Suitable algorithms are then required for spectral deconvolution into selected sub-bands and mapping of specific Raman fingerprints, as discussed in section 3. Figure 7 shows Raman maps of individual lipids in the same zone of neuronal networking shown in Figure 6. Maps are displayed for: (a) all lipids (CH2 scissoring band at 1450 ± 15 cm−1),31 (b) phosphatidylcholine (PC, 719 ± 10 cm−1), (c) phosphatidylethanolamine (PE, 1769 ± 18 cm−1), (d) phosphatidylserine (PS, 1723 ± 27 cm−1), and (e) phosphatidylinositol (PI, 576 ± 10 cm−1). Note that we observed a shift of ∼18 cm−1 in frequency for the selected PE band between pure PE and PE in neuronal cells (1744 vs 1769 cm−1, respectively, cf. Figure 1c). Such a difference is certainly due to major chemical and structural differences for the lipid in the cell membrane, which we do not have data about to properly discuss it here. Nevertheless, PE is the only lipid structure that presents a vibrational CO
Figure 7. Raman maps of: (a) all lipids, (b) PC, (c) PE, (d) PS, and (e) PI. Frequencies and their intervals in brackets locate the mapping procedure and correspond to the Raman fingerprints for phospholipid molecules described in section 3 and represented in Figure 3.
stretching band at such high frequency. Note also that the C O doublet observed for PS bulk at 1716 and 1732 cm−1 (cf. Figure 1c) is observed as a singlet centered at 1723 cm−1 in the Raman spectrum of the cells. The frequencies shown in brackets represent the Raman fingerprints located for these individual phospholipid molecules in parts a−d of Figure 3, respectively. As seen, all maps for individual phospholipids show intensifications at the cell membrane and along the linked neurites. However, there are important differences at specific locations which detail different distributions for different phospholipid species. For example, for fixed image contrast parameters, the intensity of PE remained quite weak as compared to those from other elementary phospholipids. Moreover, membrane locations with intensified emissions from PC and PI were quite similar, but they conspicuously differed from those observed for PS. Such features possess precise physiological meanings for differentiating neuronal cells, which will be discussed in the next section. Figure 8 shows Raman maps which were collected for individual dNTPs in the same area of Figures 6 and 7. The shown maps correspond to (a) dATP (1504 ± 20 cm−1), (b) dGTP (1515 ± 18 cm−1), dCTP (1551 ± 10 cm−1), and (d) dTTP (1368 ± 10 cm−1). The selected frequencies, shown in brackets, correspond to the Raman vibrational fingerprints located for individual nucleosides in parts a−d of Figure 4, respectively. Similar to the case of phospholipids, significant topographic differences could also be noted among different F
DOI: 10.1021/acschemneuro.8b00235 ACS Chem. Neurosci. XXXX, XXX, XXX−XXX
Research Article
ACS Chemical Neuroscience
Figure 9. Details of Raman spectra in the frequency interval 500−850 cm−1 from (a) location A (membrane site) and (b) location B (nucleus site), as shown in Figure 6a. The origin and the frequency of each deconvoluted band are labeled in inset.
However, it neighbors a PS+PI band at 596 cm−1 and needs to be deconvoluted from it on its high-frequency side. In parts a and b of Figure 10, deconvoluted Raman spectra are shown in
Figure 8. Raman maps of: (a) dATP, (b) dGTP, (c) dCTP, (d) dTTP, and (e) dATP oxidized. Frequencies and their intervals in brackets locate the mapping procedure and correspond to the Raman fingerprints for phospholipid molecules described in section 3 and represented in Figure 4.
nucleoside triphosphates. An overall low concentration of dCTP was recorded as compared to other triphosphates, while higher concentration gradients between membrane and nucleus could be noticed for dATP and dGTP as compared to those noticed in the concentration map of dTTP. The physiological implications of these observations will be discussed in the next section. Spectral details are explicitly given in Figures 9−11 as examples for the deconvolution of Raman fingerprint bands at selected locations in the hyperspectral maps of Figures 7 and 8. The low-frequency spectral region between 500 and 850 cm−1 (Figure 9) mainly served to locate Raman fingerprints of PC and PI (centered at 719 and 576 cm−1, respectively). This region also contained one fingerprint band for dGTP in the lipid silent zone 600−700 cm−1 (i.e., at 650 cm−1). However, this frequency was not selected for mapping in this study because it almost exactly overlapped with a relatively strong emission from dATP at 648 cm−1 (cf. Figure 2a). Spectra from two specific locations (labeled A and B in Figure 6a) are shown in parts a and b of Figure 9, respectively. These locations correspond to two typical sites in correspondence of membrane and nucleus, respectively. The PC band at 719 cm−1 clearly appears as a spectral shoulder and could easily be bound at its low-frequencies side. On the other hand, the spectral location at 576 cm−1 selected for PI stems at a relatively isolated frequency among lipids (cf. Figure 1d).
Figure 10. Details of Raman spectra in the frequency interval 1450− 1800 cm−1 from (a) location C (membrane site) and (b) location D (nucleus site), as shown in Figure 6a. The origin and the frequency of each deconvoluted band are labeled in inset.
the spectral range between 1475 and 1775 cm−1 for two selected locations C and D (cf. Figure 6a). In the spectral region depicted in Figure 10, fingerprints for dATP and dCTP can be found in the lipid silent zone between 1450−1650 cm−1. The dATP band appears at 1504 cm−1 as a clear shoulder and could be clearly isolated from the sharp dCTP band centered at ∼1551 cm−1. A comparison between the spectra in parts a and b of Figure 10 clarifies how the intensities of the dATP and dCTP bands could be clearly separated upon spectral deconvolution. In this spectral zone, Raman fingerprints can also be found for PS (at 1724 cm−1 as a single band merging the to CO contributions at 1716 and 1732 cm−1) and PE (found at ∼1760 cm−1 although expected at 1744 cm−1, cf. Figures 1d and 10b). Note that in Figure 10a, G
DOI: 10.1021/acschemneuro.8b00235 ACS Chem. Neurosci. XXXX, XXX, XXX−XXX
Research Article
ACS Chemical Neuroscience a relatively strong band located at 1615 cm−1 could be observed, which was assigned to oxidized dATP according to a Raman spectroscopy study previously published by D’Amico et al.27 An intensity map of this band was also recorded, and it is displayed in Figure 8e. Finally, parts a and b of Figure 11
propagation of action potential and (ii) in forming neural circuits are made possible by precise structural changes of the membrane structure, in turn dictated by enzymatically driven molecular signaling.32 More in detail, neurite outgrowth requires massive addition/delivery of specific proteins and lipids to the sprouting tips of newly outgrowing neurites, which occurs through membrane trafficking. Figure 12 shows a
Figure 11. Details of Raman spectra in the frequency interval 1225− 1625 cm−1 from (a) location E (membrane site) and (b) location F (nucleus site), as shown in Figure 6a. The origin and the frequency of each deconvoluted band are labeled in inset.
compare the deconvoluted Raman spectra collected in the spectral interval 1225−1625 cm−1 at locations labeled as E and F in Figure 6a. We used this spectral window to deconvolute fingerprint emissions for dGTP and dTTP at 1530 and 1368 cm−1, respectively. The region at around 1530 cm−1 could be unequivocally assumed as representative of dGTP because it is neighboring another dGTP band on the low-frequency side (at 1474 cm−1) and a quite sharp emission from dCTP on the high-frequency side (at 1551 cm−1). Note, however, that the dGTP doublet at 1474/1530 cm−1 did not preserve a constant ratio throughout mapping. For this reason, it was important to include frequencies
