Introduction
Chelerythrine (CET) is a quaternary benzo-phenanthridine alkaloid (QBA) of 348 Daltons mass and extracted from Papaveraceae, i.e. plants of the Poppy family. In solutions near physiological pH, CET is positively charged (CET+) due to an iminium group (N+=C) attracting small anions such as Cl. At increased pH values, CET+ loses its charge and becomes a pseudobase with an apparent pKa value of ∼9 [1]. In the past, QBAs such as CET and its sister alkaloids, the sanguinarines, have been widely used as medicinal remedies in insecticides, bactericides, fungicides, viricides, in ointments, mouthwash solutions, and tooth pastes, sedatives and anti-helmintics [2]. More recently, CET had been first reported as a protein kinase C (PKC) inhibitor [3] and, utilizing this mechanism, later heralded as a potential anticancer agent capable of inducing apoptosis in a variety of cancer cell lines [4,5,6]. Acting as a BH3-mimetic, CET induces apoptosis by binding to BH1-like motifs of B cell lymphoma type 2 (Bcl-2) proteins inhibiting pro-survival B-cell lymphoma extra large (BclXI) proteins and activating mitochondrial Bcl-2 associated X (Bax)/ Bcl-2 homologous antagonist-killer (BAK) protein pro-apoptotic mechanisms [7,8,9].
Recently, CET was shown to inhibit ouabain-sensitive K+ or Na+/K+ pump (NKP) influx in human lens epithelial cells (HLECs) without interference with ouabain binding and PKC-mediated NKP phosphorylation [10]. The CET structure-related sanguinarines have been known for a long time to inhibit the NKP in a variety of tissue preparations [11,12,13,14,15,16,17], without an explanation for this effect. Sequence homologies between BH1-like motifs originally reported to bind CET in Bcl-Xl proteins and found in the cytoplasmic aspect of the crystal structure of the NKP led to the hypothesis that CET may inhibit the NKP function through these BH1 motifs [10]. The amino acid sequence ARAAEILARDGPN of the first putative CET binding site resides in the A domain of the NKP and has two arginine (R) residues. Interestingly, fluorescence studies suggest interaction of R residues with sanguinarines, however, in the carboxy (C)-terminus of the NKP [18]. Moreover, sequence homologies of the NKP BH1-like motifs found in all P-type ATPases suggest other cellular metabolic pathways can be influenced by the presence of CET [10].
Although QBAs have been shown to rapidly diffuse across plasma membranes and may be used as fluorescing DNA probes [19], the mechanism of transmembrane distribution of CET is unknown and thus the question as to whether the charged monomeric CET+ or its lipid soluble pseudobase traverse the plasma membrane prior to binding and affecting intracellular structures such as the NKP remains unresolved. Traditional techniques for monitoring the amount and localization of active metabolites in biological systems (i.e., ultraviolet-visible absorption spectrophotometry, fluorescence spectroscopy, liquid chromatography, etc.) are often limited by low detection limits, increased experimental duration and cost, or indirect observation of the species of interest via a (radio) chemical “label”. Recently, surface-enhanced Raman spectroscopy (SERS) has emerged as a powerful molecular sensing technique due to its high degree of molecular specificity, low detection limits, and multiplex detection capabilities [20,21], permitting accurate determination of distribution and amount of metabolites in biological systems with very short experimental durations, a conditio sine qua non for pharmacological studies.
In order to obtain insight into the interaction of CET with the NKP, this study attempts to track the distribution of CET across the plasma membrane into the cytosol of HLECs, our model, in which CET was shown to inhibit the NKP [10]. Raman analysis of a stock CET solution exhibited a few very weak peaks; however, upon the addition of silver nanoparticles (AgNPs), a large signal enhancement was observed yielding vibration-rich spectra. The SERS enhancement of CET facilitated the construction of a calibration function for determination of the concentration of monomeric CET in the cytosol and its content in membrane particulate fractions. The alkaloid CET accumulated in the cytosol and plasma membranes at concentrations near or above diffusion equilibrium suggesting transport as an organic cation CET+, presumably the species interacting with Bcl-2 proteins and NKP [10].
Materials and Methods
Reagents
Silver nitrate (99.5%, AgNO3), sodium borohydride (99%, NaBH4), were purchased from Fisher Scientific (Fairlawn, NJ) and chelerythrine chloride (98%, CET) from Acros Organics (Fairlawn, NJ). Potassium bromide (≥ 99%, KBr) and nitric acid (Optima grade) were purchased from Fisher Scientific. All reagents were utilized without further purification. High quality water (> 18 MΩ cm) was obtained from a LabConco water purification system and utilized as solvent throughout. Cell culture media and buffers were described earlier [10]. The NKP Na+-conformational buffer consisted of the following reagents (in mM, all from Sigma Chemical, St. Louis, MO): 100 NaCl, 20 KCl, 20 MgCl2, 20 Imidazole, 1 NaF, and HALT (Pierce Biochemicals) as protease inhibitor.
Cell Culture and Preparation of Cytosol and Membrane Extracts
HLECs were obtained as a primary fetal human lens cell line (FHL124) by courtesy of Dr. John Reddan, Oakland University, Michigan, and were cultured and propagated using specialty support and media as described earlier by this laboratory [10,22,23]. Cells were grown to near confluence in 100 mm diameter Petri dishes (49.1 cm2 surface area/dish), culture media aspirated, washed twice with each 4 mL room temperature (RT)-equilibrated balanced salt solution (BSS), and exposed to 4 mL phosphate buffered saline (PBS) without or with 50 μM CET in the presence and absence of dimethylsulfoxide (DMSO) at 37 °C for 30 min, a time sufficient to nearly completely inhibit NKP [10]. Thereafter, cells were washed thrice with 4 mL ice-cold PBS, all supernatant fluids removed, and each dish covered with 0.4 mL M-PER® (Pierce Biochemicals) lysing solution containing 1:100 HALT reagent to stop proteolysis, and maintained for 10 min on a shaker at RT Cellular debris was collected into Eppendorf tubes, which were then spun for 2 min at 14,000 g in an Eppendorf RT centrifuge to separate the cytosolic supernatants from the crude membrane particulate material. It was noted that the cytosol of the CET-treated cultures was slightly yellow, suggesting that CET had entered the cells. The crude membrane pellet was resuspended and washed 4 times in fresh M-PER® solution, after which the final pellet was dissolved in 0.5 mL M-PER® solution containing 1% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS), a non-ionic detergent known to solubilize membrane proteins by replacing membrane lipids. These final solubilized crude membrane samples were thrice ultra-sonicated on ice with 3 bursts to achieve maximal membrane disintegration. All samples were spun at 14,000 g for 2 min, the supernatant kept and the small finite pellet discarded. Aliquots of 0.05 mL supernatant were removed for protein analysis by the BIORAD® technique. Samples were stored at -20 °C for further SERS-based analysis.
In order to obtain a purified membrane preparation, HLECs, after CET exposure, were washed as described above, but then collected from the Petri dishes by scraping into a HALT-containing NKP Na+-conformational buffer. The scrape was sonicated on ice as described above, and then centrifuged in a Beckman Optima XL ultracentrifuge (Beckman Co, Stanford Ca) at 4,600 g for 10 min at 4 °C to remove mitochondria and nuclear material followed by a final centrifugation at 131,000 g for 1 h at 4 °C to obtain the final plasma membrane pellet. This membrane pellet, further solubilized by sonication in NKP buffer containing 1% CHAPS and stored in aliquots at -80 °C, possesses only the α1 subunit of the NKP [10].
Estimation of Cell Volume and Calculation of Cytosolic Dilution
Based on previous estimates of endothelial and lens epithelial cell volume [23,24] of 10-11L/cell and an average cell count of 106/dish, a total cell volume of 10-2 mL/dish was estimated. As an example, given that 3 Petri dishes/control or experimental cells were available, which received a total M-PER lysing solution of 1.2 mL, the final total volume was 1.23 mL, i.e., the dilution of the original cytoplasm was estimated to be 41 fold; and 71 in another experiment.
AgNP Synthesis, Size-Selection, and Concentration
AgNPs were synthesized in a batch-wise process via a modified Creighton method [25] employing the ice-cold reduction of AgNO3 (1.0 mM, 50 mL) by NaBH4 (2.0 mM, 300 mL). The formation of spherical AgNPs of moderate size distribution (∼1-200 nm in diameter [26] was confirmed by the presence of an intense, symmetric ultraviolet-visible (UV-VIS) absorption peak at 392 nm. Once characterized, individual batches of AgNPs (350 mL each) were combined to achieve a total volume of 4.0 L. Tangential flow filtration [27] was then employed to size-select AgNPs, while eliminating the reaction byproducts/excess reagents and increasing the concentration of nanosilver for optimum SERS-based sensing. It was recently shown that the optimal size of AgNPs for SERS enhancements is ∼50 nm in diameter[28]. Briefly, 4.0 L of colloid was recirculated through a hollow fiber polysulfone filter module (50 nm pore size, 390 cm2 surface area, Spectrum Labs®) yielding a retentate of AgNPs of increased diameter (≥ 50 nm) and a filtrate containing dilute, smaller AgNPs (< 50 nm). The concentrated AgNPs (Ag50C) were stored at ∼ 4 °C until further use in SERS measurements.
Raman and SERS Measurements and Analysis
Raman and SERS measurements were conducted using a LabRamHR 800 system (Horiba Scientific, Inc.) coupled to either an internal HeNe laser (632.8 nm, 15 mW) or an external, frequency-doubled Nd:YAG laser (532.1 nm, 15 mW) with an acquisition time of 20 s and 3 s, respectively. The laser beam was focused into a 2 mL quartz cuvette carrying the Raman or SERS sample via a high stability, confocal microscope (BX 41) using a 50X objective. Serial dilutions of a CET stock solution (20 mM) were performed prior to the SERS analysis, yielding concentrations of 10-3 M through 10-9 M. Aliquots of the CET solutions were briefly incubated with the Ag50C colloid, followed by addition of 1.0 M KBr to induce AgNP aggregation and SERS “hot-spot” formation [28] prior to analysis. SERS samples of HLEC extracts of the plasma membrane and cytosol were prepared in a similar fashion, and the total CET concentration in each extract was estimated via a standard calibration curve constructed from the SERS scattering intensities of the stock CET solutions. Peak integration routines and user-defined baseline correction algorithms were performed in Origin 8.5 software for each Raman and SERS spectrum.
Results
Raman and SERS Analysis of Monomeric CET
Prior to SERS analysis, the UV-VIS absorption spectrum of 10-4 M CET revealed absorption peaks at 212, 267, 315, 338 and 406-430 nm (Fig. 1A). The extinction spectrum as a function of CET concentration, along with the molar absorption coefficients calculated for the most prominent UV absorption peaks at 267 and 315 nm, not readily available in the literature, are given in Figure 1B.
Fig. 1
A) UV-VIS absorption spectrum of monomeric CET solution (10-4 M). Dashed lines indicate the laser excitation wavelengths utilized in the SERS and Raman measurements, respectively. B) Plot of absorbance versus CET concentration along with linear fits to determine the extinction coefficients of the 267 nm and 315 nm absorbance bands in A. Uncertainties in the molar absorption coefficient (ε) are indicated parenthetically.
CET contains a highly chromophoric imine bond, which experiences pre-resonance effects under 532 nm excitation and leads to a large fluorescent background in the Raman spectrum. However, the Raman spectrum of monomeric CET exhibited weak, yet distinctive vibration modes upon excitation via the 632.8 nm HeNe laser (Fig. 2A). Characteristic functional group-vibration modes and those reported in the literature were utilized in order to provide a tentative assignment for each peak observed within the Raman spectrum (Table 1). Raman spectra of monomeric CET exhibited low signal-to-noise ratios for concentrations lower than 10-2 M, i.e., far above the physiologically active concentrations of the monomer.
Table 1
Tentative Assignments of Aqueous Raman and SERS Vibrational Modes of CETa,b,c. aMode designations: δ, scissoring; ν, stretching; ρ, rocking; σ, deformation; , ω, wagging; τ, torsion; i.p., in-plane; o.p., out-of-plane. bIntensity abbreviations: vw, very weak; w, weak; mw, medium-weak; m, medium; ms, medium-strong; s, strong. cAssignments derived from Ref [29,30,31]
Fig. 2
A) Raman and SERS spectra of monomeric CET chloride solution at 10-2 M and 10-6 M, respectively. B) Expanded Raman and SERS spectra of CET in the 100-1800 cm-1 fingerprint region. Acquisition times were 30 s and 3 s for Raman and SERS measurements under 632.8 nm and 532.1 nm excitation, respectively. The SERS spectra in A and B are intensity-shifted for comparison.
The addition of AgNP50C to stock solutions of CET not only yielded an exponential increase in scattering intensity, but also effectively quenched the fluorescence resulting from pre-resonance effects at concentrations as high as 10-4 M. The SERS spectra of monomeric CET showed a vibration-rich structure (Fig. 2B). In addition, the large increase in SERS signal observed for CET-AgNP50C mixtures yielded sub-picomolar detection limits (i.e, 10-10 M), which are well below IC50 values reported in the literature for diverse inhibitory functions by this drug. Vibration-rich SERS spectra (data not shown) were also observed for CET solutions containing the original unmodified AgNP colloid.
However, the SERS spectra using the modified Ag50C exhibited greater spectral intensities, which is in good agreement with literature reports of SERS scattering as a function of AgNP size. SERS spectral analysis of vibration bands was performed in a similar manner as for the regular Raman scattering spectra (Table 1). It should be noted that the SERS analysis of CET has been sparsely reported and the assignments provided here serve only as tentative descriptors.
The four vibration modes occurring at 430 (o.p. τ(C-C-C)), 527 (o.p. sextant σAryl), 659 (in bold in Table 1, defining stretched (C-C-C) bonds) and 778 cm-1 (o.p. (CH)) were utilized as “marker bands” of CET (Table 1) due to the large spectral intensity and lack of overlap with adjacent peaks. Therefore, the presence of any one of these four marker bands in cellular extracts can confirm the presence of monomeric CET. In addition to determining several characteristic marker bands for spectral identification, the in-plane skeletal stretching mode at 659 cm-1 was observed to have the greatest intensity of all the marker bands and was thus utilized to quantify the concentration of CET in biological samples. The constructed calibration function revealed a linear relationship between the normalized 659 cm-1 peak areas and the concentration of the CET solutions from 10-5 M to 10-10 M (Fig. 3). Despite the presence of the four marker modes, accurate peak integration parameters could not be obtained for solutions greater than 10-5 M due to the increased fluorescence of free CET molecules.
Fig. 3
Calibration plot relating the normalized 659 cm-1 peak area to the concentration of CET within each of the standard SERS samples. Error bars depict the standard deviation of three different CET solutions measured at each concentration. Error bars may be smaller than the symbols.
Small aliquots of plasma membrane and cytosolic HLEC extracts (50 - 100 µL) were quantitatively diluted with either water (Raman) or the Ag50C and KBr (SERS) to a total volume of 2 mL prior to spectral analysis. The normal Raman spectrum of the cytosolic extracts showed few, yet distinctive peaks corresponding to the DMSO solvent utilized in the extraction procedure (Figure 4A). A large fluorescent background was observed for the crude plasma membrane extract under normal Raman scattering conditions (i.e., in the absence of AgNPs), thus obscuring all vibration modes aside from a lone DMSO peak at 679 cm-1. No CET marker bands were observed in the cytosolic or plasma membrane extracts under regular Raman conditions, consistent with CET concentrations below 10-3 M in each sample (Fig. 4).
Fig. 4
A) Ordinary Raman spectra of cytosolic and crude plasma membrane extracts from LECs treated with CET in DMSO, collected using a 532.1 nm excitation line. B) SERS spectra of the same extracts in A, showing significant signal enhancement. Spectra in A and B are intensity-shifted for comparison.
The SERS spectra of CET-treated cytosolic, and crude and purified plasma membrane fractions exhibited the four distinct marker bands of CET (Fig. 4B, 5), even when using relatively short acquisition times (3 sec in each). In addition, the extensive biological matrix present in each sample did not appear to affect the intensity of each marker band. This is likely the result of a high degree of affinity between the positively charged iminium nitrogen and the predominantly negatively charged AgNPs in each sample. Previous SERS analyses of molecules of similar structure containing positively charged amino groups (e.g., rhodamine 6G) further corroborate this finding [28]. Total areas for the 659 cm-1 peak/sample were normalized, and concentrations were interpolated using the empirical relation derived from the calibration procedure and the dilution factors (Table 2).
Table 2
Estimated Concentrations of CET within HLEC Extracts.aCET concentrations were determined by accounting for a 1:20 dilution factor within the SERS samples. Values in ( ) are the SD for N = 3 individual measurements
Fig. 5
A) SERS spectra of untreated and CET- treated purified plasma membrane fractions expanded in the fingerprint region. B) SERS spectra of whole cell extracts from HLECs with and without CET exposure done in the presence of CHAPS (see Methods). Spectra are intensity-shifted for comparison purposes.
The data in Table 2 yield the CET molar concentration (column 5) measured in the 20 fold-diluted cytosol and membrane extracts (column 1). Extracts A and B are from the same experiment, and C and D from different experiments. In A/B HLECs were disintegrated by MPER, and only the insoluble crude membrane material further solubilized with 1% CHAPS. It is evident that on a mol CET μg-1 protein basis there was 3x10-2 more CET in the crude membrane fraction than in the cytosol. Based on the calculated dilution factor of the HLEC cytosol of 41 (see Methods), the [CET]i in the cytosol is computed as 43 x 10-6 M, which is close to the extracellular [CET] 50 x 10-6 M. When the crude membranes were subjected to two centrifugation steps (see Methods), the resulting plasma membrane pellet (Table 2 C) exhibited a ∼70% reduction in CET consistent with riddance from other particulate material such as endoplasmic reticulum, Golgi and mitochondrial membranes and most likely nuclear material. Still, normalized to protein, most of the CET preferred to lodge within the membranes. For the whole cell extract obtained in the presence of 1% CHAPS to solubilize all membrane material (with exception of a small but distinct CET-positive insoluble pellet), the CET content μg-1 protein was within the same order of magnitude, yielding 1.97 x 10-11 instead of 2.11 x 10-11 CET content μg-1 protein, consistent with similar cytosolic concentrations of CET in near equilibrium with the external media.
In conclusion these data suggest that SERS is capable to determine the free cytosolic concentration of monomeric CET, but also shows that a rather large CET membrane pool exists suggesting that the total CET that is associated with the cell probably exceeds diffusion equilibrium.
Discussion
Previously, it was proposed that the QBA CET inhibits NKP by binding to at least one BH1 motif discovered within the NKP's actuator (A) domain, thus, for example constraining the 120o rotation of the A domain and its temporary protection of the high energy aspartate 369-phosphate residue that confers the conformation required for the trapping of the 3 Na+ ions in an occluded state [10].
The rationale for the present study was to provide proof of concept by establishing the distribution of CET between the cytosol and the crude plasma membrane preparation using a novel technological approach, SERS, with the ultimate goal to identify the compound within the NKP structure. The use of the large signal enhancement of SERS over ordinary Raman spectra gave us the advantage to a) characterize CET based on its four marker bands (Fig. 1), b) identify it in concentrations less than 104 orders of magnitude smaller than Raman spectroscopy would have done (Fig. 2A and B) and thus establish a calibration curve (Fig. 3) over five orders of magnitude, and c) identify within cytosol and crude and purified plasma membrane fractions, CET concentrations in the micromolar range (Fig. 4 and 5, Table 2). Therefore, the greater sensitivity of the SERS method coupled to the minimal pre-resonance effects of the 532 nm laser excitation wavelength also easily surpassed by several orders other approaches such as UV-VIS spectrometry, where the limit of detection was around 10-4 to 10-5 M (c.f. Fig. 1B). Furthermore, these results show extreme promise in the ability of SERS to be used as a sensitive and specific technique for tracking small molecule cellular metabolites with visible absorption resonances, even within crude cellular matrices.
As shown in Table 2, the CET cytosolic concentrations were quite close to that of the bathing medium of 50 μM, raising the question as to whether CET might equilibrate across the plasma membrane, as a positively charged monomer (CET+) or as pseudo base that does exist at higher pH values [30]. For the sake of the argument, if CET traversed the membrane as CET+, then, according to the Nernst Equation: log [CET+]i = log [CET]o - Em/61.5, the free intracellular CET concentration, [CET]i, should have been 5 fold above equilibrium at 244 μM, assuming a membrane potential, Em, of - 40 mV typical in epithelial cells, where [CET]o is the external CET concentration = 50 μM, and 61.5 mV = 2.3 RT/zF at 37 °C with the usual meanings. If we assume that the cytosolic and membrane CET content add up altogether within one cellular compartment, the total [CET]i should be 267 μM, a value that is close to that predicted from the Nernst equation. Thus it is not improbable that indeed CET traverses as CET+ rapidly equilibrating perhaps through cationic channels that are capable to accommodate the 348 dalton compact benzo-phenanthridine ring. In this context, it is of interest that epithelial cells have at least two isoforms of organic cation channels, OCT1 and OCT2 [32], and their presence is currently evaluated in terms of their capability to transport positively charged QBAs. Moreover, it has also been shown that transport through epithelial OCTs plays a pivotal role in the targeting of positively charged QBAs to the plasma membrane. The time-dependent fate of CET cellular distribution will be further investigated to shed light on this issue.
Finally, based on the calculated number of 1013 CET molecules μg-1 crude membrane protein and assuming that at least 10% of the membrane protein is NKP with a molecular weight of 105 Daltons, one can compute a ratio of CET/NKP of about 100, a number still far exceeding the expected ratio of not more than 2 based on the putative presence of two BH1 motifs/NKP with solvent accessibility for CET. Experiments ongoing with affinity purified NKP using mouse-anti human α1 NKP subunit antibody either from plasma membrane-free cytosolic or from purified membrane fractions suggest a CET/NKP ratio of <1. Differences currently addressed may be due to a) insufficient washing of the crude membrane particles releasing all trapped CET only in the presence of CHAPS, b) the tendency of CET to partition and lodge within the structurally not unrelated cholesterol component of the membrane bilayer, c) charge interactions with negatively charged lipids and proteins with neuraminic acid residues, d) absorption to residual nuclear material in particular in the crude membrane fraction, e) trapping of the alkaloids within putative organic cation channels, and last but not least interferences of solvents and CET-binding macromolecules with AgNP augmentation of the electromagnetic signals, i.e. SERS-based nano-sensing.
Conclusions
The NKP is a target of QBAs and recent findings implicate at least two BH1-like motifs within the crystal structure of the transporter to which CET, a BH3-mimetic, might bind and inhibit function [10]. Using a novel approach, SERS-based nano-sensing, the distribution of the QBA CET between the plasma membrane and the cytosolic extracts of HLECs was studied. The data suggest that CET most likely travels as the charged monomeric CET+, because already the free cytosolic CET concentration approached the external CET concentration, and the plasma membrane contained CET exceeding that of the cytosol on a protein basis by more than 2 orders of magnitude.
Conflict of Interests
None
Acknowledgments
Part of this work was presented at the 2013 Lake Cumberland Biological Transport Meeting, Ky. Dr. Sizemore highly acknowledges the WSU start-up fund.
PubMed ID
