• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. Dec 12, 2006; 103(50): 18882–18886.
Published online Nov 29, 2006. doi:  10.1073/pnas.0609265103
PMCID: PMC1665645
Applied Physical Sciences, Medical Sciences

Mammalian pharmacokinetics of carbon nanotubes using intrinsic near-infrared fluorescence

Abstract

Individualized, chemically pristine single-walled carbon nanotubes have been intravenously administered to rabbits and monitored through their characteristic near-infrared fluorescence. Spectra indicated that blood proteins displaced the nanotube coating of synthetic surfactant molecules within seconds. The nanotube concentration in the blood serum decreased exponentially with a half-life of 1.0 ± 0.1 h. No adverse effects from low-level nanotube exposure could be detected from behavior or pathological examination. At 24 h after i.v. administration, significant concentrations of nanotubes were found only in the liver. These results demonstrate that debundled single-walled carbon nanotubes are high-contrast near-infrared fluorophores that can be sensitively and selectively tracked in mammalian tissues using optical methods. In addition, the absence of acute toxicity and promising circulation persistence suggest the potential of carbon nanotubes in future pharmaceutical applications.

Keywords: nanoparticle biodistribution, nanoparticle toxicity, luminescence spectroscopy, single-walled carbon nanotubes

Single-walled carbon nanotubes (SWNTs) are an important class of artificial nanomaterials with remarkable mechanical, thermal, electronic, and optical properties. These properties suggest diverse future biomedical uses in areas such as targeted chemotherapeutics, in vitro cell markers, diagnostic imaging contrast agents, biochemical sensors, and photoablative therapy agents (19). Before medical applications can be developed, it is necessary to explore the behavior and fate of SWNTs in mammals. However, little is currently known in this area, in part because of the challenge of detecting and tracking these all-carbon nanoparticles in complex biological environments.

SWNTs can be envisioned as sections of graphene sheets rolled up to form seamless cylindrical tubes with a variety of structures (10). Each of these structures has a well defined diameter and chiral angle and shows either semiconducting or metallic behavior. The nanotube preparations used for this study contain several dozen structural types that are ≈1 nm in diameter and ≈300 nm long. After excitation with visible light, each type of semiconducting SWNT fluoresces at a near-infrared (near-IR) wavelength between ≈900 and 1,600 nm that is characteristic of its specific structure (11, 12). We have previously exploited this fluorescence emission to study the active ingestion of SWNTs by macrophage cells in vitro (13).

Here we report the use of the intrinsic near-IR fluorescence, which is a property only of individualized SWNTs, to measure their blood elimination kinetics in rabbits and to identify the organs in which they concentrate. These methods and results provide a foundation for developing the targeted delivery of nanotubes to specific tissues for diagnostic and therapeutic uses. In contrast to alternative methods that track carbon nanotubes by linking them covalently or noncovalently to external fluorophores or chelated radioisotopes (1, 8, 14), the near-IR fluorescence technique used here provides a simple, unambiguous way to monitor chemically pristine SWNTs with sensitivity high enough to detect even a single intracellular nanotube.

Results and Discussion

Because as-produced carbon nanotubes are strongly aggregated and highly hydrophobic, they required processing to form stable aqueous suspensions suitable for i.v. administration. Raw nanotubes were ultrasonically dispersed in a solution of the artificial surfactant Pluronic F108, and then ultracentrifuged. F108 was selected because of its biocompatibility, its extension of blood circulation times of surface modified nanoparticles, and its use as a SWNT suspending agent in our prior in vitro study (13, 15, 16). Proper dispersion of the resulting sample was confirmed by its strong near-IR fluorescence, which arises from individual but not from aggregated SWNTs. Four normal New Zealand rabbits were injected with a 7.5-ml bolus of this SWNT suspension through implanted jugular vein catheters. The dose of SWNTs was 75 μg, corresponding to ≈20 μg per kg of body mass. Control animals were instead injected with the same volume of 1% aqueous Pluronic solution. At times between 0.5 and 24 h after injection, we collected 1-ml blood specimens from each animal. Blood sera from these specimens were analyzed by near-IR fluorescence spectroscopy after a 48-h sedimentation period. Rabbits were killed 24 h after SWNT injection, and their organs and tissue samples were harvested for histopathology and near-IR fluorescence imaging.

Fig. 1 compares normalized emission spectra of the Pluronic SWNT suspension before injection, a blood serum sample collected 30 min after injection, and a SWNT suspension prepared directly in rabbit serum without any exposure to Pluronic. All three traces clearly show the characteristic spectral signature of disaggregated semiconducting SWNTs. However, spectral features in the blood sample are significantly broadened and red-shifted relative to those in Pluronic. It is now well known that the positions and widths of SWNT fluorescence features vary with the nature of the surrounding medium (17, 18). We attribute the observed spectral changes, which are very similar to changes found for SWNTs inside macrophage cells (13), to the displacement of the nanotubes' Pluronic coating by blood proteins. The close match between the spectral traces in Fig. 1 for the blood-circulated SWNTs and SWNTs directly dispersed in rabbit blood serum supports this interpretation. Although a prior biosensor development study reported no displacement of polyethylene oxide SWNT coatings by proteins in vitro, the concentration of proteins used in that study was lower than in blood serum by a factor of ≈104 (19). This concentration factor accounts for the facile coating displacement by proteins in vivo.

Fig. 1.
Normalized emission spectra (using 658 nm excitation) of samples of SWNTs prepared as suspensions in aqueous Pluronic (thin blue curve), in rabbit serum (dotted curve), and in blood serum sampled 30 min after i.v. injection (thick red curve).

In separate in vitro experiments at room temperature, we monitored the displacement kinetics of a Pluronic surface coating by serum proteins through time-dependent fluorescence spectrometry. When serum proteins were added to a Pluronic suspension of SWNTs, nearly complete spectral shifts occurred within our experimental time resolution of 0.7 s. Subsequent, more subtle changes in peak positions and intensity were observed on a longer time scale, as illustrated by the kinetic data of Fig. 2. These data can be adequately fit by a biexponential decay plus asymptotic baseline model in which the two kinetic components are of comparable magnitude and have half-lives of ≈20 and 310 s. We suggest that these slower processes may reflect structural relaxation of the protein coating on the nanotube surface. In circulating blood at physiological temperature, one would expect the Pluronic displacement processes to be accelerated, so we conclude that the injected nanotubes' initial environment is replaced by a relatively stable coating of blood proteins within a period of seconds. Our pharmacokinetics results therefore reveal the fate of SWNTs surrounded by endogenous proteins rather than a synthetic surfactant. These kinetic findings on surface displacement may also be significant in understanding the properties of similar block copolymer amphiphiles used as coating agents for i.v. drug delivery (20).

Fig. 2.
Measured fluorescence intensity (symbols) for a sample of Pluronic-suspended SWNTs mixed with rabbit serum, as a function of time from mixing. The solid curve shows a best fit to a biexponential kinetic model.

To quantify the kinetics of SWNT elimination from blood circulation, we measured spectrally integrated SWNT fluorescence intensities from specimens representing different time points. Fig. 3 shows a plot of serum SWNT content vs. circulation time, averaged over measurements from four rabbits. The concentration scale was calibrated by assigning the initial signal strength to the injected mass of SWNTs divided by the estimated blood volume (120 ml). As illustrated by the solid curve, the data are well modeled by first-order (exponential) decay with a half-life of 1.0 ± 0.1 h. The absence of biexponential or multiexponential kinetic components indicates that there was no significant temporary accumulation of nanotubes in tissues that could act as reversible reservoirs.

Fig. 3.
Time dependence of blood serum SWNT concentration after injection, as measured for four rabbits. Each data point is the averaged emission intensity spectrally integrated above a linear baseline connecting the minima at 1,100 and 1,250 nm. This baseline ...

We examined the biodistribution of SWNTs among organ systems after elimination from the systemic circulation by performing near-IR fluorescence microscopy on tissue specimens. Fig. 4 shows representative micrographs of liver tissue taken from rabbits killed 24 h after SWNT injection. In Fig. 4, the near-IR fluorescence images were false-colored in green and overlaid on corresponding bright-field micrographs of conventionally stained tissues. Green-coded objects were identified as SWNTs through two optical criteria (21). One criterion involved emission spectra measured from selected objects within the microscope field. SWNT emission was recognized either from single sharp near-IR peaks at wavelengths characteristic of individual nanotube structures, or from superpositions of such peaks when the emitting object contained multiple nanotubes (12, 21). The other criterion involved the strong polarization of optical transitions along the nanotube axis. If an object was a single SWNT, then rotating the polarization axis of the excitation beam caused characteristic modulation of its emission intensity (21). Polarization and/or spectral analysis confirmed that all of the green-coded emission shown in Fig. 3 arose from SWNTs.

Fig. 4.
Micrographs at two magnifications of liver tissue from rabbits killed 24 h after i.v. administration of suspended SWNTs. (A and B) Near-IR SWNT fluorescence images with field widths of 390 μm (A) and 83 μm (B). Scattered isolated bright ...

Fig. 4A reveals numerous regions in the liver specimen with significant SWNT concentrations. The more magnified image of Fig. 4B shows one or two green clusters in addition to ≈30 diffraction-limited green spots. Each of these spots is emission from a single semiconducting SWNT. Similar analysis of tissue specimens from the kidneys, lungs, spleen, heart, brain, spinal cord, bone, muscle, pancreas, intestine, and skin revealed far fewer or no nanotubes, and the control specimens showed no emissive features identifiable as SWNTs. We conclude that, at 24 h after i.v. injection, the only significant SWNT concentration is in the liver.

During the 24-h period between exposure and death, the experimental animals displayed normal behavior and no evidence of adverse effects from i.v. SWNT administration at the ≈20 μg/kg dosage used here. In addition, pathological examination during necropsy revealed no gross organ abnormalities, and histological evaluation of tissue sections found no pathological differences between the experimental and control animals. Therefore, we deduce an absence of acute toxicity for the i.v. SWNT dosage used here.

Our results may be compared with those from a recent study by Singh et al. (14) of chemically functionalized carbon nanotubes that were covalently linked to a chelated radioactive tracer (111In). Unlike the findings described here for pristine nanotubes, it was reported that i.v. administration of the functionalized SWNTs in mice led to efficient uptake and clearance by the kidneys, with much less accumulation in the liver and other organs. The prior study deduced blood elimination kinetics from concentrations measured at just three time points. Those data were not consistent with first-order decay, and one time point was apparently discounted to obtain a half-life estimate of ≈3 h from the other two (14). We are confident that our fluorescence-based method provides significantly more reliable elimination kinetics. When improved data become available for derivatized nanotubes, it should be possible to assess the influence of chemical functionalization on SWNT pharmacokinetics. We note that intrinsic near-IR fluorescence allows unambiguous observation of unmodified nanotubes. By contrast, methods that use linked fluorophores or radioactive labels require covalent derivatization and rely on in vivo stability of the linking structure for valid tracking.

In view of the spectroscopic evidence that the nanotubes' initial Pluronic coating was very quickly displaced during circulation, we consider that the measured 1.0 ± 0.1 h half-life for pristine SWNT elimination depends on serum proteins that adhere to the nanotube surface. The observed 1-h half-life represents many circulation periods and may be sufficient to allow nanotubes that are linked to targeting moieties such as antibodies or peptides to localize in tissues of interest. Moreover, circulation time can probably be significantly extended by PEGylation [poly(ethylene glycol) wrapping] of the nanotube surfaces (16, 22). In principle, this may be achieved either by covalent sidewall derivatization or by surrounding the nanotubes with a stable sheath of noncovalently bound PEG-like molecules that will resist displacement by proteins.

Although it is often disregarded, the aggregation state of nanotubes is another potentially important factor in pharmacokinetics. The earlier in vivo report, as well as many in vitro studies, used samples that were aggregated into bundles containing at least tens of individual SWNTs (14, 23, 24). By contrast, the results reported here reveal the behavior of nanotubes that remain disaggregated in vivo, as shown by their retention of near-IR fluorescence. Our findings are therefore relevant to the many envisioned biomedical applications that exploit properties of individualized SWNTs. The absence of acute toxicity (at a low dose level) and reasonably long blood circulation time found here suggest that SWNT may well prove useful in medicine. We believe that their intrinsic near-IR fluorescence provides a powerful tool to speed development of such applications.

Materials and Methods

Nanotube Preparation.

Raw SWNTs grown by the high-pressure carbon monoxide (HiPco) method (25) were suspended by intense, direct contact ultrasonic agitation in an aqueous 1% solution of pharmaceutical-grade Pluronic F108, a nonionic poloaxamer surfactant. This processing exfoliates many individual nanotubes from bundles bound by van der Waals forces and allows the debundled individuals to be stabilized and suspended by physisorbed surfactant molecules (11). The resulting suspension was centrifuged for 4 h at 100,000 × g to remove impurities and obtain a supernatant fraction enriched in individual SWNTs at a concentration of ≈10 μg/ml. Finally, the SWNT suspension was autoclaved at 121°C for 60 min before use to ensure sterility.

Ex Vivo Spectroscopy and Microscopy of SWNTs.

The near-IR emission spectrum of each blood serum specimen was measured by using 658-nm diode laser excitation in a model NS1 NanoSpectralyzer (Applied NanoFluorescence, Houston, TX). Near-IR fluorescence microscopy was performed using a custom-built apparatus that differed from the one described in ref. 21 only by the addition of a 785-nm diode laser excitation source to supplement the 658-nm laser. Studies involving the identification of individual SWNTs used 785-nm excitation with a ×60 oil-immersion objective and a 946-nm long-pass filter in the collection path.

For the biodistribution observations, 12 different organ systems and tissues were analyzed: brain, heart, lung, liver, kidney, spleen, stomach, pancreas, small intestine, muscle, skin, and bone. Bright-field images were taken of these tissue sections by using a ×20 objective, and the tissue was surveyed for SWNT emission with 658-nm excitation. SWNT emission was recognized by its dependence on excitation polarization and characteristic spectral signature. Fluorescence imaging of individual nanotubes was performed by scanning the sample position while exciting with the 658-nm laser and collecting emission with a ×60 or ×100 oil-immersion objective followed by a 1,125-nm long-pass filter. Spectra were normally acquired only in case of an ambiguous outcome in the test based on modulation of emission intensity through rotation of the excitation beam polarization axis.

Bulk and individual SWNT spectra from within biological tissues were measured using a near-IR spectrograph (Horiba JY, Edison, NJ) model CP140) and cryogenically cooled 512-element InGaAs detector array (Roper Scientific, Trenton, NJ, model OMA-V). An optical fiber connected the microscope to the spectrograph entrance slit. Spectra were collected by averaging five accumulations, each integrated for 10 s, and subtracting backgrounds. A ×60 oil- immersion objective was used for obtaining spectra. The laser spot size was focused to 5–10 μm in diameter (10–20 pixels under ×60). Excitation lasers at 785 or 658 nm were used with a 946-nm long-pass blocking filter in the detection path.

Image Processing.

To prepare an overlay micrograph showing SWNT fluorescence and visible bright-field tissue images, a conventional bright-field color micrograph was recorded of a tissue specimen stained with hematoxylin and eosin (H&E). An adjacent (≈3 μm thick) unstained slice of the same tissue specimen was mounted in the near-IR microscope and imaged in bright-field mode. (Laser-induced heating prevented observation of SWNT near-IR emission from stained specimens.) Then, a series of 125 near-IR fluorescence images were acquired, each with 0.5-s exposure. In this series, the excitation spot was focused to a diameter of ≈20 image pixels and moved to a different region of the field between exposures such that the entire field was covered by the end of the series. These 125 images were digitally processed by using WinSpec (Roper Scientific) to remove pixels with intensities <500 units on a scale in which all identifiable SWNT features exceeded 1,000 units. This processing eliminated signals from thermal background and detector noise and allowed the 125 images to be digitally coadded to show the locations of SWNT fluorescence emission throughout the field. The resulting SWNT emission image was false-colored as green using Photoshop (Adobe, San Jose, CA). To prepare an overlay, the visible bright-field image was spatially offset and scaled until all of the gross structural features in the tissue specimen were precisely aligned in the visible and near-IR bright-field images. The false-colored near-IR fluorescence image representing SWNT emission was then digitally overlaid onto the visible bright-field image. Each of these composite images contains a number of isolated green pixels that correspond to defective sensor elements in the near-IR camera. However, all genuine SWNT features appear several pixels in size.

In Vitro Surfactant Displacement Kinetics.

To monitor time-dependent spectral shifts reflecting surfactant displacement, a programmed sequence of 230 blank-corrected fluorescence spectra were recorded by the NS1 NanoSpectralyzer (Applied NanoFluorescence) at 700-ms intervals from a 0.5-ml sample of the Pluronic suspension. During this measurement, 1 ml of a 3% BSA solution was rapidly added to the sample cell. The resulting data sequence was carefully analyzed for spectral changes. In a complementary experiment, a 0.5-ml sample of SWNTs in Pluronic suspension was quickly added to and mixed with 1.0 ml of rabbit serum in a fluorimetric cuvette at 297 K. The cuvette was then placed in the NS1's sample holder, and fluorescence spectra were measured every 5 s for a period of 900 s. From this spectral sequence, the fluorescence intensity was plotted as a function of time for emission near the prominent features at 1,285 and 1,143 nm. Each of these plots was analyzed by kinetic modeling.

Rabbit Injection and Tissue Analysis.

Before initiation of these preclinical studies, all protocols were reviewed and approved by the M. D. Anderson Institutional Animal Care and Use Committee. Adult New Zealand White rabbits (3–4 kg) were sedated with intramuscular injection of ketamine (30–50 mg/kg) and acepromazine (1–3 mg/kg). The rabbits were then intubated and maintained with isoflurane (1–3%) inhalational anesthesia. The neck of each animal was shaved and prepped with betadine surgical scrub. After cutdown, the jugular vein was cannulated with a 5 French catheter. The catheter was sutured to the vein, tunneled s.c. to the posterior neck of the animal, and affixed.

Once the catheter was in place, an initial pretreatment blood sample was drawn. Then, 7.5 ml of SWNTs in 1% Pluronic F-108 was injected through the catheter over 2 min. Catheter patency was maintained with heparinized saline (100 units/ml). The surgical wounds were reapproximated with suture. The rabbits were then allowed to emerge from anesthesia.

Blood samples were extracted from the indwelling catheter at the following time points: 30 min, 60 min, 90 min, 2 h, 3 h, 4 h, 6 h, 8 h, 12 h, and 24 h. After collection, the blood samples were stored in heparinized tubes at 4°C until the time of analysis, ≈48 h later.

At 24 h after SWNT injection, each rabbit was given 1 ml of IV Beuthanasia-D. During necropsy, the following tissue samples were obtained from representative animals: brain, spinal cord, heart, lung, liver, spleen, kidney, pancreas, stomach, small intestine, skin, muscle, and bone. The tissue samples were preserved in 10% formalin.

After formalin fixation, tissue blocks were embedded in paraffin. Sections were cut from 3 to 5 μm thick. Unstained samples were mounted on fused quartz slides for evaluation by visible and near-IR fluorescence microscopy. Hematoxylin and eosin-stained specimens were mounted on glass slides and evaluated by using visible light microscopy.

Control rabbits were treated as described above but were injected with 7.5 ml of 1% Pluronic F-108 solution without nanotubes. After i.v. injection, the SWNT-exposed rabbits recovered uneventfully. They exhibited normal behavior, including food and water intake, before being killed at 24 h. Gross examination of the organs at necropsy failed to reveal any abnormality after SWNT injection. Hematoxylin and eosin microscopic evaluation detected no pathological differences between the experimental and control animals.

Acknowledgments

We thank J. T. Willerson, S. W. Casscells III, and J. L. Conyers (Univ. of Texas Health Science Center, Houston, TX) for instrumentation support and C. Moran (Rice University) for expert measurements of the SWNT length distribution. This work was supported at M. D. Anderson by National Institutes of Health Grant CA016672 and at Rice by National Science Foundation (NSF) Grant CHE-0314270, Alliance for NanoHealth (National Aeronautics and Space Administration NNJ05HE75A), the NSF Center for Biological and Environmental Nanotechnology (EEC-0118007), and the Welch Foundation (C-0807).

Abbreviation

SWNT
single-walled carbon nanotubes.

Footnotes

Conflict of interest statement: R.B.W. holds an interest in Applied NanoFluorescence, LLC.

References

1. Kam NWS, Jessop TC, Wender PA, Dai H. J Am Chem Soc. 2004;126:6850–6851. [PubMed]
2. Lin Y, Taylor S, Li H, Shiral Fernando KA, Qu L, Wang W, Gu L, Zhou B, Sun Y-P. J Mater Chem. 2004;14:527–541.
3. Pantarotto D, Briand JP, Prato M, Bianco A. Chem Commun. 2004;2004:16–17. [PubMed]
4. Pantarotto D, Singh R, McCarthy D, Erhardt M, Briand J-P, Prato M, Kostarelos K, Bianco A. Angew Chem Int Ed Engl. 2004;43:5242–5246. [PubMed]
5. Bianco A, Kostarelos K, Prato M. Curr Opin Chem Biol. 2005;9:674–679. [PubMed]
6. Teker K, Sirdeshmukh R, Sivakumar K, Lu S, Wickstrom E, Wang H-N, Vo-Dinh T, Panchapakesan B. NanoBiotechnology. 2005;1:171–182.
7. Barone PW, Parker RS, Strano MS. Anal Chem. 2005;77:7556–7562. [PubMed]
8. Kam NWS, O'Connell MJ, Wisdom JA, Dai H. Proc Natl Acad Sci USA. 2005;102:11600–11605. [PMC free article] [PubMed]
9. Heller DA, Baik S, Eurell TE, Strano MS. Adv Mater. 2005;17:2793–2799.
10. Reich S, Janina J, Thomsen C. Carbon Nanotubes: Basic Concepts and Physical Properties. New York: Wiley; 2004.
11. O'Connell M, Bachilo SM, Huffman CB, Moore V, Strano MS, Haroz E, Rialon K, Boul PJ, Noon WH, Kittrell C, et al. Science. 2002;297:593–596. [PubMed]
12. Bachilo SM, Strano MS, Kittrell C, Hauge RH, Smalley RE, Weisman RB. Science. 2002;298:2361–2366. [PubMed]
13. Cherukuri P, Bachilo SM, Litovsky SH, Weisman RB. J Am Chem Soc. 2004;126:15638–15639. [PubMed]
14. Singh R, Pantarotto D, Lacerda L, Pastorin G, Klumpp C, Prato M, Bianco A, Kostarelos K. Proc Natl Acad Sci USA. 2006;103:3357–3362. [PMC free article] [PubMed]
15. Tan JS, Butterfield DE, Voycheck CL, Caldwell KD, Li FT. Biomaterials. 1993;14:823–833. [PubMed]
16. Moghimi SM, Szebeni J. Prog Lipid Res. 2003;42:463–478. [PubMed]
17. Weisman RB. In: Applied Physics of Carbon Nanotubes. Rotkin SV, Subramoney S, editors. Berlin: Springer; 2005. pp. 183–202.
18. Heller DA, Jeng ES, Yeung T-K, Martinez BM, Moll AE, Gastala JB, Strano MS. Science. 2003;311:508–511. [PubMed]
19. Chen RJ, Bangsaruntip S, Drouvalakis KA, Kam NWS, Shim M, Li Y, Kim W, Utz PJ, Dai H. Proc Natl Acad Sci USA. 2003;100:4984–4989. [PMC free article] [PubMed]
20. Kabanov AV, Batrakova EV, Alakhov VY. J Controlled Release. 2002;82:189–212. [PubMed]
21. Tsyboulski DA, Bachilo SM, Weisman RB. Nano Lett. 2005;5:975–979. [PubMed]
22. Izzo F, Marra P, Beneduce G, Castello G, Vallone P, De Rosa V, Cremona F, Ensor CM, Holtsberg FW, Bomalaski JS, et al. J Clin Oncol. 2004;22:1815–1822. [PubMed]
23. Radomski A, Jurasz P, Alonso-Escolano D, Drews M, Morandi M, Malinski T, Radomski MW. Br J Pharmacol. 2005;146:882–893. [PMC free article] [PubMed]
24. Woerle-Knirsch JM, Pulskamp K, Krug HF. Nano Lett. 2006;6:1261–1268. [PubMed]
25. Nikolaev P, Bronikowski MJ, Bradley RK, Rohmund F, Colbert DT, Smith KA, Smalley RE. Chem Phys Lett. 1999;313:91–97.

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Links

  • PubMed
    PubMed
    PubMed citations for these articles
  • Substance
    Substance
    PubChem Substance links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...