Quinine copolymer reporters promote efficient intracellular DNA delivery and illuminate a protein-induced unpackaging mechanism

QCR and Control Polymer Synthesis.

We sought a facile, one-pot reaction to incorporate quinine into a polymer using inexpensive, commercially available starting materials. Quinine contains several functional handles that enable it to be incorporated into a polymeric scaffold, including its secondary hydroxyl (42⇓–44), tertiary amine (45⇓–47), and α-olefin groups (48⇓⇓⇓–52) (Fig. 2A). We first pursued development of reaction conditions that would promote the direct incorporation of quinine into a polymeric backbone via radical propagation of its α-olefin group. While degradative chain transfer and steric hindrance of the bulky quinuclidine ring were necessary challenges to overcome (52⇓–54), the advantage to a direct approach is threefold: 1) Quinine can be used directly in a copolymerization reaction to enable a rapid, inexpensive, and scalable production method; 2) the quinuclidine amine can be protonated, aiding electrostatic interactions with DNA; and 3) the quinoline ring is distal from the backbone, which allows these polymer pendant groups to effectively intercalate into DNA without steric hindrance and allows reporting of binding and release via microscopic and spectroscopic methods.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Synthesis and characterization of QCRs and controls. (A) Free-radical copolymerization scheme used in this study to incorporate quinine, consisting of a bicyclic quinuclidine head group and fluorescent quinoline ring (λ_ex = 350 nm, λ_em = 450 nm), into copolymers with comonomers such as HEA (blue). (B) Structural properties of polymers used in biological studies as determined by ¹H NMR and SEC. (C) Quinine was copolymerized with various acrylamides including HEAm, and HEA was copolymerized with canonical cationic comonomers such as DMAEMA.

We found that while quinine cannot be directly homopolymerized (51), quinine was active to copolymerization with more reactive comonomers, and we fully characterized the reactivity ratios with several comonomers to enable tuning of chemical composition. We copolymerized quinine with several hydrophilic acrylate- and acrylamide-based comonomers that have not been copolymerized with quinine previously including HEA, 2-(hydroxyethyl) acrylamide (HEAm), acrylamide (Am), N,N-dimethyl acrylamide (DMAm), and N-isopropylacrylamide (NIPAm) (Fig. 2 B and C and SI Appendix, Fig. S1). The quinine copolymers we created contained up to 17% quinine by molarity, as determined by ¹H NMR (SI Appendix, Figs. S3–S7), and polymers were obtained that ranged between 9 and 22 kDa in molar mass (Fig. 2C and SI Appendix, Fig. S8 and Table S1). The reactivity ratios of quinine with comonomers HEA, HEAm, and Am were monitored via ¹H NMR (SI Appendix, Fig. S9). The reactivity ratios for all three comonomers were found to be r₁ < 0.2 and r₂ > 10. These values indicate that quinine is statistically incorporated within the polymer chains as isolated units spaced by hydrophilic repeat units. These hydrophilic comonomers acted as spacer units between quinine repeat units, which improves the solubility and reduces charge density and toxicity of cationic polymers, as shown in polyethyleneimine (PEI) derivatives and carbohydrate-based polycations (9, 55⇓–57). The architecture was further confirmed via a combination of size-exclusion chromatography (SEC), differential refractive index, ultraviolet absorption spectroscopy, and static light scattering analysis showing that quinine was equally distributed along each copolymer chain and that all polymer chains contained quinine (not a mixture of two homopolymers) (SI Appendix, Fig. S8). The molar masses, dispersities, and architectures of the synthesized copolymers are comparable to other effective polymeric delivery agents (58⇓–60). Overall, this direct approach to successfully propagate quinine via copolymerization is simple, inexpensive, reproducible, and readily scalable, making it amenable for numerous fundamental and translational research endeavors.

Quinine Facilitates Enhanced Binding to DNA in Polyplexes.

Our previous work shows that monomeric quinine interacts with DNA electrostatically via the quinuclidine moiety, as well as through intercalation via the quinoline ring (26). We hypothesized that the quinine copolymers would efficiently bind and compact pDNA and form self-assembled polyplexes in solution via both binding modes, allowing for protection against degradation and efficient cellular delivery. Electrophoretic mobility shift assays demonstrate that all of the quinine copolymer variants were able to bind pDNA (SI Appendix, Fig. S11) at a range of formulation N/P ratios as low as N/P = 0.5, which is defined as the molar ratio of quinuclidine amines (N) to DNA phosphate groups (P). Dynamic light scattering (DLS) showed that quinine copolymers self-assemble with plasmids in solution to form positively charged polyplexes between roughly 80 and 200 nm in hydrodynamic diameter (d_h) (SI Appendix, Figs. S12 and S13). Like many transfection reagents, we found that poly(quinine-co-HEA) polyplexes aggregated upon dilution in serum-free cell media due to deprotonation at neutral pH (SI Appendix, Figs. S10, S12, and S13) (60, 61). Interestingly, we discovered that the aggregation of poly(quinine-co-HEA) polyplexes upon addition of serum-free Dulbecco’s modified Eagle’s medium (DMEM) could be controlled as the mean polyplex size (d_h) increases monotonically to ∼1 µm within 60 min (Fig. 3E). Indeed, polyplex aggregation has been shown to promote transfection by increasing sedimentation of DNA, which concentrates the polyplexes at the cell surface and increases the amount DNA that is internalized (62⇓⇓–65). This control in particle size can be used to directly tune and improve biological efficacy (discussed below).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Characterization of QCR–DNA binding. (A) Dye-exclusion assay monitors pDNA compaction in polyplexes (N/P = 6) formed in acidic aqueous media (pH 3 to 4). The fluorescence of PicoGreen intercalation in polyplexes is normalized to the fluorescence from free pDNA. (B) The polyplex solution in A is diluted first in cell media (DMEM) followed by addition of FBS (10% vol/vol). The change in relative PicoGreen fluorescence upon sequential addition of solutions is used to calculate the percent decrease in compact pDNA between steps. (C) (i) Raman spectrum of unbound poly(quinine-co-HEA) polymer under acidic solution conditions and (ii) difference spectrum highlighting the spectral shift that occurs in the quinoline ring mode of poly(quinine-co-HEA) due to DNA intercalation. The difference in the quinoline ring mode frequency (3 cm⁻¹) between intercalated and deintercalated quinine can be used to monitor poly(quinine-co-HEA) polyplex unpackaging. (D) Schematic of polyplex aggregation in serum-free DMEM prior to exposure to cells (defined as formulation time). (E) Plot showing the hydrodynamic diameter of poly(quinine-co-HEA) polyplexes (N/P = 8) over time after addition of serum-free DMEM. Data for A, B, and E are represented as the mean ± SD (n = 3).

Analogous copolymer controls were created that replaced quinine with canonical amines, such as 2-(dimethylamino)ethyl methacrylate (DMAEMA) (57, 65) (Fig. 2A and SI Appendix, Fig. S2) to remove the intercalation ability for comparison. Using a dye-exclusion assay, we found that when quinine was replaced with DMAEMA in a HEA copolymer, the ability of the copolymer to compact pDNA was completely inhibited (Fig. 2A and SI Appendix, Fig. S14). The QCRs maintained more pDNA compaction upon dilution in serum-free media compared to polyplexes formed with the DMAEMA homopolymer (Fig. 2B and SI Appendix, Figs. S14 and S15). This result shows that quinine has exceptional DNA binding and compaction properties compared to canonical amine-containing monomers such as DMAEMA, which is likely due to the dual binding mechanism of electrostatics and intercalation.

The mechanism underlying the exceptional compaction of pDNA by quinine in poly(quinine-co-HEA) was further investigated using Raman spectroscopy. We previously proved that the quinoline ring symmetric stretching mode (referred to as the quinoline ring mode henceforth) quantitatively reports on the local interactions of quinine with its chemical environment (26). In the case of monomeric quinine, the quinoline ring mode characteristically upshifts in frequency upon deintercalation from DNA due to the loss of π-stacking interactions with nucleobases (SI Appendix, Table S3). To examine the binding of quinine to DNA in the copolymers, we measured the Raman spectra of poly(quinine-co-HEA) polyplexes under similar pH conditions as the DLS and dye-exclusion assays. To highlight the changes that occur due to binding, the spectral contributions of free pDNA and poly(quinine-co-HEA) polymers not bound to DNA were subtracted (SI Appendix, Fig. S26). The resulting difference spectrum (Fig. 3C, ii) shows a prominent positive feature at 1,386 cm⁻¹ that derives from the quinoline ring mode and is downshifted 3 cm⁻¹ from the same band in the unbound polymer solution (Fig. 3C, i and SI Appendix, Table S3). This frequency shift, therefore, quantitates the intercalation of quinine into pDNA and can be used as a spectroscopic marker to monitor the degree of polyplex packaging and unpackaging. Taken together, the dye exclusion and Raman spectroscopic results indicate that quinine binds to DNA through both electrostatic interactions with the phosphate backbone and π-stacking with the nucleobases, which contributes to the overall stability of the polyplex formulations. Collectively, the confluence of these characteristics confers copolymers containing quinine with considerable compaction capabilities compared to canonical cationic amine-containing constituents.

Optimization of Delivery Performance.

We first screened our library of QCRs for their ability to deliver plasmids encoding for either luciferase or ZsGreen (a green fluorescent protein) to HeLa cells. We determined transfection efficiency by either measuring the total output of protein by a culture (luciferase) or by measuring the percentage of cells positive for ZsGreen. Interestingly, when quinine was copolymerized with an acrylamide, such as HEAm, Am, NIPAm, or DMAm, the resulting polyplexes were unable to achieve transgene expression (SI Appendix, Fig. S16). Poly(quinine-co-HEA), however, showed exemplary transfection efficiencies with both reporter plasmids and provided a 58% increase in ZsGreen expression efficiency as compared to the gold-standard commercial reagent Lipofectamine 2000 (SI Appendix, Fig. S16). Upon finding a hit comonomer type, we determined the optimal molar ratio of quinine needed for transfection using a luciferase-based transfection screen with a small library of poly(quinine-co-HEA) copolymers containing between 3 and 17% quinine by molarity (SI Appendix, Table S2). We found that the variant containing 14% quinine by molarity yielded at least threefold higher gene expression compared to the other variants (SI Appendix, Fig. S16).

Focusing on optimizing biological performance with the variant of poly(quinine-co-HEA) containing 14% quinine, we looked to our previous discovery of polyplex size control (Fig. 3 C and D). By adding the polyplex solution to cells at specific periods along the aggregation time course, we could control the average diameter of polyplexes that the cells were exposed to during the transfection experiments and correlate its efficiency in transfection to particle size (Fig. 4 A–C). Using quinine’s endogenous fluorescence (SI Appendix, Fig. S18), we used fluorescence microscopy to measure and track particles of various sizes in the intracellular space (Fig. 4 C and D and SI Appendix, Figs. S19 and S20). The calculated mean diameters of the internalized particles in the wide-field images were similar to the hydrodynamic diameters measured via DLS for the corresponding formulation time (SI Appendix, Fig. S21A). In addition, microscopy verified that increased cell incubation times led to an increase in internalized particles (SI Appendix, Fig. S21B). We measured the transfection efficiency with ZsGreen for each timeframe and observed that both particle size and number of internalized particles positively correlated to ZsGreen expression efficiency as well as cytotoxicity (Fig. 4E and see SI Appendix for details). Indeed, a fine balance between transfection performance and cytotoxicity in this system exists and can be tailored by controlling the incubation periods that affect both polyplex size and number of particles internalized by cells. This tight control allows for differential transgene delivery to target cell populations without introducing variability into cargo load.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Fluorescence of intracellular polyplexes containing quinine and size-dependent activity. (A) After the “formulation time” (Fig. 3D) to achieve the desired mean particle diameter, the poly(quinine-co-HEA) polyplexes were incubated with HeLa cells for a defined time (defined as “cell incubation time”). (B and C) Images of fixed HeLa cells 48 h posttransfection with pZsGreen and poly(quinine-co-HEA) (N/P = 8) at (B) 4× magnification with poly(quinine-co-HEA) (blue) and (C) 40× magnification. (D) Images of HeLa cells fixed 6 h posttransfection with poly(quinine-co-HEA) at various formulation and cell incubation times. The sample names [i, ii] are derived from (i) formulation time (Fig. 3D) and (ii) cell incubation time (Fig. 4A). The left image in each pair is an overlay of transmission and polymer (blue), and the right image is of polymer only. (Scale bars, 10 µm.) (E) Bar graph showing the percent transfection efficiency of live cells (as determined by flow cytometry) and relative cell viability (as determined by CCK-8 cell counting kit) 48 h posttransfection. (F) HeLa cells transfected with the conditions shown for sample [30,15] in E were incubated with endocytosis inhibitors. Incubation with DMA (macropinocytosis inhibitor) gives a statistically significant reduction in transfection indicating macropinocytosis contributes to successful transfection of aggregated polyplexes. Data for E and F are represented as the mean ± SD (n = 3); *P < 0.05.

To understand why large particles enabled higher transfection efficiencies, we investigated which internalization mechanism was the greatest contributor to transgene expression of large particles by inhibiting different modes of endocytosis with small-molecule inhibitors. We probed clathrin-mediated endocytosis, caveolae-dependent endocytosis, and macropinocytosis with the inhibitors amantadine, fillipin III, and 5-(N,N-dimethyl)amiloride hydrochloride (DMA), respectively (Fig. 4F). A significant decrease in transfection efficiency for DMA-treated cells suggests that macropinocytosis contributes as an internalization route of poly(quinine-co-HEA) polyplexes that leads to gene expression. Aggregation of the polyplexes may allow the polyplexes to bypass unproductive endocytic routes limited to smaller size capacities (66, 67).

Efficient Transfection in Multiple Cell Types.

Efficient transgene delivery is a challenge in many cell-focused investigations, particularly in experiments involving primary or transfection-resistant cell types important for research and cell therapy development. To explore the utility of the QCR poly(quinine-co-HEA) variant containing 14% quinine in overcoming this barrier in other cell types, we identified the N/P ratios that balanced efficient delivery of a ZsGreen fluorescent reporter plasmid with minimal cytotoxic effects. Excellent delivery performance with poly(quinine-co-HEA) was achieved with transfection-amenable adherent cell types such as HEK 293T (Fig. 5A and SI Appendix, Fig. S22A) and HeLa (Figs. 4 B–F and 5A). In HEK 293T cells, we observed high transfection efficiency (≥92% ZsGreen+) at N/P ≥6 (Fig. 5A and SI Appendix, Fig. S22A), which is the minimum N/P ratio needed for full compaction of DNA (SI Appendix, Figs. S12 and S14). In comparison, Lipofectamine 2000 demonstrated a lower transfection efficiency (48%) but higher relative cell viability (93%; Fig. 5C and SI Appendix, Fig. S22B). We next assessed its application in a suspension cell type (lymphoblast line K562) by increasing complex dose to account for decreased surface area in suspension cells. We found that poly(quinine-co-HEA) (N/P = 6 and 8) transfection efficiencies (5.1 and 5.8%, respectively) in K562 cells exhibited a threefold improvement over Lipofectamine 2000 (1.8%) (Fig. 5B and SI Appendix, Fig. S23). Indeed, these results support previous work that sedimentation of polyplexes onto the cell surface (Fig. 4) aids optimal transgene delivery.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

QCR variant of poly(quinine-co-HEA) containing 14% quinine efficiently transfects a variety of cell types. (A and B) The transfection efficiency in the delivery of pZsGreen (4.7-kb plasmid) to adherent cell lines (HeLa and HEK 293T), keratinocytes, and suspension cell line K562 was improved significantly by using poly(quinine-co-HEA) compared to Lipofectamine 2000 as determined by flow cytometry 48 h posttransfection. (C) The cell viability of HEK 293T cells was determined 48 h posttransfection via CCK-8 assay. (D) Efficient transfection of keratinocytes using a large plasmid (10-kb, pZsGreen) was maintained using poly(quinine-co-HEA). Data in A–D are represented as the mean ± SD (n = 3).

We next tested the efficacy of poly(quinine-co-HEA)–mediated gene delivery in keratinocytes, a transfection-resistant cell type that is a vital component in the upper layer of the skin and important to the modeling and treatment of genetic skin disorders such as epidermolysis bullosa. Gene editing of keratinocytes, performed for autologous transplants to repair diseased and damaged skin, have primarily relied on viral vectors due to their high transfection efficiency (68⇓–70). Viral vectors, however, are severely restricted in cargo capacity (i.e., adeno-associated viruses are limited to ∼4.7 kb of cargo), which limits their ability to deliver large therapeutic proteins or gene editing machinery. In addition, common nonviral alternatives have proven to be inefficient in transfecting keratinocytes. We performed a ZsGreen transfection in this cell type (Fig. 5A) with Lipofectamine 2000 and found poor transfection efficiencies (<1% ZsGreen+) that corroborated the results seen in literature (0 to 4% transfection efficiency) (71⇓–73). Efficient transfection could be achieved, however, with poly(quinine-co-HEA) (58% ZsGreen+ for N/P = 8). Interestingly, we also performed transfection with a plasmid double in size (10 kb; Fig. 5D), which yielded 43% ZsGreen+ cells after transfection with poly(quinine-co-HEA) as compared to only 1.2% for Lipofectamine 2000. These results demonstrate the versatility of this vehicle for diverse and transfection-resistant cell types as well as large gene payload sizes relevant to numerous challenging diseases (i.e., dystrophin for muscular dystrophy treatment and collagen for genetic skin diseases).

HEA Facilitates Decompaction of DNA in the Presence of Protein.

We hypothesized that the interplay between polymer binding to DNA and release of the cargo plays an important role in the excellent performance of the QCR poly(quinine-co-HEA). It is known that, for in vivo applications, protein binding can cause premature unpackaging of polyplexes (61). Consistent with this, we observed that titrating increasing amounts of fetal bovine serum (FBS) into poly(quinine-co-HEA) polyplex solutions prior to transfection leads to a dose-dependent decrease in transgene expression (SI Appendix, Fig. S24). We recognized, however, that for important ex vivo cell therapy applications, serum-free media can be used during transfection. Thus, exposure to protein can be limited to the intracellular space (after endocytosis), which could be an important factor to stimulate the unpackaging of polyplexes. To examine the influence of protein binding, we added FBS to polyplex solutions diluted in cell media and monitored their unpackaging using a dye-exclusion assay (Fig. 3B). Upon addition of FBS, the pDNA bound by poly(quinine-co-HEA) decomplexed by over threefold more than poly(DMAEMA). When the acrylate in the copolymer (HEA) was replaced with an acrylamide analog (HEAm), this dramatic decomplexation by FBS was eliminated. This result indicates that the acrylate in poly(quinine-co-HEA) facilitates polymer binding to protein, which also leads to pDNA unpackaging. This finding is supported by DLS (see SI Appendix, Fig. S12B for details) and by work from Zhao et al., who observed higher levels of serum binding to the HEA polymer brushes compared to the acrylamide analog due to decreased surface hydration (74⇓–76). This investigation supports the hypothesis that while quinine is needed for efficient binding of pDNA outside the cell, the HEA comonomer facilitates pDNA release from polyplexes upon exposure to intracellular protein.

Raman Imaging Verifies Protein-Induced Unpackaging within Cells.

The utility of the inherent reporting properties of poly(quinine-co-HEA) is highlighted by confocal Raman microscopy tracking experiments that quantify quinine deintercalation and further support the hypothesis that proteins facilitate polyplex unpackaging inside cells. Using Raman microscopy to monitor quinine–pDNA binding within poly(quinine-co-HEA) polyplexes, we were able to correlate the presence of protein within the polyplexes to pDNA unpackaging during the transfection process. HeLa cells were treated with poly(quinine-co-HEA) polyplexes and fixed for Raman hyperspectral imaging 4, 24, and 48 h after transfection. Each pixel in these images contained a Raman spectrum that roughly spanned a region from 900 to 1,800 cm⁻¹ (Fig. 6). These hyperspectral images were analyzed using principal component analysis (PCA) in order to identify spectroscopic signatures of polyplex dissociation (SI Appendix, Figs. S27 and S28). All of the images that were analyzed contained at least one principal component (PC) whose loading vector showed the quinoline ring mode at 1,369 cm⁻¹ (Fig. 7 A and B), indicative of the quinine moieties in poly(quinine-co-HEA) intercalating into DNA. However, in the 48-h posttransfection images shown in Fig. 6C, we identified PCs whose loading vectors showed quinoline ring modes that were upshifted to 1,372 cm⁻¹ (Fig. 7 C and D), which signaled the deintercalation of quinine from pDNA. The PC score maps in Fig. 7 A and C were used to determine the percent deintercalation of quinine moieties (SI Appendix, Fig. S29), which allowed for the degree of polyplex unpackaging to be quantified (Fig. 7F). On average, we observed that roughly 10 to 30% of quinine moieties were deintercalated from DNA in the polyplexes shown in Fig. 6C. However, our analysis allowed us to observe unpackaging heterogeneities not only between but also within the particles themselves. For example, while many particles exhibited a fairly uniform unpackaging landscape, some particles showed most of their unpackaging on the exterior, giving rise to ring-shaped patterns (Fig. 7F, Inset). The presence of these ring-shaped patterns suggests that the unpackaging of the DNA suffuses from the outside in, akin to surface erosion.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

Representative Raman images of HeLa cells and QCR polyplex particles after (A) 4 h, (B) 24 h, and (C) 48 h posttransfection. Cells (green) were visualized using the integrated intensity of the protein Amide I band (1,660 cm⁻¹), while polyplex particles (magenta) were visualized by the quinoline ring mode of quinine (1,369 cm⁻¹). The overlay of these images shows the presence of both intracellular and extracellular particles at all timepoints. (Scale bars, 5 µm.)

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.

Results from PCA of Raman hyperspectral images for HeLa cells 48 h posttransfection. (A) PC 1 score map showing the relative concentration of quinine moieties in poly(quinine-co-HEA) polyplexes that are DNA-intercalated. (B) PC 1 loading vector showing the 1,369-cm⁻¹ quinoline ring stretching mode spectral signature indicative of quinine–DNA intercalation. (C) PC 3 score map showing the relative concentration of quinine moieties in polyplexes that are deintercalated from pDNA. (D) PC 3 loading vector showing the 1,372-cm⁻¹ quinoline ring stretching mode indicative of quinine deintercalation. (E) Raman image showing the relative concentration of protein distributed in the cells. The PC score maps shown in A and C were used to determine (F) the percent deintercalation for every pixel of the polyplex particles. (F, Inset) A magnified region containing a polyplex with a ring-shaped unpackaging behavior. (G) Correlates the percent deintercalation with respect to the relative concentration of protein as a function of the normalized distance (R/r) from the centroids of polyplex particles inside cells (see SI Appendix for details). (Scale bars, 5 µm.)

In order for the particles to undergo unpackaging from the surface inward there is likely some component in the intracellular milieu that is facilitating unpackaging. It is unlikely that the unpackaging of the polyplexes is facilitated by hydrolysis of the HEA pendant groups since HEA is stable within the intracellular pH range encountered by polyplexes (SI Appendix, Fig. S25). Instead, our dye-exclusion results (Fig. 3B) indicate that intracellular proteins are likely causing polyplex unpackaging. If this were the case, we would expect to observe a correlation between the concentration of protein and pDNA deintercalation in the polyplex particles. To investigate this, we calculated the radially averaged cross-sections of all of the individual particles to quantify the distribution of poly(quinine-co-HEA), pDNA unpackaging, and proteins as a function of distance inside the polyplexes (SI Appendix, Fig. S30). The radial cross-section of the relative polymer concentration decreased monotonically from the centroid of the particles. In contrast, the radial cross-sections corresponding to the deintercalation of poly(quinine-co-HEA) from pDNA and the relative protein concentration do not follow this monotonic behavior but instead mirror each other. In fact, as indicated in Fig. 7G, the relative concentration of protein colocalized with polyplex particles is strongly correlated (r² = 0.958) with the percentage of poly(quinine-co-HEA) quinine moieties that are deintercalated from the DNA cargo.

This remarkable linear correlation indicates that proteins dominate the unpackaging of the poly(quinine-co-HEA) polyplexes inside the cells. Despite the heterogeneity observed between the different particles, the radial cross-sections (SI Appendix, Fig. S30) show that, on average, proteins are distributed throughout the polyplexes and are strongly colocalized with deintercalated quinine moieties of poly(quinine-co-HEA) polymer chains. Our unique combination of chemical vector design and Raman chemical imaging reveals that polyplexes are porous inside cells. Indeed, this quality enables proteins to percolate into the polyplexes, thereby unwrapping pDNA (Fig. 1C) for highly efficient transcription.