Cell-modified plasmonic interface for the signal-amplified detection of Cucurbitacin E.

Cucurbitacin E (CuE) plays an important role in anticancer, antichemical carcinogenesis, and body immunity, etc., and the detection of its concentration is meaningful to pharmacological studies and clinical applications. However, the small molecular weight of CuE makes direct detection difficult through a surface plasmon resonance (SPR) sensor. In this work, we propose a cells-amplified signal strategy at the plasmonic interface, realizing the detection of CuE with ultra-low concentration. The seeded HeLa cells are modified onto the surface of the SPR sensor, and a small amount of CuE can lead to the remarkable morphology change of cells and the release of cell-related substances onto the plamonic interface, thus significantly amplifying the signal. Experimental results show that by using an unmodified SPR sensor with the bulk refractive index sensitivity of 2367.3 nm/RIU (RIU: refractive index unit), there no effective signal can be detected during the CuE concentration range of 0-100 nM; whereas, employing the proposed strategy, the signal for CuE detection can be significantly enhanced, resulting in a high detection sensitivity of 0.6196 nm/nM, corresponding to a limit of detection of 45.2 pM (25.2 pg/mL). The proposed cells-based signal amplifying strategy shows great potential applications in drug screening or bio-sensing to small molecules with low concentration.


Introduction
According to the estimates from the World Health Organization, cancer is the first or second leading cause of death before the age of 70 years in the most of the countries [1]. Meanwhile, the drugs from natural plants or products have shown great potentials in the anticancer [2]. Cucurbitacins, belonging to a large family of triterpenoid compounds isolated from cucurbitaceae plants, have been used as medicines for centuries [3,4]. Several members of cucurbitacins, such as cucurbitacin E (CuE), B, and I, have been demonstrated to exhibit a wide range of biological activities including anticancer, anti-viral, hepatoprotective, and anti-inflammatory effects [5,6]. However, the pharmacodynamic effect is a critically-related to the applied dose, and a little excess dose will cause adverse effects [7]. Taking the CuE as an example, an excess dose will lead to severe anemia, frequent vomiting and diarrhea, and even respiratory and circulatory failure and death, etc [8]. Therefore, the accurate and highly-sensitive detection to the cucurbitacins concentration is greatly-important to the pharmacological studies and clinical applications. The current detection methods focus on the high-performance liquid chromatography (HPLC). For example, Joydeb et.al demonstrated a reversed-phase HPLC method for CuE detection, and achieved the limit of detection (LOD) of 3.45 µg/mL (i.e. 6.2 µM) in the concentration range of 1-100 µg/mL (i.e. 1.8-180 µM) [9]. Zhibin et.al proposed an ultra-high performance liquid chromatography-tandem mass spectrometry method for the simultaneous determination of CuB and CuE, lowering the LOD to 1.6 ng/mL (2.86 nM) and 1.58 ng/mL (2.84 nM) for CuB and CuE, respectively [10]. Furthermore, employing the same method, Nicole et.al succeeded in extending the detection objects to CuB, E, I and E-glucoside [11]. Even though the HPLC has achieved success in detecting cucurbitacins, it is still limited due to the expensive instrument, complex operation, and the disability for the real-time and in-line detection.
Nowadays, surface plasmon resonance (SPR) sensors have been established as a powerful tool to monitor the biomolecular events occurred in the vicinity of sensor surface [12,13]. SPR, a surface sensitive characterization phenomenon, can be used for analyzing the kinetic parameters and detecting the concentrations of biomolecules with high sensitivity and selectivity [14,15]. However, SPR sensors encounter a big problem in directly detecting the molecules with small weight (∼500 Da or less) or ultra-low concentration, due to the limited sensitivity to bulk refractive index (RI) [16]. To improve the sensitivity of SPR sensor, various methods have been proposed. One of the most common methods is adding an overlayer, having a nanoscale thickness and a relatively higher RI, onto the plasmonic interface of an SPR sensor, where the two dimensional materials, for example MoS 2 [17], TiO 2 [18], graphene oxide [19], and MoSe 2 [20], are well-qualified. In recent years, engineering the metal layer to gold nanorod array [21] or multilayer metal-dielectric structure [16], thus forming the so-called metamaterials, has been demonstrated to be a greatly-promising method for overcoming the limit of sensitivity. However, the aforementioned methods either need complicated fabrication procedures or the improved-sensitivity is still not enough for the detection of small molecules such as the CuE (only 556.69 Da).
On the other hand, numerous studies have shown that the CuE can disrupt the mammalian cells cytoskeleton, which is a fiber network framework to maintaining the basic form of cell [22,23]. Consequently, the morphology of cells is quickly impaired, and the cell motility and cytokinesis are also suppressed. Emerging evidence suggests that cucurbitacins can induce the disruption of actin cytoskeleton and the increase of actin fragments in various cell types [5,24,25], which further raises the environmental RI around the cells. This RI change is detectable for SPR, compared with the small amount of CuE. In other word, the signal from CuE is amplified by the cells, providing us a promising way to detect CuE with enhanced sensitivity.
In this work, we propose and demonstrate a signal amplification strategy for the detection of ultra-low concentration CuE. Here, the HeLa cells, acting as the intermediate media for amplifying the signal, are modified on the plasmonic interface of a traditional SPR sensor to construct a cells-amplified SPR (CA-SPR) sensor. The CuE acting on HeLa cells can lead to the cells deformation and releasing the cell-related substances, thus an amplified RI change at the sensing area of sensor. Employing the proposed strategy, a high sensitivity of 0.6196 nm/nM and an LOD of 45.2 pM (25.2 pg/mL) can be achieved experimentally using a traditional SPR sensor. The proposed signal enhancement strategy features the advantages of low-cost and ease of operation, providing a promising method for the detection of small weight molecules with low concentration.

Materials and methods
HeLa cells used in the experiments were obtained from the cell bank of the Chinese Academy of Sciences (Shanghai, China). The DMEM medium (Gibco/Invitrogen), containing 10% fetal bovine serum, 100 U/ml penicillin, and 100 ug/ml streptomycin, was used for cell culture. CuE with 98% purity was purchased from Shun Bo Biological Engineering Technology Co, LTD (Shanghai, China). Ethanol solutions with different RIs, were prepared by mixing a certain amount of ethanol with DI water, and their RIs were characterized by an Abbe refractometer (NT52-975, Edmund Optics Co., Ltd.). CuE solutions with different concentrations (0, 40, 60, 80, and 100 nM) were prepared with cell culture medium.
The HeLa cells taken out from the liquid nitrogen jar were recovered by rapidly shaking in a water bath at 42℃, and then seeded in a 25 cm 2 culture flasks using DMEM. The recovered cells were cultured for a week at 37℃ in a humidified cell incubator with 5% CO 2 . During culturing, the medium was renewed in every two days. The modification of HeLa cells on SPR chips includes two steps [ Fig. 1(a)], i.e. the SPR chip fabrication and the cells modification. To fabricate the SPR chips, we used vacuum evaporating method (E6080, Sichuan Sisheng Vacuum Equipment Co. Ltd., China) to successively deposit an adhesive chromium layer (∼5 nm) and a gold layer (∼50 nm, Fig. S1) on a glass slide (K9 glass, 28×20×0.17 mm in length×width×thickness), which was first cleaned by piranha solution (H 2 SO 4 /H 2 O 2 in a 3:1 ratio) for 10 minutes. Then, the gold-deposited glass slides were rinsed sequentially in ethanol and deionized water, and were dried in a drying oven under 160°C for 2 hours for sterilization. To modify cells onto the SPR chips, six sterilized SPR chips were firstly placed in the each well of a six-well cell culture plate, respectively. Then, the cells, which are in the logarithmic growth phase and with the >95% trypan blue exclusion rate, were collected and seeded in the wells that contain SPR chips. Through cell counter, the cell density in each well was controlled to be as equal as possible (∼2×10 5 cells/well). Finally, the six-well plate was maintained at 37°C in a humidified incubator with 5% CO 2 for 24 hours to let the cells adhere onto the surface of chips naturally. Following the process described above, the fabrication CA-SPR chips is completed.
A typical Kretschmann SPR configuration was adopted for the measurement [Figs. 1(b) and 1(c)]. The CA-SPR chip was mounted on the top of a prism (K9 glass) with the assistance of RI matching cedar oil that can fill the gap between the prism and the SPR chip. Immediately, a ∼300 µL CuE solution was dropped on the chip, and the CuE action on the cells was monitored and recorded by a self-built SPR system shown in Fig. 1(c). The light from a tungsten-halogen lamp [LS-1, Ocean Optics] is firstly polarized to p-polarization light. Then, the p-polarized light passes through a focusing lens and reflects at the slide/gold interface, and the reflected light is collected by a spectrometer (USB 4000, Ocean Optics) with the aid of a lens. At the slide/gold interface, if the wave vectors of the incident light and the surface plasmon wave (SPW) match with each other, the energy of incident light will couple into SPW, which propagates along the gold film accompanying with the attenuation due to the large damp of gold material. Consequently, the output spectrum will feature an absorption dip centered at a specific wavelength, which is called the resonant wavelength. Because the strongly-confined evanescent field of SPW penetrates into and interacts with the medium over the gold surface in several hundred of nanometers, a small RI change in the vicinity of gold surface can induce a significant shift of the resonant wavelength. Moreover, the CuE-induced apoptosis and collapse of cytoskeleton for the HeLa cells modified on the gold surface causes the RI change in the sensing area. Therefore, the shift of resonant wavelength can reflect, to a certain degree, the progress of CuE acting on cells, and the total shift amount can indicate the degree of action, which is related to the concentration of CuE.  Fig. 2(b)]. The obtained RI sensitivity falls into the normal range of the Kretschmann-based prism scheme without enhancement processing [17,19], and it is far from directly detecting the bulk RI change induced by the CuE concentration, which will be demonstrated in the following experiment. The transmission spectra of an unmodified SPR chip under CuE solutions with different concentrations were measured and shown in Fig. 2(c). The resonant wavelength remains almost unchanged only with a fluctuation of 2.03 nm when the CuE concentration varies from 0 to 100 nM, and the small fluctuation can be attributed to the slight change in the light path of SPR-chip-prism component during the operation of changing CuE solution. The result confirms that the unmodified SPR chip can't be used for the CuE detection, because the RI change induced by the CuE concentration change is too small to be detected.

Sensing performance of the unmodified SPR chips
In order to access the influence of temperature, the transmission spectrum of the SPR sensor covered by distilled water under different temperatures were tested. Figure 2(d) shows that the resonant wavelength has a small blue-shift of 1.15 nm when temperature increases from 22 to 28°C . The blue-shift is caused by the negative thermo-optical coefficient (approximately -1.0×10 −4 RIU/°C) of water [26], and the corresponding theoretical blue-shift amount is 1.41 nm, which is close to the measured 1.15 nm. In our experiments, the ambient temperature is controlled at 25 ± 0.5°C, corresponding to a temperature-induced resonant wavelength uncertainty of 0.23 nm, which is much smaller compared to the resonant wavelength shift induced by the CuE acting on cells in the following experiments.

Sensing performance of the CA-SPR chips
By mounting CA-SPR chips onto the prism and then dropping CuE solutions onto the chips, we can monitor the action of CuE on the cells via monitoring the resonant wavelength in real time. During the entire test period of 3 hours, the transmission spectrum was recorded in every 5 minutes. Figures 3(a)-3(c) present the time-resolved spectra of the CA-SPR chips exposed to the CuE solutions with the concentrations of 0, 60, and 100 nM, respectively. For the case of 0 nM CuE solution, i.e. the pure culture medium, we can see that the resonant wavelength almost remains unchanged only with a standard fluctuation of 0.53 nm [ Fig. 3(a)], which can be regarded as the system noise. This indicates the cells adherent on chips can keep in good condition under the pure culture medium environment, and the growth in the 3 hours period has little influence on the resonant wavelength, thus confirming that the resonant wavelength shift shown in Figs. 3(b) and 3(c) is unrelated to the cells growth or the other analytes including fetal bovine serum, penicillin, and streptomycin. Contrarily, for the cases of 60 and 100 nM CuE solution, the resonant wavelength exhibits a significant red-shift along with time [Figs. 3(b) and 3(c)], indicating that the CuE action on cells results in a detectable RI change in the vicinity of chip surface, and the mechanism will be discussed in the following section. In addition, the wavelength shift induced by 100 nM CuE is larger than that of 60 nM [ Fig. 3(d)], and the concentration dependence on the wavelength shift provides the foundation for CuE detection.
Based on the same procedures, the 40 and 80 nM CuE action on CA-SPR chips were characterized. Considering the resonant wavelength shift at the 180th minute, we can obtain the wavelength shift depending on CuE concentration, as shown in Fig. 3(e). For the CA-SPR chips obtained by the same procedures, the CuE-induced wavelength shift increases with the CuE concentration, and they have a good linear relationship in the concentration range from 0 to 100 nM. Linear fit result shows that the sensor has a CuE detection sensitivity of 0.6196 nm/nM with a linear correlation coefficient 0.9456. The LOD can be calculated by LOD=3σ/S [27], where S is the sensitivity and 3σ (=0.028 nm) is the three times of standard deviation of the output noise, which is measured as the standard deviation of the resonant wavelength under the same conditions (including the surrounding RI, temperature, equipment parameters, etc.) for the current sensing system [28]. Moreover, all the acquired spectra are first interpolated employing the "spline" method and then smoothed by "lowess" method in Matlab, which allows us to quickly determine the resonant wavelength with a resolution of 0.002 nm. Consequently, the LOD is calculated as 45.2 pM (25.2 pg/mL), which is lower than the reported results obtained by HPLC method [9,10].

Discussions
In order to investigate the amplification mechanism of the CA-SPR sensor, the cells incubated CuE solutions were observed under a microscope for 3 hours. We can see from the Fig. 4 that: when cells are exposed to the control solution (0 nM CuE), the morphology doesn't show obvious change; however, when the CuE was applied, the cells start to change from a clear irregular polygon shape to a blurred morphology accompanying with the rupture of cell structure, and the extent of morphology change becomes greater as the increase of the CuE concentration and the action period. The change trend of cell morphology agrees well with that of the resonant wavelength obtained by measurement as shown in Fig. 3.
The shift of resonant wavelength is determined by the RI change in the vicinity of the plasmonic interface between the gold film and analyte. As we can see from the simulated distribution of electric field [Figs. 5(a)-5(c), the details for simulation can be seen in Supplement 1], the evanescent field intensity reaches the maximum at the interface of gold/analyte and penetrates into the analyte with an exponential decay. That is, only the RI change in a limited ∼300 nm height area [from Fig. 5(c)], where the interaction between the evanescent field and the analyte occurs, makes the contribution to the wavelength shift. However, because of the relatively-large size of the adherent HeLa cell (∼15 µm diameter [29]), the evanescent field can only penetrate through the cell membrane (∼7 nm) and into the cytoplasm with a limited distance, as shown in Fig. 5(d). Therefore, when the cells are exposed to the pure cell culture media without CuE, the cell-growth-induced RI change is too small to detect, well explaining the result shown in Fig. 3(a).
Once the adherent cells are applied with CuE, the cytoskeleton starts to change even to be destroyed [30], and the apoptosis of some cells start occurring [31]. Therefore, as shown in Figs. 5(e) and 5(f), more and more cell-related substance, including the cytoplasm and cell fragments, will fall into the sensing area above the gold surface within ∼300 nm height. Meanwhile, the collapse of cytoskeleton downs the cell nucleus into the sensing area as well. As a result, the RI in the sensing area will increase, leading to the red-shift of the resonant wavelength as shown in Figs. 3(b) and 3(c).
From the above discussion on the cell morphology and the electric field interaction, we can well qualitatively explain the results obtained by experiments. Though we can't directly use  the SPR sensor to detect the CuE, the modification of cells can significantly amplify the signal, achieving an ultra-low concentration detection to CuE with the LOD of 45.2 pM (25.2 pg/mL). The LOD achieved here is not only lower than those obtained by the well-established HPLC method [9][10][11], but also lower than those in detecting biological substance achieved by other signal-amplification schemes, for example two-dimensional materials modification [32,33,38], magnetic or magneto-plasmonic nanoparticle enhancement [36,39], polydopamine modification [35] (see the details from Table 1 and Table 2). Although the presented cell-based assay shows less specific than the conventional biochemical assays, it provides a convenient, sensitive, and rapid method for the CuE detection, which might offer great promise for a general platform of chip-based bio-sensing and bio-analysis such as the mode of action, pathway activation, toxicity, and the phenotypic responses of cells mediated by exogenous stimuli.

Conclusions
In summary, we have demonstrated a novel signal-amplified strategy on plasmonic interface for the ultra-low concentration detection of CuE. The plasmonic interface of an SPR sensor is modified with HeLa cells as the intermediate media to amplify the signal, based on the principle that the CuE can induce the cell morphology change and the release of cell-related substances, resulting in an enhanced interaction of the evanescent field with the sensed substance. Experimental results show that the unmodified SPR sensor (with the bulk RI sensitivity of 2367.3 nm/RIU) fails to directly detect the CuE. Whereas, after modification with cells, the sensor becomes extremely-sensitive to the CuE, exhibiting a high sensitivity of 0.6196 nm/nM in the range of 0 to 100 nM, corresponding to an LOD of 45.2 pM (25.2 pg/mL). This work firstly proposes a cell-amplified signal strategy for the SPR detection of drug concentration with a sensitive, simple, and rapid method, showing great promise for a general platform of chip-based bio-sensing and drug-screening for small weight molecules. Disclosures. The authors declare no conflicts of interest.
Data availability. Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
Supplemental document. See Supplement 1 for supporting content.