Zwitterionic near infrared fluorescent agents for noninvasive real-time transcutaneous assessment of kidney function† †Electronic supplementary information (ESI) available. See DOI: 10.1039/c6sc05059j Click here for additional data file.

Zwitterionic near infrared fluorescent agents were developed for non-invasive real-time transcutaneous assessment of kidney function.


Introduction
Evaluation of kidney function is crucial for a number of clinical situations. 1 However, current approaches for kidney function assessment are time-consuming and cumbersome, and therefore delay denitive diagnosis. 2 The glomerular ltration rate (GFR) is considered to be the best indicator for overall renal function. Plasma endogenous creatinine concentration is commonly used to estimate the GFR, but it may result in erroneous estimates due to age, gender, muscle mass and many other anthropometric variables. 3 Determination of the plasma/urinary clearance of exogenous renal markers (such as 99m Tc-DTPA, inulin and iothalamate) is invasive and cumbersome, due to the requirement of multiple blood/urine sampling steps and tedious sample analysis by HPLC. 4 Moreover, many studies revealed that creatinine and iothalamate are secreted by proximal tubule cells, while cystatin C and 99m Tc-DTPA are reabsorbed by the tubular epithelial cells, leading to a bias in the GFR. [1][2][3][4] Many attempts have been made to tackle these problems. Recently, uorescent GFR agents have gained much attention, [5][6][7] however, very few uorescent agents for GFR assessment have been reported. Although the uorescence intensity of FITC-inulin in plasma can be measured over a specic time period aer a bolus injection, the repeated blood sampling and the necessity to heat and dialyze the FITCinulin solution makes the method cumbersome and still invasive in manner. 5 Dorshow et al. reported that hydrophilic pyrazine-dicarboxylic acid derivatives can be excreted via the the kidneys, but the wavelength of pyrazine based uorescent agents is still in the green region of the spectrum. 6 Additionally, in our previous studies, we developed a non-invasive approach to assess kidney function in real-time based on a miniaturized electronic device attached to the skin and a GFR agent: FITC-sinistrin, 7 which has better water solubility than FITC-inulin and there is no need to dialyze before injection. By these means, the elimination kinetics of the uorescent agent FITC-sinistrin can be transcutaneously determined. The major advantage of this approach is the independence from blood/urine sampling and laboratory assays. This therefore allows renal function assessment in real-time and makes the evaluation of rapid changes in renal function possible, for example, in acute renal failure. Importantly, more precise results of the plasma clearance can be obtained via transcutaneous real-time measurement, which relies on a high number of data-points rather than a limited number of data-points from blood/urine sampling. 7 Nevertheless, both inulin and sinistrin suffer from inherent limitations including high cost, limited availability, tedious extraction and purication from plant roots. Also, the wavelength of the aforementioned uorescent GFR agents is still in the blue and green region of the spectrum, which further limits their application because of poor penetration and disturbance of the auto-uorescence signal from tissue. 8 One of the major obstacles encountered with non-invasive transcutaneous assessment of kidney function in vivo when using the aforementioned uorescent GFR agents with a short emission wavelength (<600 nm) is the strong intrinsic auto-uorescence background from living tissue, which signicantly compromises the accuracy of measurement under physiological conditions. 8 However, in the NIR region (650-900 nm), the absorption coefficient of tissue is greatly suppressed to a minimum level, thus it drastically reduces the noise from the auto-uorescence and increases tissue penetration. 8 Therefore, developing GFR agents in the NIR window is crucial for noninvasive transcutaneous assessment of kidney function. Zheng et al. focused on the development of renal clearable luminescent gold nanoparticles and used them to monitor the stages of kidney dysfunction, however, several terms are still unclear, such as plasma protein binding (PPB), urinary recovery, tubular reabsorption or secretion as well as potential toxicity. 9 Several other studies pointed out that conventional organic uorophores could persistently accumulate in the skin lipid membranes aer intravenous injection, due to their high lipophilicity. 9 What's worse is that the high lipophilicity leads to strong binding between uorophores and plasma protein in serum, for example, indocyanine green (ICG, Scheme 1a) exhibits extremely high PPB (99%), liver uptake and excretion, 10 and functions as a roadblock for kidney clearance. Many attempts have been made to overcome the above obstacles, for example, Hilger et al. have enhanced the hydrophilicity and decreased the PPB of dyes through increasing the number of sulfonate groups on asymmetric cyanine dyes. 11 However, having regions of negative charge makes the dyes prone to being secreted or reabsorbed by the kidney proximal tubules. 11 The fact that improving hydrophilicity and decreasing PPB comes at the expense of increasing tubular reabsorption or secretion prevents this method from gaining further application. Choi et al. reported a zwitterionic heptamethine cyanine dye, 12 and although it has a decreased PPB and can be eliminated by the kidneys, its PPB (19%) is still very high compared to those 'gold standard' renal function agents such as inulin and iothalamate (9.5%), 13 and this leads to a relatively long clearance half-life. Furthermore, its tubular reabsorption and excretion in the kidneys have not been studied. Therefore, to date, it is still an unmet challenge to develop an ideal GFR agent for non-invasive assessment of kidney function that simultaneously possesses the following features: (i) an absorption and emission wavelength in the NIR range for deeper penetration depth; (ii) high hydrophilicity and low PPB to accelerate renal clearance and avoid liver uptake; (iii) weakly charged characteristics to prevent interaction with organic transporter proteins Scheme 1 Structures of cyanine dyes and HPbCD based NIR agents. in kidney tubules; (iv) no toxicity, no metabolism and capability to be completely excreted into urine; (v) lterable via the glomerulus, no tubular reabsorption and secretion in the kidneys.
Herein, we report the rst NIR GFR agents with improved hydrophilicity and much lower PPB (<7%) through introducing zwitterionic charges and 2-hydroxylpropyl-b-cyclodextrin (HPbCD) into heptamethine cyanine dyes for assessment of kidney function in conscious rats. Relying on the attractive features of zwitterionic HPbCD based agents, which include high stability, non-toxicity and no metabolism in vivo, we demonstrate non-invasive, real-time monitoring of kidney function in an unprecedented efficient manner, with the zwitterionic HPbCD based agents being completely and rapidly excreted through the kidneys into the urine, exclusively ltered via the glomerulus, and with no reabsorption or secretion in the kidney proximal tubule. We further demonstrated that the zwitterionic HPbCD based agents can be used to assess kidney function in both healthy and nephropathic rats, which represents groundbreaking progress in the eld of kidney function assessment and preclinical diagnosis.

Design and synthesis
Based on the aforementioned criteria of an ideal GFR agent, a series of NIR agents were designed, which comprise two key functional components: HPbCD and NIR uorophores. Firstly, introduction of HPbCD is used to increase hydrophilicity, decrease PPB and accelerate excretion, due to HPbCD having better water solubility than b-cyclodextrin (bCD) and higher stability against hydrolysis by a-amylases from either porcine or human origin, 14 and it allows the encapsulation of the lipophilic uorophore into the cavity of HPbCD. The merit of its nontoxicity resulted in FDA approval ten years ago. 14 Additionally, its well-dened structure, low cost and excellent availability make it an ideal component for GFR agents in comparison with inulin and sinistrin. Secondly, NIR uorophores were employed for labeling HPbCD, thus, providing a deeper penetration depth and minimizing the disturbance of auto-uorescence from skin tissue during transcutaneous measurements. NIR uorescent agents with three different molecular surface charge characteristics (zwitterionic, anionic and cationic in Scheme 1b) were developed to systematically study the inuence of different molecular surface charges on their PPB and excretion. The detailed synthetic procedures are described in the ESI. † Three parent cyanine dyes (ZWCY, ANCY and CACY) have been modied with sulfonate groups and quaternary ammonium cations to impart negative charges and positive charges, respectively. Starting from phenylhydrazine or 4-hydrazinobenzenesulfonic acid, 12 carboxylic acid or azide-tagged NIR dyes were synthesized (Fig. S1 †) by reacting the meso chlorine atom of three parent cyanine cores with the appropriate nucleophiles. Finally, these tagged dyes were conjugated with HPbCD or propynyl-HPbCD to obtain the corresponding agents ( Fig. S2 and S4 †). The intermediates and products were characterized by 1 H-NMR, 13 C-NMR and LR-MS.

Physicochemical characteristics and optical properties
The absorption and uorescence spectra are depicted in Table  1, and Fig. 1, S5 and S6. † The parent dyes such as ZWCY, ANCY and CACY display an absorption peak at around 780 nm and an emission peak at roughly 810 nm with a small Stokes-shi of 30 nm. As expected, modication with an aromatic amine moiety at the chloro atom position signicantly increased the Stokes-shi up to 80 nm. The introduction of HPbCD moieties on those dyes does not cause any shis in the spectrum. Notably, their spectra effectively match with the conguration of the transcutaneous device, which consists of two light-emitting diodes with an excitation wavelength of 700 nm and a photodiode for emission wavelength detection at 790 nm ( Fig. S10 †). Moreover, the absence of reagent and free dye peaks in the optical spectrum and HPLC curves (Fig. 1, S5 and S9 †), as well as in the NMR spectra, proves the high purity of ABZWCY-HPbCD and AAZWCY-HPbCD.
The log D value (distribution coefficient) at pH 7.4 for each compound was calculated using the JChem plugin of ChemAxon (Table 1). It indicated that all of the compounds have high hydrophilicity, especially for the HPbCD based agents. The net charge of the compounds spans from À2 to +3, depending on the number of attached sulfonate and/or tetraethyl ammonium groups. The degree of labeling (DOL) is dened as the average number of dye molecules coupled to HPbCD. To preserve the properties of HPbCD, the DOL of ABZWCY-HPbCD and AAZWCY-HPbCD was controlled below 0.05 with small variations (AE0.004) between different batches of products (Table S3 †), suggesting good reproducibility of such labeling procedures. Molar extinction coefficients and uorescence quantum yields for each compound are shown in Table 1, except for the HPbCD based agents, as their extinction coefficients and quantum yields are dependent on their DOL.

Plasma protein binding (PPB)
The interaction between uorescent dyes and plasma proteins is particularly important as it affects the capability of the dyes as GRF agents in the following aspects: bio-distribution and clearance pathways as well as the rate of excretion in vivo. 15 For example, in our previous study, we found that the commercial anionic dye IRDye800CW has a high PPB up to 41%. 16 It is also reported that IRDye800CW exhibits skin accumulation aer intravenous injection and cannot be excreted rapidly by the kidneys. 9b We rst evaluated the PPB values for free dyes, anionic dyes and cationic dyes, which exhibit very high PPB values, as shown in Table 1, for example, ABANCY and AAANCY show 95% and 80% PPB, respectively. ABCACY and AACACY display 54.5% and 60% PPB, respectively. Meanwhile, the zwitterionic dyes ABZWCY and AAZWCY have much lower PPB (19% and 15.7%, respectively) than all other free dyes, which indicated that zwitterionic dyes possess much lower binding abilities towards plasma proteins than anionic and cationic dyes. Similar PPB values were also observed in the mixture of free dye (ABZWCY and AAZWCY) and HPbCD, with values of 17.7% and 13.7%, respectively. Attaching HPbCD to free dyes could further decrease the PPB, for example, anionic HPbCD based agents (ABANCY-HPbCD and AAANCY-HPbCD) and cationic HPbCD based agents (ABCACY-HPbCD and AACACY-HPbCD) exhibit PPB values ranging from 20% to 30%.
Surprisingly, the zwitterionic agents ABZWCY-HPbCD and AAZWCY-HPbCD have much lower PPB (3.7% and 6.5%, respectively), even lower than some 'gold standard' agents.  Dynamic light scattering experiments further demonstrated that no aggregation occurred on either zwitterionic agents or their mixture with rat plasma protein (Fig. S8 †). The above results demonstrate that HPbCD modication could signicantly decrease the PPB for all three types of charged free dyes. Additionally, PPB is not only dependent on the dye hydrophilicity, but is also related to the molecular charge number and distribution. Since albumin is alkalotic, acidic agents will preferentially bind to albumin, then to lipoprotein aer albumin becomes saturated. Basic agents will bind to the acidic alpha-1 acid glycoprotein. 17 Thus, anionic and cationic compounds are more inclined to bind to proteins than zwitterionic compounds. Therefore, we will mainly focus on zwitterionic compounds in the following parts.

Stability studies in esterase and cell viability evaluated by MTT assay
In order to evaluate the stability of conjugation bonds between HPbCD and NIR uorophores, we co-incubated ABZWCY-HPbCD and AAZWCY-HPbCD with porcine liver esterase (PLE) for 24 h, respectively, and then tracked their stability with HPLC ( Fig. 1d and e). A peak for ABZWCY-HPbCD and AAZWCY-HPbCD in both PBS and PLE appeared at 5.8 min and 6.0 min, respectively, while a unimodal peak for free dye ABZWCY and AAZWCY in PBS appeared at 8.0 min and 9.0 min, respectively. No degradation was observed for both NIR agents that were coincubated with PLE, as indicated by there being no appearance of a free dye peak in the HPLC chromatographs. This supports the observation that both the triazole unit of AAZWCY-HPbCD from a classic click reaction and the ester bond of ABZWCY-HPbCD are highly stable in esterase conditions. The cytotoxicity of ABZWCY-HPbCD and AAZWCY-HPbCD was evaluated using a 3-(4,5-dimethyl-2-thiazoly)-2,5-diphenyltetrazolium bromide (MTT) assay, with the results (Fig. 1f  and g) showing that HK-2 human proximal tubular cells maintained high viability aer being incubated with varying concentrations of these two agents, suggesting no cytotoxic effect from ABZWCY-HPbCD and AAZWCY-HPbCD.

Transcutaneous measurement of kidney function in a healthy rat model
Having demonstrated low PPB and no cytotoxicity of the zwitterionic ABZWCY-HPbCD and AAZWCY-HPbCD NIR agents, we proceeded to study whether these zwitterionic compounds can be excreted by the kidneys in healthy rats (8-10 weeks), by conducting non-invasive transcutaneous measurements. The principles and methods of the transcutaneous technique are described in Fig. S10. † Firstly, as a control, IRDye800CW was chosen to be injected intravenously into rats and its elimination curves were measured transcutaneously. However, its clearance curves show no decay even at 90 min post-injection (Fig. 2f), consistent with previous ndings. 9b A long retention time of IRDye800CW in vivo was attributed to its higher PPB (41%) and negative charge feature, leading to skin accumulation and slow renal excretion. In contrast, the anionic HPbCD based agent ABANCY-HPbCD exhibited an obvious decay in the clearance curves ( Fig. 2a) and had a half-life of 55.2 AE 5.86 min (Table 2), which is attributed to its lower PPB (19.7%) than IRDye800CW. To assess whether the anionic agent ABANCY-HPbCD is reabsorbed or secreted by organic anion transporter (OAT) proteins in the kidney tubules, probenecid, an inhibitor of OAT proteins, 18 was administered (30 min prior to injection of ABANCY-HPbCD) to block tubular reabsorption and the secretion pathway, then, elimination curves were measured for the same rats. However, ABANCY-HPbCD has severe tubular secretion, as evidenced by slower clearance curves and a longer clearance half-life (110.5 AE 5.4 min, Table 2 and Fig. 2b and g). We also measured the cationic HPbCD based agent in the same manner and the results are similar to those of anionic HPbCD based agents. Next, the free dye ABZWCY was injected intravenously and its elimination curves were also recorded transcutaneously. As shown in Fig. 2c, its uorescence kinetics curves declined, but did not return to background levels at 2 h post-injection, suggesting it was not excreted completely during this period. The half-life of the free dye ABZWCY (52.6 AE 8.8 min) is similar to that of ABANCY-HPbCD. This is in agreement with their similar PPB values ($20%). To determine whether ABZWCY has tubular reabsorption or secretion in addition to ltration, probenecid was administered to block tubular reabsorption and the secretion pathway. It was found that the half-life of ABZWCY (48.7 AE 5.4 min, Table 2 and Fig. 2g) only slightly changes in the presence of probenecid, demonstrating that tubular reabsorption or secretion is a minor factor in half-life alterations. This result also indicated that both anionic and cationic compounds are more likely to be secreted by the kidney tubules than zwitterionic compounds are. In addition, a mixture of ABZWCY and HPbCD was studied, with the clearance curves (Fig. 2e) and half-life (54.2 AE 7.4 min) being similar to those of ABZWCY only, indicating that free HPbCD has no effect on the excretion of ABZWCY. Surprisingly, compared to anionic agents and free zwitterionic dye, a much faster clearance decay and a much shorter half-life were observed for ABZWCY-HPbCD and AAZWCY-HPbCD either with or without probenecid treatment ( Table 2 and Fig. 3).
Those results indicated that zwitterionic HPbCD based agents have no skin accumulation. Their shorter clearance half-life ($30 min) was attributed to their evenly distributed charges, higher hydrophilicity and lower PPB (<7%). Importantly, a negligible difference in the clearance half-life (Fig. 3f) with and without probenecid treatment was observed with both zwitterionic HPbCD based agents, indicating that they have no severe tubular reabsorption or secretion in the kidneys. We also found that the half-life of the mixture of ABZWCY and HPbCD is much longer than that of ABZWCY-HPbCD and AAZWCY-HPbCD, therefore, ABZWCY and its HPbCD mixture are not superior to ABZWCY-HPbCD and AAZWCY-HPbCD as suitable candidates for kidney function assessment. Considering that ABZWCY-HPbCD and AAZWCY-HPbCD possess a positively charged quaternary ammonium group, it is necessary to determine whether these zwitterionic HPbCD based agents can be reabsorbed or secreted by organic cation transporter (OCT) proteins in kidney tubules. 18 AAZWCY-HPbCD was injected into healthy rats and elimination curves and clearance half-life values were measured in the presence of an OCT inhibitor, cimetidine. The results show that the OCT inhibitor has no effect on its clearance half-life (30.9 AE 4.1 min, Fig. 3e and f). Based on all of the above results, we conclude that zwitterionic HPbCD based agents can be ltrated via the glomerulus efficiently, and exhibit slight differences in the clearance half-life by both tubule OAT and OCT proteins in kidneys.

Urinary recovery
An ideal kidney function agent should have no metabolism in vivo and be recovered completely in urine. With this in mind, we investigated the recoveries of injected doses using in vivo experiments with metabolic cages. High urinary recoveries were determined for the free dye ABZWCY and its mixture with HPbCD (99.4 AE 4.8% and 96.5 AE 5.8%, respectively, Fig. 4 and Tables 2 and S6 †). Similar tendencies of urinary recoveries of ABZWCY-HPbCD and AAZWCY-HPbCD with values of 97 AE 3.9% and 103.3 AE 4.2%, respectively, were also observed (Fig. 4, and Tables 2 and S7 †). Indeed, urinary recoveries of the given dose were almost completed at 9 h post-injection for ABZWCY-HPbCD and AAZWCY-HPbCD, which is faster than the free dye ABZWCY and in agreement with their shorter clearance halflife. The results demonstrated that all of the injected zwitterionic agents are excreted rapidly into urine. However, ABANCY-HPbCD has a relatively high PPB and long clearance half-life and exhibits severe kidney tubular secretion, so it is not considered for further urinary recovery studies.

Biodistribution
To conrm the distribution of the zwitterionic HPbCD based agents aer intravenous injection in healthy rats, we investigated the uorescence distribution of ABZWCY-HPbCD and AAZWCY-HPbCD using small animal imaging. As expected, the uorescence images of the organs obtained from the control rat displayed almost no uorescence signal ( Fig. 5a and b). Notably, uorescence signals can be mainly observed in the kidneys and bladder from the rats that were administered with these two zwitterionic HPbCD based agents, with only slight signals in the intestines and no signicant nonspecic uptake in the other organs and tissues ( Fig. 5c-g). These experiments further conrm that these two zwitterionic HPbCD based agents were excreted through the kidneys to the urine.

Metabolism studies
To determine whether these agents can be metabolized in vivo, urine samples were collected and investigated by HPLC. The results (Fig. S12 †) indicated that no metabolites were found from the HPLC curves. To better understand these results, we performed an additional experiment using MALDI-TOF. The obtained MALDI data (Fig. S12 †) show that the mass distribution of ABZWCY-HPbCD and AAZWCY-HPbCD recovered from urine samples is the same as that before injection. These results further conrmed that these two zwitterionic HPbCD based agents have no metabolism in vivo.

Transcutaneous measurement of kidney function in a nephropathic rat model
Finally, transcutaneous assessments of kidney function were conducted in a transgenic rat (TGR) model (24 weeks) and agematched wild-type rats. A TGR model with overexpression of the human Ang II type 1 receptor (hAT1R) in podocytes was used. 19 In this TGR based nephropathic model, the damage progressed to nephron loss via focal segmental glomerulosclerosis, leading to the degeneration of both the glomerulus and tubules, which is associated with a reduction in the glomerular ltration rate, tubular necrosis and protein leakage in urine. 19 The loss of  kidney function in the nephropathic model was testied by the urinary parameters including protein and albumin excretion in Table 3 and Fig. 6a and b, consistent with previous ndings. 19a The clearance half-life of ABZWCY-HPbCD in AT1R transgenic rats (42.88 AE 3.97 min, Table 3 and Fig. 6e) is much longer than that observed in the wild-type rats (32.98 AE 4.35 min, Table 3 and Fig. 6d), which is consistent with the fact that urinary protein and albumin excretion of AT1R transgenic rats are much higher than those in the wild-type rats. Additionally, we found that the clearance half-life of wild-type rats (32.98 AE 4.35 min) is slightly longer than that of the afore-measured rats (30.1 AE 2.7 min), which is attributed to older rats and an agerelated decline in kidney function. These results show the potential of NIR zwitterionic agents as exogenous markers for evaluating kidney function in a kidney disease model.

Conclusions
In summary, zwitterionic NIR GFR agents ABZWCY-HPbCD and AAZWCY-HPbCD were rationally designed and synthesized. They show favourable uorescence properties, deeper penetration depths, high hydrophilicity and extremely low PPB.
Furthermore, they also exhibit high stability in esterase and non-toxicity. By taking advantage of the above properties, we have demonstrated that these zwitterionic NIR agents outperform the existing commercial NIR dye (IRDye800CW) in the context of skin accumulation and PPB. More importantly, these two zwitterionic HPbCD based agents can be excreted efficiently through the kidneys into urine without severe reabsorption and secretion in the kidney tubules. Urinary recovery and uorescence distribution investigations by small animal imaging experiments further demonstrated that they could be completely and rapidly excreted through the kidneys without in vivo metabolism. Studies in rat models of both healthy models and those with kidney disease showed that zwitterionic HPbCD based agents are promising markers for evaluating kidney function. To the best of our knowledge, this is the rst report on assessing kidney function based on near infrared GFR agents and the rst example of using zwitterionic charge characteristics for developing GFR agents. Relying on these zwitterionic NIR agents and transcutaneous uorescence detection techniques, we demonstrate a rapid, robust and convenient approach for non-invasive real-time assessment of kidney function, without the need for time-consuming blood/urine  6 Urinary albumin excretion (a, ***P < 0.001), protein excretion (b, ***P < 0.001) and clearance half-life of ABZWCY-HPbCD (c, **P < 0.01) in wild-type rats and AT1R transgenic rats. Elimination curves of ABZWCY-HPbCD by transcutaneous measurements in wild-type rats (d) and AT1R transgenic rats (e).
sample preparation. We believe that this work represents a signicant progress towards highly efficient kidney function assessment, and holds great promise in kidney disease diagnosis in the future. Formal preclinical development studies are in progress.