Analysis of Cationic-Lipid
7240
A N A LY T I C A L C H E M I S T R Y / O C T O B E R 1 , 2 0 0 7
© 2007 AMERICAN CHEMICAL SOCIET Y
Plasmid-DNA Complexes Complexes between plasmid DNA and various cationic lipids are promising vehicles to deliver genetic information into cells for gene therapy or vaccines.
C. Russell Middaugh Joshua D. Ramsey University of Kansas
T
he idea of using genes as drugs is both exciting and controversial. In most gene therapies, the proteins that are coded for by the DNA have a therapeutic effect, but the genetic material itself is delivered to the patient, generally in two different ways. The first uses modified viruses that contain the gene(s) of interest. This takes advantage of the fact that viruses have evolved over hundreds of millions of years to efficiently deliver their genomes into cells and have their content expressed. This can work fairly well, but the problems of immunogenicity and toxicity, even to the extent of death, have led to serious concerns about this approach (1, 2). The second method attempts to build a synthetic virus—an approach that provides better control of the delivery system’s properties. These nonviral vectors usually are created by combining a bacterial plasmid containing the gene(s) of interest with a positively charged (usually polymeric) partner to form a molecular complex. Plasmids themselves are manipulated easily to contain any gene of interest as well as on-and-off elements and other regulatory signals. Preparation of large quantities in a pure form is also straightforward. Positively charged complexing agents that have been used so far include basic peptides, polyethylenimine, amino dendrimers, various synthetic block copolymers, and cationic lipids (CLs) (3–7). Anionic lipids in combination with multivalent cations have been explored as an alternative to CLs (8). The CL-containing vectors are, however, the most thoroughly investigated. They have been shown to have some efficiency in the clinic, although it is significantly less than that of their viral counterparts; cytotoxicity and immunogenicity are still problematic (9, 10). Reviews have addressed characterization and the biological aspects of CLs for gene delivery (11–13). Traditionally, gene delivery vectors have been analyzed in terms of their biological effects, such as their ability to enter cells and direct the expression of selected encoded genes. This process is known as transfection, and its measurement lacks the accuracy and precision expected for the analysis of a material that will be used as a drug in humans. Furthermore, the complexity of biological fluids makes physical characterization of vectors in such environments difficult. Characterization O C T O B E R 1 , 2 0 0 7 / A N A LY T I C A L C H E M I S T R Y
7 2 41
of vectors in situ typically requires the vector to be radio- or fluorescently labeled. Thus, measurements are performed most often on pure, buffered solutions of DNA–polycation complexes. The vectors themselves are large, complex, and usually very heterogeneous in structure. This makes their physical and chemical characterization difficult, even in a pure state. Recent attempts have been made to treat such vectors with rigorous physicochemical methods to produce stabler, more effective, and safer gene-based drugs. In this article, we take CL–DNA complexes (CLDCs), or “lipoplexes”, as prototypical examples and discuss the progress and future of their physical and chemical analysis. Originally described in 1987 (7 ), CLDCs are the most thoroughly studied nonviral gene delivery agents and serve as a good introduction to the analytical problems involved in their characterization.
Primary structure and composition
ject to a variety of chemical changes. The simplest approach is to separate the DNA from the CL by methods based on size, charge, or density. This can work only if the complexes can be dissociated quantitatively. Because the interactions between the CL and DNA are primarily electrostatic in nature, in principle, separation can be achieved by adding salt, but the presence of additional, less clearly defined interactions makes this difficult if not impossible. If separation can be accomplished, analysis of the individual components is easy with conventional methods. For example,
(I) C 0 < O (II) �
When highly negatively charged DNA is mixed with virtually any polycation, some degree of condensation of the polynucleotide occurs. In solution, plasmid DNA usually exists as a highly supercoiled, covalently closed circle of DNA with small amounts of open circular (nonsupercoiled) and linear (cleaved in both polynucleotide chains) contaminants. If you FIGURE 1. A depiction of two distinct pathways from the (left) lamellar phase to the take a rubber band and twist it several times, (right) columnar inverted hexagonal phase of CLDCs. you will observe its collapse into a much smaller (Left) Helical DNA strands (blue ribbons) between layers of CLs (green head groups with volume. This is essentially what happens when yellow aliphatic chains) and neutrally charged helper lipids (white head groups with yellow the negative charges on the phosphate groups aliphatic chains). (Center top) Along pathway I, the natural curvature C0 of the CL monolayer is driven negative by the addition of the helper lipid DOPE. The CL DOTAP is cylindrical, of the DNA are neutralized by polycations, whereas DOPE is conical, thus leading to the negative curvature. (Center bottom) Along which relieve the electrostatic repulsive forces pathway II, the transition is induced by adding helper lipids consisting of mixtures of DOPC within the DNA and subsequently reduce its and the cosurfactant hexanol, which reduces the membrane bending rigidity . (Adapted with permission from Ref. 17.) volume. In the original bacterial cell used to produce the plasmid, the DNA is collapsed by an energy-dependent en- DNA can be hydrolyzed, and the nucleotide composition can be zymatic reaction in which the strands are twisted into a more determined by LC and MS methods. A number of techniques condensed, highly tensioned state. The lipids in the complex exist to analyze specific degradation products as well (14). are thought to be in a bilayer form with the apolar lipid tails Perhaps surprisingly, however, the composition of CLDCs on the inside and the cationic head groups on the outer sur- is not usually determined at all—rather, the composition is face. The question is what happens when the DNA and CL are simply assumed to be defined by the identity and amounts of combined. Regular and irregular particle-like structures are the components originally introduced. In the future, however, formed with a size range of a few tens to a few hundreds of as gene therapy agents approach pharmaceutical reality, more nanometers, but the resultant structures display significant size rigorous criteria likely will be required. (and, presumably, compositional) heterogeneity. Whether such mixtures can be analyzed, even in principle, by any type of Secondary, tertiary, and quaternary structure rigorous procedure is certainly a reasonable question. We will Microscopy. Under certain conditions, polymers almost always argue that they can be, although ambiguities in interpretation manifest local interactions within their chains that lead to are significant. highly organized structure. The best known examples of this The actual composition of lipoplex formulations is more in biochemical systems are the -helices and sheets of proteins, difficult to define than expected. Although only two macro- the various forms of the DNA double helix, and the bilayer molecular components may be present (e.g., DNA and CL), structure of amphipathic lipids. The latter two structures are they can be present either as pure species or as complexes of the most relevant. The structures of DNA and CL bilayers unknown composition. Furthermore, both molecules are sub- have been observed both separately and in complex by X-ray 7242
A N A LY T I C A L C H E M I S T R Y / O C T O B E R 1 , 2 0 0 7
diffraction; small-angle X-ray scattering; and transmission electron, atomic force (AFM), and scanning tunneling (STM) microscopies (15–20). From such work, a fairly uniform picture has emerged in which aligned strands of the helical DNA are sandwiched between alternating monolayers of CLs (Figure 1, left; 16). When neutrally charged helper lipids (so called because they enhance transfection efficiency) are included, a remarkable transformation takes place and an inverted hexagonal phase of the CL part of the complex is formed (Figure 1, right; 17 ). In this state, the previously buried apolar side chains of the lipids now extend outward from the DNA chains, forming a uniform apolar coat around each strand of DNA. Although this second form of the complex has been accepted to transfect more efficiently, research has shown that factors other than structure may account for differences in efficiency of gene delivery. In one case, increasing the charge density was shown to increase the transfection efficiency of lamellar complexes to levels observed with the inverted hexagonal complex (21). A separate study showed that an optimum charge density may exist that balances the structural stability of the complex with the release of the DNA (22). Unfortunately, the methods used to achieve these insights into CLDC structure are not applicable to complexes in pharmaceutical formulations. They do, however, clearly establish the helical nature of the DNA and the local structure of the CLs. A quite different problem occurs when imaging methods are used to probe the structure of lipoplexes. Whether negative staining combined with electron microscopy or AFM or STM methods are used, one almost always sees images containing a collection of amorphous entities described as anything from nebulous globs to structures that resemble spaghetti and meatballs (19). At present, the utility of such data for rigorous analysis remains to be established. Although the helical nature of the DNA can sometimes be seen, typically little further detail is apparent. Dynamic light scattering (DLS). In addition to direct imaging methods, the size and shape of lipoplexes can be characterized by DLS. In this method, the fluctuations in the intensity of scattered light due to the Brownian motion of particles are analyzed by autocorrelation methods to yield a diffusion constant and ultimately a size in the form of a hydrodynamic diameter. Although the polydispersity of the complexes formed by CLs and plasmid DNA makes a rigorous analysis of the resulting data somewhat difficult, a mean diameter can still be obtained or the data deconvoluted in such a way as to obtain distinct populations of particles as a function of approximate size ranges. When DLS is used in the presence of an electromagnetic field, an estimate can be made of the charge on the particle or the potential (the voltage at the surface of shear). The intensity of the scattered light as a function of the angle of detection
also can be used to directly measure the molecular weight of the particle as well as its radius of gyration (which is based on the distribution of mass of the particle rather than its hydrodynamic behavior). The ratio of the radius of gyration to the experimentally determined or calculated hydrodynamic radius provides a measure of the asymmetry of the particle. DLS studies have identified CL–DNA particles in the size range 50–300 nm and confirmed their intrinsic heterogeneity (23–25). Their potentials range from the highly negative (an excess of DNA) to the highly positive (an excess of CL). One of the few correlations between the physical properties of particles and their ability to transfect cells is based on DLS measurements where it has been found that smaller (90–150 nm), positively charged lipoplexes generally produce good gene expression. This probably reflects the mechanism by which such particles enter cells; initially, the mechanism appears to involve electrostatic interactions with highly negatively charged cell-surface proteoglycans (26, 27 ). Analytical ultracentrifugation. This method can be used in one of two modes. In equilibrium sedimentation, material is sedimented at such a speed that equilibrium is reached in the centrifuge tube. The resultant distribution of mass of the solute in the tube is then optically monitored. This distribution can be analyzed and converted to an absolute molecular weight, which can in turn be used to calculate a size on the basis of assumptions about the density and shape of the particles. Densities (partial specific volumes) can be measured experimentally with oscillating U-tubes or pycnometers. In the second approach, the rate at which the particles sediment is measured (sedimentation velocity); this rate permits an estimate of hydrodynamic size to be made. Unfortunately, the usual heterogeneity of lipoplexes has made it difficult to apply either method. A simple but less informative method is based on the sedimentation of lipoplexes in preformed density gradients of inert solutes, such as sucrose or dextrose (28, 29). This method has been used to provide empirical measures of particle behavior that are not related simply to their physical properties. As the homogeneity of pharmaceutical preparations of lipoplexes increases, both sedimentation velocity and equilibrium experiments have the potential to play important roles as analytical tools. Gel electrophoresis. Although sufficiently porous size-exclusion matrices exist, this method has yet to be widely applied to the structural characterization of lipoplexes. The gel retardation assay is occasionally used to determine the positive-tonegative charge ratios that minimize the amount of unbound DNA (30). Because of the relative charge and size differences between lipoplexes and plasmid DNA, electrophoresis of the two produces widely separated bands. The gel retardation assay shows that with increasing charge ratios the amount of unbound DNA decreases and the quantity of lipoplexes retarded O C T O B E R 1 , 2 0 0 7 / A N A LY T I C A L C H E M I S T R Y
7243
near the top of the gel increases. Complete retardation of DNA often does not occur until an excess of positive lipid charge to negative DNA charge exists; this probably indicates that not all of the positive charge associated with the lipid is involved in the formation of the complex. Farhood et al. attributed this observation to steric hindrance between the two species due to the bulky nature of the liposome and superhelical DNA (31).
� (degrees L mol –1 cm –1)
10,000
DNA 0.25 0.5 1.0 1.5
DNA 0.25 0.5 1.0
5000
1.5
0
–5000
–10,000 200
220
240
260
280
300
320
340
Wavelength (nm)
FIGURE 2. Representative CD spectra of DOTAP–DNA complexes. The charge ratios are indicated. Open symbols are CLDCs below charge neutrality, and the closed symbol is a CLDC above charge neutrality. (Adapted with permission from Ref. 22.)
Circular dichroism (CD). This straightforward technique focuses primarily on the helical structure of DNA by measuring the difference in absorption of left- and right-handed circularly polarized light. Because of the regular helical array of the DNA bases, very strong optical activity is present within DNA molecules. For the most common type of DNA, known as Bform, a very characteristic spectrum displays a strong positive peak near 275 nm and a negative signal of similar intensity at ~245 nm. When CLs are added to DNA, a red shifting of both peaks and a fairly dramatic reduction in intensity of the 275 nm feature are seen (Figure 2). These large spectral changes were originally thought to be due to a change from B-form DNA (10 bp/turn) to C-form (9 bp/turn). Later studies, however, based on IR and Raman measurements and molecular dynamics simulations, strongly suggest that the changes seen are due to small alterations in the interactions between the bases while the DNA essentially remains in the canonical B-form (32). What is important from an analytical point of view is that the CD spectra of the complexes are very sensitive to their structure and that therefore CD can monitor the state of the DNA in lipoplexes and polyplexes (cationic polymer-based complexes; 33–35). FTIR. This method can characterize the structure of both the DNA and CL components of complexes (36). A disadvantage of this technique is that 10–20-fold higher concentrations are required for solution studies; an advantage is that 7244
A N A LY T I C A L C H E M I S T R Y / O C T O B E R 1 , 2 0 0 7
very concentrated and even dried materials can be examined easily (37, 38). In such studies, peak positions corresponding to vibrational modes of a wide variety of molecular groups, including vibrations of the DNA bases and phosphate groups, as well as peak positions corresponding to lipid methylene and carbonyl stretching bands are measured as a function of CLto-DNA ratios. Titration of one component into the other clearly reveals distinct peak positions for various stages of complex formation—this supports the use of this method for structural characterization and verification. Figure 3 presents two examples in which the CLs DOTAP and DDAB were titrated into a solution of plasmid DNA. During these titrations, the positions of peaks corresponding to various vibrational modes and stretching bands were monitored. The data show shifts in the peak positions occurring over a range of CL-to-DNA weight ratios; these correspond to distinct structural states in the complexes. Similar, but preliminary, studies have been undertaken with Raman spectroscopy (32), but that approach is limited because of the higher concentrations needed to obtain spectra. Fluorescence spectroscopy. This technique has been used to examine various aspects of lipoplex structure. The most commonly used method is the simple displacement of fluorophores bound to plasmid DNA. Typically, dyes are either intercalated between the bases of the DNA or bound within the minor groove. The normally solvent-quenched quantum yield of fluorescent dyes is dramatically enhanced when they are bound to DNA. When a polycation such as a CL micelle is added to such labeled DNA, the fluorescent dye often is displaced through competition with the former CL. This displacement of the dye leads to a reduction in fluorescence quantum yield that is easily measured. If a combination of dyes is used along with titration experiments, lipoplex-induced structural (especially topological) changes in the DNA can be detected, and changes in the solvent accessibility of the minor groove are seen (39). Thus, simple measurements of dye fluorescence intensity and wavelength emission maximum can be used to obtain at least comparative structural information. The major disadvantage of this approach is that, unlike CD and FTIR studies, a label is needed. Furthermore, the label is usually added to the DNA before complex formation occurs. Therefore, this method cannot be used to characterize an intact lipoplex directly. Another fluorescence-based method that suffers from some of the same problems but is of significantly higher resolution is fluorescence resonance energy transfer (FRET). When two fluorescent groups have the property that the emission spectrum of the donor overlaps the absorption spectrum of the acceptor, it is possible to excite the donor but see emission from the acceptor (or quenching of the donor; 40). By measuring the efficiency of such events, one can estimate the distance between the donor(s) and acceptor(s). FRET is a versatile technique that is increasingly seen in a wide variety of applications (41). In the case of lipoplexes, it has been possible to label the DNA by either intercalation of a dye between the bases or binding of a dye to the minor groove (donors). Then, a different dye can be placed within
Stability of complexes
(g) 2923
2925
2921
2924
2919
2923
2917
(b) 2855 2854
Wave number (cm–1)
the bilayer of the CLs (acceptors); this makes it possible to estimate the mean distance between various sites within the complexes (23, 42). Although they have only been explored to a limited extent, versions of this approach may turn out to be quite useful. For example, FRET measurements with respect to time can be used to obtain the kinetics of complex formation even at very early times, permitting one to see critical structural rearrangements as complexes are created (43, 44).
(a) 2926
(h) 2853
2853
2852
2852
2851
(c) 1746
2850
1742 1738
(i) 1717.5
1734
1716.5
(d) 1720 1718
1715.5 1714.5
If lipoplexes are to be used as pharmaceutical 1716 agents, methods to characterize their stability (j) 1493 1714 in real time (storage) and accelerated modes 1491 will be necessary. The idea is that by increas(e) 1492 ing temperature, changing pH, agitating the 1489 complexes, or applying other stresses, one 1490 1487 can increase degradation to the point that it can be detected at short time intervals and 1488 the drug’s environment adjusted accordingly. (f) 1227 In addition to the methods described below, (k) 1227 these changes can be determined by the 1225 1225 methods described earlier (36, 45–48). 1223 1223 Gel electrophoresis. Agarose gel electrophoresis combined with intercalating dyes is 1221 1221 0 1 2 3 0 1 2 3 4 a common method to visualize the size and Weight ratio (CL/DNA) Weight ratio (CL/DNA) conformation of plasmid and linear DNA. Plasmid DNA commonly exists as three isoforms (linear, open circular, and supercoiled) FIGURE 3. The effect of (a–f) DOTAP:DNA and (g–k) DDAB:DNA weight ratios on DNA that migrate through the gel at different rates and CL vibrational modes. based on their shapes. Because plasmid DNA The peak positions were plotted versus the weight ratio of CL to DNA with 1 mg/mL DNA undergoing oxidative degradation often in each sample. Stretching bands: (a, g) lipid asymmetric methylene; (b, h) lipid symmetric shows a transition from predominantly su- methylene; (c) lipid carbonyl; (d, i) guanine/thymidine carbonyl; (e, j) guanine/cytosine, inpercoiled to open circular and eventually to plane; (f, k) DNA asymmetric phosphate. (Adapted with permission from Ref. 37.) linear forms because of cleavage of the phosphodiester backbone, monitoring the change in the relative structure (52, 53). One such probe, KMnO4, selectively oxiamounts of the isoforms can serve as a simple stability assay. dizes unpaired thymine residues, thus making the DNA susThe stability of intact complexes has also been characterized ceptible to piperidine cleavage of the modified phosphodiester with the gel retardation assay described earlier (49). Complexes backbone. This is resolved with polyacrylamide gel electrothat have undergone increasing levels of degradation produce phoresis. In contrast to a bulky enzymatic probe, the relatively bands corresponding to more unbound DNA. small size of KMnO4 allows detection of changes as minor as A related method can be used to characterize the capacity strand separation at the base pair level and can therefore be of the CL to protect the genetic material from degradative used to detect DNA melting within a lipoplex. One such study enzymes that are likely to be encountered inside and outside found that upon initial interaction between DNA and CLs, the the cells (50). Here, lipoplexes are exposed to DNase I, which DNA undergoes a partial unwinding (54). cleaves unbound and/or unprotected DNA into linear fragIsothermal titration (ITC) and differential scanning ments. A detergent is then used to separate the DNA from (DSC) calorimetries. One indirect method that has been exthe CL, and the components are detected by agarose gel tensively used to characterize the stability of complexes is ITC electrophoresis. (55–57). In this approach, a concentrated solution of either Chemical probes. DNA can organize into a variety of the DNA or the CL is added in small increments to the other, complex secondary structures, such as left-handed DNA, cru- and the small amount of heat released or absorbed is measured ciforms, and other multistranded structures (51). A number by compensating for the small temperature changes produced of techniques that use chemical probes have been developed by the binding interaction. The results are analyzed by fitting to recognize and detect minor changes within the secondary the resultant data to various binding models, and in principle, O C T O B E R 1 , 2 0 0 7 / A N A LY T I C A L C H E M I S T R Y
7245
60,000
Molar heat capacity (kcal/K mol)
for short pieces of DNA, the more extensive interactions that occur within a plasmid molecule (many thousands 50,000 of DNA base pairs in length) produce a very high Tm h (>90 °C). g The results of DSC experiments with lipoplexes pro40,000 duce the transitions expected for the CL and DNA f components (58). In the case of complexes that contain 30,000 e CLs that undergo phase transitions within a measurable range, a sharp transition (Figures 4d and 4g) is typid 20,000 c cally seen. When the bilayer is stabilized by the DNA, this transition is usually shifted to a higher temperature b 10,000 (Figures 4b to 4c and 4e to 4f). a This phase transition also can be seen directly with 0 IR spectroscopy by analyzing the methylene stretch40 50 60 70 80 90 100 110 120 ing vibrations or through the use of fluorescent probes Temperature (°C) which insert into the bilayer. A broader transition is FIGURE 4. DSC thermograms of CLDCs and cationic liposomes alone. seen at >100 °C that reflects the melting of the plasmid (Figures 4a to 4c, 4e to 4f, and 4h to 4i). It is possible (a) DNA; (b, c) DDAB–DNA complexes at positive-to-negative charge ratios of to go above 100 °C, because the experiments can be (b) 0.75:1 and (c) 1.0:1; (d) DDAB liposomes alone; (e, f) distearoyltrimethylammonium propane (DSTAP) –DNA complexes at (±) ratios of (e) 0.75:1 and (f) 1.0:1; conducted under pressure, which will raise the boiling (g) DSTAP liposomes alone; (h, i) DOTAP–DNA complexes at (±) ratios of (h) 0.75:1 point of water. A complex third transition is also seen at and (i) 1.0:1. (Adapted with permission from Ref. 58.) 60–70 °C, which reflects the presence of contaminatthe equilibrium constant, stoichiometry (molecular ratio of the ing linear and open circular forms of DNA (Figure 4a). two components in the complex at saturation), enthalpy, and Both high-sensitivity ITC and DSC are now available in autoentropy of binding are determined. Unfortunately, the interac- mated modes; this makes them applicable to high-throughput tions between a plasmid DNA molecule and a CL particle are lipoplex analyses. complex; this makes the determination of thermodynamic paUV absorption. Because the absorption of UV light is rameters difficult, if not impossible. Irreversibility of binding, sensitive to changes in base pair interactions, the absorption aggregation, and structural changes imply a lack of equilibrium spectrum of DNA strongly reflects the degree of secondary and the occurrence of multiple events—situations that pro- structure formation. Recent technical advances in the use of hibit an unambiguous analysis of ITC data. Furthermore, this diode array spectrophotometers combined with computermethod is not applicable to preformed complexes, the desired aided data analysis now permit derivative spectra of ±0.01 nm targets of most stability studies. resolution to be easily obtained, thereby allowing signals from Another form of calorimetry, DSC, is of direct utility for the different types of bases to be resolved. Because of base pair stability analyses. The lipoplexes are heated simultaneously interactions that occur within the ordered secondary structure with a reference sample (buffer), and sufficient thermal energy of double-stranded DNA, UV absorption is less than that is supplied to maintain the solutions at the same temperatures. predicted solely on the basis of nucleotide concentrations (45). If a change in structure occurs in the target material, the com- As DNA undergoes a loss of secondary structure (melting), pensating energy required to maintain the same temperature UV absorption increases. Monitoring absorption as a function in the two samples can be converted to an energy associated of temperature is especially useful for the generation of DNA with the structural alteration. In the case where the transition melting curves. One can quickly and easily generate melting is reversible, the enthalpy of the structural alteration can be curves and measure the Tm of different lipoplexes over a range obtained. Although this condition is usually not the case, it is of charge ratios with a variety of automated instrumentastill possible to measure a “melting temperature” Tm, which is tion. For example, Zhang and colleagues showed that the CL related to the thermal stability of the complex. DDAB increased the Tm of double-stranded calf thymus DNA Two types of thermal transitions can occur within lipoplexes. by ~8 °C (59). The first corresponds to a transition from the gel phase to the Empirical phase diagrams (EPDs). The analysis and charliquid crystal phase; during this process, the ordered alkyl side acterization of lipoplexes are difficult because of their intrinsic chains of the CLs are converted at elevated temperature to a complexity and heterogeneity. In the case of better-defined more fluid state. The phenomenon may or may not be displayed pharmaceuticals, facile methods such as the various types of by a particular CL bilayer. The second type of transition is pro- HPLC are quite useful. Methods such as MS and NMR can duced by the melting of the two chains that make up the DNA provide very detailed structural pictures because of their high double helix. These two strands are held together primarily by information content. Their current applicability to lipoplexes hydrogen bonds between the bases as well as by interactions be- is, however, low. In contrast, lipoplexes require analysis by a tween their planar surfaces. Although this disruption of DNA wide variety of methods that probe various aspects of their structure is expected to occur at low temperatures (25–60 °C) structure and stability. Multiple measurements have the po7246
i
A N A LY T I C A L C H E M I S T R Y / O C T O B E R 1 , 2 0 0 7
150 125 100 75
(b) Ionic strength (mM)
Ionic strength (mM)
(a)
150 125 100 75
Ionic strength (mM)
Ionic strength (mM)
tential to provide high informa50 50 tion content, but until recently, nobody had attempted to syn25 25 thesize such complex data sets 10 10 into a comprehensive picture. 4 5 6 7 8 4 5 6 7 8 Such a method has recently been pH pH developed and applied to lipoplexes and other gene delivery (c) (d) systems (25). This approach was 150 150 originally developed to analyze 125 125 peptides and proteins and was 100 100 eventually applied to viruslike particles, viruses, and recently 75 75 intact bacteria (60–63). 50 50 In this method, the macromolecular entity or complex is 25 25 represented as a multidimen10 10 sional vector in which the com4 5 6 7 8 4 5 6 7 8 ponents of the vector are meapH pH surements made as a function of variables such as temperature, pH, and ionic strength. Thus, FIGURE 5. Ionic-strength–pH EPDs of various nonviral gene delivery complexes generated from DLS, an n-dimensional vector is cre- CD, and fluorescence studies. ated on a pH–temperature or (a, c) DOTAP lipoplex and (b, d) DOTAP–DOPE lipoplex at positive-to-negative charge ratios of (a, b) 0.5 and pH–ionic-strength plane, where (c, d) 4. Regions of similar color represent single phases, whereas the change of color defines the conditions the n-dimensional vector is com- under which the structure of the gene delivery complex alters. (Adapted with permission from Ref. 25.) posed of n measurements. The data from the various techniques are then normalized, per- at positive-to-negative charge ratios of 0.5 and 4.0 (Figure 5). mitting the techniques to be compared directly. Although The resultant 5D, colored EPDs display three to five varinormalization allows direct comparision of the different tech- ably resolved regions that correspond to lipoplexes displaying niques and prevents any one technique from dominating the structures with varying degrees of size, DNA collapse, and secresulting analysis, normalization of data that exhibit very little ondary structure. Interpretation of the colored regions within change results in the amplification of noise. the EPD, however, requires analysis of the original data (25). The resulting vector field is expressed as a matrix, which The light blue-green regions observed at pH 7 for DOTAP is then used to generate a density matrix. From the density lipoplexes (Figure 5c) and at pH 7 and 8 for DOTAP–DOPE matrix, the eigenvectors and eigenvalues of the data are de- lipoplexes (Figure 5d) at high charge ratios reflect changes in termined and used to transform the original vector field into the size of the lipoplexes. The largest region in each of the a new, orthogonal vector system. The three dominant eigen- EPDs appears at pH 5–8 at ionic strengths >25. Analysis of vectors are assigned the colors red, green, and blue, and the the fluorescence data reveals that the gradient of colors within transformed vector field is now represented as a combination this region is the result of differences in DNA condensation. of the three contributing color components. Each vector (i.e., CD measurements related to DNA secondary structure show the resultant color) is then plotted as a function of the vari- little change with increasing ionic strength over most of the ous stress variables. The resultant color maps then represent pH range. One exception, however, is the CLDCs at pH 4. At the various physical states present as regions of uniform color. this pH, the complexes undergo secondary structural changes These pictures are known as EPDs. over the range of ionic strength explored. These changes are In earlier studies using this approach, the most common reflected in the color gradients appearing in the pH 4 column variables were temperature and pH; a wide variety of experi- of the EPDs shown in Figure 5. Thus, a fairly comprehensive, mental methods such as CD, fluorescence, FTIR, and light intuitive picture of the complexes is produced. scattering were used to construct the vectors. In the cases of The use of this approach, however, has two important calipoplexes, pH and ionic strength were the most useful pertur- veats. First, the word “empirical” is deliberately chosen. EPDs bations because of the electrostatic nature of the complexes. should not be thought of as thermodynamic (equilibrium) The techniques used to construct the vectors were DLS (to phase diagrams, because reversibility across the “pseudo” (or measure the size of the complexes), CD (to characterize the apparent) phase boundaries (the regions of abrupt color secondary structure of the DNA), and the dye YOYO-1 (to change) is not implied. These are simply phenomenological monitor the topology of the DNA). The CL DOTAP in the representations of the behaviors of CLDCs as a function of presence and absence of the helper lipid DOPE was examined changing pH, ionic strength, and charge ratio. Nevertheless, O C T O B E R 1 , 2 0 0 7 / A N A LY T I C A L C H E M I S T R Y
7247
EPDs’ high content of information about lipoplex behavior is of broad utility. Second, the absolute colors cannot be compared between diagrams because of the data normalization procedure used. Such comparisons could be permitted, however, if all of the data were handled simultaneously, but this has not been done here. The analysis of CLDCs is in its infancy. No doubt, as such complexes, including those containing other cationic partners, are prepared in more homogenous, better defined forms, more rigorous approaches will become applicable. The intrinsic complexity of any such gene delivery vehicle can be expected to present continued challenges that will call for novel approaches yet to be imagined. C. Russell Middaugh is a professor and Joshua D. Ramsey is a postdoc at the University of Kansas. Middaugh’s research interests include proteins, gene therapy, biophysical chemistry, drug formulation and delivery, and vaccines. Ramsey’s research involves drug and gene delivery, protein engineering, and pharmaceutical formulation. Address correspendence about this article to Middaugh at University of Kansas, 2030 Becker Dr., Lawrence, KS 66047 (
[email protected]).
(29) (30) (31) (32) (33) (34) (35) (36)
(37) (38) (39) (40) (41) (42) (43) (44) (45)
(46)
References (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25) (26) (27) (28) 7248
Hollon, T. Am. J. Ophthalmol. 2000, 129, 701. Kohn, D. B.; Sadelain, M.; Glorioso, J. C. Nat. Rev. Cancer 2003, 3, 477–488. Gottschalk, S.; et al. Gene Ther. 1996, 3, 448–457. Boussif, O.; et al. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 7297–7301. Kukowska-Latallo, J. F.; et al. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 4897– 4902. Wolfert, M. A.; et al. Hum. Gene Ther. 1996, 7, 2123–2133. Felgner, P. L.; et al. Proc. Natl. Acad. Sci. U.S.A. 1987, 84, 7413–7417. Liang, H.; Harries, D.; Wong, G. C. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 11,173–11,178. Tousignant, J. D.; et al. Hum. Gene Ther. 2000, 11, 2493–2513. Dow, S. W.; et al. J. Immunol. 1999, 163, 1552–1561. Pedroso de Lima, M. C.; et al. Adv. Drug Deliv. Rev. 2001, 47, 277–294. Dass, C. R. J. Mol. Med. 2004, 82, 579–591. Wasungu, L.; Hoekstra, D. J. Controlled Release 2006, 116, 255–264. Ravanat, J. L.; et al. Chem. Res. Toxicol. 1995, 8, 1039–1045. Cui, L.; Zhu, L. Langmuir 2006, 22, 5982–5985. Radler, J. O.; et al. Science 1997, 275, 810–814. Koltover, I.; et al. Science 1998, 281, 78–81. Fang, Y.; Yang, J. J. Phys. Chem. B 1997, 101, 441–449. Sternberg, B.; Sorgi, F. L.; Huang, L. FEBS Lett. 1994, 356, 361–366. Huebner, S.; et al. Biophys. J. 1999, 76, 3158–3166. Lin, A. J.; et al. Biophys. J. 2003, 84, 3307–3316. Caracciolo, G.; et al. Langmuir 2007, 23, 4498–4508. Wiethoff, C. M.; et al. J. Biol. Chem. 2002, 277, 44,980–44,987. Wiethoff, C. M.; et al. J. Pharm. Sci. 2004, 93, 108–123. Ruponen, M.; Braun, C. S.; Middaugh, C. R. J. Pharm. Sci. 2006, 95, 2101– 2114. Mislick, K. A.; Baldeschwieler, J. D. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 12,349–12,354. Wiethoff, C. M.; et al. J. Biol. Chem. 2001, 276, 32,806–32,813. Smith, J. G.; et al. Pharm. Res. 1998, 15, 1356–1363. A N A LY T I C A L C H E M I S T R Y / O C T O B E R 1 , 2 0 0 7
(47)
(48)
(49) (50) (51) (52) (53) (54) (55) (56) (57) (58) (59) (60) (61) (62) (63)
Xu, Y.; et al. Biophys. J. 1999, 77, 341–353. Lesage, D.; et al. Biochim. Biophys. Acta 2002, 1564, 393–402. Farhood, H.; et al. Biochim. Biophys. Acta 1992, 1111, 239–246. Braun, C. S.; et al. Biophys. J. 2003, 84, 1114–1123. Choosakoonkriang, S.; et al. J. Pharm. Sci. 2003, 92, 1710–1722. Lobo, B. A.; et al. J. Pharm. Sci. 2003, 92, 1905–1918. Braun, C. S.; et al. J. Pharm. Sci. 2005, 94, 423–436. Choosakoonkriang, S.; et al. In Nonviral Vectors for Gene Therapy: Methods and Protocols; Findeis, M. A., Ed.; Humana Press: Totowa, NJ, 2001; Vol. 65, pp 285–317. Choosakoonkriang, S.; et al. J. Biol. Chem. 2001, 276, 8037–8043. Choosakoonkriang, S.; et al. J. Pharm. Sci. 2003, 92, 115–130. Wiethoff, C. M.; et al. J. Pharm. Sci. 2003, 92, 1272–1285. Förster, Th. Ann. Phys. 1948, 437, 55–75. Selvin, P. R. Nat. Struct. Biol. 2000, 7, 730–734. Madeira, C.; et al. Biophys. J. 2003, 85, 3106–3119. Zhang, Y.; et al. Biochim. Biophys. Acta 2003, 1614, 182–192. Braun, C. S.; et al. Biophys. J. 2005, 88, 4146–4158. Braun, C. S.; Kueltzo, L. A.; Middaugh, C. R. In Nonviral Vectors for Gene Therapy: Methods and Protocols; Findeis, M. A., Ed.; Humana Press: Totowa, NJ, 2001; Vol. 65, pp 253–284. Lobo, B. A.; et al. In Nonviral Vectors for Gene Therapy: Methods and Protocols; Findeis, M. A., Ed.; Humana Press: Totowa, NJ, 2001; Vol. 65, pp 319–348. Wiethoff, C.; Middaugh, C. R. In Nonviral Vectors for Gene Therapy: Methods and Protocols; Findeis, M. A., Ed.; Humana Press: Totowa, NJ, 2001; Vol. 65, pp 349–376. Ausar, S. F.; Joshi, S. B.; Middaugh, C. R. In Gene Therapy Protocols, 3rd ed.; Walker, J. M., Le Doux, J., Eds.; Humana Press: Totowa, NJ, 2007, in press. Singh, M.; Kisoon, N.; Ariatti, M. Drug Deliv. 2001, 8, 29–34. Moret, I.; et al. J. Controlled Release 2001, 76, 169–181. Bates, A. D.; Maxwell, A. DNA Topology, 2nd ed.; Oxford University Press: New York, 2005. Johnston, B. H.; Rich, A. Cell 1985, 42, 713–724. Palecek, E. Crit. Rev. Biochem. Mol. Biol. 1991, 26, 151–226. Prasad, T. K.; Gopal, V.; Rao, N. M. FEBS Lett. 2003, 552, 199–206. Matulis, D.; Rouzina, I.; Bloomfield, V. A. J. Am. Chem. Soc. 2002, 124, 7331–7342. Lobo, B. A.; et al. Arch. Biochem. Biophys. 2001, 386, 95–105. Pozharski, E.; MacDonald, R. C. Biophys. J. 2002, 83, 556–565. Lobo, B. A.; et al. J. Pharm. Sci. 2002, 91, 454–466. Zhang, Z.; et al. Biophys. Chem. 2002, 97, 7–16. Kueltzo, L. A.; et al. J. Pharm. Sci. 2003, 92, 1805–1820. Ausar, S. F.; et al. J. Biol. Chem. 2006, 281, 19,478–19,488. Ausar, S. F.; et al. Mol. Pharm. 2005, 2, 491–499. Zeng, Y.; Fan, H.; Joshi, S. B.; Middaugh, C. R. 2007, unpublished results.
