at SciVerse ScienceDirect

Biomaterials 34 (2013) 3479e3488

Contents lists available

Biomaterials

journal homepage: www.elsevier .com/locate/biomater ia ls

The effect of crosslinking agents on the transfection efficiency, cellular andintracellular processing of DNA/polymer nanocomplexes

Hao Zheng, Cui Tang, Chunhua Yin*

State Key Laboratory of Genetic Engineering, Department of Pharmaceutical Sciences, School of Life Sciences, Fudan University, Shanghai 200433, China

a r t i c l e i n f o

Article history:Received 13 November 2012Accepted 19 January 2013Available online 8 February 2013

Keywords:NanocomplexesCrosslinking agentsGene deliveryCellular and intracellular processingTransfection efficiency

* Corresponding author. Tel.: þ86 21 6564 3797; faE-mail address: chyin@fudan.edu.cn (C. Yin).

0142-9612/$ e see front matter � 2013 Elsevier Ltd.http://dx.doi.org/10.1016/j.biomaterials.2013.01.072

a b s t r a c t

Cellular and intracellular processing of DNA/polymer nanocomplexes was optimized by tailoring thecomposition of crosslinking agents for improving in vitro and in vivo transfection efficiency. Nano-complexes composed of trimethyl chitosan-arginine conjugate (TMC-Arg), plasmid DNA (pDNA), anddifferent proportions of sodium tripolyphosphate (TPP) and poly(g-glutamic acid) (g-PGA) were pre-pared. All TMC-Arg nanocomplexes (TANC) possessed similar particle sizes and preferable protection ofpDNA against degradation. The Zeta potentials of TANC decreased with increasing amount of TPP, whichwere positively correlated to their cellular uptake levels. The composition of crosslinking agents affectedtheir internalization mechanisms, wherein the addition of g-PGA changed from clathrin-mediatedendocytosis to caveolae-mediated one. The increment of TPP amount in TANC was responsible fortheir reduced binding affinity to pDNA and rapid pDNA release, which was related to their subcellulardistribution and in vitro and in vivo transfection patterns. More compact TANC were associated witha delayed protein expression while easily dissociated ones gave a faster onset of action and higher shortterm gene transfer. However, TANC that dissociated too readily had the inability of gene transfectionowing to pDNA degradation in the endolysosomes. Therefore, tailoring the composition of crosslinkingagents in nanocomplexes may provide a feasible tool for improving transfection efficiency.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Gene therapy holds great promises to prevent and treat geneticand acquired human diseases [1]. However, ferrying therapeuticexogenous genes into target cells safely and efficiently remains tobe an insurmountable barrier. In view of safety concerns for potentviral vectors that have restricted their applications [2,3], the diversecationic polymers have emerged as competitive and attractivealternatives due to their enhanced biodegradability and bio-compatibility, unrestricted payload sizes, and potential of repeatedadministration [4,5]. To mediate highly effective transfection, pol-ymeric vectors have to traverse multiple obstructions includingbeing internalized, escaping from degradative endosomal com-partments, dissociating appropriately to release genetic payload,and escorting nucleic localization and accumulation [6]. Unfortu-nately, most polycations would fail at one or even more of theobstacles mentioned above, resulting in the inability of gene de-livery [7].

Chitosan is a promising vehicle for gene delivery due to its bio-compatibility and biodegradability. The strong interaction between

x: þ86 21 5552 2771.

All rights reserved.

chitosan and plasmid DNA (pDNA) can protect their payload fromnuclease degradation, however, which can also prevent their dis-sociation within the cells, thereby precluding efficacious trans-fection. Researchers have demonstrated that the addition ofa negatively charged crosslinking agent could improve in vitrotransfection efficiency of chitosan through addressing variousintracellular challenges. The incorporation of poly(g-glutamic acid)(g-PGA) in chitosan/pDNA nanocomplexes can promote cellularuptake, change the internalization pathways, and improve intra-cellular release of pDNA, leading to the enhanced in vitro trans-fection efficiency [8]. By adsorbing pDNA onto the surface ofchitosan/alginate nanoparticles, a relationship between reducedstrength of chitosan-pDNA interaction and improved in vitrotransfection efficiencyhas beenobserved [9]. However, to the best ofour knowledge, the effects of the addition of a negatively chargedcrosslinking agent on in vivo gene expression are unavailable.

Trimethyl chitosan-arginine conjugate (TMC-Arg) was devel-oped as a pDNA delivery vector, aiming at combining good solu-bility of trimethyl chitosan (TMC) and enhanced membranepermeability as well as nuclear localization of arginine residues[10e13]. In this study, nanocomplexes composed of TMC-Arg,pDNA, and different proportion of crosslinking agents (sodiumtripolyphosphate (TPP) and g-PGA) were prepared via ionic

H. Zheng et al. / Biomaterials 34 (2013) 3479e34883480

gelation. A series of nanocomplexes with different cellular andintracellular processing were obtained by modulating the compo-sition of crosslinking agents. Their cellular and intracellular pro-cessing was monitored in terms of cell binding, cellular uptake,internalization mechanisms, pDNA release, and subcellular distri-bution. Furthermore, the influences of the composition of cross-linking agents on the extent and kinetics of transfection both onHEK293 cells in vitro and after intramuscular administration in vivowere investigated.

2. Materials and methods

2.1. Materials and animals

Chitosan (deacetylationdegreeof85%andmolecularweight (Mw)of100kDa)wasobtained from Golden-shell Biochemical Co., Ltd. (Zhejiang, China). L-arginine hy-drochloride, 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride(EDC),N-hydroxysuccinimide (NHS), TPP, fluorescein isothiocyanate (FITC), and ethidiumbromide (EB)were fromSigma (St. Louis,MO,USA). g-PGA (Mwof200kDa) andDNaseI were obtained from Vetan (Taiwan, China) and Worthington (Lakewood, NJ, USA),respectively. pDNA encoding Enhanced Green Fluorescent Protein (pEGFP) wasamplified in E. coli and purified by EndoFree Maxi Plasmid Kit (TianGen, Beijing,China).

HEK293 cell lines (human embryonic kidney cells) were obtained from theAmerican Type Culture Collection (ATCC, Rockville, MD, USA) and cultured in Dul-becco’s modified Eagle’s medium (DMEM, Grand Island, NY, USA) supplementedwith 10% fetal bovine serum (FBS).

Male Kunmingmice (6 weeks old, body weight 20� 2 g) were obtained from theAnimal Centre of Fudan University. Animal experiments were performed accordingto the Guiding Principles for the Care and Use of Experiment Animals in FudanUniversity. The study protocol was reviewed and approved by the InstitutionalAnimal Care and Use Committee, Fudan University, China.

2.2. Synthesis and characterization of TMC-Arg

TMC-Arg was synthesized as previously described with a slight modification[10,14]. Firstly, chitosan was reacted with CH3I for 120 min to obtain TMC withquaternization degree of around 30%. Secondly, TMC was further modified witharginine through amide reaction between the amino groups on TMC and the car-boxyl groups on arginine catalysed by EDC and NHS. Briefly, the carboxyl groups ofarginine were activated by EDC and NHS in TEMED/HCl solution (pH 4.8) for 1 h andthen conjugated to the amino groups of TMC under stirring for 10 h at room tem-perature. The resultant product was purified by dialysis against water for 3 days(MWCO, 3500 kDa) followed by lyophilization. TMC-Arg was characterized by 1HNMR in D2O on AVANCE DMX 500 NMR spectrometer (Burker, Germany).

2.3. Nanocomplexes formation and characterization

Ionic gelationmethodwas employed toprepare TMC-Argnanocomplexes (TANC).In brief, pDNA solution (0.2 mg/mL) was mixed with g-PGA solution (1 mg/mL) andTPP solution (0.2mg/mL) at variousweight ratios as listed in Table 1. Themixturewasfurther added into TMC-Arg solution (2 mg/mL) dropwise under stirring. TANC wereincubated at room temperature for 30 min before use.

The particle sizes and Zeta potentials of TANC were determined by ZetasizerNano (Malvern, Worcestershire, UK). The association efficiency of TANC wasassessed by electrophoresis on 1% (w/v) agarose gel in the presence of EB witha current of 130 V for 30 min.

2.4. pDNA protection

TANC containing 1 mg of pDNAwere incubated with 5 units of DNase I or 5 mL ofmouse serum at 37 �C for 30 min. The DNase I or serum was then inactivated byadding 2.5 mL of EDTA (200mM) and incubating at 80 �C for 5 min. Heparinwas used

Table 1Composition and characterization of TMC-Arg nanocomplexes (TANC).

Weight ratio(TMC-Arg/pDNA/TPP/g-PGA)

Particle size (nm)a Zeta potential(mV)

TANC1 30/3/0/0 135.0 � 6.1 (0.243) 30.4 � 2.7TANC2 30/3/1/2 132.3 � 2.6 (0.243) 25.9 � 1.1TANC3 30/3/2/1 104.9 � 3.7 (0.211) 23.9 � 1.7TANC4 30/3/3/0 149.8 � 0.8 (0.152) 20.5 � 2.8

a Values in parentheses represented the polydispersity index (P.I.).

to dissociate pDNA and the integrity of pDNA was assessed by agarose gel electro-phoresis. Naked pDNA treated with DNase I or serum was served as a positivecontrol.

2.5. Cell viability

To determine the cytotoxicity of TANC, HEK293 cells were seeded on 96-wellplate at the density of 1 � 104 cells per well and cultured for 24 h prior to addingTANC to each well at the concentration of 2 mg pDNA/mL. Cells were incubated for24, 48, 72, and 96 h followed by methyl tetrazolium (MTT) assay. Cells treated with0.2 M PBS (pH 7.4) were served as controls.

2.6. Cell binding and cellular uptake

For determination of cell binding, HEK293 cells were harvested and resus-pended in isotonic buffer (44.4 g/L glucose, 200 mg/L KCl, 2.90 g/L Na2HPO4∙12H2O,200 mg/L KH2PO4; pH 7.4). Then, cells were incubated with TANC under agitation at37 �C for 2 h before determining the Zeta potential.

FITC-pDNA was synthesized as previously described [15], and allowed to formTANC as described in Section 2.3. HEK293 cells were seeded on 24-well plate at thedensity of 4 � 104 cells per well and incubated at 37 �C for 24 h. TANC at the dose of2 mg FITC-pDNA/mL were added to each well. After incubating for 4 h, cells werewashed with 0.2 M PBS (pH 7.4) three times to remove TANC adhering onto thecytoplasmic membranes and lysed with 0.5% (w/v) SDS (pH 8.0). The cell lysate wasquantified for FITC-pDNA using VARioSKAN Flash microplate reader (Thermofisher,USA) (lex ¼ 488 nm, lem ¼ 519 nm) and protein content by Lowry method. Uptakelevels were expressed as the amount of FITC-pDNA associated with 1 mg of cellularprotein.

To investigate the mechanisms involved in internalization, cells were pretreatedwith genistein (200 mg/mL) or chlorpromazine (10 mg/mL) for 30 min andthroughout the 4-h uptake experiment at 37 �C. Results were expressed as theuptake percentage of the control where cells were incubated without inhibitortreatment.

2.7. Heparin displacement

Different concentrations of heparin were added to TANC and the mixture wasallowed to incubate at room temperature for 2 h to dissociate pDNA, after which itwas loaded onto a 1% (w/v) agarose gel in the presence of EB and run at 130 V for30 min to visualize the bands.

2.8. EB exclusion

TANC were stained with EB (0.1 mg/mL) at room temperature for 10 min beforemonitoring its fluorescence intensity (lex¼ 518 nm, lem ¼ 605 nm). The results weredenoted as the relative fluorescence where EB alone and pDNA combined with EBserved as 0 and 100, respectively.

2.9. In vitro release

TANC containing FITC-pDNA were suspended in PBS (pH 7.4) or acetate buffer(pH 5.0) and then incubated at 100 rpm and 37 �C. At each predetermined time, thesuspension was centrifugated at 13,300 rpm for 30 min and the supernatant wascollected to quantify for FITC-pDNA content by fluorimetry.

2.10. Subcellular distribution

Subcellular distribution of TANC was evaluated according to the modifiedmethod described previously [16]. In brief, HEK293 cells were collected by cen-trifugation after incubating with TANC containing FITC-pDNA for 0.5, 1, 2, 4, 8, and12 h, respectively. The cell pellet was resuspended in 300 mL of TM-2 buffer (10 mMTriseHCl, 2 mM MgCl2, 0.5 mM PMSF; pH 7.4), followed by incubating at roomtemperature for 1 min and in ice for 5 min. Cells were then lysed with 0.5% (v/v)Triton X-100 in ice and centrifugated at 800 rpm for 10min. The pellet was dissolvedin the lysis buffer (0.5% SDS, pH 8.0) to determine the pDNA level in the nucleus andthe supernatant was directly used to measure the pDNA level in the cytoplasm.Results were expressed as the relative percentage of pDNA internalized.

2.11. In vitro transfection

HEK293 cells were seeded on 24-well plate at the density of 4 � 104 cells perwell and incubated for 24 h. The cell culture medium was replaced with FBS-freeDMEM prior to adding TANC at the concentration of 2 mg pDNA/mL. Following in-cubation for 4 h, the medium was changed to DMEM containing 10% FBS, anda further incubation was allowed for another 20 h, 44 h, 68 h, and 92 h, respectively.Cells were then trypsinized and suspended in ice-cold 0.2M PBS (pH 7.4) for analysisof transfected cells by flow cytometry (FACScan, BD, USA). Cells treated with nakedpDNA and Lipofectamine 2000/pDNA nanocomplex were served as negative and

H. Zheng et al. / Biomaterials 34 (2013) 3479e3488 3481

positive controls, respectively. For the qualitative observation, transfected cells at48 h were visualized using IX71 fluorescence microscopy (Olympus, Japan).

2.12. In vivo gene expression

TANC were injected into the posterior tibialis muscles of mice at a dose of 10 mgof pDNA permouse. Naked pDNA and Lipofectamine 2000/pDNA nanocomplex wereserved as a negative and positive control, respectively. After 3 and 7 days, mice weresacrificed and the muscles were isolated. Following homogenized with RIPA buffer(50 mM TriseHCl (pH 7.4), 150 mM NaCl, 0.1% SDS, 0.1% NP40), the lysate wascentrifugated at 12,000 rpm for 15 min and the supernatant was quantified forfluorescence intensity by spectrofluorimetry (lex ¼ 485 nm, lem ¼ 538 nm) andprotein content by the Lowry method. The results were presented as fluorescenceintensity associated with 1mg protein. For the qualitative observation, at 3 days postadministration, muscles were cryofixed in OCT medium and cut into section (10 mmthickness) using CM1900 cryomicrotomy (Leica, Heerbrugg, Switzerland). GFPexpression was observed using fluorescence microscopy (CLSM, Olympus, Japan).

2.13. Statistical analysis

All data were expressed as the mean with its standard deviation (mean � SD).Student’s t-test was employed to make comparison between two groups. One-WayAnalysis of Variance (ANOVA) was used to test among three or more groups. Dif-ferences were judged to be statistically significant when P < 0.05.

3. Results

3.1. Synthesis and characterization of TMC-Arg

As depicted in Fig. 1, the peaks at 3.40 ppm in 1H NMR spectra ofTMC and TMC-Arg corresponded to the protons of quaternizedmethyl groups. The peak at 2.45 ppm in 1H NMR spectra of TMC-Arg was assigned to the protons of alkyl groups in Arg, confirm-ing the successful synthesis of TMC-Arg. The quaternization de-grees of TMC and TMC-Arg were around 30% as calculatedaccording to the method described previously [17].

Fig. 1. 1H NMR spectra of TMC and TMC-Arg. 3.40 ppm: quaternized methyl groups ofTMC and TMC-Arg (*); 2.45 ppm: alkyl groups of Arg (arrow).

3.2. Nanocomplexes formation and characterization

Four kinds of TANC, namely TANC1, TANC2, TANC3, and TANC4,were prepared through adding various proportions of TPP andg-PGA as listed in Table 1. Although the positively charged TANCpossessed similar particle sizes ranging from 100 nm to 150 nmwith uniform size distribution (polydispersity index less than 0.3),their Zeta potentials decreased with the increase of TPP amount(Table 1). Agarose gel electrophoresis was used to examine theassociation efficiency of pDNA in TANC, and complete retardation ofpDNA demonstrated their desirable condensation capacity towardspDNA regardless of the composition of TANC (Fig. 2a).

3.3. pDNA protection

The protection effect of TANC to pDNA against DNase I and serumwas displayed in Fig. 2b and c, respectively. Naked pDNA was com-pletely digestedwith nobands observed. By contrast, clearmigrationbands of pDNA loaded into TANC were noted, suggesting efficaciousprotection effect of TANC to pDNA regardless of their composition.

3.4. Cell viability

As shown in Fig. 3a, negligible cytotoxicity was observed for allTANC at the same condition applied for transfection experiment,which provided a warranty of successful gene transfer.

Fig. 2. (a) Agarose gel electrophoresis of TMC-Arg nanocomplexes. Lane 1e5 repre-sented naked pDNA, TANC1, TANC2, TANC3, and TANC4, respectively. (b) Protection ef-fect of TMC-Arg nanocomplexes to pDNA against DNase I. Lane 1e6 represented nakedpDNA, naked pDNAþDNase I, TANC1þDNase I, TANC2þDNase I, TANC3þDNase I, andTANC4þ DNase I, respectively. (c) pDNA protection of TMC-Arg nanocomplexes againstserum. Lane 1e6 represented naked pDNA, naked pDNA þ serum, TANC1 þ serum,TANC2 þ serum, TANC3 þ serum, and TANC4 þ serum, respectively.

Fig. 3. (a) Viability of HEK293 cells analysed by MTT assay after being exposed to TMC-Arg nanocomplexes. Cells treated with PBS were served as controls. Indicated values weremean � SD (n ¼ 6). (b) Cell binding of TMC-Arg nanocomplexes onto HEK293 cells. Cells treated with PBS were served as controls. Indicated values were mean � SD (n ¼ 3).(c) Cellular uptake of TMC-Arg nanocomplexes in HEK293 cells following incubation for 4 h at 37 �C. Indicated values were mean � SD (n ¼ 3). * Statistically significant differencesobserved from TANC1 (*P < 0.05, **P < 0.01). (d) The effects of inhibitor treatment on the internalization of TMC-Arg nanocomplexes. Indicated values were mean � SD (n ¼ 3).*P < 0.05.

H. Zheng et al. / Biomaterials 34 (2013) 3479e34883482

3.5. Cell binding and cellular uptake

The binding of TANC onto the surface of HEK293 cells was con-firmed by the rise of Zeta potential after incubating with TANC(Fig. 3b). Preferable cellular uptake was found in TANC with higherZetapotential, as evidencedby the cellular uptake amount increasingin theorderof TANC4< TANC3<TANC2< TANC1 (Fig. 3c). Therefore,the composition of crosslinking agents affected the Zeta potentials ofTANC, thusmaking an effect on cellular uptake levels. Cellular uptakeinhibitors were adopted to investigate the internalization mecha-nisms of TANC. As indicated in Fig. 3d, treatment with chlorproma-zine yielded a depression of internalization by 69.9% and 61.7% forTANC1 and TANC4, respectively, while genistein induced an inhibi-tion of internalization by 75.7% and 58.0% for TANC2 and TANC3,respectively. These results indicated that the composition of TANCexerted remarkable influence on their internalization pathways.

3.6. pDNA binding and release

To assess the influence of the composition of crosslinking agentson binding affinity of TANC towards pDNA, various concentrationsof heparin were incubated with TANC and the resultant mixtures

were electrophoresed on agarose gel to observe the bands of dis-sociated pDNA. The concentration of heparin required to disruptthe structure of TANC1 was 1 mg/mL, while that of the other TANCwas 0.2 mg/mL (Fig. 4a). To further discriminate the bindingstrength between TMC-Arg and pDNA, EB exclusion assay wasutilized [18]. The fluorescence quenching percentage of TANC1,TANC2, TANC3, and TANC4 was 76.2%, 72.6%, 63.5%, and 52.1%,respectively, suggesting a decreased interaction between TMC-Argand pDNA as the amount of TPP in TANC increased (Fig. 4b). Bothheparin displacement and EB exclusion assay indicated that thebinding affinity of TANC towards pDNA was dependent on theircomposition of crosslinking agents.

Different binding strength obviously affected in vitro releaseprofiles of pDNA from TANC, which were determined both underneutral and acidic conditions to simulate the environment in thecytoplasm and endolysosome, respectively. As shown in Fig. 4c andd, the release rate and extent of pDNA was positively correlatedwith increasing amount of TPP in TANC regardless of the releasemedium, which was accorded well with EB exclusion assay. Inaddition, the amount of pDNA release in acidic medium wasremarkably lower than that in neutral medium, suggesting thatpDNA was not easily dissociated in endolysosomal compartments.

Fig. 4. (a) Heparin displacement assessment. Lane 1e7 represented naked pDNA, TMC-Arg nanocomplexes incubated with heparin with the concentration of 0.2, 0.5, 1, 2, 4, and6 mg/mL, respectively. (b) EB exclusion assay to determine the binding strength. Indicated values were mean � SD (n ¼ 3). *Statistically significant differences observed from TANC1(**P < 0.01, ***P < 0.001). (c) Accumulative release profiles of TMC-Arg nanocomplexes in 0.2 M PBS (pH 7.4). Indicated values were mean � SD (n ¼ 3). (d) Accumulative releaseprofiles of TMC-Arg nanocomplexes in 0.2 M acetate buffer (pH 5.0). Indicated values were mean � SD (n ¼ 3).

H. Zheng et al. / Biomaterials 34 (2013) 3479e3488 3483

3.7. Subcellular distribution

Fig. 5 showed subcellular distribution of TANC in HEK293 cells.Nuclear accumulation of TANC1 increased with time up to 12 h,while that of the other TANC declined after reaching the peakvalues at different time point. Release profiles of TANC, which werestrongly correlated with the composition of crosslinking agents,might contribute to this difference in subcellular distribution pat-terns. Besides, highest accumulation amount of pDNA in the nu-cleus was achieved by TANC3 at 4 h.

3.8. In vitro transfection

Fig. 6a showed the extent and kinetics of gene expression ofTANC in HEK293 cells. The transfection efficiency mediated byTANC1 increased with time, while that of TANC2 reached a plateauat 72 h. The transfection efficiency of TANC3 reached the peak valueat 48 h, which was significantly higher than the other TANC(P < 0.05), but experienced a continuous decrease afterwards.TANC4 exhibited inferior ability of persistent gene transfection,however, its transfection efficiency was high at 24 h. The releaseprofiles of TANC, which were governed by the composition ofcrosslinking agents, were responsible for their difference inin vitro transfection patterns. Naked pDNA mediated negligible

gene expression and Lipofectamine 2000/pDNA nanocomplex wasinferior to TANC3 in terms of in vitro transfection efficiency. Thequalitative observation was well consistent with the quantitativeresults obtained at 48 h (Fig. 6b).

3.9. In vivo gene expression

To investigate the impacts of the composition of crosslinkingagents on in vivo gene transfection, intramuscular protein expres-sion was analysed in the duration of 3 and 7 days. After 3 days postnanocomplexes administration, transgene expression increased inthe order of TANC1< TANC2< TANC3 (Fig. 7a). However, after 7days, the GFP expression levels of TANC1 and TANC2 increased by2.2 and 1.5 folds, respectively, which were comparable to those ofTANC3. TANC4 and naked pDNA evoked an unappreciable GFPexpression in vivo, and the transfection efficiency of Lipofectamine2000/pDNA nanocomplex was also marginal. Fluorescence micro-scopy images confirmed the quantitative results of 3 days postTANC administration (Fig. 7b).

4. Discussion

Cationic polymer-based nanocomplexes suffered from unsatis-factory transfection efficiency, which severely limited their

Fig. 5. Subcellular distribution of TMC-Arg nanocomplexes in HEK293 cells from 0.5 h to 12 h. Indicated values were mean � SD (n ¼ 3).

H. Zheng et al. / Biomaterials 34 (2013) 3479e34883484

therapeutic applications. Therefore, gaining insight into the key pa-rameters and rate-limiting steps involved in the polymer-based de-livery systems was of vital importance. It appeared that differentintracellular processing of cationic polymer-based nanocomplexeswould trigger the discrepancies in transfection efficiency [8,19e21].Therefore, in the current study, TANC with different cellular andintracellular processing were obtained by tailoring the compositionof crosslinking agents, thus allowing us to investigate their differ-ences in transfection efficiency both in vitro and in vivo.

TMC-Arg was synthesized by conjugating the carboxyl groups ofarginine to the amino groups of TMC as evidenced by 1H NMRspectra (Fig. 1). In the previous investigation, TMC with Mw of15 kDa and quaternization degree of around 70% was adopted tosynthesize TMC-Arg [13]. In this study, to enhance the pDNAcompaction capacity, TMC with Mw of 100 kDa was chosen [22].Meanwhile, the quaternization degree declined to 30% to reducethe cytotoxicity caused by extra positive charges of trimethylgroups [10].

TMC-Arg bore positive charges in neutral condition, and couldinteract with negatively charged pDNA, TPP, and g-PGA throughelectrostatic interaction to form nanocomplexes. The resultantTANC possessed similar particle sizes, but exhibited decreased Zetapotentials with the increase of TPP amount (Table 1). TANC couldeffectively condense pDNA and protect pDNA from enzymatic deg-radation as evidenced by the gel retardation assay, which indicatedthat adding TPP and g-PGA exerted unappreciable influence on thecondensation capacity and protective effect of TMC-Arg (Fig. 2).Adequate protection of pDNA integrity allowed the possibility ofevaluating the effects of cellular processing on transfection

efficiency both in vitro and in vivowhile eliminating the implicationsof pDNA degradation in the extracellular environment.

Cytotoxicity of TANC in HEK293 cells was firstly assessed andno damage to cell viability was observed at the same conditionapplied for transfection experiment, which guaranteed the execu-tion of subsequent cellular procedures (Fig. 3a). To accomplishgene transfer, a sequential of cellular barriers lay on the pathway,including cellular association and uptake, endolysosomal escape,timely dissociation, and nuclear transportation. All of these obsta-cles would have implications on both in vitro and in vivo trans-fection efficiency. Adherence to cell surfaces was considered as thefirst step to trigger transfection [23]. Positively charged TANCexhibited high binding affinity to negatively charged cell mem-branes of HEK293 cells through electrostatic interaction (Fig. 3b).Following adsorption onto the cell membranes, TANC proceeded tobe internalized via assorted endocytosis pathways. The cellularuptake levels of TANC showed a positive correlation with theirsurface charges that were dependent on the composition ofcrosslinking agents (Fig. 3c).

Internalization pathways were associated with intracellulartrafficking of nanocomplexes and the subsequent gene expression[24]. Therefore, investigations into cellular uptake pathways wereconducted to shed light on transfection efficiency. A remarkablereduction following chlorpromazine treatment (P< 0.05) suggesteda clathrin-mediated process dominant in the internalization ofTANC1 and TANC4 [25], implying that adding TPP did not alter theinternalization pathways of TANC. By contrast, the addition ofg-PGA changed the internalization pathway to caveolae-mediatedendocytosis as evidenced by significant uptake inhibition after

Fig. 6. (a) In vitro transfection kinetics of TMC-Arg nanocomplexes in HEK293 cells from 24 h to 96 h. Cells treated with naked pDNA and Lipofectamine 2000/pDNA nanocomplexwere served as negative and positive controls, respectively. Indicated values were mean � SD (n ¼ 3). # Statistically significant differences observed from TANC3 at 48 h (##P < 0.01,###P < 0.001). (b) Fluorescence microscopy images of transfected cells at 48 h.

H. Zheng et al. / Biomaterials 34 (2013) 3479e3488 3485

genistein treatment (P< 0.05) [25], and the degree of inhibitionwaspositively correlated with the g-PGA amount (Fig. 3d). The uptakemechanisms of nanocomplexes were dependent on their compo-sition and physiochemical properties [26,27]. A previous reportindicated that nanocomplexes with diameters below 200 nm werelikely to be internalized via clathrin-mediated endocytosis [28],which was in accordance with the uptake mechanisms of TANC1,TANC3, and TANC4. As for the uptake pathways of TANC2 andTANC3, the involvement of caveolae-mediated endocytosis could beattributed to the addition of g-PGA in the nanocomplexes [20]. Forefficacious transfection, caveolae-mediated endocytosis might beadvantageous over clathrin-mediated one owing to its possibleavoidance of trafficking to the hostile endolysosomal vesicles [29].

Collectively, the composition of crosslinking agents in TANCaffected their cellular uptake and internalization pathways, both ofwhich would induce their disparities in the transfection efficiencyboth in vitro and in vivo.

After entering into cytoplasm, nanocomplexes had to unpackagetimely to liberate their genetic payload to initiateprotein expression,which implied that the degree of polymer-pDNA interaction exertedremarkable influence on eventual transfection efficiency [30]. In thisstudy, heparin displacement and EB exclusion were simultaneouslycarried out to assess the binding affinity of TANC towards pDNA.Heparin could liberate pDNA from nanocomplexes via competitiveinteraction with cationic polymers, and the amount of heparinrequired to displace pDNA was positively correlated with bindingstrength [31]. EB exclusion could reflect pDNA compaction ability bymeasuring the reduction in fluorescence [18]. Both experimentsconfirmed that TANC with distinct binding affinity for pDNA wereobtained by merely modulating the proportion of TPP and g-PGA.Moreover, an increased amount of TPP was responsible for thereduced binding strength (Fig. 4b). In vitro release assay (Fig. 4c andd) suggested that an elevated release rate and extent both underacidic and physiological conditions strongly correlated with the

Fig. 7. (a) In vivo transgene expression of TMC-Arg nanocomplexes for 3 and 7 days, respectively. Indicated values were mean � SD (n ¼ 4). *P < 0.05. #Statistically significantdifferences observed from Lipofectamine 2000/pDNA nanocomplex for 3 days (###P < 0.001). $ Statistically significant differences observed from Lipofectamine 2000/pDNAnanocomplex for 7 days ($$$ P < 0.001). (b) At 3 days post administration, fluorescence microscopy images of GFP expression in posterior tibialis muscles of mice. Scale bars ¼ 50 mm.

H. Zheng et al. / Biomaterials 34 (2013) 3479e34883486

decreased binding strength. Although both anionic TPP and g-PGAcould competewithpDNA to interactwithpolycations, their abilitiesto reduce the association efficiency of nanocomplexes towardspDNA were different. Compared to micromolecular TPP, the chainentanglement effect of macromolecular g-PGA with pDNA mightmake TANC more compact to some extent, leading to its weakeneddisassociation capacity. Therefore,modulating the proportion of TPPand g-PGA, especially for TPP, could produce differentiation inbinding affinity towards pDNA, which resulted in disparate extentand kinetics of pDNA release. In addition, the accumulative releaseamount of pDNA in neutral medium (cytoplasm) was significantlyhigher than that in acidic conditions (endolysosome), implying thatthemajority of pDNA encapsulated in TANCmight be released in thecytoplasm for successive procedures.

Thereafter, pDNA was required enter into the nuclei to initiatetranscription and the subsequent protein expression. The ability oftransporting pDNA to localize and accumulate in the nucleus wasquantitatively determined for the TANC (Fig. 5). A correlation be-tween pDNA release profiles and the nucleic accumulation patterns

of TANC was found herein. TANC1 with higher pDNA binding af-finity underwent a steady increase of nucleic accumulation duringa 12 h period, owing to its slow release rate. By contrast, other TANCwith relatively “loosened” conformations could make more rapidlyliberated pDNA accumulate in the nuclei and the faster the pDNAreleased from TANC, the sooner it reached the summit. This dis-crepancy in nucleic accumulation of TANC would definitely have animplication on in vitro and in vivo transfection efficiency.

As discussed above, changing the composition of crosslinkingagents elicited distinct cellular uptake levels, internalizationmechanisms, pDNA release profiles, and subcellular distributionpatterns of TANC. These differences ultimately affected in vitrotransfection efficiency of TANC on HEK293 cells. TANC3 displayedthe highest transfection efficiency at 48 h (P< 0.05), whichmight beascribed to preferable cellular adsorption and uptake levels, an in-termediate release profile, favourable nucleic accumulation, anda specific internalization pathway dominated by caveolae-mediatedendocytosis which avoided endolysosomal degradation [25]. Bycontrast, an insufficient uptake amount and a “fast” release rate in

H. Zheng et al. / Biomaterials 34 (2013) 3479e3488 3487

combination with undesirable clathrin-mediated endocytosiswhich led to entering endolysosomes, made TANC4 exhibit theunappealing gene transfer ability. In spite of being internalized viaclathrin-mediated endocytosis, TANC1 possessed higher uptakeamount and a slow release rate, thus possibly allowing the majorityof pDNA payload to enter into the cytoplasm and initiate persistentgene expression subsequently.

When and where nanocomplexes liberate genetic payload wascentral to transfection efficiency for polymeric delivery vectors. Thein vitro release behaviour of TANC not only affected their intra-cellular distribution, but also impacted their kinetics of transfectionefficiency from 24 h to 96 h (Fig. 6). A previous report had proventhat nanocomplexes with more stable structures allowed for a rel-atively longer duration of gene expression due to persistent releaseof pDNA [32]. Therefore, TANC1 with strong binding affinity to-wards pDNA experienced a continuous increase in transfection ef-ficiency from 24 h to 96 h. By contrast, TANC with more facilelydissociated structures mediated a faster onset of action and pro-voked a higher gene expression in comparatively shorter time [33].This was confirmed by the fact that transfection efficiency of TANC2and TANC3 achieved peak values at 72 h and 48 h, respectively.TANC4 with the most “loosened” structure showed prematuredissociation prior to reaching the action site of pDNA, possiblyleading to the degradation of dissociated pDNA in the endolyso-somes. Consequently, TANC4 mediated inferior gene transfectionexcept at 24 h. Such results suggested that the degree of interactionbetween polymer and its pDNA payload affected the intracellularand endosomal release of pDNA, which was crucial to the kineticsof in vitro transfection efficiency.

Inert in vivo transfection efficiency has frustrated the applicationof most cationic polymer gene delivery systems. The release be-haviours of TANC, which were controlled by the composition ofcrosslinking agents, also affected the in vivo transfection efficiencypatterns. TANC with faster release rates presented a higher in vivogene expression over the duration of 3 days, as evidenced byincreased protein expression levels from TANC1 toTANC3. At 7 daysfollowing administration, an increase in gene expressionwas notedfor both TANC1 and TANC2, which was comparable to that ofTANC3. This could be explained by the fact that relatively stablenanocomplexes might be advantageous to long term transfectionin vivo. TANC4 liberated pDNA too early to ensure sufficient intactcoding sequence to initiate protein expression, leading to inferiorgene transfer ability in vivo. Additionally, insufficient uptake levelsmight be another contributing factor.

5. Conclusions

Tailoring the composition of crosslinking agents in TANC wasshown to affect their cellular and intracellular processing includingcellular uptake, internalization pathways, pDNA release, andnucleic accumulation, which allowed us to investigate the key pa-rameters for the extent and kinetics of transfection both in vitro onHEK293 cells and in vivo after intramuscular administration.The high transfection efficiency of TANC3 might be ascribed totheir preferable cellular adsorption and uptake levels, caveolae-mediated endocytosis involved in, an intermediate release profile,and favourable nucleic accumulation. These findings could beserved as guidelines for the rational design of polymeric vectors forimproving transfection efficiency.

Acknowledgements

The authors are thankful for the financial support from theNational Natural Science Foundation of China (No. 81172995).

References

[1] GrigsbyCL, LeongKW.Balancing protection and release of DNA: tools to addressa bottleneck of non-viral gene delivery. J R Soc Interface 2010;7:S67e82.

[2] Luo D, Han E, Belcheva N, Saltzman WM. A self-assembled, modular DNAdelivery system mediated by silica nanoparticles. J Control Release 2004;95:333e41.

[3] Salem AK, Searson PC, Leong KW. Multifunctional nanorods for gene delivery.Nat Mater 2003;2:668e71.

[4] Niidome T, Huang L. Gene therapy progress and prospects: non-viral vectors.Gene Ther 2002;9:1647e52.

[5] Zhang XQ, Wang XL, Zhang PC, Liu ZL, Zhuo RX, Mao HQ, et al. Galactosylatedternary DNA/polyphosphoramidate nanoparticles mediate high gene trans-fection efficiency in hepatocytes. J Control Release 2005;102:749e63.

[6] Wang BQ, He CB, Tang C, Yin CH. Effects of hydrophobic and hydrophilicmodifications on gene delivery of amphiphilic chitosan based nanocarriers.Biomaterials 2011;32:4630e8.

[7] Gabrielson NP, Pack DW. Acetylation of polyethylenimine enhances genedelivery via weakened polymer/DNA interactions. Biomacromolecules 2006;7:2427e35.

[8] Peng SF, Yang MJ, Su CJ, Chen HL, Lee PW, Wei MC, et al. Effects ofincorporation of poly(g-glutamic acid) in chitosan/DNA complex nano-particles on cellular uptake and transfection efficiency. Biomaterials 2009;30:1797e808.

[9] Douglas KL, Piccirillo CA, Tabrizian M. Effects of alginate inclusion on thevector properties of chitosan-based nanoparticles. J Control Release 2006;115:354e61.

[10] Kean T, Roth S, Thanou M. Trimethylated chitosans as non-viral gene deliveryvectors: cytotoxicity and transfection efficiency. J Control Release 2005;103:643e53.

[11] Kim T, Ou M, Lee M, Kim SW. Arginine-grafted bioreducible poly(disulfideamine) for gene delivery systems. Biomaterials 2009;30:658e64.

[12] Gao Y, Xu ZH, Chen SW, Gu WW, Chen LL, Li YP. Arginine-chitosan/DNA self-assemble nanoparticles for gene delivery: in vitro characteristics and trans-fection efficiency. Int J Pharm 2008;359:241e6.

[13] Morris VB, Sharma CP. Folate mediated in vitro targeting of depolymerisedtrimethylated chitosan having arginine functionality. J Control Release 2010;348:360e8.

[14] Liu WG, Zhang JR, Cao ZQ, Xu FY, Yao KD. A chitosan arginine conjugate asa novel anticoagulation biomaterial. J Mater Sci Mater Med 2004;15:1199e203.

[15] Ishii T, Okahata Y, Sato T. Facile preparation of a fluorescence-labeled plasmid.Chem Lett 2000;4:386e7.

[16] Zhao YX, Gao JQ, Sun XY, Chen HL, Wu LM, Liang WQ. Enhanced nucleardelivery and cytotoxic activity of hydroxycamptothecin using o/w emulsions.J Pharm Pharm Sci 2007;10:61e70.

[17] Thanou MM, Kotze AF, Scharringhausen T, LueBen HL, de Boer AG, Verhoef JC,et al. Effect of degree of quaternization of N-trimethyl chitosan chloride forenhanced transport of hydrophilic compounds across intestinal Caco-2 cellmonolayers. J Control Release 2000;64:15e25.

[18] Ren Y, Jiang X, Pan D, Mao HQ. Charge density and molecular weight of pol-yphosphoramidate gene carrier are key parameters influencing its DNAcompaction ability and transfection efficiency. Biomacromolecules 2010;11:3432e9.

[19] Wang J, Gao SJ, Zhang PC, Wang S, Mao HQ, Leong KW. Polyphosphoramidategene carriers: effect of charge group on gene transfer efficiency. Gene Ther2004;11:1001e10.

[20] Peng SF, Tseng MT, Ho YC, Wei MC, Liao ZX, Sung HW. Mechanisms of cellularuptake and intracellular trafficking with chitosan/DNA/poly(g-glutamic acid)complexes as a gene delivery vector. Biomaterials 2011;32:239e48.

[21] Ho YP, Chen HH, Leong KW, Wang TH. Evaluating the intracellular stabilityand unpacking of DNA nanocomplexes by quantum dots-FRET. J ControlRelease 2006;116:83e9.

[22] Mao S, Sun W, Kissel T. Chitosan-based formulations for delivery of DNA andsiRNA. Adv Drug Deliver Rev 2010;62:12e27.

[23] Zhao X, Yin LC, Ding JY, Tang C, Gu SH, Yin CH, et al. Thiolated trimethylchitosan nanocomplexes as gene carriers with high in vitro and in vivotransfection efficiency. J Control Release 2010;144:46e54.

[24] Khalil IA, Kogure K, Akita H, Harashima H. Uptake pathways and subsequentintracellular trafficking in nonviral gene delivery. Pharmacol Rev 2006;58:32e45.

[25] Douglas KL, Piccirillo CA, Tabrizian M. Cell line-dependent internalizationpathways and intracellular trafficking determine transfection efficiency ofnanoparticle vectors. Eur J Pharm Biopharm 2008;68:676e87.

[26] Pichon C, Billiet L, Midoux P. Chemical vectors for gene delivery: uptake andintracellular trafficking. Curr Opin Biotechnol 2010;21:640e5.

[27] Zhang LW, Monteiro-Riviere NA. Mechanisms of quantum dot nanoparticlecellular uptake. Toxicol Sci 2009;110:138e55.

[28] Rejman J, Oberle V, Zuhorn IS, Hoekstra D. Size-dependent internalization ofparticles via the pathways of clathrin-and caveolae-mediated endocytosis.Biochem J 2004;377:159e69.

[29] Rejman J, Bragonzi A, Conese M. Role of clathrin- and caveolae-mediatedendocytosis in gene transfer mediated by lipo- and polyplexes. Mol Ther2005;12:468e74.

H. Zheng et al. / Biomaterials 34 (2013) 3479e34883488

[30] Forrest ML, Meister GE, Koerber JT, Pack DW. Partial acetylation of poly-ethylenimine enhances in vitro gene delivery. Pharm Res 2004;21:365e71.

[31] Kwok A, Hart SL. Comparative structural and functional studies of nanoparticleformulations for DNA and siRNA delivery. Nanomedicine 2011;7:210e9.

[32] Lavertu M, Methot S, Tran-Khanh N, Buschmann MD. High efficiencygene transfer using chitosan DNA nanoparticles with specific combinations

of molecular weight and degree of deacetylation. Biomaterials 2006;27:4815e24.

[33] Koping-Hoggard M, Varum KM, Issa M, Danielsen S, Christensen BE,Stokke BT, et al. Improved chitosan-mediated gene delivery based on easilydissociated chitosan polyplexes of highly defined chitosan oligomers. GeneTher 2004;11:1441e52.