2.1. Dynamic Light Scattering
DNA and PAMAM dendrimers (generation 5) were acetylated to various extents, mixed in 10 mM NaBr solutions, and studied by dynamic light scattering. According to our titration experiments (data not shown here), the number of charges on non-acetylated G5 dendrimers is 114, instead of 128, the number of positive charges on a perfectly synthesized dendrimer [
17]. Therefore, assuming that defects that lead to the reduced charge occur randomly, the standard deviation of number of charges on a non-acetylated dendrimer can be approximated by:
, where N is the maximum number of charges possible, which equals 128, and N
k is the measured mean value of 114. If the acetylation of dendrimer terminal groups is also taken to be random, the standard deviation of the number of charges on an individual acetylated dendrimer can also be approximated by the equation above. Thus, the numbers of positive charges, and their standard deviations computed by propagation of errors using both sources of variation mentioned above, on dendrimers acetylated to 0%, 15%, 30%, 50%, 65%, and 85%, are 114 ± 3.5, 97 ± 5.2, 80 ± 6.0, 57 ± 6.5, 40 ± 6.2, 17 ± 4.0, respectively. The relaxation time distributions of pure dsDNA as well as DNA-dendrimer complexes were obtained from autocorrelation functions of scattering light intensities using analysis with CONTIN 2DP. Selected relaxation time distributions of DNA-dendrimer complexes are shown in
Figure 1 with scattering angle θ fixed at 50°. The apparent hydrodynamic radii of the DNA-dendrimer complexes were computed using the peaks of the relaxation time distributions along with the Einstein-Stokes equation, and plotted in
Figure 2. Note that
rcharge here and in the following discussion is defined as the ratio of the total number of positive charges on all dendrimers to the total number of negative charges on all DNA molecules in the solution, that is,
rcharge = [NH
3+]/[PO
4−]. All the primary amine groups (pK
a = 9.0 [
18] ~10.77 [
19]) were assumed to be protonated in pH 7~8 in this study. This assumption was supported by titration experiments and the observation that solutions become cloudy when
rcharge defined above is close to unity. In this paper, we limit ourselves to the cases with
rcharge less than 1 to study DNA-dendrimer complexes before phase separation takes place.
Figure 1.
Relaxation time distributions of PAMAM dendrimer/DNA samples measured at scattering angle θ = 50°. (a) non-acetylated G5/DNA. (b) 30% acetylated G5/DNA. (c) 50% acetylated G5/DNA. (d) 65% acetylated G5/DNA. (e) 85% acetylated G5/DNA.
Figure 1.
Relaxation time distributions of PAMAM dendrimer/DNA samples measured at scattering angle θ = 50°. (a) non-acetylated G5/DNA. (b) 30% acetylated G5/DNA. (c) 50% acetylated G5/DNA. (d) 65% acetylated G5/DNA. (e) 85% acetylated G5/DNA.
The relaxation time distribution of 0.15 mg/mL pure salmon sperm DNA (2000 ± 500 bp) in 10 mM NaBr solution is plotted in
Figure 1e, where
rcharge is 0. The two peaks of the DNA relaxation time distribution correspond to the internal (shorter τ) and translational (longer τ) modes of DNA. The apparent hydrodynamic radius of DNA computed based on the translational peak is 110 nm, in agreement with previous studies [
1,
6]. DNA concentrations for all dynamic light scattering measurements were fixed at 0.15 mg/mL, while the dendrimer concentration was varied to obtain the specified r
charge. The apparent hydrodynamic radius of the G5 PAMAM dendrimer was determined to be around 3 nm by DLS, indicating that aggregation of dendrimer in the absence of DNA was negligible.
As shown in
Figure 1a, non-acetylated G5 dendrimer-DNA complexes give sharp peaks, indicating that the complexes size distribution is relatively narrow. Comparing the positions of these peaks to those of the pure DNA, we can conclude that the former, which correspond to faster relaxation, and therefore smaller objects, represent condensed DNA-dendrimer complexes. It should be noted that values of
rcharge of 0.28, 0.57, 0.86, and 1.14 correspond to
rmolar values of 10, 20, 30, and 40, where
rmolar is defined by [dendrimer]/[DNA]. The decay time distributions of the acetylated denrimer-DNA complexes with the same molar ratios (
rmolar) are plotted in
Figure 1b–e, except for the pure DNA curve at the bottom of
Figure 1e. For the 30% acetylated dendrimer (
Figure 1b) and 50% acetylated dendrimer (
Figure 1c), the decay time distributions are much broader than for the non-acetylated dendrimer in
Figure 1a, probably because the number of positive charges on the acetylated dendrimers varies somewhat from dendrimer to dendrimer, as estimated in the standard deviations in numbers of charges given above. For the dendrimers with the highest acetylation ratios (65%, 85%), the relaxation time distributions exhibit two peaks, the shorter of which is presumably contributed by internal motion within the DNA complex, while the slower mode corresponds to translational motion of the complexes. However, the slower mode for 85% acetylated dendrimer-DNA is almost the same as for pure DNA, indicating that the weakly charged dendrimer (~15 charges per dendrimer) was not able to condense the DNA significantly. This result agrees with a previous study of DNA/Poly-L-lysine [
3], which showed that a transition of the DNA-polycation complex conformation occurs when the number of charges on the polycation is decreased. Highly charged polycations can bend and condense dsDNA while weakly charged polycation only attach to it without producing compaction.
The relaxation time distributions of acetylated dendrimer-DNA complexes with higher molar ratios (above 40) are not presented here, since for these cases a small fraction of the most highly charged dendrimers from the polydisperse charge distribution can condense the DNA even when the average charge per dendrimer is low and this small fraction dominates the dynamic light scattering signals.
The apparent hydrodynamic radii of the DNA-dendrimer complexes are plotted in
Figure 2. For highly charged dendrimers (0%, 30%, 50% acetylated), the hydrodynamic radii of the complexes are around 50 nm. Therefore, this level of acetylation of primary amine groups on the PAMAM dendrimer does not significantly reduce its ability to condense DNA, despite the lessening of the dendrimer charge, which is needed for condensation [
14]. Since reduction of the charge density on the dendrimer can reduce the cytotoxicity of dendrimer significantly [
12], partial acetylation of the primary amine groups of PAMAM dendrimer might have significant potential in gene therapy. For highly charged dendrimers, as
rcharge increases, the apparent hydrodynamic radius of the complexes also increases (
Figure 2a), which is likely the result of formation of large complexes containing more than one DNA chain. However, for weakly charged dendrimers (65% acetylated), the complex radius decreases when
rcharge is increased up to 0.5, as the weakly charged dendrimer at low concentration binds to DNA without compaction, but compaction apparently increases with increasing numbers of bound dendrimers per DNA molecule. Based on our dynamic light scattering experiments, to condense DNA in the gene delivery process, PAMAM dendrimers with around half of their primary amine groups acetylated might be a good choice. To determine whether the polydispersity of charges on acetylated dendrimer will affect the gene delivery, transcription experiments should be performed.
Figure 2.
Apparent hydrodynamic radii of dendrimer/DNA complexes measured at θ = 50°. (a) non-acetylated G5/DNA. (b) 30% acetylated G5/DNA. (c) 50% acetylated G5/DNA. (d) 65% acetylated G5/DNA. Some typical error bars are given.
Figure 2.
Apparent hydrodynamic radii of dendrimer/DNA complexes measured at θ = 50°. (a) non-acetylated G5/DNA. (b) 30% acetylated G5/DNA. (c) 50% acetylated G5/DNA. (d) 65% acetylated G5/DNA. Some typical error bars are given.
2.2. Steady-State Fluorescence Spectroscopy
To determine the fraction of free DNA in the DNA-dendrimer mixture, steady-state fluorescence spectroscopy experiments were carried out 10 min after mixing DNA-dendrimer complex solutions with nucleic acid stain GelStar
®. The excitation wavelength was fixed at 493 nm, and emission light intensity was recorded at 527 nm. The normalized emission light intensity of dendrimer-DNA complexes for various dendrimer concentrations are plotted in
Figure 3.
Figure 3.
Emission light intensity (527 nm) versus rcharge for DNA condensed by dendrimers with various acetylation ratios. DNA-0% acetylated dendrimer (squares); DNA-15% acetylated dendrimer (circles); DNA-25% acetylated dendrimer (diamonds); DNA-50% acetylated dendrimer (up triangles); DNA-70% acetylated dendrimer (down triangles).
Figure 3.
Emission light intensity (527 nm) versus rcharge for DNA condensed by dendrimers with various acetylation ratios. DNA-0% acetylated dendrimer (squares); DNA-15% acetylated dendrimer (circles); DNA-25% acetylated dendrimer (diamonds); DNA-50% acetylated dendrimer (up triangles); DNA-70% acetylated dendrimer (down triangles).
GelStar
® was assumed to bind only to free DNA, that is to portions of the DNA not blocked by dendrimer [
1]. After reacting GelStar
® with DNA, the emission light intensity increased dramatically although the emission from free GelStar
® was negligible. Previous studies [
1,
6] assumed that GelStar
® is not able to react with dendrimer-bound DNA. This is validated by our non-acetylated G5-dendrimer results, for which the normalized emission light intensity is roughly equal to (1-
rcharge); see
Figure 3. Therefore, GelStar
® only binds to the free DNA segments which have not been neutralized by dendrimer, since the interaction between DNA and GelStar
® is also dominated by electrostatic force. However, the relation
I/I0 = (1 −
rcharge) does not hold for DNA-acetylated dendrimers. This is apparently because the binding affinity of acetylated dendrimers to DNA decreases as the acetylation ratio increases. One implication is that the DNA segments condensed by dendrimer with fewer charges might still be accessible to proteins.
2.4. Molecular Combing Assay
Linearized λ-DNA has a hydrophobic 12-base overhang on each end, which allows it to stick and anchor at each end to a polystyrene-coated cover glass [
22,
23]. When the PS coated cover glass is pulled out from DNA solution, the YOYO-1 stained free DNA molecules are aligned on the cover glass surface by the high air-water surface tension, and then are visualized by a fluorescence microscope with blue excitation (
Figure 5b). We also fluorescently labeled dendrimers with amine-reactive tetramethylrhodamine isothiocyanate (TRITC) dye molecules. TRITC-labeled PAMAM dendrimer-DNA complexes were visualized by sequentially imaging DNA and dendrimer molecules using blue and green excitation, respectively. The overlays of DNA and dendrimer images are shown in
Figure 5c. Since the λ-DNA molecules were condensed by dendrimer, even the high surface tension (up to 4.0 × 10
−10 N) is not able to stretch the DNA out to its full contour length (16 μm). Thus, we are able to image the condensed form of λ-DNA directly. Here, we only present images of dendrimer-DNA complexes at low molecular ratio (100 dendrimers per DNA) where the dendrimer-DNA complexes are somewhat condensed but do not form dense globular particles, and a few complexes stick to the PS surface so that we can image them. For dendrimer-DNA complexes with higher molecular ratio (for example, around 1,000 dendrimer/DNA), the complexes do not stick to the surface, possibly because the two hydrophobic ends of the λ-DNA at these ratios are trapped within the denser complex particle, preventing the DNA from binding to the PS surface.
Figure 5.
(a) Experimental setup for molecular combing. (b) Immobilized, aligned, YOYO-1 stained λ-DNA on PS surface. (c) λ-DNA/G5 PAMAM dendrimer (TRITC labeled) complexes deposited on PS coated cover glass surface (green: λ-DNA, red: PAMAM dendrimer).
Figure 5.
(a) Experimental setup for molecular combing. (b) Immobilized, aligned, YOYO-1 stained λ-DNA on PS surface. (c) λ-DNA/G5 PAMAM dendrimer (TRITC labeled) complexes deposited on PS coated cover glass surface (green: λ-DNA, red: PAMAM dendrimer).
In the dendrimer-DNA complexes, as shown in
Figure 5c, the dendrimer molecules concentrate along some regions of the DNA, while other regions remain dendrimer free. This is consistent with a previous study [
1] in which dendrimer-bound DNA was shown to coexist with dendrimer-free DNA. These earlier results were interpreted as evidence that the dendrimer molecules bind to the DNA in a “cooperative” manner, in which dendrimers have higher binding affinity to DNA with dendrimers already attached to it than to bare DNA. In what follows, we suggest that instead of, or along side of, cooperative binding, that irreversible diffusion-limited binding might help account for these results. In any event, the dendrimer-DNA complexes, once formed, are strong enough to resist unraveling during combing, implying a resistance to a force of around 500 pN per complex [
23].
We also incubated λ-DNA molecules aligned on the PS surface with a TRTIC-labeled G5 dendrimer solution, expecting dendrimers might slide one dimensionally along the DNA deposited on the surface, since a previous study [
24] revealed that electrostatic forces were sufficient to confine the charged nanoparticle in one dimension. However, rather than experiencing 1D diffusion along DNA as some DNA-binding proteins do, G5 dendrimers either stick to the DNA or the PS-coated surface or performed simple 3D Brownian diffusion (data not shown). The failure to undergo 1D diffusion along DNA might be due to the high binding energy [
14] of the irreversible interaction of charged dendrimers with DNA [
11].
2.5. Flow Stretching Assay
After biotinylated λ-DNA with one end attached to a neutravidin monolayer was deposited on the flow channel surface, the YOYO-labeled DNA was stretched by a shear flow. Based on parabolic flow assumption, at the flow rate imposed, the shear rate at the free end of the DNA, which based on the DNA coil size is estimated to be about 1 µm above the cover glass surface, is around 300 s−1 , which is able to stretch the DNA almost to its full contour length.
A TRITC-labeled G5 dendrimer (non-acetylated) solution was then introduced into the flow cell to interact with the tethered λ-DNA. The binding process was visualized and recorded by fluorescence microscopy using green excitation. We observed that the dendrimer bound and decorated the entire λ-DNA molecule, as opposed to the partial binding we observed in the molecular combing assay. When the λ-DNA is fully stretched out, each segment of the DNA apparently has almost the same probability of binding a dendrimer molecule. Apparently because the binding is irreversible and the well-mixed dendrimer solution is continuously injected into the flow channel, the dendrimers can cover the whole DNA chain evenly. After formation of the dendrimer-DNA complex, we used TE buffer to wash out the flow channel. No significant fluorescence decrease was observed on the tethered DNA until photocleavage or neutravidin detachment occurred, usually around 5~10 min after illumination, which confirmed that dendrimer attached to DNA irreversibly.
Figure 6c,d show the length distributions of both free DNA and dendrimer-DNA complexes. 77 λ-DNA lengths as well as 62 dendrimer-DNA complexes lengths were recorded for these distributions. The error bar was approximated using equation:
, where
Nk is the number of data points of
kth bin and
N is the total number of data points. The lengths distributions are wide, which is mainly due to the photocleavage of λ-DNA, which shortens some of the DNA molecules before imaging is complete. However, these two histograms indicate that flow-stretched tethered DNA molecules are condensed by G5 dendrimer bound to the DNA chains, agreeing with optical tweezer experiments [
11].
The interaction between flow-stretched λ-DNA and TRITC-labeled G5 PAMAM dendrimers with 50% primary amine groups acetylated were studied using the same procedures. Because the neutravidin was physically attached to the cover glass, it could easily detach the surface under high drag force. Thus, we only imaged the DNA in flow for less than 30 min. Within this time, binding of 50% acetylated dendrimer to DNA was not observed. This indicates a significantly reduced binding affinity upon partial acetylation of the dendrimer, in agreement with our dynamic light scattering results as well as fluorescence spectroscopy results discussed above.
Figure 6.
(a) Experimental setup for imaging dendrimer binding to flow-stretched DNA with one end tethered to the surface. (b) Green: tethered YOYO-labeled λ-DNA in flow; red: tethered TRITC-labeled dendrimer-DNA complex in flow. (c) Distribution of free λ-DNA molecules lengths. (d) Distribution of dendrimer-DNA complex lengths.
Figure 6.
(a) Experimental setup for imaging dendrimer binding to flow-stretched DNA with one end tethered to the surface. (b) Green: tethered YOYO-labeled λ-DNA in flow; red: tethered TRITC-labeled dendrimer-DNA complex in flow. (c) Distribution of free λ-DNA molecules lengths. (d) Distribution of dendrimer-DNA complex lengths.