Molecular MRI of collagen
Citation for published version (APA):
Sanders, H. M. H. F. (2009). Molecular MRI of collagen. Technische Universiteit Eindhoven.
https://doi.org/10.6100/IR651820
DOI:
10.6100/IR651820
Document status and date:
Published: 01/01/2009
Document Version:
Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)
Please check the document version of this publication:
• A submitted manuscript is the version of the article upon submission and before peer-review. There can be
important differences between the submitted version and the official published version of record. People
interested in the research are advised to contact the author for the final version of the publication, or visit the
DOI to the publisher's website.
• The final author version and the galley proof are versions of the publication after peer review.
• The final published version features the final layout of the paper including the volume, issue and page
numbers.
Link to publication
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal.
If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:
www.tue.nl/taverne
Take down policy
If you believe that this document breaches copyright please contact us at: openaccess@tue.nl
A catalogue record is available from the Eindhoven University of Technology
Library
ISBN: 978-90-386-1971-2
Printed by:
Drukkerij de Rijk, Uden, The Netherlands
Sieberg 317, 5403 WC, Uden, tel. 06-41796022
Cover design: Erik Sanders
Molecular MRI of collagen
PROEFSCHRIFT
ter verkrijging van de graad van doctor aan de
Technische Universiteit Eindhoven, op gezag van de
rector magnificus, prof.dr.ir. C.J. van Duijn, voor een commissie aangewezen door
het College voor Promoties in het openbaar te verdedigen
op maandag 21 september 2009 om 16.00 uur
door
Honorius Marinus Henricus Franciscus Sanders
geboren te Uden
Dit proefschrift is goedgekeurd door de promotor:
prof.dr. K. Nicolay
Copromotoren:
dr.ir. G.J. Strijkers
en
dr. M. Merkx
Chapter 1: Introduction: Molecular imaging of collagen ...1
Chapter 2: Lipid-based nanoparticles as MRI-contrast agents: characterization and application ...9
Chapter 3: Morphology, binding behaviour and MR-properties of paramagnetic collagen binding liposomes ...43
Chapter 4: High-Resolution NMR imaging of paramagnetic liposomes targeted to a functionalized surface ...59
Chapter 5: CryoTEM study on binding of CNA35-based contrast agents to assembled collagen fibrils ...73
Chapter 6: Molecular MR imaging of collagen in mouse atherosclerosis using CNA35-micelles ...85
Chapter 7: Towards improved MR contrast agents, reduction of toxicity and improvement of detection sensitivity ...105
Chapter 8: Concluding remarks ...119
Summary...125
Dankwoord ...127
Curriculum Vitae ...130
1.1 Introduction
Molecular imaging is a relatively new discipline that aims to gather information on molecular pathways underlying pathologies by visualizing molecular processes in living organisms, preferably without disturbing them. One of the most interesting potentials of this field is the (early) diagnosis of diseases, such as cancer and cardiovascular diseases. Next to identifying disease areas, molecular imaging is capable of phenotyping diseases by the assessment of disease specific molecular pathways. This field also promises to contribute to the evaluation of disease treatment and might even be able to improve treatment by image guided drug delivery and the specific monitoring of the effects of therapy.
Instead of probing intrinsic tissue properties, as traditional imaging techniques do, molecular imaging exploits the interaction of disease biomarkers with tailor made probes that alter the signal in the area of interest as a function of the concentration of these disease markers. Two traditional imaging modalities in the field of molecular imaging are the nuclear techniques PET and SPECT, which exhibit phenomenal sensitivity, but a low spatial resolution and they both provide no anatomical background. MRI on the other hand exhibits a good anatomical reference and offers a high spatial resolution, typically 1 mm for clinical scanners and down to 50 µm for high-field animal scanners. Non-targeted MR contrast agents are already frequently used in hospital examinations to enhance local tissue contrast. However, a considerable drawback of these MR-contrast agents to render them suitable for molecular imaging purposes is the poor detection sensitivity of MR-techniques for the contrast agent induced signal alterations. This key issue can be dealt with in several ways, such as creating more potent probes by either increasing their relaxivity or by creating new nanoparticulate probes, which carry a high payload of contrast generating material. In this thesis collagen is the biological target of interest, which is, compared to most other targets, available in relatively high amounts, which also helps to overcome the problem of the low intrinsic sensitivity of MRI.
In this thesis a variety of MRI-probes are described which are suitable to image collagen. Next to biochemical and biophysical characterizations, several techniques were used to interrogate the probes on their magnetic properties and collagen binding capability. Finally, these probes were examined for their ability to detect collagen in a mouse model of atherosclerosis and myocardial infarction.
1.2 Collagen and its role in atherosclerosis and myocardial infarction
Collagens are the most abundant proteins in the human body, making up about 25 – 35% of the whole body protein content.1-3 Up to now 27 different collagen types have been identified. All these collagens can be characterized by the repeating Gly-X-Y sequence, which is required to form the unique triple helical structure, that consists of three polypeptide chains. The X and Y positions are mostly occupied by proline and hydroxyproline, which ensures triple helix stability by steric restrictions.
Collagens can be subdivided in different families according to their supramolecular assembly. In this thesis the focus lies on the fibrillar collagens (examples are type I and III), which are mainly located extracellularly. Fibrillar collagens, often aided by non-collagenous structures, are capable of assembling into complex supramolecular structures, such as microfibrils and fibrils, which is schematically shown in figure 1.1. The basic component of fibrillar supramolecular structures, the collagen triple helix, is around 300 nm long and 1.5 nm in diameter. These structures assemble into microfibrils, which in turn join into fibrils. Fibrils
show a characteristic D-band pattern (67 nm repeat), originating from the staggered arrangement of triple helical collagen molecules.
The wide diversity in collagen suprastructures creates a large set of building blocks, which are responsible for maintaining the structural integrity of organs and tissues. Next to its crucial role in normal tissue maturation, collagen is also involved in wound and fracture healing and in important pathological processes such as atherosclerosis and myocardial infarction.4-8
In atherosclerosis, collagen is considered to be a major indicator for plaque stability.9-12 About 60% of the total protein content in atherosclerotic plaques is collagen, mainly consisting of type I and III.4, 11, 13 Collagen is of major importance in all stages of plaque development and may even be involved in retention of low density lipoprotein, calcification and in the promotion of atherogenesis.9, 13 In general, plaques with a more favorable stable phenotype have a high collagen content, although this may ultimately lead to arterial stenosis.4, 10 Instable plaques suffer from a collagen deficit due to excessive breakdown by matrix metalloproteinases, making these plaques prone to rupture.4, 10
Figure 1.1. Left: Cartoon of collagen assembly.14 Right: Image of a collagen fibril taken with CryoTEM.15
The formation of collagen is a key event during tissue remodeling after myocardial infarction (MI). Cardiac infarction is a frequently occurring cardiovascular event, frequently caused by a ruptured atherosclerotic plaque. Over the past decades early mortality has considerably decreased due to the success of acute intervention strategies.16 Patients surviving this acute phase of myocardial infarction enter a phase of cardiac wound healing, ultimately resulting in the formation of scar tissue. Nevertheless, in the long run a major complication is heart failure that results from factors such as inadequate healing, hypertrophy, structural remodeling and dilation. During the first phases of wound healing collagen levels are strongly elevated and remain significantly higher throughout the process.6, 7, 16, 17 Imaging collagen level and distribution in the early stages after infarction could serve as an important diagnostic tool to detect eminent heart failure or rupture likelihood of the cardiac wall.18
1.3 Collagen binding ligands
An essential part of a collagen specific probe suitable for the field of molecular imaging is the collagen binding ligand. This ligand should be specific for collagen, exhibit reversible binding, not interfere with tissue development and ultimately also be inexpensive to produce.
Antibodies directed against collagen are attractive candidates. However a major drawback is that they are very expensive, which becomes a problem when large quantities are needed. Furthermore antibody binding to collagen can become that strong that it carries the risk of interfering with normal tissue development.
Dichlorotriazinyl aminofluorescein (DTAF) has been reported as a small fluorescent probe for visualizing collagen fibers in living organisms.19-21 However, DTAF is rather unspecific for collagen as it reacts with amines and forms a covalent bond.22-26
Cells in tissues use integrins to interact with the collagen matrix. These integrins exhibit natural collagen binding specificity and are therefore attractive candidates for a collagen targeting ligand. Krahn et al. modified the human integrin α1I-GST with a fluorescent moiety and showed specific binding to collagen with a dissociation constant of ~50 µM.26 Next to its moderate affinity also cross-reactivity with other extra cellular matrix proteins was observed, rendering this integrin less suitable as a targeting ligand.
Figure 1.2. Left: Cartoon of hypothetical binding model of CNA35 to a collagen triple helix, the
“collagen hug model”. (A) The collagen triple helix is initially associated with the N2 domain. (B) The collagen is then wrapped by the N1–N2 linker and the N1 domain. (C) The N1 domain interacts with the N2 domain via multiple hydrophobic interactions and finally the C-terminal latch is introduced in the N1 domain to secure the ligand in place. Right: Crystal structure of CNA35 bound to collagen-like triple helical peptide.27
Krahn and colleagues also exploited the use of collagen adhesion proteins from bacteria. CNA35, the collagen binding domain of a collagen adhesion protein from Staphyloccoccus
aureus28 was expressed in E.coli and produced in large quantities. Subsequently the protein was modified with fluorescent Oregon green moieties and showed specific binding with a biphasic binding behavior. The high affinity binding had an apparent dissociation constant KD of 0.5 µM. Next to that additional binding occurred at higher protein concentrations. The binding mechanism, shown in figure 1.2, of CNA35 has been studied extensively by Zong et
al.27 The crystal structure of CNA35 bound to a synthetic collagen like triple helical peptide was analyzed. From this data it was suggested that binding occurs according to the “collagen hug” model. First the N2-domain, also known as CNA19, which exhibits a moderate affinity
for collagen itself as well29, interacts with the collagen triple helix. In a second step the N1 domain folds around the collagen triple helix. Finally the C-terminus complements a β-sheet on the N1 domain thus locking the CNA35-protein in place. This binding model suggests that CNA35 possesses a higher affinity for single triple helices as compared to triple helices imbedded in fully assembled collagen fibrils.
Oregon green labeled CNA35 (CNA35-OG488) has been used in tissue engineering experiments26, 30 and Megens and coworkers used fluorescently labeled CNA35 to study collagen in atherosclerotic lesions.12 They concluded that the vascular endothelium in organs and healthy arteries represents a non-permeable barrier for CNA35-OG488. However in pathologies like atherosclerosis this barrier is compromised, allowing CNA35-OG488 to penetrates into the extravascular space. Megens et al. showed that CNA35-OG488 binds to collagen in atherosclerotic plaques, demonstrating its potential as a molecular imaging probe for this disease process.
Xu et al.28 have described that the Y175K mutation of CNA35 reduces the collagen binding affinity significantly, rendering this mutant-CNA35 an excellent control ligand to active CNA35.31
Another attractive possibility to identify collagen specific binding ligands is to use phage display techniques. Caravan and coworkers used this approach to find a peptide-sequence specific for collagen type I.32 They found a peptide, of which they improved the binding characteristics by introducing a biphenylalanin and rendered it detectable by MRI by coupling three Gd-DTPA moieties (see figure 1.3). The dissociation constant reported was 1.8 µM. By changing 1 cysteine residue from the L- to the D-form an increase of the dissociation constant to 400 µM was observed, making it suitable as control ligand.
Figure 1.3. A Collagen-targeting peptide equipped with Gd chelates; L-amino acids are designated by one letter code, except where noted; Gd chelates are appended through the N terminus, through branched Lys-Gly residues at the N terminus, and through a Lys side chain within the cyclic portion of the peptide.18
Recently, Helms et al. used the phage display technique as well to identify a collagen specific peptide.33 Highly homologues sequences were found with a consensus sequence H-V-F/W-Q/M-Q-P/A-P/K. The HVWMQAP-heptapeptide, in its monomeric form, showed a dissociation constant of KD = 61 ± 5 μM, while a scrambled or the reversed sequence showed no significant binding. Titration experiments for a pentameric phage analog displaying five of these sequences yielded a 100-fold improvement in collagen affinity, showing the same affinity as that of CNA35 (KD = 550 ± 100 nM).
1.4 Collagen detection by MRI
Collagen detection by MRI has been pioneered by Navon and coworkers using a method based on double quantum (DQ) coherences, which are formed due to interactions of water protons with collagen.34 This elegant technique is dependent on the presence of structurally organized tissues, such as collagen. Like in chemical exchange saturation transfer (CEST), images of 1H DQ coherences can be overlaid on normal MR images to get an anatomical reference. Unfortunately, all structurally aligned protons contribute to the signal, also non-exchanging protons attached to proteins for example. Next to that this technique suffers from a low resolution compared to standard MR images. The active collagen remodeling that occurs in many pathologies causes a considerable fraction of the collagen fibers to be disorganized, which limits the applicability of the DQ coherence technique. Therefore, this thesis proposes MRI detection of collagen using collagen specific contrast agents.
Recently, Caravan and coworkers reported a collagen specific MRI contrast agent consisting of 3 Gd-DTPA moieties bound to a peptide with a moderate affinity for collagen type I.35, 36 Using this contrast agent they were able to discriminate scar tissue formed after myocardial infarction in a mouse from viable myocardium and blood, which is shown in figure 1.4.
Figure 1.4. MR imaging of myocardial of a mouse scar (40 days after MI) following injection of the
collagen specific EP-3533 (A–D) and the control EP-3612 (E–H, image acquired two days later in same mouse). First image (A, E) is a black blood T2-weighted image acquired before injection of the contrast agent to delineate the anatomy. Images B–D and F–H are inversion recovery images taken prior to, 1 min after, and 40 min after injection of the contrast agents.18
This study illustrates that the high collagen concentration present in infarcted tissue is an attractive molecular target for MR imaging even with small imaging probes.35, 36 Using a similar probe (except that the DTPA moieties were replaced by DOTA moieties), Spuentrup
et al. showed that this probe enables prolonged, high-contrast, high-spatial-resolution
visualization of myocardial perfusion defects by binding to collagen.37 The main advantage of a small collagen binding probe over traditional blood pool agents is the prolonged residence time of the probe at the side of interest, which enables a larger window for imaging. Normal blood pool agents only show a first pass effect, while this collagen targeted probe showed contrast in normal (well perfused) tissue for at least 60 minutes. In well perfused tissues, which will contain a lot of collagen bound probe, signal intensity will increase, whereas in non-perfused tissue signal intensity will remain at its original value. In future studies, one has
to consider, however, whether redistribution of the probe at later time points might cause signal enhancement by binding to collagen in hypoperfused regions as well.
1.5 Aim of this thesis
This thesis describes the development and characterization of collagen targeted MRI contrast agents and their subsequent utilization in an in vivo mouse model of atherosclerosis. Chapter 2 gives a general introduction to the synthesis, characterization and utilization of lipid-based contrast agents. Contrast agents based on paramagnetic liposomes (chapter 3 and 7), micelles (chapter 6) and micellular iron oxide particles (chapter 7) were synthesized and thoroughly characterized using a broad range of techniques, such as dynamic light scattering, cryogenic transmission electron microscopy (cryoTEM) and nuclear magnetic resonance.
A new technique called high resolution NMR depth profiling (chapter 4) was used to aid the characterization of contrast agents in their bound state. Furthermore, new insights in the targeting behavior of collagen specific contrast agents were obtained using advanced cryoTEM studies (chapter 5).
Collagen-targeted paramagnetic micelles were subsequently employed to phenotype atherosclerotic lesions in live mice using high-resolution MRI, as described in chapter 6. Chapter 7 discusses the initial results on potential modifications of the contrast agents presented in this thesis. Finally, chapter 8 gives some concluding remarks on this thesis.
Reference list
1. Lullo, D. The Journal of Biological Chemistry 277, 4223-4231 (2002). 2. Lehninger, A. L. Biochemistry (1993).
3. Hall, D. A. International Review of Connective Tissue Research (1964). 4. Rekhter, M. D. Cardiovasc Res 41, 376-84 (1999).
5. Libby, P. et al. Circulation 105, 1135-43 (2002). 6. Jugdutt, B. I. Circulation 108, 1395-403 (2003). 7. Cleutjens, J. P. et al. Cardiovasc Res 44, 232-41 (1999). 8. Myllyharju, J. et al. Ann Med 33, 7-21 (2001).
9. Plenz, G. A. et al. Circulation 108, 129-130 (2003).
10. Rodriguez-Feo, J. A. et al. Curr Pharm Des 11, 2501-14 (2005). 11. Katsuda, S. et al. J Atheroscler Thromb 10, 267-74 (2003). 12. Megens, R. T. et al. Mol Imaging 6, 247-60 (2007). 13. Barnes, M. J. et al. Exp Gerontol 34, 513-25 (1999). 14. Campbell. Biology (1995).
15. Sanders, H. M. H. F. et al. In preparation (2009).
16. Blankesteijn, W. M. et al. Acta Physiol Scand 173, 75-82 (2001). 17. Jugdutt, B. I. et al. Circulation 94, 94-101 (1996).
18. Caravan, P. et al. Angew Chem Int Ed Engl 46, 8171-3 (2007). 19. Davison, P. F. et al. Invest Ophthalmol Vis Sci 27, 1478-84 (1986). 20. Davison, P. F. et al. Invest Ophthalmol Vis Sci 26, 1202-9 (1985). 21. Woo, H. M. et al. Br J Ophthalmol 85, 345-9 (2001).
22. Blakeslee, D. et al. J Immunol Methods 13, 305-20 (1976). 23. Brinkley, M. Bioconjug Chem 3, 2-13 (1992).
24. Letourneur, D. et al. J Biomater Sci Polym Ed 4, 431-44 (1993). 25. Logeart, D. et al. Eur J Cell Biol 74, 376-84 (1997).
26. Krahn, K. N. et al. Anal Biochem 350, 177-85 (2006). 27. Zong, Y. et al. Embo J 24, 4224-36 (2005). 28. Xu, Y. et al. J Infect Dis 189, 2323-33 (2004). 29. Symersky, J. et al. Nat Struct Biol 4, 833-8 (1997). 30. Boerboom, R. A. et al. J Struct Biol 159, 392-9 (2007). 31. Reulen, S. W. et al. J Am Chem Soc 131, 7304-12 (2009). 32. Helm, P. A. et al. 864 (ISMRM, 2007).
33. Helms, B. et al. J Am Chem Soc (2009). 34. Navon, G. et al. NMR Biomed 14, 112-32 (2001). 35. Caravan, P. Chem Soc Rev 35, 512-23 (2006). 36. Mulder, W. J. et al. Bioconjug Chem 15, 799-806 (2004). 37. Spuentrup, E. et al. Circulation 119, 1768-75 (2009).
Lipid-based nanoparticles as MRI contrast agents: characterization
and application
Based on:
H.M.H.F. Sanders, S. Hak, G.J Strijkers, K. Nicolay, Lipid-based nanoparticles as MRI
contrast agents: Characterization and Application, Encyclopedia of Nanoscience and
2.1 Introduction
Lipid-based nanoparticles, such as micelles, liposomes and emulsions, have been widely used as drug delivery vehicles in the last decades. They are composed of naturally occurring phospholipids or closely related synthetic lipids and are therefore considered biocompatible. The nanoparticles may carry considerable payloads of hydrophilic or hydrophobic pharmaceutical agents, protecting them from interactions with plasma proteins and deactivation, which leads to a markedly improved bioavailability. Prolonged circulation times and favorable biodistribution of these nanoparticles were achieved by coating them with hydrophilic polymers, such as polyethylene glycol (PEG).1 Furthermore, conjugation of targeting ligands to home the carriers to diseased tissues has lead to improved delivery of the enclosed pharmaceuticals. Research on the incorporation of contrast generating materials for imaging applications, especially MR-detectable probes, has tremendously increased over the last few years, which has resulted in powerful MR contrast agents. These characteristics, combined with the ease of preparation, ensure that multi-component lipid-based nanoparticles serve as an excellent and diverse platform for nanoparticulate MR contrast agent design. This renders these particles suitable for application in the field of molecular and cellular imaging.
In this chapter, first the characteristics of amphiphiles and basic principles of lipid aggregation are described. Next, a variety of possibilities for introducing functionality, like contrast generating material and targeting ligands, is presented. In a second part an extensive overview of the different experimental techniques used for characterization of nanoparticulate lipid-based contrast agents of which many were used in the studies described in the following chapters is given. To conclude, applications of a wide variety of lipidic MR-contrast agents are reviewed.
2.2 Amphiphiles
Amphiphiles are molecules that contain both a hydrophobic (non-polar) and a hydrophilic (polar) domain. Fig. 2.1 shows the molecular structure of two typical and frequently-used amphiphiles, distearoylphosphatidylcholine (DSPC) and cholesterol.
Figure 2.1. Two typical amphiphiles. The commonly used phospholipid DSPC (A) has two fatty acyl
chains in the hydrophobic domain, while the hydrophilic part is composed of the phosphate moiety with the surrounding nitrogen and oxygen atoms. In the case of the sterol cholesterol (B) the hydrophobic part is represented by the hydrocarbon ring structures, while the hydrophilic part solely consists of the hydroxyl group.
A
In an aqueous solution, at very low concentrations, amphiphiles are present as single molecules. Above a certain concentration, however, which is termed the critical aggregation concentration (CAC), the dualistic character of amphiphiles and the entropically unfavorable contact between water molecules and the hydrophobic domain induce their self-association into a variety of aggregates. In these aggregates the hydrophobic domains are clustered together in order to minimize the contact with water. These hydrophobic clusters are decorated with a hydrophilic surface capable of forming hydrogen bonds with the surrounding water molecules, consequently lowering the entropic stress of the water molecules.
There is a large diversity in both the hydrophilic and the hydrophobic domains of amphiphiles. The hydrophobic part could vary in length and size; for example, it could exist of a single or multiple carbon chains. The hydrophilic headgroup may vary in charge (anionic, cationic or nonionic) and size as well. Such variations in the two different domains determine the ratio between the size of the polar and the non-polar part. Together with parameters like pH, temperature and concentration, this is determining the geometry of the aggregates formed. In general, smaller aggregates are entropically favored over larger structures. On the other hand, smaller structures are energetically less favorable.2, 3
Phospholipids and cholesterol are naturally occurring amphiphilic molecules that represent the most important structural elements of biological membranes (Fig. 2.1). In recent years, many phospholipid-like structures have been synthesized to benefit from the amphiphilic character and used to achieve a wide variety of aggregates.
Amphiphile aggregation
At high amphiphile concentrations, cubic, lamellar and hexagonal phases may be formed 4, 5, whereas in the low concentration regime, structures such as liposomes, spherical micelles, thread-like micelles, disc-like micelles and rod-like micelles predominate. In practice, it is very desirable to be able to predict which of these structures are formed, for example when synthesizing new amphiphiles or after a chemical modification of an amphiphile.
For one-component systems, the critical packing parameter (CPP) in most cases correctly predicts the type of aggregate formed. The CPP is defined as:
c l a v CPP ⋅ = 0 Eq. 2.1
where a0 is the optimal surface area of the hydrophilic headgroup, v is the volume of the hydrocarbon chain and lc is the critical chain length. The volume occupied by the hydrocarbon chain is assumed to be fluid and incompressible. The critical chain length is the maximum length the chains of the amphiphile can extend to. These two parameters can be estimated using c c n l =1.5+1.265⋅ , and Eq. 2.2 c n v=27.4+26.9⋅ , Eq. 2.3
where nc is the number of carbon atoms per saturated hydrocarbon chain. In the case of a dual tailed amphiphile, v is doubled. Equations 2 and 3 were phenomenologically deducted from experimental data.4, 6
A limitation of this prediction method is that a0 is difficult to estimate, because it is not only a property of the molecule itself but also depends on amongst other, the salt concentration, lipid concentration and pH.4 Moreover, it should be noted that repulsive forces of charged
headgroups, hydration forces and steric headgroup hindrance, are not included in this simple model. a0-values in aqueaous environment are ~64 Å2 for phosphatidylcholine (PC) and ~72 Å2 for phosphatidylethanolamine (PE).3
Predicting the type of aggregates formed in the case of multi-component systems is not as straightforward as for single component systems. The CPP can still be used for a first order prediction, though additionally to the dimensions of each component, the energy and entropy differences between mixing and separating the different components must be taken into account. In practice, it is rewarding to consult literature on lipid mixtures similar to the ones of interest in order to obtain first estimates on the influence of the individual components on the geometry of the aggregate formed.
Figure 2.2. Lipidic aggregates which can be formed with variable CPP. The left hand side (A) shows a
schematic representation of relevant amphiphiles, including single and double-tailed phospholipids, PEGylated double-tailed phospholipid and cholesterol, together with their CPP. The right hand side shows schematic pictures of possible aggregates. These are: (B) spherical micelles; (C) sterically stabilized micelles; (D) thread-like micelles; (E) disc-like micelles; (F) PEG-stabilized discs; (G) conventional liposomes; (H) PEG- and cholesterol-stabilized liposomes.
Aggregate types
Fig. 2.2 illustrates a variety of lipidic aggregates that could be formed depending on the CPP. For lipids to assemble into spherical micelles their optimal surface area a0 must be sufficiently large and their hydrocarbon volume v sufficiently small, such that the radius of the micelle will not exceed the critical chain length lc, corresponding to a CPP ≤ 1/3. For spherical micelles, the critical micelle concentration (CMC) is typically in the micromolar range.7 Above the CMC, usually a very narrow size distribution, or low polydispersity, is found, and the average micelle size is very insensitive to concentration changes.4
Cylindrical (rod-like) micelles will be formed when decreasing the headgroup size, or when increasing the hydrocarbon volume, such that 1/3 ≤ CPP ≤ 1/2. Rod-like micelles may become very large and usually display a considerable polydispersity. Furthermore, unlike spherical micelles, the dimensions of cylindrical micelles are extremely sensitive to concentration changes.4
When the CPP is further increased, i.e. 1/2 ≤ CPP ≤ 1, bilayers will be formed. In this situation, the volume of the hydrophobic domain is increased relative to the surface area of the headgroup. This is typically the case when amphiphiles contain two hydrophobic carbon chains, as for dual tailed phospholipids. In this case, the CAC is drastically lower compared to when spherical micelles are formed.4
An example of a bilayered structure is a disc-shaped micelle, which is normally very short-lived due to the unfavorable edge interactions of the hydrophobic domains with water.8 Upon closure of unstabilized discs, liposomes could be formed. Liposomes are defined as spherical, self-closed structures, formed by one or several concentric lipid bilayers with an aqueous phase inside and in between the lipid membranes.9 Unilamellar vesicles usually range in diameter from 800 nm down to 50 nm. In the case of dual tailed phospholipids, liposomes are most likely the dominant aggregates formed.
When CPP > 1, usually the case for double tailed lipids with a0 < 42 Å2, amphiphiles will start to precipitate from aqueous solutes. An example of an amphiphile with CPP > 1 is cholesterol. However, cholesterol is widely used in multicomponent systems, where its behavior is quite different.
Multicomponent aggregates
In practice, often, multicomponent systems are used in order to obtain nanoparticles with specific, desired properties. When in such systems large amounts of lipids with large headgroups are present, e.g. the widely used PEG-lipid (CPP ~ 0.05), micelles will be formed most likely.7 PEG-lipids are phospholipids with long hydrophilic PEG-polymers conjugated to the phosphate moiety. Due to their bulky hydrophilic headgroup, the CMC of these PEG-lipids is very low. For example, the CMC of PEGylated distearoyl-phosphoethanolamine (DSPE) with different molar mass of the PEG tail (2000, 3000 and 5000 Da) varies from 0.5 to 1.5 µM, with higher CMC for longer PEG chain length.7
In multicomponent mixtures containing PEG-lipids and bilayer-forming lipids, discs can also be obtained as a stable structure.10 It seems likely that the PEG-lipids will preferentially accumulate at the rim of the disc, such that the large PEG-polymers shield the hydrophobic carbon chains from surrounding water molecules, preventing the discs from aggregation or closure.10, 11 However, experimental evidence for this suggestion is lacking. A rule of thumb is that discs will be formed in PEG-lipid/lipid mixtures when the lipid is in the gel phase.5, 10, 12 Some other observations are that threadlike micelles are commonly observed in PEG-lipid/lipid samples when the lipid is in the liquid crystalline phase, while discs are observed in
such mixtures supplemented with cholesterol.13, 14 When cholesterol is incorporated in a lipid bilayer, the hydroxyl group will be situated at the hydrocarbon-water interface, while the stiff ring system and the hydrocarbon tail are embedded in the hydrophobic interior of the membrane.15 This orientation of the cholesterol molecule will decrease the curvature of the bilayer. Depending on the thermodynamic phase of the lipid membrane and the amount of cholesterol incorporated, cholesterol can also alter the packing order in the membrane in several manners, reducing the membrane permeability.16 In addition, cholesterol increases the membrane’s elastic bending rigidity when included in phospholipid bilayers.17-19
Liposomes are typically obtained in lipid mixtures containing dual chain phospholipids, and less than 7 % PEG-lipid, which sterically stabilizes the liposomes.20-22 These formulations are often supplemented with over 30 % cholesterol, for further bilayer stabilization.21, 23
Another class of multicomponent aggregates are core-sized lipid-based particles. These particles consist of an inorganic core, e.g. an iron oxide nanoparticle or quantum dot, coated with a single- or bilayer of lipids.24-26 One of the advantages of covering iron oxide and quantum dot-based nanoparticles with lipids, is that additionally to the functionality of the inorganic nanocrystal itself, a range of other functionalities can be introduced relatively easily. These include fluorescent lipids, Gd(III)-loaded lipids and lipids to which ligands for biomarker specificity can be conjugated, thus creating a flexible platform for multimodality, targeted imaging. Another advantage is that the biocompatibility and circulation times can be tuned using an appropriate lipid coating.
The same strategy can be used to create multimodal emulsions. For instance, iodinated or fluorinated hydrophobic molecules can be enclosed by a single amphiphile layer to create a suspension that is stable in an aqueous environment. Iodinated particles can be used for computer tomography (CT) purposes 27, 28, while fluorinated molecules can be used in 19 F-MRI.29
2.2.1 MR-properties
So far, the general properties of amphiphiles and amphiphile aggregates have been discussed. In this section, the strategies employed to obtain lipidic nanoparticles endowed with properties for MRI detection will be reviewed. The aggregates discussed above consist of hundreds to hundreds of thousands of individual lipids and therefore the incorporation of large amounts of Gd(III)-containing amphiphiles is relatively straightforward and the most widely used approach to obtain highly potent, lipid-based, nanoparticulate MR contrast agents. In the case of liposomes, an alternative strategy to obtain paramagnetic nanoparticles is encapsulation of low molecular weight paramagnetic species in the aqueous interior. Another approach is the incorporation of superparamagnetic iron-oxide in either a lipidic mono- or bilayer, or in the liposomal lumen. Furthermore, 19F containing lipidic nanoparticles can be produced, which can be utilized in 19F-MR imaging. A recent development is the enclosure of chemical exchange saturation transfer (CEST) agents in the liposomal lumen in order to obtain so called LIPOCEST agents. An overview of the different approaches is given below.
Paramagnetic nanoparticles
The first lipid-based MR contrast agents reported on, were liposomes which entrapped low-molecular weight, hydrophilic paramagnetic agents in the aqueous interior. Paramagnetic agents that have been incorporated in liposomes are Mn(II) 30, 31, Gd(III)DTPA 32-34, Mn(II)DTPA 35, Gd(III)DTPA-BMA 36, Gd(III)HP-DO3A 37 and Mn(II) bound to serum proteins.38 Although several successful in vivo applications have been reported, the
applicability of this type of contrast agent is limited due to very low ionic relaxivities of the liposomal paramagnetic species. The water permeability of the liposome entrapped membranes is low, resulting in limited water exchange between the liposomal lumen and the bulk water, thereby severely quenching the relaxivity.34, 39, 40 Several attempts have been made to increase the water permeability of the bilayer, which can be achieved by varying the saturation level and length of the lipid carbon chains and by the reduction of the cholesterol content.20, 40, 41 Liposomes with enhanced permeability, however, are usually not stable in serum.5 A second complication of this type of liposomal contrast agent is that upon degradation, the hydrophilic paramagnetic agents can easily diffuse away from the site of release, and this makes the interpretation of the observed contrast enhancement problematic.
Figure 2.3. Two different types of Gd(III) containing amphiphiles. In Gd(III)DTPA-BSA (A) the
hydrophilic part solely consists of the Gd(III)-chelate. Note that the aliphatic chains are directly conjugated to atoms playing a role in the coordination of Gd(III). In Gd(III)DOTA-DSPE (B) the highly stable DOTA-based Gd(III)-chelate is conjugated to the primary amine of the phospholipid DSPE.
Another approach to obtain paramagnetic nanoparticles is the incorporation of Gd(III)-bearing amphiphiles. As the aggregates contain a large number of individual lipids, a high payload of Gd(III) per particle can be achieved. A wide variety of Gd(III)-containing amphiphiles have been developed in the last decades.42-47 In general two types of Gd(III)-bearing amphiphiles are distinguished. One type in which the hydrophilic part solely consists of the Gd(III)-chelate, which is directly conjugated to one or two fatty acyl chains. In the other type, the Gd(III)-chelate is conjugated to the hydrophilic part of an existing amphiphile like for example PE. Fig. 2.3 shows an example of both types of Gd(III) bearing amphiphiles. Several studies demonstrated that the paramagnetic amphiphiles, in which the Gd(III)-chelate is conjugated to an existing amphiphile, result in higher relaxivities compared to the chelates where the hydrophilic domain solely consists of the Gd(III)-chelate.47-49
Hak et al. incorporated both lipids displayed in Fig. 2.3 in an identical liposomal formulation and demonstrated that the relaxivity of the Gd(III)DOTA-DSPE is roughly two times higher than the relaxivity of Gd(III)DTPA-bistearyl amide (Gd(III)DTPA-BSA).48 At 25 ºC and 1.41 T the longitudinal relaxivities (r1) were 12 mM-1s-1and 7 mM-1s-1, respectively. A similar observation was made by Winter et al., who compared the relaxivities of Gd(III)DOTA-PE to
A
the relaxivity of Gd(III)DTPA-dioleoyl amide (Gd(III)DTPA-DOA) when incorporated in emulsions.47, 49 Gd(III)DTPA-DOA has a double bond in both the aliphatic chains, in contrast to its saturated analogue Gd(III)DTPA-BSA. They also demonstrated roughly a doubling of the relaxivity in the case of the PE conjugated chelates when compared to Gd(III)DTPA-DOA. In these studies, this difference in relaxivity was mainly explained by a much higher water exchange rate in the case of the PE conjugated Gd(III)-chelates. These increased water exchange rates can be explained by the overall structure of the amphiphiles. When localized in a lipid layer, the phosphate group is presumably positioned at the interface of the lipid layer and the bulk water. Consequently, the Gd(III)-chelate extends from the membrane, in contrast to the chelates in amphiphiles where the hydrophilic part solely consists of the Gd(III)-chelate. This results in better access for water molecules to the paramagnetic complex as compared to Gd(III)DTPA-BSA and Gd(III)DTPA-DOA.
Winter et al. additionally showed that a longer spacer between the Gd(III)-chelate and the hydrophobic domain results in a higher relaxivity.47 With increasing length of the spacer, the accessibility of the chelate for water molecules might increase. However, at the same time, the rotational correlation time of the Gd(III)-chelate could decrease due to increased molecular flexibility, which would result in lowering of the relaxivity. The optimal length of the spacer probably depends on the formulation used. Variation of the distance between the Gd(III)-chelate and the hydrophobic domain thus may provide an attractive way to optimize the relaxivity and hence, lower the detection limit of amphiphilic Gd(III)-chelates.
In liposomes, another important parameter determining the relaxivity of Gd(III) containing amphiphiles is the degree of saturation of the hydrophobic domain. Strijkers et al. demonstrated that the relaxivity in liposomes composed of unsaturated phospholipids is higher when compared to liposomes with saturated lipids.20 Laurent et al. also showed a higher relaxivity for the unsaturated Gd(III)DTPA-DOA in dipalmitoyl phosphocholine (DPPC) liposomes when compared to its saturated analogue Gd(III)DTPA-BSA.41 This was explained by the higher water permeability of lipidic bilayers (partially) composed of unsaturated lipids compared to bilayers composed of saturated lipids only. As a consequence, water accessibility of the Gd(III)-chelates located in the inner leaflet was increased, which led to an increase in relaxivity.
An issue which has recently drawn a lot of attention is transmetallation of Gd(III)-chelates. Transmetallation of Gd(III)-chelates is the process in which the Gd(III)-ion is replaced by endogenous ions like e.g. Zn(II), Cu(II) or Ca(II). The linear, open-chain Gd(III)-chelates such as Gd(III)DTPA and Gd(III)DTPA-BMA were shown to be very susceptible to transmetallation.50, 51 Administration of these Gd(III)-chelates to patients with renal failure has recently been related to nephrogenic systemic fibrosis (NSF).50-52 Although the exact cause of this severe disorder has not been determined yet, transmetallation could play a role and therefore it is of great importance to identify alternatives for the linear DTPA and DTPA-derived Gd(III)-chelates, especially in the case of targeted lipid-based contrast agents, which have prolonged blood circulation and tissue residence times. The macrocyclic chelates, such as Gd(III)DOTA and Gd(III)HP-DO3A, have been demonstrated to be very stable and essentially immune to transmetallation under in vitro conditions.50, 51 In recent years several amphiphilic Gd(III)-chelates based on Gd(III)DOTA have been developed.45-48
Superparamagnetic nanoparticles
Iron-oxides have been incorporated in liposomes and micelles as well. Micellular iron-oxides consist of a hydrophobic iron oxide core coated with a monolayer of lipids. These particles
have been used in vitro to target apoptotic cells 25 and to non-specifically load cells with iron oxide.53
Liposomes containing iron oxide nanoparticles are called magnetoliposomes.54 In these structures, water soluble iron-oxide particles reside in the liposomal lumen. Two types of magnetoliposomes exist. The first type is a hydrophilic iron-oxide nanoparticle of ~15 nm coated with a bilayer of lipids, resulting in a particle with a diameter of around 40 nm.24 The second type is several hundreds of nm in size and contains multiple water soluble iron-oxide particles in the liposomal lumen.55-57 No molecular imaging studies have been performed using magnetoliposomes. However, several studies employed these particles for in vitro cell labeling 57, and as a non-specific, in vivo MR contrast agent.55, 57
Due to the limited use of these particles, very limited research has been done to investigate parameters which influence the relaxivity and, hence the detection limit. LaConte et al. showed that the thickness of the lipid coating of micellular iron oxide particles influences their relaxivity.58 The transverse relaxivity r
2 decreases, and r1 increases with increasing thickness of the lipid layer.
LIPOCEST
Incorporation of CEST-agents into the lumen of liposomes has recently led to the new class of LIPOCEST-agents. Aime et al. introduced the first LIPOCEST agent by incorporating the shift reagent [Tm(dotma)]- in liposomes.59 They demonstrated a detection limit for the liposomes of 90 pM, when suspended in buffer. The frequency shift (δ) of the water protons inside the LIPOCEST-agent as compared to the bulk water is the most important factor for the efficacy of a LIPOCEST-agent. Larger shifts will improve the in vivo detection as the LIPOCEST effect needs to be larger than susceptibility differences and natural occurring chemical shifts. Furthermore, it will facilitate the detection of multiple LIPOCEST-agents in the same imaging voxel. Terreno et al. showed an increase in this frequency shift from δ = - 4 ppm to -45 ppm when the CEST-agent was contained in osmotically shrunken, non-spherical liposomes, due to bulk magnetic susceptibility effects.60 An elegant feature of LIPOCEST is the possibility to detect multiple CEST-agents with a different chemical shift independently from each other within the same imaging voxel. This was also shown by Terreno et al., who demonstrated the first ex vivo co-localization of two LIPOCEST-agents.61 In the field of molecular imaging this is a valuable feature, because it may enable the simultaneous detection of different molecular epitopes of interest in vivo.
This new type of contrast agent is still in its infancy and before in vivo applications are possible, several problems have to be overcome. One of these problems is that natural in
vivo occurring chemical shifts can interfere with the detection of these agents. Furthermore, high-power RF-pulses are needed to detect these agents, which can lead to exceeding the specific absorption rate limitations (SAR-limitations) for in vivo experiments.
19F-MRI
The main advantage of 19F-MR contrast agents over the use of paramagnetic and superparamagnetic contrast agents is that there is no native 19F in the human body, which opens the possibility of quantitative determinations of targeted fluor containing probes. The group of Wickline and Lanza introduced paramagnetic perfluorocarbon (PFC) emulsions and used it in a number of molecular MRI studies.62-64 Additionally to 1H-MRI, they recently employed 19F-MRI to detect this nanoparticle in vivo.29, 64, 65 Particles were detectable with 1 H-MRI, which allowed for localization on an anatomical background, and subsequent
quantification was achieved with 19F-MRI. Nevertheless, a limitation of 19F-MRI is the low sensitivity. The detection limit is in the millimolar range, which may not be sufficient to detect sparse molecular epitopes in vivo. On the other hand, the technique is still in its infancy and Flögel et al. recently reported a detection limit in the sub millimolar range.66 Because of its quantitative character 19F-MRI might become a valuable tool in molecular MRI.
2.2.2 Multimodal lipid-based platforms
Besides the relative ease of incorporation of high payloads of contrast generating material, another excellent quality of lipid-based MR contrast agents is the straightforward integration of multiple imaging labels in a single nanoparticle, creating a so-called multimodal contrast agent. Fig. 2.4 shows a selection of possible multimodal lipidic nanoparticles. The ability of multimodal detection of a nanoparticle has the benefit that the strengths of several imaging modalities can be combined. This is often critical to the understanding of the in vitro and in
vivo behavior of the agent.
Figure 2.4. A selection of possible multi-modal lipid-based nanoparticles (right) along with their
building blocks (left column). The nanoparticles are: (A) a multimodal liposome, composed of DSPC, DSPE-PEG, Gd-bearing lipid, a fluorescent lipid and cholesterol; (B) an emulsion encapsulated in a lipid monolayer comprised of DSPC, DSPE-PEG, a fluorescent lipid and a Gd-bearing lipid; (C) a hydrophobic quantum dot decorated with a monolayer of DSPE-PEG and Gd-bearing lipids; (D) micellular iron-oxide composed of DSPE-PEG and a fluorescent lipid with an iron oxide core within the lipidic shell; (E) a micelle with a gadolinium-containing lipid, DSPE-PEG and a fluorescent moiety. All particles are functionalized with a targeting ligand at the distal end of a PEG-chain.
In terms of resolution and sensitivity, optical imaging is highly complementary to MRI, and therefore optical properties have widely been integrated in lipid-based MR contrast agents. By incorporation of tracer amounts of commercially available fluorescent lipids, like rhodamine-PE and fluorescein-PE, formulations for both MR and optical detection can be obtained. As fluorescence techniques are much more sensitive than MRI in the detection of low concentrations of contrast agent, tracer amounts are sufficient, and hence the influence of the optical label on the morphology of the nanoparticle is negligible. Mulder et al. demonstrated the successful use of αvβ3-targeted paramagnetic liposomes, labeled with
rhodamine-PE in tumor bearing mice. In vivo MRI demonstrated uptake into the tumor and ex
vivo fluorescence microscopy revealed the specific targeting of angiogenic endothelium.67 Lanza et al. introduced the use of perfluorocarbon (PFOB) emulsions as paramagnetic multimodal tracers.68 Their emulsions consist of a perfluorocarbon core, surrounded by a lipid monolayer containing amphiphilic Gd(III)-chelates. Besides imaging with conventional 1 H-MRI, these particles thus can be detected with 19F-MRI, which is expected to allow a more straightforward quantification of these particles.69 Additionally, the authors incorporated fluorescent lipids in the lipidic monolayer to obtain a particle detectable with fluorescent techniques as well.29
CT is another imaging modality which has been combined with MRI. CT provides high spatial resolution and is very fast, which for example enables very sensitive and specific detection of coronary artery stenoses. Zheng et al. loaded liposomes with the water soluble MRI contrast agent gadoteridol and the CT contrast agent iohexol.70 They showed that this contrast agent was detectable in vitro as well as in vivo with both CT and MRI. Cormode et al. synthesized high density lipoprotein (HDL) in which amphiphilic Gd(III)-chelates were incorporated in the lipid monolayer and the cholesterol ester/triglyceride core was replaced with a gold nanoparticle which was shown to be detectable with CT.71 Furthermore they incorporated a fluorescent lipid to enable optical detection.
The new generation of LIPOCEST agents can also be considered a multimodal platform. Aime et al. demonstrated Gd(III)-loaded liposomes to be a combined T1/T2-lowering and CEST agent.72 A possibility could be to add fluorescent detection by incorporation of amphiphilic fluorophores in this formulation.
Combining the high sensitivity of positron emission tomography (PET) or single photon emission computed tomography (SPECT) with the high resolution of MRI may also represent an attractive multimodal combination. The low resolution, but very sensitive, nuclear techniques can provide information on the biodistribution of the nanoparticles. Whole body scans are possible and thus areas of interest can thus be identified relatively rapidly. Thereafter, these areas may be investigated in more detail with high-resolution MRI. To that end, Lijowski et al. prepared a PFOB based emulsion labeled with both Gd(III) and the SPECT label 99mTc(III).73
Another interesting feature of lipid-based nanoparticles is their ability to act as drug carrier vehicles. Liposomes and micelles have been utilized extensively in the last decades for this purpose. Hydrophilic drugs have been incorporated in the liposomal lumen and hydrophobic drugs can be incorporated in the lipidic mono or bilayer and in a micellular core.22, 74-76 The modification of these lipid-based nanoparticles into contrast agents has led to the possibility of image guided drug delivery. Lobatto et al. administered paramagnetic liposomes, with glucocorticoids encapsulated in the lumen, to atherosclerotic rabbits. They demonstrated abundant uptake in the atherosclerotic plaque with T1-weighted MRI.77
2.2.3 Functionalization of lipid-based nanoparticles
Untargeted contrast agents, like Gd(III)-DTPA, are often used to perform contrast enhanced magnetic resonance angiography (CE-MRA) of the vascular system or dynamic contrast enhanced MRI to study non-specific accumulation in, for example tumor tissue. To perform targeted molecular imaging, the contrast agents need to be equipped with targeting ligands that specifically bind to, for instance, disease specific markers. Lipid-based contrast agents can be decorated with targeting ligands in multiple ways, as will be described in this section.
The most common method is to covalently attach a targeting ligand to the surface of a preformed lipidic nanoparticle via a surface functionality. Many lipids with functional reactive groups, such as maleimide, thiol, amine and carboxylic acid moieties exist and most of them are commercially available. In the case of PEGylated nanoparticles, it is best to have these functionalities at the distal end of a PEG-lipid, as this enhances the efficiency of both conjugation and targeting capabilities.
The functional group of choice, of course depends on the options to perform chemistry on the targeting moiety. If a free thiol is present, one can use for instance a pyridyldithiopropionyl (PDP)-functionality. A drawback of using PDP is that the disulfide bond formed between PDP and the free thiol easily reduces under in vivo conditions.78 Thiol-groups react easily with maleimide groups at the distal end of a PEG-lipid to form a stable bond. The latter strategy is commonly used and widely applicable.9, 79
Not all targeting ligands express a free thiol, though it is often possible to introduce one, for example via the functionalization of amines with succinimidyl-S-acetyl-thiol-acetate (SATA) 79 or N-hydroxysuccinimidyl-3-(2-pyridyldithio)propionate (SPDP).78 To obtain a free thiol after SATA-modification, the protecting acetyl group is cleaved off using hydroxylamine. In the case of SPDP, a reducing agent like dithiothreitol (DTT), has to be used in the second step. This condition is, however, less mild than the use of hydroxylamine.
Another approach can be the use of activated carboxyl groups; these can react with amine groups of the ligand. The coupling reaction is carried out in the presence of 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDAC) and N-hydroxysulfosuccinimide (NHS) to form an acyl amino ester, which will subsequently react with the primary amine of the ligand, yielding an amide bond.78
Although very easy to perform, next to possible inactivation, the use of amines results in a heterogeneous coupling of ligands. To circumvent these problems, several approaches might be taken. First, it is possible to introduce a free and unique thiol via recombinant techniques. Next to that, native chemical ligation (NCL) was proposed as a site specific ligation strategy, using a C-terminal thioesther moiety on a protein after cleavage of a self-cleavable intein domain.80, 81 This requires a cysteine to be present as the functional group on the distal end of the PEG-lipid.
Next to incorporating a free thiol or a C-terminal thioester, it is also possible to couple a biotin moiety to the targeting ligand. This enables one to exploit the high-affinity biotin-avidin interaction to non-covalently attach ligands to nanoparticles. Avidin can be added to lipid-based particles containing a biotin moiety at the distal end of PEG-lipids. In a second step, the biotinylated ligand can be mixed in, leading to a functional particle. Of course this strategy could also be used for a multistep pre-targeting approach in vivo, in which first the biotinylated targeting ligand is injected, followed by avidin and ultimately by the biotinylated nanoparticle .78, 82
Another approach to circumvent the non-specific modification of proteins during conjugation, is the use of so called docking proteins as intermediates between the liposome and the targeting protein. Examples include the use of protein G as a docking site for IgG 83 and the use of RNase I as a docking site for proteins with a C-peptide tag.84-86 Although this results in a more homogenous presentation of targeting ligands on the liposomes, this approach requires additional conjugation steps and depend on a non-covalent bond between targeting ligand and docking protein, which limits the efficacy of this strategy.
A different approach to coupling a target ligand to preformed contrast agents, is to put a hydrophobic anchor on the targeting ligand, thus creating an amphipile, ready to mix in with the amphiphile mixture during preparation of the nanoparticle.3 The advantages are to have
the possibility to purify the modified ligand and to be able to achieve a high incorporation efficiency. A drawback is that synthesizing and purifying an amphiphilic ligand could be time consuming and difficult. For incorporation in liposomes, an additional drawback of this strategy is that about half of the targeting ligand will be located inside the liposomal lumen, preventing it from contributing to the particles targeting capabilities. For more information about coupling strategies, a number of excellent books and reviews are available.3, 5, 9, 78, 79
2.3 Characterization
In recent years, many studies have reported on novel lipid-based contrast agents. In order to understand and improve the design and the in vivo application of these nanoparticles, it is of key importance to experimentally characterize the particles properly. In this section, the suitable experimental techniques to characterize lipid nanoparticles are discussed.
2.3.1 Morphology
One of the most important features of the nanoparticles is morphology. It is of major importance in verifying the robustness of the used preparation protocol, it provides information on the stability of the particles, it could predict aspects of the in vivo behavior, like clearance mechanisms and targeting efficiency and it sometimes proves to be crucial to explain unexpected experimental results.
This paragraph will describe two techniques that are very suitable to study the nanoparticulate morphology: cryogenic transmission electron microscopy (cryo-TEM) and dynamic light scattering (DLS). Other techniques such as small angle neutron scattering (SANS), small angle X-ray scattering (SAXS) and atomic force microscopy (AFM) can also provide information on morphology; however these will not be covered here.
2.3.2 Cryo-TEM
Electron microscopy (EM) is a very versatile technique of which many forms exist, including scanning EM, freeze fracture EM and transmission EM (TEM). The most powerful EM-technique to investigate lipid-based nanoparticles is cryo-TEM. Cryo-TEM is TEM performed on samples of sub-micron thickness at cryogenic temperatures. The most important advantage of cryo-TEM over normal TEM is that, if samples are properly prepared, there are no dehydration effects, which might lead to clustering or collapsing of the nanoparticles. An aqueous lipid dispersion is put on an ordinary copper EM-grid. After removing the excess of the dispersion by gentle blotting, under controlled temperature and at 100% humidity, a thin film is created. Next, the sample is brought rapidly into melting ethane, resulting in nearly instant freezing of the liquid film, thus arresting the movements of contrast agents. Rapid freezing results in the formation of vitreous ice rather than crystalline ice. Vitreous ice is less electron dense than crystalline ice, enabling TEM in the first place. Performing cryo-TEM on the sample, which is continuously kept at temperatures well below -135 ºC (the de-vitrification temperature of pure water), allows to study the contrast agent in the natural hydrated state.
Since cryo-TEM samples of biological specimens can be very sensitive to electron beam damage it is important to work in low electron dose conditions, meaning that only a limited amount of electrons can be used for imaging, before the specimen is destroyed. This usually results in somewhat noisy pictures compared to normal EM.
A typical example of a cryo-TEM picture of Gd(III)-bearing liposomes is shown in Fig 2.5. Circular structures with a dark rim can be recognized. The dark rim is the phospholipid bilayer. The hydrophilic outer parts of the bilayer are more electron dense compared to the aliphatic inner part, hence the bilayer can be observed. The bilayer is only visible at the rim of the liposomes, because there the electrons pass parallel to the bilayer, whereas for the remainder of the bilayer, the electrons pass perpendicular or quasi perpendicular through the membrane.
Figure 2.5. Typical cryo-TEM picture of paramagnetic liposomes composed of 25% Gd-DTPA-BSA, 33% cholesterol, 37% DSPC and 5% PEG-DSPE.
The interior of the liposomes is somewhat dark as well. This is not only caused by lipids, but primarily by the fact that the liposomes are thicker than the ice layer they are embedded in. Hence, fewer electrons will pass through the liposomes than through the surrounding water layer, creating darker regions in the images. Next to unilamellar vesicles, also three multi-lamellar vesicles can be distinguished in Fig 2.5, while there is no clustering visible and all vesicles seem circular. Because only a small fraction of the particles is viewed in a single image, multiple images have to be made before hundreds of vesicles can be statistically analyzed and quantitative conclusions on the morphology can be drawn, which makes this technique time-consuming.
Several studies have shown the effectiveness of cryo-TEM for characterizing lipid-based nanoparticles.8, 10, 12, 13, 20, 87-89 Fig. 2.6A demonstrates the value of cryo-TEM in a study by Vucic et al.90, in which a two component lipid-based MR-contrast agent composed of Gd(III)DTPA-BSA and P2FA2 was investigated. P2FA2 is an apolipoprotein E derived peptide, which functions as a strong detergent. P2FA2 is hypothesized to interact with cells through receptors for low density lipoprotein (LDL) and proteoglycans and was additionally labeled with fluorescein. P2FA2 has been shown to efficiently incorporate into micelles and enhance their uptake into rat brain capillary endothelial cells in vitro.
P2FA2 was mixed with Gd(III)DTPA-BSA in different ratios. At high concentrations of P2FA2 only circular micelles were observed. Upon increasing the Gd(III)DTPA-BSA content, plate-like structures started to appear, eventually evolving into long ribbon-plate-like structures when further increasing the Gd(III)DTPA-BSA content. A further increase in Gd(III)DTPA-BSA resulted in precipitation. From this series of images it is clear that cryo-TEM can provide detailed information about the structures at hand.
Figure 2.6. (A) Cryo-TEM series of a mixture of P2fA2 and Gd(III)DTPA-BSA with changing molar
ratios between the different components. The conversion from micelles to threads can be observed. (B) DLS results of the same lipid mixtures at various molar ratios between the two components.90 Next to assessing the overall structure of lipid-based nanoparticles, cryo-TEM is a valuable tool to monitor potential alterations in the size and morphology upon storage. This provides information on the stability of the system and the achievable shelf-life of the nanoparticles.
2.3.3 DLS
Cryo-TEM is one of the most precise methods in determining the size and size distribution of lipid-based nanoparticles. However this technique is expensive, and time-consuming. A widely-used, relatively simple and fast method to determine the size distribution of lipidic colloids is dynamic light scattering (DLS). DLS is also known as photon correlation spectroscopy (PCS) or quasi elastic light scattering (QELS).
During a DLS measurement, laser light (typically 632.8 nm) is scattered by particles in solution. By measuring the fluctuations of the scattered light, induced by Brownian motion (random diffusion) of the particles in the laser beam, the diffusion coefficient can be determined mathematically. For detailed information two excellent monographs are available.91, 92 In short, smaller particles move faster than larger particles, which can be expressed in a different diffusion constant. From the diffusion coefficient, the hydrodynamic radius of the particles in dispersion is calculated via the Stokes-Einstein equation
D T k R b h πη 6 = Eq. 2.4
where D is the diffusion coefficient of the particles causing the fluctuation, kb is the Boltzmann’s constant, T is the temperature, η is the solvent viscosity and Rh is the mean hydrodynamic radius. Fig 2.6B shows the hydrodynamic radii found by Vucic et al. on the same system as presented in Fig 2.6A (cryo-TEM). It is clear that the subtle changes in the regime from 20-100% of P2FA2 are detected with DLS. However, from the hydrodynamic radius alone no information about the precise physical condition of the sample could be deduced. In the low concentration regime of P2FA2 clear differences were found with DLS.
This DLS analysis made the authors decide to elaborate with a detailed cryo-TEM investigation to get information on morphology.
One of the features of light scattering is that smaller particles scatter less compared to large particles. For particles much smaller than half the incident wavelength, the particles scattering intensity is proportional to Rh.6 This means that large scatterers dominate the light scattering intensity so that small, and hence, weaker scatterers are hardly detected at low concentrations. A few percent of large scatterers can thus dominate the experimentally measured intensity. Upon interpretation of results, this effect has to be taken into account.
2.3.4 Particle composition
Most lipid-based MRI-contrast agents consist of multiple constituents in order to obtain stable formulations, optimal circulation times, and to introduce both target specificity and multimodality. Often, simply by mixing the different components, the optimal ratio is obtained. However, it is important to verify whether no separation has occured in the suspension and that the particles created indeed have the desired ratios between the different components. Often, a targeting ligand is coupled to preformed nanoparticles via surface chemistry. Unreacted ligands are usually removed by washing methods, like ultracentrifugation or dialysis. It is important to verify whether the coupled ligand is indeed present on the contrast agent and whether it is still functional. For these reasons, below a number of methods are described to determine contrast agent composition, as well as ligand coupling and targeting efficacy.
Lipid composition
Common constituents of lipid-based MRI-contrast agents are PEG-DSPE, cholesterol, phospholipids, such as DSPC and DSPE, and to generate contrast, Gd(III)-containing lipids or iron-oxide particles. Multiple methods exist to determine phosphate, gadolinium and iron content.
As a gold standard for determining phosphorous-, gadolinium- and iron-content, a method called inductively coupled plasma (ICP) in combination with mass spectrometry (MS) or atomic emission spectrometry (AES) is often used. Before performing ICP analysis it is necessary to destruct the sample in such a way that the element of interest is present in its ionic form, which is not straightforward as different contrast agents may require different destruction protocols. After destruction, ICP is used to efficiently transfer the ions into vacuum, and the content of the ion of interest (the detection limit is in the order of picomoles of material) is determined with MS or AES using a calibration curve.
ICP analysis requires dedicated equipment and is therefore not always the method of choice. In combination with destruction, it is possible to use other methods to determine the concentration of the ion of interest. A popular and easy method for the determination of phosphate content in lipid-based contrast agents is the Rouser method 93. In this colorimetric approach, the samples are destructed in 70% perchloric acid, the inorganic phosphate is colored using a molybdate solution, and quantified using a spectrophotometer and a calibration curve.
As an alternative method to determine the amount of gadolinium present in a sample, a relaxometric approach could be taken. After complete destruction, free Gd(III) ions are present in solution. By measuring the longitudinal relaxation rate (R1) it is possible to determine the gadolinium concentration by making use of a calibration curve with e.g. GdCl3.
