Renal Clearance of Nanoparticles

Although targeted nanoparticles hold promise for the detection and treatment of human disease, toxicity - either potential or real - remains the major roadblock to clinical translation.¹ Historically, the U.S. Food and Drug Administration (FDA) has required that agents injected into the human body, especially diagnostic agents, be cleared completely, in a reasonable amount of time. This policy makes sense in that total body clearance minimizes the area under the exposure curve. It also minimizes the chance that the agent will interfere with other diagnostic tests. For example, gold, used extensively in the nanotechnology literature, has a linear attenuation coefficient 150-fold higher than even bone, and at doses injected intravenously would likely preclude accurate computed tomographic (CT) scanning, especially in organs such as the liver, where it eventually accumulates.

Against this backdrop is the inherent stability of most nanoparticles. Indeed, a recent study suggests that quantum dots (QDs) with the appropriate organic coating are retained in the body for at least two years and remain fluorescent.² When considering that many nanometer-sized objects proposed for clinical use contain heavy metals, regulatory approval of such stable particles is unlikely, and the type of long-term toxicity studies that would be required for such approval will continue to discourage clinical translation.

A potential solution to this conundrum is to focus on the physiology underlying biodistribution and clearance of agents injected intravenously into the body. For globular proteins, a hydrodynamic diameter (HD) of approximately 5–6 nm is associated with the ability to be cleared rapidly from the body via renal filtration and urinary excretion (Table 1). Nanoparticle toxicity would be minimized, if not eliminated, if there were a way to clear them from the body. However, currently it is unknown what the renal filtration threshold is for metal-based nanometer-sized objects, and which organic coatings are compatible with renal clearance. In this study, we employ fluorescent QDs as a model system to define the HD and surface charge combination that permits rapid body elimination of nanometer-sized objects.

In order to define the renal filtration threshold, a precise size-series of extremely small QDs was synthesized. All QDs had a CdSe/ZnS core/shell, which was coated with anionic (dihydrolipoic acid; DHLA), cationic (cysteamine), zwitterionic (cysteine), or neutral (DHLA-connected polyethylene glycol; DHLA-PEG) small molecules (Fig. 1a). The charge of the coating had a profound effect on the adsorption of serum proteins as well as the HD; purely anionic or cationic charge was associated with an increase in HD of over 15 nm after incubation with serum (Fig. 1b). Although neutral (PEGylated) QDs did not bind serum protein, it was not possible to synthesize them with an HD less than ≈ 13.2 nm. Shorter PEG lengths resulted in insoluble QDs.³ Surprisingly, zwitterionic coatings, in this case the amino acid cysteine (Cys), prevented serum protein adsorption while producing the highest solubility and the smallest possible HD.

Cys-coated QDs (QD-Cys) were characterized by three independent sizing methods (Fig. 1c). Transmission electronic microscopy (TEM) was used to measure core/shell diameter, which as expected, increased with peak fluorescence emission (Fig. 1d). Dynamic light scattering (DLS) and gel-filtration chromatography (GFC) were highly correlated with respect to HD after addition of the organic coating (Fig. 1c and Supplementary Fig. 1 online). Using these methods, it was possible to create a series of QD-Cys, ranging in final HD from 4.36–8.65 nm and differing only in peak emission wavelength (515–574 nm; i.e., QD515–QD574; Fig. 1e and Supplementary Fig. 2 online), whose HD and stability were unchanged after 4 hr incubation with 37°C serum. Electron paramagnetic resonance (EPR) spectroscopy on QD-Cys samples incubated with spin trapping 5,5-dimethyl-1-pyrroline-N-oxide (DMPO) confirmed that radicals were not produced during photon absorption (see Supplementary Fig. 3 online).

Initial studies focused on the smallest fluorescent nanoparticle, QD515. After dilution into sterile saline and intravenous injection into rat at a dose of 10 pmol/g animal weight, the renal excretion of Cys-coated QD515 could be visualized directly, as could its transport down the ureters bilaterally and into the bladder (Fig. 2a and Supplementary Video). Comparison of all QD HDs in mice revealed that at 4-hr post-injection, only QD515 (HD = 4.36 nm), QD534 (HD = 4.99 nm), and QD554 (HD = 5.52 nm) could be found excreted into the bladder (Fig. 2b), with intensity decreasing as HD increased. The larger QDs including DHLA-, cysteamine-, and DHLA-PEG-coated ones, were never found in the bladder, but instead were trapped in the liver, lung, and spleen in large amounts (data not shown).

Since in vivo fluorescence imaging provides only qualitative/semi-quantitative measurement of QD behavior in vivo, we developed a technique to label the QD-Cys surface covalently with a chelated form of the gamma ray-emitting isotope ^99mTc. Despite non-covalent interaction of Cys with the QD surface, no aggregates were detectable by HPLC analysis, and the amount of unbounded isotope was less than 10% in each preparation. Intravenous injection of ^99mTc-labeled QDs revealed that changes in final HD resulted in dramatic changes in blood half-life (Fig. 3a). Indeed, based on the β-phase terminal half-life (t_1/2β), the blood half-life of QDs ranged from 48 minutes to 20 hr as the HD increased from 4.36 to 8.65 nm (Fig. 3b).

Using ^99mTc-labeled QDs, it was possible to follow distribution and clearance in all organs, and from the body itself, over time. Importantly, radiolabeled QDs remained intact, even after excretion into urine (see Supplementary Fig. 4 online), suggesting that gamma ray emission could be used as a reliable surrogate for QD distribution. Four-hours post-intravenous injection of QD515 (HD = 4.36 nm), the dominant signal was in the bladder, with the only appreciable organ distribution being liver (4.5 ±1.0 %ID) and kidney (2.6 ±0.4 %ID), with the latter likely representing QDs in the process of being excreted (Fig. 3c). In contrast, QD574 (HD = 8.65 nm) exhibited high uptake in liver (26.5 ±3.9%ID), lung (9.1 ± 4.0%ID), and spleen (6.3 ±2.4%ID) and a proportionally lower signal in bladder (Fig. 3d).

By measuring the radioactivity of both excreted and pre-excreted urine, as well as the entire remaining carcass excluding bladder and urethra, it was possible to define the relationship between HD, renal clearance, and total body retention. As shown in Fig. 3e, this relationship was sigmoidal, with the 50% point for total body clearance of QDs at 4 hr being an HD ≈ 5.5 nm.

Inorganic, metal-containing nanoparticles require a solubilizing organic coating for biological (i.e., aqueous) compatibility. In the case of QDs, this coating is also a requirement for maintaining photoluminescence efficiency. Unfortunately, in order to achieve biocompatibility, organic coatings often result in a significant increase of the final HD. Our results suggest that both the molecular weight and charge of the solubilizing ligand contribute to final HD in vivo, and that pure charge (anionic or cationic) is associated with unexpected serum protein adsorption. This adsorption did not appear to affect solubility, but increased HD by almost 15 nm.

Nanoparticle HD is a critical design parameter in the development of potential diagnostic and therapeutic agents. The mammalian vasculature has an average pore size of ≈ 5 nm. Below this value, there is relatively fast equilibrium between agents injected intravenously and the extracellular space, but above this value, transport across the endothelium is remarkably slow. For example, human IgG (HD = 10 nm) requires approximately 24 hr to equilibrate between the vascular and extracellular spaces after intravenous injection.^4,5 To a first approximation, glomerular filtration in the kidney is controlled by similar effective pore sizes, and renal excretion is a strong function of HD (Table 1). For nanoparticles that do not biodegrade in vivo into biologically benign components, the only other major route of excretion from the human body is through the liver, into bile, and into feces. The problems with liver excretion for nanoparticles are three-fold. First, the liver is designed specifically to capture and eliminate nanoparticles (e.g., viruses) over ≈ 10–20 nm in HD. Hence, special coatings such as PEG are required just to prevent first-pass extraction from the reticuloendothelial system (RES; liver, spleen, and bone marrow), but these coatings necessarily increase HD. Dense PEGylation might increase blood half-life, but it also precludes elimination from the body. Second, excretion of intact nanoparticles into bile is an extremely slow and inefficient process. Third, long-term retention in the RES leads to a large area under the exposure-time curve and increases the likelihood of toxicity. At the very least, our data suggest that analysis of bodily fluids, including urine and bile, should be part of human risk assessment after environmental exposure to nanoparticles, and can help estimate total retained dose if the exposure dose is known.

The importance of total body clearance of nanometer-sized objects is not a trivial one. Consider, for example, what would occur if retention affected other medical tests. Radiological tests are particularly susceptible. Metals with high atomic numbers deposited in organs would interfere with x-ray imaging (i.e., plain films, fluoroscopy, and CT) due to changes in linear attenuation coefficient, magnetic resonance imaging due to proton-free voids, ultrasound due to increased echogenicity, and possibly even SPECT and PET due to photon attenuation (both) and/or effects on positronium production (PET). Renally-cleared nanoparticles, even metal-containing ones, would minimize or eliminate these problems.

Several caveats, though, must be considered. First, the absolute value for the renal filtration threshold (HD = 5.5 nm in this study) depends on the calibration curves of the sizing techniques. Since the reported HDs of the molecules used for calibration vary in the literature by as much as 10%, so too must our confidence in the 5.5 nm cutoff. For this reason, it will be important to test each nanometer-sized object independently in vivo, and not rely solely on in vitro predictions. Second, our study has not addressed the effect of nanoparticle shape on renal filtration. For example, quantum rods⁶ and other non-spherical nanoparticles will need separate in vivo analyses. If eventual results correlate with those from long rigid proteins, it is likely that the long axis of the particle will dominate its in vivo behavior. Finally, our study suggests that molecularly targeting QDs and other nanoparticles requires careful attention to HD. Indeed, our results suggest that only extremely small targeting molecules will be of use for in vivo targeting if renal excretion is to be maintained. Fortunately, the increased affinity associated with multimerization of such ligands on the QD surface⁷ may permit a small number of targeting ligands to be used relative to coating ligands.

Thus far, few nanometer-sized objects are being actively translated to the clinic. Our study suggests that in order to satisfy both patient safety and regulatory review, nanoparticle biodistribution and clearance must be carefully considered. We propose three criteria for distinguishing a nanoparticle that has potential clinical utility: 1) a final HD ≤ 5.5 nm to permit complete elimination from the body and/or 2) a formulation with completely non-toxic components and/or 3) biodegradability to clearable components. We suggest that without satisfying these criteria, metal containing nanoparticles will have limited clinical utility.