Evaluation of adsorption of DNA/PEI polyplexes to tubing materials

Abstract

Nucleic acid drugs hold great promise for potential treatment of a variety of diseases. But efficient delivery is still the major challenge impeding translation. Nanoformulations based on polymers and lipids require preparation processes such as microfluidic mixing, spray drying or final filling, where pumping is a crucial step. Here, we studied the effect of pumping on the com-ponent and overall loss of a binary polyplex formulation made of DNA and polyethyleneimine (PEI). We varied tubing length and material with a focus on subsequent spray drying. Interestingly, product loss increased with the length of silicon tubing. Losses of DNA were prevented by using Pumpsil. The following spray drying process did not affect DNA content but caused PEI loss. Characterization of the different tubing materials revealed similar hydrophobicity of all tubing materials and showed neutral Pumpsil surface charge, negative Santoprene surface charge, and a positive Silicon surface charge. Hence, adsorption of DNA onto tubing material was concluded to be the root cause for DNA loss after pumping and is based upon an interplay of ionic and hydrophobic interactions between polyplexes and tubing material. Overall, selecting the appropriate tubing material for processing nucleic acid nanoparticles is key to achieving satisfactory product quality.

Introduction

Nucleic acid therapeutics have become a new reality for the treatment of diseases such as muscular dystrophies, amyloidosis and hepatic porphyria [1], with billions of doses of mRNA vaccines having been administered since 2020 [2]. But efficient delivery of nucleic acids to target sites is still hampered by their chemical nature: molecule size, negative charge and susceptibility to nucleases are the main challenges to be overcome. Several strategies have been developed including polymer based nanoparticle delivery [3]. Positively charged polymers such as polyethyleneimine (PEI) can interact with the negative phosphate groups of nucleic acids and form polyplexes of around 100 nm. These polyplexes can be taken up by the cells for transfection [4]. While polycationic materials such as PEI are often associated with cellular toxicity [5], they offer much higher encapsulation efficiency compared to nanoparticle formulations based on polymers such as poly(lactid-co-glycolid) (PLGA) [6, 7].

Pulmonary delivery has attracted great interest in the research community to tackle diseases such as asthma, chronic obstructive pulmonary disease and cystic fibrosis. Local treatment of these diseases is preferred over systemic parenteral application as less drug is required with lower risks for adverse effects due to direct tissue targeting [8]. It is known that only particles with an aerodynamic diameter between 1 and 5 μm are able to follow the airstream and deposit in the lower lung areas. While particles greater than 5 μm deposit in the throat or the upper airways, particles below 1 μm are often exhaled [9]. Therefore, polyplexes need to be formulated into microparticles with excipients suitable for inhalation for example by spray drying [10, 11]. During spray drying, solutions/suspensions can be sprayed via a nozzle into a drying chamber where heat is applied to dry the formulation gently.

We recently observed losses of around 30% bulk DNA (bDNA) and 20% PEI upon spray drying of model nanoparticles [11]. We assumed that these losses were attributed to the spray drying process and that the different extent to which the two nanoparticle components were lost was caused by particle instability. Comparable losses of nucleic acids and their formulations after spray drying were also reported by others [12–14]. However, further experiments to identify the root cause for the observed effects were not conducted so far. Considering that spray drying involves also pumping of the liquid feed, this study focused on pumping as an essential step in liquid product handling for filling, filtration, and transfer. While piston pumps have a high dosing accuracy on the one hand, steel particle shedding as well as drug aggregation and shear are of concern. [15] Diaphragm pumps are similar to piston pumps in means of dosing accuracy and shear forces, but no steel particle shedding is observed [16, 17]. Peristaltic pumps have lower dosing accuracy, but they are well suited for shear sensitive products, and as single use systems the risk for cross contamination is minimized [18]. Syringe pumps are characterized by constant mean flow rate, high dosing precision and are also ideal for shear sensitive drugs but are limited in volume and thus not applicable for industrial production [19].

In the last years substantial research was conducted on the use of pumps in fill and finish of biopharmaceutics focusing on the effect of shear stress on protein adsorption and aggregation [15–17, 20]. To the best of our knowledge, no reports are available investigating the behavior of nucleic acid nanoformulations upon pumping. In this study, peristaltic pumps were utilized because of their standard use in the pharmaceutical industry during fill and finish processes, liquid transfer and drug supply for spray dryers.

Since silicone tubing is commonly used in the manufacturing of biopharmaceuticals due to low cost and availability, our herein presented study investigated the effect of tubing quality on the nanoformulation: a simple, commercially available as well as a pre-sterilized and certified tubing were compared. Other commonly used materials in pumping of biopharmaceuticals are thermoplastic vulcanisates due to higher longevity and higher chemical resistance compared to silicone. Santoprene, a representative of this material type, was hence included into this study design.

Due to our previous observations, this study aimed at gaining deeper insights into the impact of peristaltic pumping on polyplexes both with respect to loss and composition. Data sets were recorded for three different tubing materials, which are regularly used in pharmaceutical and food industry: high- and low-quality silicone tubings as well as a thermoplastic elastomer. The effects of only pumping or pumping and spray drying were evaluated based on polyplex size, polydispersity index (PDI) and the quantitative composition of the final product. Also, different tubing pretreatments and the addition of polysorbate 20 (PS20) were tested to evaluate their effects on polyplex quality and quantity. PS20, one of the surfactants frequently applied in FDA and EMA approved products for parenteral application, was chosen to investigate the effects of a surfactant on the adsorption of polyplexes on tubing material [21]. PS20 is known to accumulate at interfaces and hence could compete with polyplexes, possibly preventing their adsorption on the tubing material.

These findings are of interest for spray drying polyplexes but are also important for any process utilizing peristaltic pumping in the field of nucleic acid nanoformulations.

Materials

Hyperbranched polyethyleneimine (PEI) (25 kDa) was obtained from BASF (Ludwigshafen, Germany). Bulk DNA sodium salt from salmon sperm (BP2514-250) (bDNA) and black 96 well plates (10307451) were purchased from Fisher Scientific (Schwerte, Germany). Heparin from porcine intestinal mucosa (H3393, >180 units/mg, grade I-A), picrylsulfonic acid (TNBS) (P2297), TRIS EDTA Buffer Solution 1x (93283) and PS20 (P9416) were obtained from Sigma Aldrich (Munich, Germany), SYBR™ Gold dye from Life Technologies (Carlsbad, CA, U.S.A.), ethylene glycol from Grüssing (Filsum, Germany) and n-hexadecane from Merck (Darmstadt, Germany). Peroxide cured silicone tubings (inner diameter 2.0 mm, outer diameter of 4.0 mm) were bought from VWR International (228-0704, Darmstadt, Germany), Pumpsil® tubing (inner diameter 1.6 mm) was a kind gift from Watson-Marlow (Rommerskirchen, Germany) and Santoprene® tubing (inner diameter 1.6 mm) was obtained from AET Lézaud (Wendel, Deutschland).

Methods

Polyplex preparation

Polyplexes in a total volume of 5 ml containing 10 μg of bDNA were prepared at an N/P ratio of 10 as described in [11].

For polyplex preparation, 10 μg of bDNA and the respective amount of PEI (13.1 μg) were diluted to 500 μL with highly purified water (HPW), each, and mixed together by pipetting. After 10 min incubation, polyplexes were further diluted with HPW to 5000 μL. After additional 10 min incubation, 70 μL were used for DLS analysis. The remainder was pumped at 1.2 ml/min and collected for further analysis. For experiments evaluating the effect of PS20 on adsorption effects, polyplexes were prepared as described above in a 0.02% PS20 HPW solution (also see ‘Tubing preparations’).

Tubing preparations

Silicon, Pumpsil® and Santoprene™ tubings were all rinsed with 50 mL, 40°C warm water and air-dried thereafter. Additionally, some sets of tubings were used after additional treatment: One set of tubings, i.e. three tubings of every kind, were used without any further treatment – no treatment (NT).

Additionally, another complete set of tubings was autoclaved at 121°C for 15 minutes to reflect this typical treatment in pharmaceutical manufacturing (AC).

Also, one set of tubings was rinsed with RNase Zap to avoid any possible contamination with RNases or DNases. Following this treatment, the set was rinsed again with HPW to flush out all remaining RNase Zap.

For the PS20 containing sample, a set of tubings was used which had no further treatment (NT-PS20).

Spray drying and pumping

For spray drying experiments, polyplexes were prepared as described in ‘Polyplex preparation’ using 10% m/v trehalose instead of HPW. A B-290 spray dryer was used (Büchi Labortechnik, Essen, Germany) at 1.2 ml/min feed rate (5 % pump rate), equipped with a two-fluid nozzle at an airflow of 470 NL/h, aspirator preset to 70 % and an inlet temperature of 65°C resulting in an outlet temperature of 40°C. Upstream of the spray dryer, a DeltaTherm-dehumidifier (DeltaTherm, Germany) was used to decrease the humidity within the drying gas. Before drying the samples, the spray dryer was equilibrated for 20 min by feeding HPW. The choice of excipient and its concentration as well as the spray drying conditions are based on our previous study [11] to allow for direct comparison.

Samples were spray dried after pumping through 60 cm silicone tubing using the built-in pump of the Büchi B-290 “PP60_SD”. Alternatively, 60 cm (“SP60_SD”) or very short 2 cm (“SP2_SD”) silicone tubing pieces were connected to the spry dryer using a syringe pump (KDS-220-CE, KD Scientific, Holliston, MA, USA). Additionally, samples were pumped through the silicone tubing only, without spray drying (“PP60”).

In another setup, Pumpsil® and Santoprene™ tubings were connected with a Masterflex L/S (7520-47), equipped with the Easy-Load II head module (77201-60, Cole-Parmer, Wertheim, Germany), because the thicker tubing did not fit the built-in pump. All Pumpsil® and Santoprene™ tubings as prepared in ‘Tubing preparations’ had a length of 60 cm. All experiments were performed in triplicates.

Z-average, PDI and zeta potential measurements

For Z-average, PDI and zeta potential measurements, 70 μL of freshly prepared or pump stressed polyplexes were placed into a disposable cuvette (Brand, Wertheim, Germany) or 700 μL of freshly prepared polyplexes into a folded capillary cell (DTS1070) and analyzed by a Zetasizer Nano ZS (Malvern Instruments, Malvern, U.K.) at 173° back scatter. Measurements were performed in triplicates with 15 sub-runs for size and PDI or up to 100 sub-runs for zeta potential.

bDNA quantification

Nucleic acid quantification was performed as described elsewhere [11]. In short, 90 μL of sample was diluted to 150 μL, and 75 μl of a 2.33 mg/ml heparin solution in TRIS-EDTA-buffer was added. After 2h of incubation, samples were diluted 1:2 with HPW, 100 μL were pipetted into a black 96 well plate, 30 μL of a 4x SYBR gold solution was added and an analysis was performed at 485 nm excitation of 520 nm emission with a FLUOstar® Omega microplate reader (BMG LABTECH, Ortenberg, Germany). Samples were analyzed in triplicates. To verify accuracy and precision of the analysis, an internal standard (iS) was prepared as described under ‘Polyplex preparation’.

PEI quantification

For polymer quantification, a TNBS assay as described previously was used [11]. In short, 9 standards ranging from 0 to 19.14 μg PEI/ ml were prepared. Of each sample, 100 μL were pipetted into a separate 0.5 ml vial. Subsequently, 30 μL of 0.088% (m/V) TNBS in 0.1 M borax buffer were added. Samples were incubated for 1 h and absorbance was measured with a quartz cuvette in a UV-1600PC spectrophotometer (VWR International, Darmstadt, Germany) at 405 nm. Again, an internal standard was prepared for verification. Analysis was performed in triplicates for calibration points, samples and internal standard. For samples containing PEI and bDNA, a calibration line was prepared with the corresponding amount of bDNA added.

Surface potential analysis

The surface potential of the tubing materials was estimated according to Altenor et al. [22]. Tubing pieces of 0.5 cm were hardened in liquid nitrogen and milled with a Pulverisette 14 classic line (Fritsch, Idar-Oberstein, Germany) equipped with a 0.5 mm sieve ring operated at 10 000 rpm. Subsequently, 500 mg powdered tubing was dispersed in 50 ml 0.01 M NaCl, pH 5.8. After 24 h incubation, the pH value of each solution was measured using an Accumet AB-150 pH electrode (Fisher Scientific, Germany) and compared to the control buffer.

Surface free energy determination

To estimate polar and dispersive parts of the three tubing materials, surface free energy was determined using a Kruess Drop Shape Analyzer DSA25 (Kruess, Hamburg, Germany). Contact angles of 2 μl of water, ethylene glycol or n-hexadecane were measured on the outer tubing wall. After 20 s of equilibration time, three measurements were performed. The drop shape was fitted by the Ellipse (Tangent-1) method with manual baseline adjustment. Surface free energy was calculated based on the test liquids’ characteristics (Table 1) via the Owens-Wendt-Rabel-Kaelble analysis with the ADVANCE software v1.1.0.2. Analysis was performed in triplicates.

Table 1. Surface tensions and surface free energy components [mN/m] of the test liquids according to Ström et al. [39].

Liquid	Surface Tension	Polar component	Dispersive component
Water	72.8	51.0	21.8
Ethylene glycol	47.7	16.8	30.9
n-Hexadecane	27.6	0.0	27.6

Scanning electron microscopy

The tubing inner surface structure was visualized via scanning electron microscopy (SEM) using a Jeol JSM-6500F instrument (Tokyo, Japan) with Inca Software (Oxford Instruments, Oxfordshire, UK). The tubing was cut into pieces and fixed on aluminum stubs (Plano, Wetzlar, Germany). After sputter coating with carbon using a CCU- 010 compact coating unit (Safematic, Zizers, Switzerland), micrographs were collected at a magnification of 200x and an accelerating voltage of 2.0 kV.

Data presentation, graphics and statistics

For spray drying experiments, results are calculated according to ‘bDNA Quantification’ and ‘PEI Quantification’ and presented in μg. For tubing material evaluation, results are presented as adsorption and calculated according to following formula:

$$adsorption=\frac{loss}{l\ast \pi \ast d}$$ (Eq. 1)

Where loss is the difference between feed and calculated amount of bDNA or PEI according to ‘bDNA quantification’ and ‘PEI quantification’ in mg, respectively, and l is the length and d the inner diameter of the tubing in m.

If not stated differently, data were processed by Graph Pad Prism5 software (Graph Pad Software, San Diego, CA, USA), presented as mean ± standard deviation and analyzed for statistical difference by One-Way ANOVA or Two-Way ANOVA with Bonferroni post-test p<0.05.

Results and Discussion

Upon spray drying of polyplexes substantial material loss has been observed independent of the reduction in yield due to losses in the spray dryer. We found up to 40 % less DNA and 25 % less PEI in the solid content for bDNA / PEI polyplexes spray dried with 10% trehalose [11]. To identify the root cause of the component loss, a 10% m/v trehalose solution containing polyplexes at N/P 10 was spray dried with the feed provided either via a syringe or a peristaltic pump using silicone tubing. In line with our previous work [23], spray drying with the peristaltic pump using a 60 cm tubing (PP60_SD) resulted in losses of approx. 33 % bDNA and 28.0 % PEI (Figure 1).

Figure 1.

Changes of (A) bDNA and (B) PEI in % after spray drying of polyplexes fed with a syringe pump equipped with a 2 cm (SP2_SD) or with a 60 cm silicon tubing (SP60_SD) or with a peristaltic pump with a 60 cm silicon tubing (PP60_SD) or polyplexes after only one peristaltic pump cycle through a 60 cm long silicon tubing without spray drying (PP60).

Accounting for such losses, the spray dried powders performed at least as efficiently as their freshly prepared counterparts in vitro and ex vivo [23]. Exchanging the peristaltic pump for a syringe pump (SP60_SD) to reduce shear forces did not significantly affect bDNA or PEI losses. Hence the kind of pump and shear forces seem to play only a minor role. Interestingly, spray drying with a much shorter tubing of only 2 cm (SP2_SD) resulted in 21.7% loss of PEI and no loss of bDNA. It was shown by Boeckle et al that polyplexes consisting of PEI have an excessive amount of polymer which is not fully bound to the polyplex but free and unbound in solution [5]. The above shown difference between PEI and bDNA losses may be explained by the fact that polyplexes at an N/P ratio of 10 contain such free unbound polymer [5]. The free PEI may be lost during the pumping or spray drying process, but not the intact polyplexes as indicated by an unchanged amount of nucleic acid. One pump cycle with a 60 cm long silicone tubing without spray drying (PP60) resulted in a bDNA loss (23.8%) similar to the one detected after spray drying utilizing the same length of tubing independent of the type of pump. These findings highlight the loss of bDNA content as a result of adsorption to certain tubing material rather than by spray drying or pumping. The b DNA loss within the 60 cm silicone tubing can be interpreted as adsorption of 1.25 ± 0.18 mg/m².Thus the total loss is related to the tubing and depends on material, length and diameter. Spray drying however, evolved into the main reason for changes in the PEI content as spray dried samples (SP2_SD, SP60_SD, PP60_SD) showed PEI losses between 22 and 38% whereas the non-spray dried formulation (PP60) showed no change in PEI content (Figure 1B). Expecting a constant loss of PEI through spray drying, equal or at least similar amounts of PEI should be detected for SP2_SD, SP60_SD and PP60_SD. However, depending on the tubing length, different PEI losses were observed. Although these differences are not significant, reversible binding of PEI cannot be excluded completely: If a short tubing is used, losses are smaller in comparison to longer tubings. The adsorption of PEI to the tubing surface in turn seems to destabilize the polyplexes. And even though PEI is recovered from the pumped but not spray dried samples, the destabilization of the polyplexes leads to losses in DNA, no matter if the samples are spray dried or not. Other reasons such as binding of DNA to the tubing material resulting in a destabilization of polyplexes and hence increased losses during spray drying are also possible.

To further elucidate the effect of adsorption to the tubing material, in addition to silicone tubing a high-quality low abrasion tubing (Pumpsil®) and a tubing consisting of thermoplastic vulcanizates (Santoprene™) were used for single pump cycles of polyplexes with a peristaltic pump without consecutive spray drying. Santoprene™ is a material which consists of an ethylene propylene diene monomer rubber embedded in a polypropylene matrix and is highly hydrophobic. Pumpsil® and Silicon tubings consist of methylated silicon oxide chains and might show slightly less hydrophobic properties. Furthermore, different pretreatment washing regimes were tested to investigate a potential effect of contaminants such as DNase. In addition, formulations containing PS20 as surfactant to reduce adsorption effects were evaluated [24, 25].

Adsorption of PEI and bDNA to silicone tubing did not change with pretreatment. Thus, contaminations from the tubing manufacturing and handling do not play a significant role in the observed losses of the polyplex components. PEI and bDNA adsorption was not prevented by addition of 0.02% PS20 to the formulation independent of the tubing material (Figure 2). PS20 can prevent adsorption events in the biopharmaceutical field [26]. But it can interact with the polyplexes and destabilize them as indicated by the higher standard deviations of z-average and PDI of the polyplexes in presence of PS20 (Figure 3). Less tightly packed polyplexes offer a greater possibility for single constituting molecules to interact with the tubing material and could enhance adsorption effects.

Figure 2. Adsorption of bDNA (red) and PEI (blue) in mg/m² without pretreatment (NT), after autoclaving (AC), after treatment with RNase Zap (ZAP) or untreated but with addition of 0.02% PS20 in the formulation (NT_PS20), n=3.

Figure 3.

Hydrodynamic size and PDI of polyplexes before (fresh) and after a single pump cycle (pumped). Different kinds of tubings were used which underwent pretreatments such as autoclaving (AC) or treatment with RNase Zap (ZAP), or they were not treated (NT) or untreated but with addition of 0.02% PS20 in the formulation (NT_PS20).

The tubing material had a marked effect on component loss. No adsorption of bDNA or DNA containing polyplexes was detected on untreated Pumpsil® whereas Santoprene™ and silicone tubings showed significant adsorption. This is surprising as Pumpsil® and silicone consist both of methylated silicon-oxide groups, but Santoprene™ is made of monomer rubber in a polypropylene matrix. Hence, the high quality, low abrasion tubing apparently had a positive effect regarding the prevention of polyplex/DNA adsorption. In contrast, PEI adsorption was independent of the tubing material, which could be a result of the abundance of free excess PEI.

To better understand the adsorption of the polyplex components to the tubing surface and shed a light on the discrepancy between both methylated silicon oxide materials, we characterized surface charge and surface free energy of the tubing materials. In general, a high surface free energy is related to higher polarity and surface charge. Interaction between materials with high surface free energy and adsorbing species is mainly driven by hydrophilic interactions (hydrogen bonds and ionic). Vice versa, interaction between materials with low surface energy and drugs is dictated by hydrophobic interactions (i.e. mostly van der Waals). This is of special interest as the PEI-bDNA polyplexes own a positive surface charge of +33.6 ± 3.7 mV and might therefore be prone to ionic interaction. Consequently, the surface free energy can elucidate the type of adsorption.

All three tubing materials showed similarly low surface energy of approximately 20 mN/m with high dispersive and hardly any polar contribution (Figure 4A). Consequently, marked hydrophobic interactions are expected. The values are in line with published data [27, 28]. Silicone and Pumpsil®, mainly consist of a structure based on poly(dimethylsiloxane) where the Si-O linkages seem to be shielded sufficiently by the methyl groups, resulting in an overall surface tension of 22.8 mN/m with polar components of 1.1 mN/m [27, 28]. Santoprene™ is a thermoplastic vulcanizate and consists of ethylene propylene diene monomer rubber particles embedded in polypropylene matrix absent of any polar groups. Literature claims an overall surface tension of 29 mN/m for the polypropylene matrix which is slightly higher than the values determined here (22.2 ± 3.22 mN/m) [29].

Figure 4. Characterization of tubing material and inner surface morphology.

Evaluation of (A) surface free energy and (B) surface charge. (C) Determination of surface morphology of the inner tubing surface.

Although surface free energy results suggest hydrophobic interactions of tubing material and polyplexes, materials can also be net positively or negatively charged. These charges contribute to the adsorption. Especially net negative surface charges could lead to a significant interaction with the positively charged polyplexes used within this study. For surface charge determination the materials were shredded and incubated in buffer at the pH of the polyplex solution [22]. Although the tubing was shredded, we expect that the data represents the inner tubing surface as the tubing is made from a single polymer melt. Pumpsil® did not show a shift of pH suggesting a net neutral surface charge (Figure 4B). The regular quality silicone tubing, however, showed a slight decrease in pH indicating a positively charged surface, whereas Santoprene™ showed the opposite characteristics. Unfortunately, exact surface charge values cannot be obtained by this approach. The obtained relative estimations are in good agreement with literature values for the isoelectric points for silicone tubing or Santoprene™ like materials of around 5.2 and 3.5, respectively [30]. Thus, Pumpsil® will less likely interact via ionic interactions whereas Silicone is likely to interact with negatively charged molecules such as nucleic acids and Santoprene™ with positively charged molecules such as the positively charged polymer on its own or the whole polyplex with its positive surface charge. The microstructure of the inner tubing surface could influence total adsorbed amount due to differences in accessible surface area. To examine surface morphology, SEM images were obtained (Figure 4C). Interestingly, the inner tubing surface of silicone based tubings had a flaky appearance while the Santroprene tubing had a porous morphology. Higher adsorption to Santoprene tubing might therefore originate from the increased surface area. Linking the observation from material characterization to the adsorbed amounts no clear hypothesis was set up for the adsorption of bDNA, PEI and polyplex. But an interplay between mainly hydrophobic interactions and also contribution of electrostatic interactions is expected. Pumpsil® showed no significant difference in the composition of the dispersive and polar components compared to Santoprene™ and silicone tubing. Results from surface free energy analysis indicate that all tubings are of hydrophobic nature and are therefore prone for short-range interactions. Only Pumpsil® owns a neutral surface charge and shows no significant adsorption of bDNA after a single pump cycle. This leads to the assumption that both, electrostatic and hydrophobic interactions are required for adsorption of bDNA. In fact, literature claims hydrophobic as well as electrostatic interactions as main factors for adsorption of free and complexed DNA. It is known that DNA can adsorb nonspecifically to hydrophobic surfaces by contribution of electrostatic forces depending on surface charge density and hydrophobic interactions [31–33]. Especially in these low ionic strength formulations, electrostatic interactions dominate the hydrophobic ones for DNA adsorption [32]. Although DNA owns negative charges due to the phosphate groups in its backbone, it also owns hydrophobic regions provided by its nucleobases which can interact with the hydrophobic surface [34]. It was shown, that the complexation of DNA with cationic agents leads to irreversible compaction on hydrophobic surfaces and in fact leads to even increased DNA adsorption of positively charge particles in comparison to free, negatively charged DNA [35, 36]. Therefore, binding of DNA to the tubing materials may as well be a result of polyplexes binding to the tubing. Upon pumping, attraction between polyplexes and the tubing material could lead to a reversible attachment of the whole polyplex on the surface. In case of such events, hydrophobic forces could drive a release of bDNA out of polyplexes and result in adsorption. Bengali et al. and Segura et al. presented that the adsorption of polyplexes for surface-mediated delivery of DNA showed higher immobilization of polyplexes on hydrophobic surfaces and a reduced DNA release compared to the hydrophilic analog. Release rate reduction was due to irreversible binding and aggregation of DNA complexes while hydrophilic surfaces showed rather reversible binding mechanisms [37, 38].

Conclusions

This study showed that losses during spray drying of polyplexes have to be interpreted in the light of pumping. The drying process itself appears to lead to PEI loss, without losses of DNA. A DNA loss is based upon adsorption effects and depends on material and length of the tubing utilized to feed the liquid to the nozzle. Furthermore, we state that there is no difference between peristaltic and syringe pumps concerning the quality and quantity of polyplexes obtained after pumping despite their different pumping mechanisms. The addition of surfactants, here PS20, does not improve the performance but accounts for destabilization leading to increased adsorption effects. In the scope of this study, it was not possible to clearly identify the driving mechanism for adsorption as no distinct differences in surface characteristics were found for both silicone tubings despite differences in adsorbed mass. However, neutral surface charge seems to be favorable as no adsorption of DNA was observed on neutrally charged Pumpsil but on positively and negatively charged Silicon and Santoprene, respectively.

We showed adsorption effects for N/P 10 PEI/bDNA polyplexes, but the phenomenon observed is expected to also apply to other polyplex systems containing DNA or RNA. Depending on N/P ratio, polymer or lipid component, hydrophobicity and surface charge, adsorption effects may vary quantitatively and need to be investigated individually. Determining the amount of both the nucleic acid component and the carrier losses individually is crucial for product quality. Overall this study highlights that tubing characteristics and lengths strongly affect product recovery. Therefore, it is recommended to keep the contact area to polymeric tubing materials during production of nucleic acid delivery systems as little as possible for minimum product loss and quality assurance.