A microfluidic approach for sequential assembly of siRNA polyplexes with a defined structure-activity relationship

Core

The amount of siRNA is the key parameter determining quantities of all other reagents in polyplex formation. For measurements and in vitro experiments, polyplexes with a final concentration of 0.025 mg/ml siRNA were produced. A nitrogen to phosphate (N/P) ratio of 12 was used to determine the amount of core oligomer CO (Fig. 1A) relative to the amount of siRNA. The N/P ratio sets the number of primary and secondary amines in the oligomer’s structure in relation to the number of phosphates in the RNA’s backbone. The azide-bearing core oligomer CON was handled the same way as CO and is described when the reference system is introduced (cf. Characterization of Core—PEG-Ligand Polyplexes)

The conventional method of polyplex preparation was done with pipettes and rapid mixing in a batch wise process. The solvent—if not noted differently—was HEPES buffer pH 7.4 with 5% glucose (HBG). This buffer was used because it does not rely on salts to be isotonic, since polyplex formation relies on charge interactions that could be hampered by ions. Here, CO solution (c_CO = 0.504 mg/ml) was added quickly to a siRNA solution (c_siRNA = 0.05 mg/ml) of equal volume and mixed by rapid pipetting, achieving a final siRNA concentration of 0.025 mg/ml. Subsequently, the formulation has been incubated for 45 min.

For automated polyplex production at a T-junction, siRNA in HBG (c_siRNA = 0.05 mg/ml) and CO in HBG (c_CO = 0.504 mg/ml) or HBG with 50% acetone were loaded into two separate syringes (one ml, Hamilton) that were connected with silicon tubes (SE-200; ProLiquid) to a T-junction (PP-T-Tüllenverbinder; ProLiquid). Each syringe was driven by a separate syringe pump (LA-120, LA-160) that run at the same speed (flowrates (FR) for each pump were 0.5, 1.0, 2.0, 5.0, and 30.0 ml/h) except for experiments with a final acetone concentration of 2.5% (c_siRNA = 0.027 mg/ml; FR_siRNA = 0.917, 1.833, 4.583, 9.167, 55.000 ml/h; c_CO = 3.026 mg/ml; FR_CO = 0.083, 0.167, 0.417, 0.833, 5.000 ml/h). The final product was collected and incubated for 45 min before use. siRNA concentration in the final formulation was 0.025 mg/ml.

For controlled core polyplex production using microfluidics, the double meander channel (DMC) in Fig. 1B was used albeit without the second meander and without both S2 inlets. siRNA in HBG (c_siRNA = 0.033 mg/ml) was loaded into S4 and CO in HBG or HBG with 50% acetone (c_CO = 3.025 mg/ml) was loaded into S3. Both syringes were driven by separate syringe pumps. FR were 100 µl/h for S3 and 900 µl/h for S4, respectively. The final product was diluted with HBG to reach c_siRNA = 0.025 mg/ml.

Addition of lipid anchor and lipid anchor—PEG-ligand oligomers

It was determined before that 20 mol % lipid anchor oligomer (LA or LAE) or lipid anchor—PEG-ligand oligomer in relation to n_CO offered an optimal balance between efficacy and aggregation of the final product (data not shown).

Lipid anchor or lipid anchor—PEG-ligand oligomers were added in two different ways to core polyplexes. If the complete product is assembled in one continuous process, the DMC in Fig. 1B will be used. siRNA in HBG (c_siRNA = 0.033 mg/ml) was loaded into S4 (FR = 900 µl/h) and CO in HBG or HBG with 50% acetone to retard siRNA compaction (c_CO = 3.025 mg/ml) was loaded into S3 (FR = 100 µl/h). Lipid anchor or lipid anchor—PEG-ligand oligomers in HBG with 50% acetone to facilitate solvent exchange were loaded into S2. The flow rate of each syringe S2 was 50 µl/h at a total flow rate of 1,100 µl/h, resulting in a flow rate ratio of lipid anchor oligomer to core polyplex of 1:11. The final product was diluted with HBG to c_siRNA = 0.025 mg/ml.

Alternatively, conventionally (i.e., with pipettes) prepared core polyplexes (c_siRNA = 0.032 mg/ml, c_CO = 0.319 mg/ml) were fed into both inlets connected to syringe S1 (single meander channel (SMC)) with the lipid anchor oligomers filled into syringe S2. In this case, flow rates were 126.5 µl/h for S2 and 600 µl/h for each S1 resulting in a flow rate ratio of 1:10.5. The final product was diluted with HBG to c_siRNA = 0.025 mg/ml. The difference in flow rates between the two set-ups is due to separate optimization steps. Both set-ups resulted in large volumes of core solution and only a thin stream (see Fig. 1B) of lipid anchor solution at the junction, accelerating the solvent exchange from 50% to 4.8% acetone and facilitating the association of the hydrophobic lipid anchor with the fatty acids in the core’s structure. It is always indicated which method for producing core—lipid anchor—PEG-ligand polyplexes was used.

DLS measurement

For dynamic light scattering (DLS) measurements, samples were prepared to contain 1.5 µg siRNA in 60 µl HEPES buffered glucose pH 7.4 (HBG) at 25 °C and the corresponding amount of oligomer. Refractive index and viscosity of the solution were calculated using the solvent builder integrated into the software (Zetasizer family software update v7.12). Viscosities and refractive indices (RI) are reported in Table 1. RI of all particles was estimated to be 1.45. In case of a CO core with N/P 12 and 20 mol % of LA, 16.6 and 2.8 µg were used, respectively. For size measurements, light scattering was measured at a 173° angle (backscatter) with a flexible attenuator with a Zetasizer Nano ZS ZEN 3600 (Malvern Panalytical Ltd, Malvern, UK) in DTS1070 micro cuvettes (Malvern Panalytical Ltd, Malvern, UK). Samples were measured three times with 12–15 sub runs each. The mean z-average in nm of those three runs is reported with error bars corresponding to the 95% confidence interval of the three runs. The underlying intensity distribution is depicted as violin plots in order to gain a better understanding of the formulation’s size distribution. The extension of the violin plot in x direction corresponds to the percentage of the total intensity measured at the specific hydrodynamic diameter depicted on the y-axis.

If zeta potential is measured, the sample will be taken from the cuvette after the size measurement, diluted with HBG to 800 µl and reloaded into the same cuvette. Light scattering was measured at a 90° angle with a flexible attenuator. Samples were measured three times (main runs) with enough sub runs to gather more than 10,000 total counts (usually 12–15). The mean zeta potential of those three runs is reported with error bars corresponding to the zeta deviation’s mean of each main run.

Stability of the core formulation

Core polyplex formulations were prepared using the SMC (Fig. 1B) set up as described above. CO was diluted in HBG with 50% acetone and siRNA was diluted in HBG only. c_siRNA of the final solution was 0.025 mg/ml. Size, polydispersity index (PDI), and zeta potential were measured as described under “DLS Measurement.” This protocol, however, was changed in the following way to allow for multiple measurements over time: Two samples with 60 µl each were prepared. The first sample was used to measure size and PDI. The second sample was diluted with HBG to 800 µl to enable zeta potential measurements. Both samples were measured directly after each other for 90 min.

Gel shift

A 1% (w/w) suspension of agarose in Tris/Borate/EDTA (TBE) buffer (149 mM TRIS, 89 mM boric acid, two mM EDTA in demineralized water) was heated until the agarose was dissolved. After a short cooling period, 0.1% GelRed^® 10000× (Biotium Inc., Fremont, CA, USA) was added. The mixture was cast into its mold and a comb was added to create wells. After 30 min, the solidified gel was placed in an electrophoresis chamber and completely immersed in TBE buffer. Polyplexes were prepared as described above. Naked siRNA was used as positive control. C_siRNA was 0.025 mg/ml in all samples, sample volume was 20 µl. Four µl loading buffer (8.21 mM glycerol, 60 mM EDTA, 0.003 mM bromophenol blue in purified water) was added to every sample (V_total = 24 µl) and each was pipetted in a well in the solidified gel. The gel was run for 60 min at 80 V.

For serum gel shifts, polyplexes were produced with higher siRNA concentration (c_siRNA = 0.25 mg/ml) and diluted afterwards with FBS 1:10 to reach the desired c_siRNA = 0.025 mg/ml. Samples containing FBS were incubated at 37 °C up to 24 h until the loading buffer was added and they were pipetted into the gel’s wells. ImageJ (v. 1.52n) (Schindelin et al., 2012) was used to conduct a densitometry analysis of the siRNA bands. To this end, ImageJ was used to extract gray values from the respective siRNA stains. The sum of gray values as a function of the gel’s extension in y (width of the stains) and x (length of the whole gel) direction was plotted with ImageJ to produce the desired analysis. The plot’s arbitrary values on the y-axis correspond to the sum of all gray values over the full width (y) at a given length position (x). The length position x is plotted on the x-axis.

FRET—experiments

Polyplexes were prepared conventionally (cf. Polyplex Preparation), albeit with a 1:2 siRNA-Cyanine 5 (Cy5):siRNA mixture. Lipid anchors (LA or LAE) were incubated with 0.75 eq. DBCO-PEG4-Atto488 (relative to azide content) over night at room temperature. Afterwards, the modified lipid anchor solution was diluted 1:2 with unmodified lipid anchor solution, resulting in a theoretical degree of labeling of 37.5%. The lipid anchor was added to the polyplexes using the SMC (cf. Addition of Lipid Anchor and Lipid Anchor—PEG-ligand Oligomers). The final siRNA concentration was c_siRNA = 0.1 mg/ml. Therefore, the final Cy5 and Atto488 concentrations were 6.1 and 21.3 µmol/l, respectively. A total of 30 µl of each sample was filled into a 96 well plate and measured with a TEKAN pleat reader (Tecan Trading AG, Switzerland, Spark 10M, SparkControl V 2.1) with the following set of filters: Cy5: excitation wavelength: 625 nm, bandwidth 35 nm; emission wavelength: 680 nm, bandwidth 30 nm; Atto488: excitation wavelength: 485 nm, bandwidth 20 nm; emission wavelength: 535 nm, bandwidth 25 nm; Förster resonance energy transfer (FRET): excitation wavelength: 485 nm, bandwidth 20 nm; emission wavelength: 680 nm, bandwidth 30 nm. Measured fluorescence is divided by gain’s value to exclude amplifier effects.

Polyplex compaction and heparin competition assay

Core polyplexes were prepared conventionally (cf. Polyplex Preparation). Solvents were HBG and HBG with 50% acetone for core polyplexes and lipid anchor oligomers, respectively. 20 mol % of indicated lipid anchor oligomers were attached to the polyplexes via solvent exchange inside the microchannel (Fig. 1B, SMC). The final solvent was HBG with 3.3% acetone. A total of 20 µl of this mixture containing siRNA (0.025 mg/ml), CO (0.252 mg/ml), and lipid anchor (LA: 0.022, LAE: 0.023 mg/ml) were pipetted into a 96 well plate and incubated with 10 µl heparin solution (11.0; 55.0; 110.0; 165.0 IU/ml in HBG) or HBG for 15 min. Afterwards, 80 µl of a 0.5 µg/ml EtBr solution in HBG were added and the samples were incubated for another 5 min. When EtBr intercalates into DNA or RNA it emits a strong signal when excited. This process can be inhibited by compacting the nucleic acid with polycations. Therefore, EtBr’s fluorescence correlates with the compaction efficiency of target oligomers. The addition of heparin tests the formulation’s resistance against anionic stress. The fluorescence of all samples was measured with a TEKAN plate Reader (Spark 10M, SparkControl V 2.1; Tecan Trading AG, Männedorf, Switzerland) utilizing the following set of filters: Excitation wavelength: 535 nm, bandwidth 25 nm; emission wavelength: 590 nm, bandwidth 20 nm. The well containing only siRNA and EtBr served as positive control and was also used to choose optimal gain and Z-position settings. All readings were normalized to samples containing free siRNA and EtBr only (positive control) and are presented here in “(%) of positive control.”

Transmission electron microscopy

Core polyplexes were prepared conventionally or inside the SMC (cf. Polyplex Preparation). Solvents were HBG and HBG with 50% acetone for core polyplexes and lipid anchor oligomers, respectively. A total of 20 mol % of indicated lipid anchor oligomers were attached to the polyplexes using solvent exchange inside the microchannel (Fig. 1B, SMC). The final solvent was HBG with 3.3% acetone. Carbon coated copper grids (300 mesh, 3.0 mm O. D.; Ted Pella, Inc., Redding, CA, USA) were hydrophilized with a plasma cleaner under argon atmosphere (420 V, 1 min). The grid’s activated surface was placed face down on a 10 µl sample droplet for 3 min. Afterwards, the sample was removed with a filter paper and five µl staining solution (1.0% uranyl formate in purified water) was placed on the grid and immediately removed to wash the sample off. Staining was performed with the same staining solution for 5 s. Afterwards, it was siphoned off with a filter paper and the remaining liquid was left to evaporate for 20 min. Grids were stored at room temperature. Samples were measured with a JEOL JEM-1100 electron microscope at 80 kV acceleration voltage.

PDMS channels

The microfluidic channels design was realized on a silica wafer with soft lithographic methods. The master microstructure was designed with the LPKF CAD/CAM software (LPKF Laser and Electronics) and made using SU8 process on silicon wafer. The microstructure of ~72 and ~90 µm thickness for single- and double-meandering channel, respectively, was rastered using LPKF ProtoLaser LDI UV-laser (LPKF Laser and Electronics). Utilized SU-8 3000 photoresist was processed in accordance with the manufacturer’s instructions. The SU-8 master was subsequently silanized in an evacuated desiccator for 12 h with tri-chloro(1H,1H,2H,2H-perfluorooctyl)silane. The PDMS elastomer was mixed with 10% (w/w) crosslinker, degassed, poured onto the wafer, and cured (75 °C, 4 h). Subsequently, PDMS was peeled from the wafer, holes for the inlets were pierced at the designated positions with a biopsy puncher, and it was bonded to a glass slide by oxygen plasma-induced oxidation (Diener Electronic; 10 W high frequency generator power, 12 s, Pico Model E). The chip was left alone for 1 h to allow the reaction to complete. Afterwards, polyethylene tubes (length = 110 mm, inner diameter = 0.38 mm) were fitted into the holes in the PDMS and everything was covered with another layer of PDMS treated in the same way as mentioned above to seal the in- and outlets completely. A to-scale model of both channels’ layout can be found in the Supplemental Information together with a detailed description of the channel’s dimensions and calculations of Reynold’s and Dean’s numbers (3. Channel layout, Figs. S10 and S11).

Culture

We used KB cells (cervix carcinoma, derived from HeLa cells) for all in vitro experiments. KB wild type cells were bought from DSZM (Braunschweig, Germany) and they were subsequently modified to code for a GFP-luciferase fusion mRNA by A. Cengizeroglu. The modified cell line is stably transcribing and translating the fusion mRNA to an eGFP-Luciferase fusion protein, which consists of two functional proteins, GFP and luciferase. The fusion protein’s expression can be silenced by any siRNA that is complementary to the GFP-luciferase fusion mRNA. Here, we used siGFP. The construct’s transfection process was described in A. Cengizeroglu’s thesis (Cengizeroglu, 2012) and first use was demonstrated by Dohmen et al. (2012). For each experiment, cells were freshly thawed from a liquid nitrogen storage tank and passaged at least four times before experiments were conducted. Cells were subcultured when 70–90% confluency was reached. Culture conditions were 37 °C and 5% CO₂. KB cells were cultured in RPMI-1640 supplemented with 10% FBS and 1% penicillin/streptomycin (five ml with 100 U/ml and 100 µg/ml, respectively).

Transfection

Cells were seeded into 96 well plates 1 day prior to transfection. All wells were pre-treated with 40 µl collagen solution per well (0.1 mg/ml, removed after 30 min, 37 °C). Afterwards, cells were seeded with 4,000 cells/well in 100 µl folate free Gibco™ RPMI 1640 (Fisher scientific, Hampton, NH, USA) supplemented with 10% FBS. The next day, the medium in all wells was replaced with 80 µl fresh medium (RMPI 1640, FolA free) and 20 µl sample solution or HBG (negative control) was added. Samples were prepared completely inside the microfluidic channel (cf. Polyplex Preparation & Fig. 1B, DMC). siRNA concentration was five µg/ml in each well. Samples were always prepared in quintuplicates. Medium was exchanged again after 4 h, total incubation time was 48 h at 37 °C, 5% CO₂.

Luciferase assay

Plates were taken from the incubator and all media was removed. A total of 100 µl/well lysis buffer (Luciferase Cell Culture Lysis 5X Reagent, Promega, diluted 1:10 with purified water) was added and incubated for another 45 min at room temperature. Plates were frozen at −80 °C until measurement. A total of 35 µl/well of the cell lysate were transferred to white, opaque 96 well plates (BertholdTech, Bad Wildbad, Germany) and measured with a Centro LB 960 luminometer (BertholdTech CENTRO, Driver V. 1.21, MikroWin, V. 5.2, 10 s integration/well). A total of 100 µl LAR buffer per well (20 mM glycylglycine, 1.0 mM MgCl₂, 0.1 mM EDTA, 3.29 mM DTT, 0.548 mM ATP, 1.30 µM coenzyme A, adjusted to pH 8.5 with NaOH) were automatically added by the machine. The output of this measurement is relative light units (RLUs) per well. The raw data was handled the following way. The mean value from each sample was calculated and was set in relation to the mean value of the respective negative control. Results are depicted in “RLU (%) of HBG.” Error bars represent 95% confidence intervals of five samples.

MTT assay

Plates were taken from the incubator and 10 µl/well 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT; Carl Roth, Karlsruhe, Germany, five mg/ml in PBS) were added and everything was incubated for another 2 h at 37 °C. Afterwards, the fluids were removed and the plates were frozen at −80 °C for at least 1 h. A total of 100 µl/well DMSO were added and the plates were gently shaken at 37 °C for 20 min to dissolve the purple formazan dye. The absorbance at 590 nm of each well against the reference wavelength (630 nm) was measured with a TEKAN plate reader (Spark 10M, SparkControl V 2.1; Tecan Trading AG, Männedorf, Switzerland). The raw data was handled the following way. The mean value from each sample was calculated and was set in relation to the mean value of the respective negative control. Therefore, results are depicted in “(%) of HBG.” Error bars represent 95% confidence intervals of five samples.

Dose titration

Core polyplexes were prepared conventionally with pipettes as described under “Polyplex Preparation.” siRNA concentrations were chosen to have a final amount of 100, 250, 500, 750, and 1,000 ng/well. CO concentrations were adjusted accordingly. To be precise, siRNA concentrations in 20 µl transfection volume were (mg/ml): 0.0050, 0.0125, 0.0250 0.0375, 0.0500. CO concentrations were (mg/ml): 0.0458, 0.1145, 0.2291, 0.3436, 0.5041. A total of 20 µl/well of each sample was transfected as described under “Transfection.” Samples were transfected in quintuplicates. The formulations’ effect on luciferase activity and metabolic activity was evaluated with a luciferase assay and a MTT assay as described above.

Size

Core polyplexes (CO + siRNA) with comparable properties were generated either by conventional bulk mixing, or at a T-junction, or with microfluidics. Integrating lipid anchor or lipid anchor PEG-ligand oligomers increased size and PDI moderately.

Hydrodynamic diameter, zeta potential, and PDI of polyplexes was measured by DLS (Figs. 2 and 3) and sizes were confirmed with transmission electron microscopy (TEM) (Fig. 4). Compacting siRNA conventionally with core oligomers (CO) by rapid pipetting yields particles with a mean hydrodynamic diameter (d_Z) of 84 nm (Fig. 2A). The PDI is very low (PDI < 0.20; Fig. 2B). Increasing control over this process either at a T-junction or with a microfluidic device, however, needs certain additional conditions to be met in order to produce similar particles. At a T-junction, the total flow rate needs to be very high (here: 60 ml/h) to generate particles with a hydrodynamic diameter of 97 nm and a PDI < 0.20. Addition of acetone does only lead to comparable particles and PDIs when flow rates of both components are identical, and CO is dissolved in 50% acetone as depicted in Fig. 2 (d_Z = 104 nm, PDI < 0.23). This approach, however, results in an acetone concentration of 25% in the final product requiring additional efforts by evaporation or dialysis to remove the organic solvent when using it in vitro or in vivo. The complete data set (influence of flow rate, acetone and flow rate differences of 1:2 or 1:10) can be found in the Supplemental Information (6. Polyplex production at a T-junction, Fig. S14).

When preparing polyplexes, it is paramount to decrease diffusion lengths or to increase time needed for efficient siRNA compaction. Otherwise, influence on kinetically controlled polyplex formation is decreased and nanoparticles’ size and polydispersity increases. Diffusion lengths inside the micro channel were minimized by flow rate differences > 1:10 and acetone was used to retard siRNA compaction. Previous experiments have shown that feeding the two outer channels with diluted siRNA solution and the middle channel with concentrated CO solution generates particles in an acceptable size range (data not shown). Here, a substantial difference in PDI and hydrodynamic diameter was observed when polyplexes were generated with (d_Z = 95 nm, PDI < 0.14) or without (d_Z = 149 nm, PDI < 0.11) additional acetone (Fig. 2).

The influence of lipid anchor and lipid anchor—PEG-ligand oligomers on core polyplexes was investigated. To this end, LA or LAE with or without their respective PEG-ligand oligomers (Fig. 1A) were attached to conventionally prepared core (CO + siRNA) polyplexes inside the microchannel (SMC, Fig. 1B). As described in detail in the methods section, it is essential to use concentrated lipid anchor solutions and diluted core polyplex solutions. This setting ensured that only a thin stream of lipid anchor solution is flowing through the Y-junction, accelerating the solvent exchange from 50% to 4.8% acetone and facilitating the association of the hydrophobic lipid anchor with the fatty acids in the core’s structure. Polyplexes’ particle size, PDI, and zeta potential were measured by DLS (Fig. 3). In order to gain a better understanding of the intensity distribution, violin plots are provided in addition to z-average values. Therefore, z-average values can be better assessed based on the underlying distribution, be it mono- or multimodal. The expansion in x direction represents the relative frequency the respective size has been measured. The z-average is located close to the position with the largest expansion in x direction for mono modal distributions (e.g., in the panel labeled “core”). When the distribution is multimodal, however, z-average’s position can be quite misleading (e.g., in the panel core-LAE, sample P12-48F) and the intensity distribution needs to be considered.

The effect of adding 20 mol % (relative to n_CO) lipid anchor or lipid anchor—PEG-ligand oligomers to core polyplexes depended on the PEG-ligand’s length on the respective lipid anchor. The addition of LA containing oligomers to the core formulation (d_Z = 123 nm, PDI < 0.13) increased hydrodynamic diameters of resulting nanoparticles moderately from 131 nm (LA alone) to 169 nm (LA: P12-48F). Additionally, PDI decreased with the addition of LA (PDI < 0.11) or LAE (PDI < 0.10) oligomers and increased from PDI < 0.12 to PDI > 0.20 with longer PEG-ligands. The hydrodynamic diameter of LAE containing polyplexes was generally ~15 nm smaller than in LA containing formulations. Although, LAE oligomers with longer PEG-ligands were more likely to form aggregates (LAE: P12-48F). As expected, zeta potential of core polyplexes alone in HBG was positive with ZP = 24 mV due to the high N/P charge ratio. Incorporation of 20 mol % LA or LAE with or without PEG-ligands had only limited effect on the particles’ zeta potential except for particles with P12-48F PEG-ligands (Fig. 3C). Incorporation of LA:P12-48F or LAE: P12-48F decreased mean zeta potential to 14 and 10 mV, respectively.

Finally, after having scrutinized all steps independently, core—lipid anchor—PEG-ligand polyplex production from its single components inside one microchannel was investigated. The DMC (Fig. 1B) was used. Syringe S3 was filled with siRNA in HBG and S4 with CO in HBG with or without 50% acetone. Syringes S2 were loaded with four different oligomers in HBG with 50% acetone: LA, LA: P12-24F, LAE, or LAE: P12-12F. Eight runs were conducted, each lipid anchor or lipid anchor—PEG-ligand oligomer was mixed with core polyplexes produced with or without the aid of acetone. Sizes were comparable to core—lipid anchor—PEG-ligand polyplexes with conventionally prepared cores when no acetone was used in the core production step. When CO was dissolved in 50% acetone, however, polyplexes completely prepared with microfluidics had a smaller hydrodynamic diameter and PDI (Supplemental Information, 7. Controlled Production of Core—Lipid Anchor—PEG-Ligand Polyplexes from their Single Components, Fig. S15).

Lipid anchor integration

LA and LAE integrate into core polyplexes.

Investigation of LA and LAE integration into core polyplexes was carried out by TEM and FRET measurements. TEM measurements revealed that formulations from siRNA and CO form spherical particles with a diameter <100 nm (Fig. 4B) which is in good agreement with DLS measurements (Fig. 3A). There was no obvious difference when particles were produced conventionally or with microfluidics (Supplemental Information, 8. TEM: Comparisons of polyplexes produced with pipettes or with the DMC, Fig. S16). LA and LAE with or without covalently bound PEG-ligands alone form tubular or fibrous structures on the TEM grid that could not be found when formulated together with core polyplexes. This finding suggests that lipid anchor oligomers are indeed interacting with core polyplexes.

These findings obtained by TEM were supported by FRET measurements. Receiving measurable FRET signals implies a distance <10 nm between chromophores (Clegg, 1995; Förster, 1948). Here, 50% siRNA with one molecule Cy5 on the sense strand was used for conventional core polyplex formation. Lipid anchor oligomers were modified with 0.75 equivalents (relative to lipid anchors’ azide) DBCO-PEG4-Atto488 and subsequently deposited on the conventionally prepared core polyplex using solvent exchange inside the micro-channel (SMC). These polyplexes emitted strong FRET signals when Atto488 dyes were excited and fluorescence was measured from Cy5 dyes alone (Fig. 4A). When CO was missing from the formulation, polyplex formation did not occur making energy transfer between dyes a function of their dilution only (sample “siRNA + LA” in panel “FRET” in Fig. 4A). All control experiments (FRET measurements from polyplexes with only one dye and fluorescence measurements of both dyes separately) can be found in the Supplemental Information (9. FRET control experiments, Fig. S17).

Stability

Lipid anchors do not influence stability of core polyplexes.

Polyplex stability was assessed with two different methods. The general ability of polyplexes to compact and hold siRNA back under the influence of an electric field was investigated with an agarose gel shift assay and its densitometry analysis to simplify comparison of bands. Ability to compact siRNA and resist polyanionic stress was tested with an EtBr displacement assay with or without additional heparin.

Polyplexes from CO and siRNA were prepared conventionally in HBG and lipid anchors ± PEG-ligands were attached inside the micro channel (SMC). Samples were diluted 1:10 with HBG or serum (FBS). Additionally, samples containing serum were incubated at 37 °C for up to 24 h to assess stability under the influence of body temperature and serum components. There was no visible difference between all formulations at t = 0 h with or without additional FBS. At the 4 h mark, only small differences between samples were visible, while all samples retained most of their payload. After 24 h, core—lipid anchor or core—lipid anchor—PEG-ligand formulations revealed a slight decrease in siRNA retention capability in comparison to the core formulation alone. At this time, the core-LAE formulation seemed to be better at retaining siRNA than the core-LA formulation. When lipid anchors coupled with PEG-ligands were used, however, core-LA—PEG-ligand formulations retained siRNA better than their LAE containing counterparts (Supplemental Information, 9. Gel shift assay, Figs. S18 and S19).

Polyplexes (CO + siRNA) were prepared conventionally and lipid anchors were added inside the SMC for the EtBr displacement assay with and without heparin competition. In this assay, LA and LAE containing polyplexes showed an unaltered protection against dye displacement behavior. Fluorescence without additional heparin for core formulation, core-LA formulation and core-LAE formulation was 14%, 18%, and 11% of the positive control, respectively. One IU/ml heparin increased fluorescence to 37%, 39%, and 22%. Total displacement was observed at heparin concentrations above five IU/ml (Supplemental Information, 9. EtBr displacement assay, Fig. S20).

Toxicity

Core (CO + siRNA)—lipid anchor—PEG-ligand polyplexes do not alter the metabolic activity profile of KB cells in comparison to core polyplexes alone.

Different fatty acids in oligo-amidoamines have been shown to induce membrane leakage in erythrocytes and to increase cell death in in vitro cell assays (Klein et al., 2016; Reinhard, Zhang & Wagner, 2017). Influence of target formulations completely prepared with microfluidics on metabolic activity of KB cells was assessed by MTT assay to account for any apparent effects on cell survivability. The MTT assay correlates metabolic activity to the amount of formazan dye produced by oxidoreductase enzymes while consuming NAD(P)H. All formulations tested in this assay showed no reduction of formazan absorption relative to untreated KB cells (Supplemental Information: 11. MTT assay of core-lipid anchor—ligand polyplexes, Fig. S21).