Preparation of Hydrothermal Carbon Quantum Dots as a Contrast Amplifying Technique for the diaCEST MRI Contrast Agents

Abstract

The discovery of exogenous contrast agents (CAs) is one of the key factors behind the success and widespread acceptability of MRI as an imaging tool. To the long list of CAs, the newest addition is the chemical exchange saturation transfer (CEST)-based CAs. Among them, the diaCEST CAs are the safer metal-free option constituted by a large pool of organic and macromolecules, but the tradeoff comes in terms of smaller natural offset. Another major challenge for the CEST CAs is that they need to operate in the tens of millimolar concentration range to produce any meaningful contrast. The quest for high efficiency diaCEST agents has led to a number of strategies such as use of hydrogen bonding, use of equivalent protons, and use of diatropic ring current. Here, we present carbon quantum dot formation using hydrothermal treatment as a new strategy to amplify diaCEST contrast efficiency. We show that while the well-known analgesic drug lidocaine hydrochloride when repurposed as a diaCEST CA produces no contrast at the physiological pH and temperature, the carbon dots prepared from it elevate the physiological contrast to a sizable 11%. Also, the maximum efficiency at an acidic pH gets amplified by a factor of 2 to 46%. The study showed that the enhancement in CEST efficiency is reproducible and the pH response of these carbon dots is tunable through variation in synthesis conditions such as temperature, duration, and precursor concentration.

Introduction

Excellent spatial and temporal resolution provided by magnetic resonance imaging (MRI) along with its non-invasive nature make it one of the most successful diagnostic imaging techniques for soft tissues.¹ While there exist techniques for in vivo imaging that provide superior sensitivity such as positron emission tomography, they often lack in resolution.² Optical imaging methods on the other hand suffer from poor in-depth penetration. The widespread popularity of MRI can be attributed to a large extent to the discovery of exogenous contrast agents (CAs).^3,4 These are small molecules that create artificial contrast between the area of investigation and the surrounding by selectively altering the water signal intensity. Thus far, all the CAs approved for clinical usages are relaxation-based ones, which either alter the T₁ or the T₂ of water in the proximity.^4,5 The relaxation-based CAs have, however, reservations of their own—the most important one being that contrast cannot be turned-off once administered. Also, the long term safety⁶ is a matter of debate for the relaxation-based CAs which are mostly gadolinium (III)-based chelates.⁷ In 2000, chemical exchange saturation transfer (CEST) was utilized as an alternative MR imaging contrast generation technique in order to address the limited sensitivity, switchability, and safety of CAs.⁸ CEST CAs alter water intensity by transferring saturation to the accessible water through chemical exchange after being selectively saturated by the application of a long low-power radio-frequency pulse train.^9,10 As the contrast is generated only when the solute exchangeable proton pool is first selectively saturated, it becomes rather straight forward to either turn-on or turn-off the contrast. However, the requirement of selective saturation of the solute demands that the solute exchangeable protons resonate far from water protons. A class of CEST CAs known as the paraCEST CAs excels in creating large offsets (Δω) due the presence of a paramagnetic metal center ligated with multi-dentate ligands.^9,11 A large offset helps in two more ways. First, it helps to avoid overlapping with almost all endogenous exchangeable protons present inside the body. Second, a large offset allows protons with relatively higher exchange rates (k_ex) to produce contrast remaining in the slow to medium exchange regime (k_ex ≤ Δω).¹² A larger k_ex then in turn produces high efficiency (larger contrast).

diaCEST CAs,¹³ on the other hand, do not contain any metal and hence are considered a safer alternative. However, they suffer heavily both in terms of offset and efficiency. The quest for efficient diaCEST CA has prompted testing of almost all classes of organic compounds containing labile protons in the slow to medium exchange regime, including endogenous CAs such as glucose,¹⁴ glycogens,¹⁵ glycosaminoglycan,¹⁶ protamine,¹⁷ glutamate,¹⁸ urea,¹⁹ nucleic acids,²⁰ peptides,²¹ and so forth present inside the human body. Few of the already known clinical agents have also been repurposed as diaCEST CAs.^22,23 However, even then the sensitivity or efficiency of diaCEST CAs remains an ongoing challenge.

The efficiency of CEST CAs depends, in an intertwined manner, on a number of molecular properties such as the optimality of exchange rate, offset from water, and availability of more than one equivalent exchangeable protons. A synergy between these factors produces the most efficient diaCEST agents. For example, salicylic acid²⁴ not only gains from a large down-field offset of ∼9 ppm, it belongs to a group of compounds in which either intramolecular²⁵ or intermolecular^22,26 hydrogen bonding slows down the exchange rate so optimally that the contrast efficiency increases. On the other hand, few porphyrin derivatives²⁷ along with a large up-field shift of almost −8 ppm benefit from two equivalent exchangeable protons similar to iopamidol,²⁸ and certain polymers,²⁹ thereby naturally producing good contrast. Finally, a few diacetamide derivatives²⁶ exploit the benefit from both equivalent exchangeable protons and finely tuned exchange rates through intermolecular hydrogen bonding. Nonetheless, a diaCEST CA would still require tens of millimolar concentration to produce any meaningful contrast. The problem is further compounded by poor water solubility of a few compounds requiring prohibited quantity of material to be injected.

Here, we present conversion to water-soluble carbon quantum dots (CDs) as a strategy to address together the low efficiency and solubility problems of diaCEST CAs. Nano-particles are not new to the world of medicine³⁰ and imaging.³¹ Among different nano-particles, CDs have recently gained tremendous popularity as they are cost effective, generally safer than many other metal-containing nano-particles, highly fluorescent, and many of them show better water solubility than their respective precursors.³² In 2019, an arginine-modified glucose-based CD was introduced as a new class of diaCEST CAs by Liu’s group.³³ Their microwave-synthesized CDs showed good contrast at physiological pH owing to the presence of both hydroxyl and guanidinium groups on the surface. As amides form the basic block of several diaCEST CAs, we chose to test if some amide group-containing molecule shows enhancement in diaCEST efficiency post CD formation. For the current study, we selected Lidocaine hydrochloride (L-HCl for brevity), a clinically approved analgesic drug that hitherto has not been reported as a diaCEST CA. We employed hydrothermal treatment for CD formation. We show that while L-HCl does not produce any contrast (<1%) at the physiological condition (pH 7.4, 37 °C), the L-HCl CD shows a good 11% contrast at the same concentration. Also, the peak efficiency of L-HCl gets amplified by almost 100% upon formation of CD.

Methods and Materials

Instrumentation and Materials

All NMR experiments for preparation of Z-spectra were performed at 310 K using a BBFO broadband probe on a 9.4 T (400 MHz) Bruker AVANCE-III Nanobay NMR spectrometer. All ¹H and ¹³C spectra were recorded at 298 K using a Bruker 700 MHz (16.4T) Avance-III HD liquid-state NMR spectrometer equipped with a triple-channel cryoprobe with z-gradients. D₂O-filled capillaries were placed inside the NMR tubes for achieving deuterium lock. A Jasco V-730 spectrophotometer was used to record the steady-state absorption spectra. The emission spectra were recorded on a Shimadzu RF-6000 fluorescence spectrophotometer. High resolution TEM (HR-TEM) images were captured using a JEOL (JEM-2100, 200 kV) electron microscope.

Synthesis of Carbon Dots

L-HCl-based carbon dots were synthesized using the hydrothermal method under variable conditions starting from different precursor concentrations. 2.5 and 5 mg/mL stock aqueous solutions of L-HCl were prepared by dissolving, respectively, 250 and 500 mg of compound into 100 mL of deionized water. The mixtures were subjected to a rotary-shaker until clear colorless solutions were obtained. Each of the solutions was then placed inside a hot air oven after separately taken and transferred to a teflon-lined steel autoclave reactor of 50 mL capacity. With the 2.5 mg/mL stock, two CDs were prepared at 200 °C with different treatment durations—one for 24 h, and the other for 10 h. The 5 mg/mL stock, on the other hand, was subjected to hydrothermal treatment for three different temperature–duration pairs: (200 °C, 10 h), (200 °C, 24 h), and (180 °C, 24 h). Certain portions of the stock solutions were kept aside for experiments as reference compound and for comparison. After the solution went through their respective stipulated treatments, the autoclave was taken out of the oven and was allowed to cool down at room temperature. The resulting clear solutions showed different shades of yellow. The change of color acted as an indicator for the formation of the CDs. The color changed the least for the treatment at 180 °C. Each sample was passed through sterilized 0.2 μm chromatography filter to ensure the removal of all suspended large particles. The CD solutions showed blue to cyan characteristic fluorescence upon subjecting to UV light.

Preparation of CEST Z-Spectra

CEST experiments were performed with selective solute saturation at a resolution of 0.25 ppm between offsets −8 and 8 ppm with respect to the water. Irradiation radiofrequency of 213 Hz (5 μT) field strength was used for the saturation duration of 3 s. The saturation duration and irradiation frequency were kept constant throughout to facilitate better comparison of different compounds. Z-spectra were generated by plotting the normalized water peak intensity (100 × M_z/M₀) as a function of irradiation offset frequency, where M_z and M₀ are the water peak intensities with and without on-resonance selective saturation. CEST contrast efficiency was given by the asymmetric magnetization transfer ratio (MTR_assym)¹²

M(Δω) and M(−Δω) are water peak intensities, respectively, after saturation at the site of the exchangeable peak (offset Δω) and saturation at a negative offset (offset −Δω) present on the other side of the water peak. All post-acquisition processing and plotting were done using in-house MATLAB (R2014b) scripts.

Exchange Rate Calculation

Exchange rate calculations were performed using a method introduced by Dixon et al. in which the value of M_z/(M₀ – M_z) is linearly fitted against 1/ω₁² from a set of Z-spectra acquired with variable saturation fields. The linear fit gives the value of exchange rate k_ex when the x-axis intercept is equated to −1/(k_ex²).³⁴ ω₁ is given by γB₁, where γ is the gyromagnetic ratio of proton and B₁ is the r.f. amplitude. This method requires conditions such as minimal water direct saturation (DS), saturation power less than the offset Δω (both in Hz), and a saturation duration long enough for complete saturation. To fulfil these criteria, a long saturation duration of 6 s and low direct saturation producing Sinc pulse-train were used. Taking Δω of the compound into account, maximum saturation powers were adjusted. All fittings and plotting were done using in-house codes in MATLAB 2014b.

Results and Discussion

Lidocaine Hydrochloride as a diaCEST CA

Lidocaine is an amphiphilic compound generally used as a local anaesthetic either as a topical solution or in the injection form. It contains an amide group adjacent to a phenyl ring. The diatropic ring current produces a useful down-field shift (Δω = ∼4.5 ppm with respect to water) for the amide proton, making it suitable for CEST contrast generation. Additionally, it contains a terminal tertiary amine group, which also shows diaCEST contrast when protonated. As the protonated form of local anaesthetics has better stability and solubility than its free base, the hydrochloride salt form is used in almost all medicinal applications.³⁵ The white crystalline L-HCl salt (1 in Figure 1a) is highly water-soluble and exists in the monohydrate form. Figure 1b shows the ¹H NMR spectrum of 1 in water (Figure S1 for ¹³C spectrum). The exchangeable protons are exchange-broadened and are missing. A 5 mg/mL aqueous solution (17 mM) of 1 produces appreciable CEST contrast in a wide acidic pH range between pH 4.5 and pH 7.0 (Figure 1c), reaching a maximum efficiency of 26.1% at pH 5.5. The MTR_assym plots (Figure 1d) show interesting features of 1. While around neutral pH (up to pH 6.5) the quaternary ammonium group Et₂NH⁺ predominantly showed CEST at around 2.25 ppm downfield to water, the CEST peak slowly moved to 4.25 ppm in the acidic pH range when the amide protons (-CONH) started exchanging with water. At a very acidic pH (below pH 5), the amide proton of course stopped exchanging and the overall CEST efficiency rapidly dropped. In spite of the fact that 1 has been proven safe, is easily soluble in water, and produces an appreciable 26% CEST contrast at pH 5.5, the lack of contrast at the serum physiological pH makes it unsuitable as a general purpose CA. We now show how CD formation by hydrothermal treatment of 1 helps to tune the CEST properties more toward a favorable pH range and produces an overall enhancement of the contrast efficiency. We also show how different conditions during the hydrothermal treatment such as precursor concentration (c in mg/mL), temperature (T in °C), and duration (d in hours) can be used as individual tuning parameters for optimization of the CEST contrast profile of the CDs. Carbon dots prepared by hydrothermal treatment of “c” mg/mL solution of 1 at “T” °C for “d” hours would be denoted as ^TCD_d^c.

Figure 1.

Characterization and CEST profiles of lidocaine hydrochloride: (a) structure and (b) ¹H ¹D-NMR spectrum in water. (c) Z-spectra and (d) MTR_assym as a function of pH of the solution in the range from pH 4.5 to pH 7.4. CEST efficiency at the physiological pH (7.4) is specifically mentioned.

Optimization of Hydrothermal Synthesis Conditions

Condition tuning led us to the most favorable carbon dot formation with highest physiological contrast when we used 5 mg/mL aqueous solution of 1 and treated it at 200 °C for 24 h (²⁰⁰CD₂₄⁵,Figure 2-center). The maximum efficiency of ²⁰⁰CD₂₄ reached nearly double of that of 1 (45.7% at pH 5.5). More importantly, the efficiency at the physiological condition showed a good 11% from nearly no contrast (<1%) by 1. From this favorable parameter set, when we varied the duration (²⁰⁰CD₁₀⁵,Figure 2-top left), the temperature (¹⁸⁰CD₂₄,Figure 2-bottom left), the concentration (²⁰⁰CD₂₄^2.5,Figure 2-top right), and the concentration–duration pair (²⁰⁰CD₁₀,Figure 2-bottom right), we found that the physiological contrast dropped drastically back to near zero. Among the three conditions, we found that duration and concentration have more drastic effects on the pH profile than temperature. Both ²⁰⁰CD₁₀⁵ and ²⁰⁰CD₂₄ show contrast 10% or more only at pH 5.5 or below. On the contrary, ¹⁸⁰CD₂₄⁵ showed nearly 20% contrast even at pH 6.5. Overall, the CD synthesis conditions turned out to be extremely sensitive tuning parameters for the pH profile of the contrast. Despite the differences, the CDs showed some common features too. First, the peak efficiency for all of the CDs remained around 45% ensuring a consistent enhancement in efficiency over the precursor 1. Second, the CEST MTR_assym plot for all the CDs showed a sharp rise around the offset 2.25 ppm with an asymmetric slow decaying tail toward higher offset. While the sharp primary CEST peak is coming from the protonated tertiary amine group, the asymmetry is arising most likely due to a small contribution from the amide proton at the offset 4.25 ppm. It is rather expected that the amide protons would not participate in CEST much as the bulky terminal amine group on the surface of the CDs would restrict the solvent access for the amide protons placed closer to the phenyl ring and hence to the core of the CD.

Figure 2.

One-pot hydrothermal synthesis of carbon quantum dots from lidocaine hydrochloride and their CEST profiles as a function of pH: Carbon dot synthesized by hydrothermal treatment of “c” mg/mL solution of the precursor at temperature “T” for “d” hours is denoted as ^TCD_d^c. The inset of each z-spectrum shows the corresponding MTR_assym (CEST efficiency). ²⁰⁰CD₂₄ (center) shows the maximum contrast at serum physiological pH 7.4 and temperature 37 °C (indicated over MTR_assym plots). Lowering (denoted by the down arrow) of values for any of the synthesis conditions from that set caused the physiological contrast to drop.

Characterization of ²⁰⁰CD₂₄⁵

After establishing that ²⁰⁰CD₂₄⁵ gives the best contrast at serum physiological condition, we went onto characterizing it and performing in-depth comparison with its precursor. To start with, post CD formation, the color of the solution changed to pale yellow from a colorless solution of 1. No precipitation or turbidity was observed over a long period of time. ²⁰⁰CD₂₄ showed blue fluorescence under UV light (Figure 3a). The absorbance and emission spectra further confirmed the fluorescence properties in the carbon dot solution (Figure 3b). As observed for most CDs, the emission maximum of ²⁰⁰CD₂₄⁵ also showed a distinct redshift with increasing excitation wavelength. Blue fluorescence is generally produced by relatively smaller size CDs.³⁶ TEM confirmed that the size distribution has two peaks (Figure 3c). One at 3.4 nm (σ = 0.55 nm), and the other at 5.2 nm (σ = 0.65 nm). These dot sizes are consistent with the typical blue emission. The crystal lines had a spacing of 0.20 nm as shown in the inset of Figure 3c. The ¹H NMR spectrum of ²⁰⁰CD₂₄ showed significant peak shifts in the aromatic region as compared to the aliphatic region (Figure 3d). This indicates that while the phenyl rings participate in the core formation (change of environment leading to change in chemical shifts), the environment of the side chain functional groups on the surface of the CDs, especially the tertiary amine group, remains intact. The structural changes were further confirmed by the ¹³C spectrum (Figure S14). X-ray photoelectron spectroscopy (XPS) measurements gave further insights toward the doping and surface groups. The wide-scan XPS spectrum of ²⁰⁰CD₂₄⁵ shown in Figure 3e clearly shows three peaks corresponding to C 1s, N 1s, and O 1s, respectively, around 285, 402, and 532.0 eV. A distinctive peak for N 1s confirms the nitrogen doping into the carbon dots. Deconvolution of the C 1s peak in Figure 3f shows peaks centered at 285.5, 287.1, and 288.5 eV, respectively, corresponding to C=C, C–O/C–N, and C=O groups. Figure 3g shows that the N 1s peak can be constructed by three components corresponding to pyrrolic N, graphitic/quaternary N, and amide N (N–C=O), respectively, at 400.6, 402.86, and 404.9 eV. A high resolution O 1s spectrum (Figure 3h) was fitted with three peaks. While the peaks at 531.7 and 532.8 eV correspond to, respectively, C–O and C=O, the small peak at 534.8 eV is attributed to O present in H₂O. The presence of protonated tertiary amine and carbonyl groups was further confirmed by peaks at 2360 and 1630 cm^–1 in the IR spectrum of ²⁰⁰CD₂₄ (Figure S15).

Figure 3.

Characterization of ²⁰⁰CD₂₄⁵: (a) solution under visible and UV illumination. (b) Absorption and emission spectra. (c) TEM image of the carbon dots. Top inset: crystal line spacing (0.2 nm). Bottom inset: particle size distribution with double Gaussian fitting. Mean particle size: μ₁ = 3.4 nm, σ₁ = 0.55 nm and μ₂ = 5.2 nm, σ₁ = 0.65 nm (d) comparison of ¹H ¹D-NMR spectra of 1 (red) with that of ²⁰⁰CD₂₄ (blue). NMR, absorption, and emission spectra of other CDs are given in Figures S2–S13. (e) XPS spectrum with the expansions of C 1s (f), N 1s (g), and O 1s (h) regions.

Exchange Constant Measurement and Comparison

In order to find a possible explanation behind the fact that hydrothermal treatment amplifies the CEST efficiency in all the reaction conditions that were tried, we decided to measure exchange constants (k_ex) of both the precursor (1) and the ²⁰⁰CD₂₄⁵. To start with, we plotted and compared the CEST efficiencies of both the compounds as a function of pH at two offsets: 2.25 ppm for the protonated tertiary amine group and 4.25 ppm for the amide proton. This was to find a common pH where both compounds give enough CEST at a particular frequency, so that k_ex can reliably be measured and compared for any one type of proton. As seen from Figure 4a, both compounds give maximum contrast at pH 5.5, but ²⁰⁰CD₂₄ gives it at 2.25 ppm offset (blue inverted triangle) and 1 gives at 4.25 ppm (yellow triangle) offset. Therefore, a k_ex measurement at pH 5.5 would compare two different functional groups. Fortunately, both compounds show enough CEST efficiency at 2.25 ppm (red square and blue inverted triangle) around pH 6–6.5, so that k_ex of the protonated tertiary amine group can be measured and compared reliably. We chose pH 6.5 for our k_ex study as it is closer to the neutral pH. Figure 4b shows the linear fitting of M_z/(M_z – M₀) against 1/ω² for 1 (red) and ²⁰⁰CD₂₄⁵ (blue). Clearly, the slope is much higher, and hence, the x-axis intercept for 1 (corresponding to a higher k_ex of 6080 rad s^–1) is much smaller than that of ²⁰⁰CD₂₄ (blue, k_ex of 2750 rad s^–1). We believe that the drop of k_ex to a more suitable value post hydrothermal treatment is one of the major factors leading to the enhancement of CEST efficiency. For a small offset of 2–4 ppm, a k_ex of 6080 rad s^–1 for 1 is rather high, leading to ineffective saturation transfer.

Figure 4.

Comparison of pH response of CEST efficiency and exchange coefficient for 1 and CDs: (a) CEST efficiency of amide (measured at 4.25 ppm offset) and tertiary amine (2.25 ppm) protons as a function of pH. (b) Linear regression of M_z/(M_z – M₀) against 1/ω² for calculating k_ex of 1 (red) and ²⁰⁰CD₂₄⁵ (blue). (c) Highest CEST efficiency shown by 1 and three other CDs irrespective of offsets.

CEST Comparison of CDs and Their Precursor

As the precursor and its CDs give maximum CEST at different offsets (Figure 4a) and also different CDs give slightly different profiles (Figure 2), comparison of CEST efficiency at any particular frequency either at 2.25 or at 4.25 ppm does not do justice to any of them. To capture a true comparison, we plotted the pH dependence of highest CEST efficiencies, irrespective of offsets, for various samples having the same concentration (5 mg/mL): 1, ¹⁸⁰CD₂₄⁵, ²⁰⁰CD₁₀, and ²⁰⁰CD₂₄⁵ (Figure 4c). Clearly, the pH response and efficiency of a compound can be very different post hydrothermal treatment. More importantly, the synthesis conditions can be used for further fine tuning of the pH profiles. Among all the CDs, ²⁰⁰CD₁₀ gives the highest CEST at 46.3% but only at a highly acidic pH of 4.5. The other two CDs sport a similar pH profile (peaking at pH 5.5) in comparison to 1, but only ²⁰⁰CD₂₄⁵ gives any meaningful contrast at the medically important neutral and physiological pH. Overall, increasing the temperature and duration enhanced the physiological contrast (¹⁸⁰CD₂₄ < ²⁰⁰CD₁₀⁵ < ²⁰⁰CD₂₄).

Conclusions

The key challenge of CEST CAs is their inherent insensitivity. For diaCEST agents, the problem is compounded by the low offsets of the exchangeable protons with respect to water as the protons having a slightly larger exchange constant cease to produce good contrast. A number of strategies have been tried and tested to overcome the sensitivity issue. Among them, use of equivalent exchangeable protons, use of inter- and intramolecular hydrogen bonding as a control over the exchange constant, and searching for natural high offset molecules have shown great success in recent times. Here, we present a new strategy in the form of converting a compound to carbon quantum dots (CDs) using hydrothermal treatment. We demonstrate that CDs show large amplification of CEST efficiency in comparison to its precursor at the same concentration. Moreover, different hydrothermal synthesis conditions such as precursor concentration, reaction temperature, and reaction time work as tuning parameters for the pH response of the resulting CDs. In the current study, we have first repurposed a well-known analgesic drug lidocaine hydrochloride as a diaCEST CA. While it showed very good CEST (∼26%) contrast predominantly in the acidic pH, the efficiency dropped drastically below 1% at the all-important physiological pH and temperature. We demonstrated that in the most optimized conditions, the carbon dots amplify the physiological efficiency several folds to 11% and the overall efficiency by a factor of nearly 2 to 46%. The one-pot synthesis was easy to perform, and the amplification was reproducible across batches. Also, the formation of CDs makes the CEST peak sharper as opposed to the wide profile produced by the precursor, facilitating a better exploitation of dose administered. In the quest of finding a plausible explanation for the impressive amplification in contrast efficiency shown by the CDs, we measured the exchange constants (k_ex) of both the precursor and the CDs. We found that the k_ex of the CDs is better suited for a small offset of the exchangeable protons leading to a better efficiency. However, the suitability of exchange constant is perhaps just one of many factors and other factors such as some inherent property of CDs cannot be ruled out. It becomes therefore important that further investigation of CDs as CEST CAs continues. Overall, through this study, we show that carbon dots as diaCEST agents have great potential that needs immediate exploration.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.2c02911.

• ¹³C NMR spectrum of the precursor and additional characterization spectra and plots (¹H/¹³C NMR, absorption spectrum, emission spectrum, and IR spectrum) for the carbon dots (PDF)

Author Contributions

S.P. conceptualized the project and prepared the carbon dots. S.P. and R.G. performed all the experiments and analyzed data using in-house MATLAB codes written by A.G. Overall supervision and manuscript drafting were done by A.G.

The authors declare no competing financial interest.