Synthesis, Characterization and Biocompatibility of a Multifunctional Gold Nanoparticle System for the Delivery of Single-Stranded RNA to Lymphocytes

The use of RNA macromolecules as therapeutic agents for HIV and other infectious diseases is promising but limited by suboptimal delivery to the target site. With HIV infection, this is particularly challenging since lymphocytes are particularly difficult to transfect. This paper describes an innovative strategy for the intracellular delivery of a novel single-stranded RNA (oligoribonucleotide) with putative anti-HIV activity. This strategy is based on a PEGylated gold nanoparticle scaffold covalently linked to the thiol-modified oligoribonucleotide via a cleavable N-succinimidyl 3-(2-pyridyldithio) propionate (SPDP) linker molecule. The nanoparticle was then coated with a cationic polymer (polyethyleneimine) to facilitate cell entry and endosomal escape. A synthetic anti-CD4 cyclic targeting peptide was attached to the polyethyleneimine-coated nanoparticle via an SPDP linker molecule, in an attempt to enhance uptake and selectivity. Synthesis, characterization, SPDP and RNA loading, cytotoxicity and antiviral activity of the nanoparticle are described. Approximately 45 000 strands of RNA were taken up per lymphocyte. Uptake was limited by relatively inefficient loading of RNA onto the gold nanoparticle surface (1 strand per 4.8 nm of nanoparticle surface area) and significant aggregation of the nanoparticle in physiological solutions. No antiviral activity was demonstrated, possibly due to insufficient intracytoplasmic delivery of the RNA.


Introduction
Various strategies exist for the use of RNA macromolecules as therapeutic agents for HIV and other infectious and noninfectious diseases.6] The challenges with efficient delivery of nucleic acids include their short half-life in plasma, [4][5][7][8][9] excretion by the kidneys, entrapment by the reticuloendothelial system, 8 formation of aggregates with serum proteins, 10 inability to cross biological membranes (including the cell membrane) because of strong negative charge 10 , and entrapment within vesicles within the cytoplasm. 5,8Lymphocytes, in particular, are difficult to transfect, 11 which significantly limits the potential of RNA therapeutics for the treatment of HIV infection.
In an attempt to address these challenges, numerous methods have been developed for RNA delivery including viral vector based strategies, direct delivery by physical methods, chemical methods and nanotechnology approaches.However, finding a safe and effective method to achieve systemic and targeted delivery of RNA has remained an elusive goal, [12][13] and siRNA delivery has only recently entered clinical trials. 13is paper describes the synthesis, characterization and biological interaction of a multifunctional nanoparticle construct which attempts to address some of these challenges.The construct is based on the following facets (Fig. 1): 1.5] A similar system, based on inorganic gold particles and biodegradable polycations, has proven to be successful in safely and effectively delivering DNA, [16][17] and more recently RNA, into cells. 15[19][20][21][22][23][24][25] Furthermore, gold nanoparticles enhance the transfection efficiency of polyethyleneimine. 26

PEG polymer to enhance biocompatibility
PEGylation reduces aggregation and non-specific interactions of gold nanoparticles with plasma proteins (i.e.9] Insertion of a PEG 'spacer ' molecule between the gold surface and the RNA also prevents reaction of the nanoparticle with the RESEARCH ARTICLE R. Parboosing, T. Govender, G.E.M. Maguire and H.G. Kruger, 1 S. Afr.J. Chem., 2018, 71, 1-14, <http://journals.sabinet.co.za/content/journal/chem/>.disulphide bond between the RNA and the linker molecule. 15

A cleavable linker molecule (N-succinimidyl 3-(2-pyridyldithio) propionate or SPDP) to allow conjugation and release of the RNA
The thiol modified RNA was conjugated to the gold nanoparticle via a disulphide bond using the linker molecule SPDP.The disulphide bond is stable in the ionic, extracellular environment but is cleaved in the reductive conditions within the cytoplasm (thus releasing the RNA from the nanoparticle). 15More specifically, the disulphide bond is cleaved by intracellular glutathione, which is found in much higher concentration in intracellular vs. extracellular environments and this facilitates the selective intracellular release of the RNA. 30

A single-stranded RNA molecule designed to inhibit HIV packaging
The sequence of the 16-mer single-stranded RNA is based on the packaging signal of HIV-1.It is designed to inhibit the encapsidation of the HIV genome by a decoy mechanism. 31

Polyethyleneimine to enhance cell entry and facilitate endosomal escape
Polyethyleneimine, a synthetic cationic polymer, is the prototype non-viral gene delivery system and a popular choice for RNA transfection, since it protects RNA from enzymatic degradation, enhances cell entry and facilitates endosomal escape. 32Polyethyleneimine facilitates interaction between RNA and the cell membrane (both of which are negatively charged).4][35][36][37][38][39][40][41][42][43][44] Polyethyleneimine then decondenses and separates from the nanoparticle, thus allowing release of the RNA. 39,45However, its use is somewhat limited by toxicity (probably due to unstable complexes in serum, resulting in release of unbound polyethyleneimine). 32old nanoparticles, by comparison, are far less toxic, but their use as transfection agents is hampered by endosomal entrapment and inefficient release of its nucleic acid cargo into the cytoplasm (due to high binding affinity between RNA and the surface of gold nanoparticles). 462][53][54][55] Also, an experimental layer-by-layer electrostatic approach has been used to overcome aggregation issues 42 .We therefore postulated that the use of polyethyleneimine-coated GNP may overcome the barrier to successful delivery of RNA to the cytoplasm of hard-to-transfect lymphocytes.

Peptide functionalization
A synthetic cyclic peptide, designed to target the CD4 receptor, was attached to the surface of the gold nanoparticle via an SPDP linker molecule that facilitated conjugation to the branched polyethyleneimine. 56The design of the peptide was based on the amino acid sequence of an anti-CD4 monoclonal antibody.8][59][60][61] In this paper, it is hypothesized that the presence of a targeting ligand on the surface of the gold nanoparticle would facilitate its selective uptake by CD4 lymphocytes via a receptor-mediated endocytosis pathway.

Synthesis of Gold Nanoparticle
Gold nanoparticles (diameter 30-40 nm) were synthesized using the well-described method of Turkevich, in which chloroauric acid (HAuCl 4 .3H 2 O) is reduced by trisodium citrate (Na 3 C 6 H 5 O 7 .2H 2 O). 65,68Briefly, 500 mL of a 1 mM solution of chloroauric acid in double-distilled water was mechanically stirred and brought to a vigorous boil under reflux.Then 50 mL of 19.5 mM trisodium citrate was rapidly added to the boiling solution.The colour of the solution changed from dark purple to wine red and the reaction was continued for a further 15 min.The solution was allowed to cool to room temperature and stored at 4 °C.All glassware was thoroughly washed with aqua regia, rinsed with double-distilled water and oven-dried overnight prior to use.The size and concentration of the gold nanoparticles was determined by UV-vis spectrophotometry using the method of Haiss. 69The concentration of the gold nanoparticle solution was adjusted to ~30 nM and stored at 2-8 °C until use.

PEGylation of Gold Nanoparticle
The heterobifunctional polymer thiol-PEG-amine contains a thiol group that reacts readily with the citrate-capped surface of the gold nanoparticle, and an amine group that attaches to N-hydroxysuccinimide ester of the SPDP linker in the next step.The thiol-PEG-amine was reacted with the gold nanoparticle using the same number of moles of PEG per nm 2 surface area of gold nanoparticle as Anderson's group. 15In a typical reaction, 32 mg of thiol-PEG-amine was added to 100 mL gold nanoparticle solution (~0.5 nM) and incubated at 25 °C for 12 h with constant mechanical stirring.Tween-20™ at a concentration of 0.01 % (v/v) was added to the solution at the beginning of the reaction to prevent adherence of the nanoparticles to the surface of the reaction vessel.The particles were washed four times by centrifugation at approximately 20 000 × g, followed by removal of the supernatant and re-dispersion in PBS containing 0.01 % Tween-20™.Re-dispersion was aided by vortexing and bath sonication for approximately 10 s each.The absence of aggregates was confirmed by visual inspection and by light microscopy.Successful PEGylation was confirmed by the colloidal stability test. 70Briefly, bare gold nanoparticles aggregate and form a purplish solution, while PEGylated nanoparticles retain their wine red colour when exposed to highly saline conditions (10x phosphate buffered saline).Furthermore, PEGylated gold nanoparticles migrate differently compared to bare gold nanoparticles on gel electrophoresis, due to loss of surface charge upon PEGylation.See Supplementary Material (Appendix A) for further details on the colloidal stability test and gel electrophoresis.The size and concentration of the NH 2 -PEG-gold nanoparticles was determined by UV-vis spectrophotometry. 15,69The solution was stored at 4 o C until ready for use.

Addition of the Linker Molecule (SPDP) to PEGylated Gold Nanoparticle
SPDP was reacted with the PEGylated gold nanoparticle solution using the same number of moles of SPDP per nm 2 surface area of gold nanoparticle as Anderson's group. 15In a typical reaction, 500 µL SPDP in PBS (3 mM, solubilized with 10 % DMSO) was incubated for 40 min with 500 µL NH 2 -PEG-AuNPs in PBS (2.7 nM, with 0.01 % T Tween-20™, pH 7.4) at 25 °C with vigorous vortexing.The unconjugated SPDP was removed by repeated centrifugation and re-dispersion as described for the PEGylation reaction above.

RNA Conjugation
In this step, the 5'thiol of the RNA reacts with the 2-pyridylthio group of SPDP. 15The RNAse-free, HPLC-purified 5' thiol modified, 3' fluorescein labelled RNA construct (Dharmacon, Pittsburgh Pennsylvania, USA) was reduced 71 and de-protected 72 as recommended by the manufacturer.The sample was dried using a centrifugal evaporator (miVac DNA Sample Concentrator, Genevac, Stone Ridge, New York, USA) and re-suspended in Borate Buffered Saline (BBS) with 0.01 % Tween-20™ (pH 8.3-8.5).The BBS was incubated with RNAsecure Reagent (Life Technologies, Grand Island, New York, USA) for 10 min at 60 °C, to remove potential RNAses prior to use.The RNA was quantified by obtaining an absorbance at 260 nm.The software provided by the manufacturer allows an adjustment to be made to account for the effect of fluorescein on the absorbance reading (BioSpec Nano, Shimadzu Corporation, Kyoto, Japan).The SPDP-NH 2 -PEG-gold nanoparticle was re-suspended in 2 mL borate buffered saline containing the RNA (15 µM RNA, 2.5 M NaCl solution, 30 mM borate, pH 8.5, 0.01 % Tween-20™) and incubated at 25 °C for 40 h with vigorous vortexing.The particles were then washed four times by centrifugation at 15 700 × g for 15 min 42 at 4 °C73 , removal of the supernatant and re-dispersion in sterile phosphate buffered saline (137 mM NaCl solution, 10 mM phosphate, 2.7 mM KCl, pH 7.4, 0.01 % Tween-20™).In the final wash, the nanoparticle solution was adjusted to a concentration of 2 µM and re-suspended in 10 mM NaCl solution.Re-dispersion was aided by vortexing and bath sonication for approximately 10 s each.The RNA was protected from light.

Polyethyleneimine Coating
In this step, branched polyethyleneimine with an average molecular weight of 25 KDa was added at a final concentration of 1.0 mg mL -1 to a stirring solution of RNA-gold nanoparticles in 10 mM NaCl solution.This was then incubated at room temperature for 30 min, in order to form a polyelectrolyte layer around the outer surface of the gold nanoparticle.The nanoparticles were then washed by centrifugation and re-dispersion in 10 mM NaCl solution. 42

Size, Polydispersity, Zeta-potential and Concentration
Size, polydispersity, zeta-potential and concentration of the nanoparticles were determined at the end of each step of the synthesis process.The size and concentration of the gold nanoparticles was determined by UV-vis spectrophotometry using the method of Haiss. 69The hydrodynamic diameter, zeta-potential and polydispersity were determined using the Malvern Zetasizer (Malvern Instruments, Worcestershire, United Kingdom).A general-purpose analysis model was selected for size measurements.The angle of detection was 175 °.For zeta potential measurements, the Smoluchowski F(ka) value of 1.5 was used for the Henry's equation.For both size and zeta potential, the automatic setting was selected to allow the software to determine the optimal duration of measurement.

Visualization of Nanoparticles by Electron Microscopy
For visualization of gold nanoparticles, a drop of the solution was placed on a copper grid and allowed to dry and then visualized on a JEOL 2100 High Resolution Transmission Electron Microscope. 82A representative image was imported into Image J and transformed into a monochrome 8-bit format.The scale of the image was set and the 'brightness' threshold adjusted appropriately.The mean, range and standard deviation of the diameter of at least 30 particles was determined using a manual particle sizing method. 83

Aggregation
To study the factors associated with aggregation of polyethyleneimine-coated nanoparticles, the final product was spun down and re-suspended in solutions adjusted to a range of pH values and with FBS (at varying percentages 0, 5, 10 and 20 %) and Tween-20™.Aggregation was quantified visually by dispensing the aliquots of the various nanoparticle solutions into the well of a clear flat-bottomed microtitre plate, which were then viewed under a Leica DMIL light microscope (Leica, Wetzlar, Germany) at ×200 magnification.Images were captured on a Zeiss Axiocam 105 camera (Carl Zeiss, Oberkochen, Germany).The images were imported into Image J software in an 8-bit grayscale format, and the black: white threshold was adjusted by the maximum entropy method. 84The number of black pixels (representing aggregated nanoparticles) was enumerated by Image J software and expressed as a percentage of the total number of pixels.This percentage corresponds to the degree of aggregation visible by light microscopy.The method was validated by preparing serial dilutions of an aggregated nanoparticle and generating a standard curve of percentage aggregation visible by microscopy vs. percentage aggregation by serial dilution.See Supplementary Material (Appendix C) for further details of the method.

SPDP Loading
The loading (i.e. the mean number of SPDP linker molecules per nm 2 surface area of GNP) was determined by a cleave-andanalyze approach.First, a reducing agent (in this case excess dithiothreitol or DTT) was used to cleave the pyridine-2-thione moiety from each molecule of SPDP conjugated onto the surface of the gold nanoparticle.The concentration of pyridine-2-thione (assumed to be equivalent to that of SPDP) was then determined by measuring its absorbance at 343 nm. 85Excess (non-conjugated) SPDP was removed by repeated centrifugation and washing, prior to measurement.See Supplementary Material (Appendix D) for further details of the method and calculations.

RNA Loading
The number of strands of RNA per nanoparticle, prior to polyethyleneimine coating was determined using a detachand-analyze approach, prior to coating with polyethyleneimine. 15Aliquots of the RNA-gold nanoparticle (1 mL each) were incubated for 30 min at 30 o C with either 0.05 M DTT solution in PBS (pH 7.4, 0.01 % Tween-20™) or PBS (pH 7.4, 0.01 % Tween-20™) alone without DTT (as a control).DTT is a reducing agent that cleaves disulphide bonds thus releasing RNA from the surface of the gold nanoparticle.Fluorescence signals from RNA prior to release are quenched by the gold nanoparticle surface plasmon resonance.A 200 µL aliquot was taken for fluorescence measurement after incubation with DTT and PBS control.The aliquot was centrifuged at 15 700 × g for 15 min at 4 °C to pellet the gold nanoparticles thus avoiding quenching effects.Fluorescence of the supernatant was then measured on a Glomax Multimode Detection System, using the Blue filter (Excitation: 490 nm, Emission: 510-570 nm) (Promega Corporation Madison, Wisconsin, USA).The concentration of RNA was calculated by plotting the difference in fluorescence (DTT aliquot less control (PBS only) aliquot) on a standard curve of fluorescently labelled RNA concentration vs. relative fluorescence units. 86See Supplementary Material (Appendix E) for further details of the method and calculations.

Cell Uptake
Uptake of the nanoparticle by MT4 lymphocytes was assessed by electron microscopy, fluorometry, flow cytometry, and fluorescent and light microscopy.

Cells
MT4 lymphocytes (obtained from the NIH AIDS Reagent Program, Division of AIDS, NIAID, NIH from Dr Douglas Richman) were grown at 37 °C in 5 % CO 2 in RPMI 1640 medium (GIBCO/Life Technologies, Grand Island, New York, USA) containing sodium bicarbonate and L-glutamine, supplemented with 10 % heat-inactivated fetal calf serum (Biochrom GmbH, Berlin, Germany).Cells were split 1:10 every 3 days and regularly monitored for viability and contamination.Cells were used at the exponential phase of growth and with viability at least 95 %.The Countess™ Automated Cell Counter (Life Technologies, Grand Island, New York, USA) was used to determine cell counts and viability.

Electron Microscopy
To study uptake of gold nanoparticles, 250 µL of the final nanoparticle product was added to MT4 lymphocytes (5 mL at 3 × 10 5 cells mL -1 in a 25 cm 2 cell culture flask) and incubated overnight at 37 °C with 5 % CO 2 .The cell suspension was then centrifuged for 5 min at 300 × g to form a pellet.The culture medium was removed without disturbing the pellet and replaced with excess 2.5 % glutaraldehyde.The mixture was incubated at room temperature for 10 min, spun again and then resuspended in fresh glutaraldehyde and incubated at 4 °C overnight.The sample was then washed 3 times with phosphate buffer and incubated for 1 h with 0.5 % osmium tetroxide.The specimen was thereafter washed 3x with phosphate buffer, dehydrated with increasing concentrations of acetone, infil-trated with epoxy resin, embedded in fresh resin and then polymerized in an oven for 8 h at 70 °C.Several sections were cut and visualized on a JEOL 2100 High Resolution Transmission Electron Microscope. 87

Flow Cytometry
MT4 lymphocytes (50 000 cells/well) were suspended in wells of a microtitre plate with 200 µL complete RPMI.The cells were then incubated overnight with 25 µL of untreated control (RPMI only), RNA control (10 µM RNA only), uncoated nanoparticle, without polyethyleneimine (3 nM), complete nanoparticle (3 nM) and 10 % DMSO (dead cell control).The cells were then harvested, washed twice by gentle centrifugation, fixed with paraformaldehyde, resuspended in PBS and stored at 4 °C until analysis by flow cytometry (within 24 h).Uptake was determined on BD FACSCanto™ II (BD Biosciences San Jose, CA, USA) instrument.Uptake of RNA was quantified as the percentage of cells that were fluorescent. 88

Fluorescent Microscopy
Cells were treated, harvested and washed as for flow cytometry and then added to poly-L-lysine-coated microscope slides.The slides were fixed with paraformaldehyde and then viewed and photographed under a Nikon fluorescent microscope (Nikon, Tokyo, Japan). 89

Fluorometry
Flow cytometry and epifluorescent microscopy may fail to detect uptake if the fluorescent signal per cell is below the limit of detection for either technique.This may occur if only a small number of gold nanoparticles are taken up per cell, for example, or if there is quenching of the fluorescent signal of the fluorescein-tagged intracellular RNA.In this case, the fluorescent signal may be amplified by examining the cumulative uptake of a large number of cells (e.g.³10 million cells).To achieve this, 10 × 10 6 cells in a cell culture flask were exposed to treatment and the cumulative fluorescent signal quantified by pelleting and lysing the entire volume of cells post-incubation.This method differentiates intracellular from extracellular RNA by thoroughly washing the cells post-incubation (to remove extracellular RNA, whether bound to nanoparticles or not).Furthermore, the method differentiates 'free' RNA vs. RNA bound to the surface of the gold nanoparticle, since the latter is removed by centrifugation prior to measurement.The details of the method are as follows: MT4 lymphocytes were centrifuged at 200 × g and then re-suspended in colourless RMPI with 10 % FCS at a concentration of 1 × 106 cells mL-1.The following nanoparticles and controls were sonicated and added to flasks containing the MT4 lymphocytes in 10 mL RPMI/10 % FCS: The RNA and nanoparticles were incubated with the cells for varying periods (1, 6, 12 and 24 h) .At the end of the incubation period, the cells were washed 4 times (by centrifugation and resuspension in PBS).An aliquot of the supernatant from the final wash was stored for measurement of residual background fluorescence.The cell pellets were frozen at -70 °C for at least 24 h.The cells were then thawed at room temperature and lysed with CyQuant cell lysis buffer as per manufacturer's instructions (Invitrogen/Life Technologies, Grand Island, New York, USA).The sample was centrifuged at 15 700 × g for 15 min at 4 °C to pellet the gold nanoparticles which cause quenching of the fluorescence signal.The fluorescence of the supernatant was measured post-lysis.The background fluorescence was subtracted from this value and the number of RNA strands released per cell was quantified with the use of an RNA standard curve (See Supplementary Material (Appendix E) for detailed calculations).

MTT Cytotoxicity (Methylthiazolyldiphenyl-tetrazolium Bromide) Assay
The MTT assay was used to determine cytotoxicity (50 % inhibitory concentration, IC 50 ), as previously described. 90Briefly, 10 µL of nanoparticle solution from each stage of synthesis was added (in triplicate) to 90 µL complete media in wells of a microtitre plate.The ' outer ' wells of the plate were filled with PBS to avoid 'edge' effects.Serial 10-fold dilutions of the nanoparticles were made.No treatment was added to the last column, which served as the untreated control.Then 60 µL of MT4 lymphocytes (at 6 × 10 5 cells mL -1 ) was added to each well.The plates were incubated for 5 days at 37 °C in 5 % CO 2 in a humidified incubator.At the end of the incubation period, 15 µL of MTT salt (7.5 mg mL -1 dissolved in PBS by sonication) was added to each well using a multichannel pipette.The plate was incubated for 1 h at 37 °C in a CO 2 incubator.Then 100 µL of media was removed without disturbing the cells.The cells were lysed and the formazan crystals solubilized with the addition 100 µL of acidified Tri-ton™X-100 in isopropanol (50 mL isopropanol with 3 mL Trito™n-X100 and 200 µL hydrochloric acid).The plates were placed on a vibrating shaker for 10 min.Absorbance was read on Glomax Absorbance Module (wavelength 540 and 690 nm).The absorbance at 690 nm was subtracted from the absorbance at 540 nm and the results plotted on a graph.The IC 50 was determined by linear extrapolation using the percentage cytotoxicity at each concentration (% cytotoxicity at a given concentration = absorbance at that concentration ¸absorbance of untreated cells × 100).Untreated cells (i.e. cells treated with PBS only) were deemed to have 100 % viability.Growth medium without phenol red was used in the assay to improve sensitivity of absorbance readings. 91

Antiviral Assay
Stock virus was prepared by harvesting supernatant of HIV-infected MT4 lymphocytes at day 5 post-infection.The virus was titrated by conventional TCID methods and stored at -80 °C until use. 90The infectivity and cytopathic effect of the virus used in the assay was confirmed by ensuring that the viability of cells infected by HIV was at least 5-fold less than that of mock-infected cells. 90] The antiviral assay to assess the cytoprotective (antiviral) effect (50 % effective concentration, EC 50 ) and selectivity index (SI) of the nanoparticle was performed as previously described. 90The final nanoparticle solution (10 µL) was added to 90 µL complete media in 6 wells of a column of a microtitre plate.Serial 10-fold dilutions were made.No treatment was added to the last column, which served as the untreated control.Then 60 µL of MT4 lymphocytes (at 6 × 10 5 cells mL -1 ) was added to all wells (except outer wells, to avoid edge effects).To assess cytotoxicity, 50 µL of RPMI was added to each well in the 'upper' half of the plate (mock infection) while 50 µL of HIV IIIB at 300 Tissue Culture Infective Dose (TCID) 50 was added to the 'lower' half (HIV infection) to assess antiviral effect.The plates were incubated for 5 days at 37 °C in 5 % CO 2 in a humidified incubator.At the end of the incubation period, the MTT assay, as described above, was performed to assess toxicity in mock (uninfected) cells and cytoprotective effect in HIV-infected cells.The absorbance readings were plotted on a graph and the IC 50 and EC 50 were determined by linear extrapolation.
The IC 50 was determined as described above.The percentage protection at each dilution was calculated as follows. 90rcentage protection at given concentration = 100 × (HT -HC)/Mc -HC) where HT = average absorbance of treated, HIV infected cells at that concentration, HC = average absorbance of untreated, HIV infected cells, and Mc = average absorbance of untreated, mock infected cells The selectivity index = IC 50 /EC 50 Untreated, uninfected cells and untreated, HIV-infected cells were deemed to have 100 % and 0 % viability, respectively. 90AZT (Azidothymidine) was used as a control (which required simultaneous preparation and incubation of a separate plate, using the same reagents, cell preparation and virus stock).

Characterization
It was necessary to synthesize several batches of gold nanoparticles for the experiments in this paper.The mean core diameter of nanoparticles in all batches fell within the range 30-40 nm.Size, zeta potential and UV-vis absorption spectra (absorbance and ë spr ) of the nanoparticle at each stage of synthesis are shown in Table 1 and Fig. 2. The average diameter in a typical batch, as measured by electron microscopy was 30.08 nm (range 19.07-47.44nm, standard deviation 6.57 nm).Nanoparticles from a typical batch are shown in Fig. 3. Addition of the cyclic peptide-SPDP conjugate to the gold nanoparticle solution resulted in extensive, irreversible macroscopic aggregation, precluding any further characterization, uptake or antiviral studies.
There was significant aggregation of the polyethyleneiminecoated nanoparticles when exposed to FBS (Figs. 4, 5 & 7).The gold nanoparticles referred to in this experiment were studied in its entirety, i.e. with PEG, SPDP, RNA and polyethyleneimine coating, as described in the Synthesis section.Fig. 4 shows a dose response effect between the degree of aggregation and the percentage FBS in the medium.It is important to note that aggregation occurred even at 10 % FBS (the typical concentration of serum used in cell culture media).Aggregation of   nanoparticles was visible by light microscopy within a few minutes of exposure to cell culture medium and plateaued within 24 h.Adjusting the pH of media to pH < 5 and pH >11 (by addition of 10 mM hydrochloric acid or sodium hydroxide, respectively) reversed the aggregation, while Tween-20™ (0.01 %) did not.The presence of live cells did not have any effect on aggregation.

Loading
An average PEGylated gold nanoparticle was decorated with an average of 20350 SPDP linker molecules (or ~2.6 SPDP molecules per nm 2 nanoparticle surface area).This is in keeping with published data on the density of PEG molecules per nm 2 nanoparticle surface area of gold nanoparticle (even if one assumes that SPDP binds to every available PEG on the surface of the nanoparticle).Estimates of PEG loading vary from <1 to ~5 PEG per nm 2 nanoparticle surface area, depending on the molecular weight of the PEG, the shape and size of the nanoparticle and measurement method. 28In particular, the molecular weight (or length) of the PEG molecule drastically influences grafting density, possible due to steric hindrance with higher chain lengths, 29 so that the density varies from 0.32 PEG nm -2 to 3.93 PEG nm -2 for PEG as the molecular weight of the PEG decreases from 51400 to 2100 respectively.Therefore, the loading of SPDP is in keeping with these data (taking into account measurement errors and possibly a ratio other than 1:1 of SPDP conjugated to PEG).
On average, each nanoparticle contained ~1050 strands of RNA (or 1 strand per ~4.8 nm 2 of nanoparticle surface area), prior to polyethyleneimine coating.Therefore, on average there were 26 SPDP molecules per 10 nm 2 nanoparticle surface area, and only ~2 (i.e. 8 %) of these were conjugated to RNA (assuming that RNA is conjugated to SPDP and not directly to the gold nanoparticle surface).By comparison, Anderson's group et al. 15 found that there were 30-40 strands of RNA per 15 nm diameter nanoparticle (or, by our calculations, approximately 1 strand of RNA per 18-24 nm 2 of nanoparticle surface area), suggesting that the loading of RNA in their study was even less efficient.However, any comparisons must take into account that the RNA in Anderson's study was double rather than single-stranded, and that a different method of RNA quantification was used).
Increasing the concentration of RNA did not improve the loading.Typically, <10 % of the RNA added to the reaction mixture was conjugated to the gold nanoparticle and the rest remained unbound and was detectable in the supernatant.Although Lee et al. do not explicitly determine the proportion of RNA that conjugates to the GNP, 15    added to the reaction was conjugated to the surface of the GNP.Therefore, the efficiency of RNA conjugation in our experiments is in keeping with that of Lee et al. 15 Further details may be found in the section on RNA concentration in the Supplementary Material (Appendix F).

Cell Uptake
Fluorometric measurements revealed that approximately 43 polyethyleneimine-coated nanoparticles were taken up per cell (corresponding to ~45 000 strands of RNA) (Table 2).This finding is consistent with electron microscopy enumeration of uptake (~30 nanoparticles per cell, for example, in Fig. 7C).It is difficult to compare these results to published data, for several reasons.Firstly, there are relatively few studies that quantify the number of gold nanoparticles taken up per cell. 94In most studies, uptake is gauged by methods such as flow cytometry, fluorescent microscopy or functional effects, which do not explicitly quantify the number of nanoparticles taken up per cell.8] In particular, uptake appears to be significantly diminished in suspension cultures (as in this paper) as opposed to adherent monolayers. 99Thirdly, the method of enumeration may yield divergent results. 96It is therefore not surprising that estimates of nanoparticle uptake vary widely, from as low as 45 nanoparticles (nanorods) per cell 100 (comparable to the findings of this paper), to hundreds 101 or thousands of nanoparticles per cell. 22,67,94,96,102here was no significant uptake of free RNA and RNA bound to nanoparticles not coated with polyethyleneimine (0 and <600 strands per cell, respectively).Uptake of RNA was below the limit of detection by epifluorescent microscopy and flow cytometry (Fig. 6).A series of experiments were undertaken to improve uptake, without success.See Supplementary Material (Appendix F) for further details of these experiments.
The preceding discussion is based on the assumption that the RNA was efficiently and completely released from the nanoparticle surface.If this was not the case, then uptake may have been underestimated by fluorescent methods (due to quenching of the fluorescent signal by the gold nanoparticle).This may explain, at least partly, the negative results on flow cytometry and epifluorescent microscopy.
Light microscopy and electron microscopy showed significant aggregation of the nanoparticles in the extracellular medium     (Fig. 7).Another study using polyethyleneimine-coated gold nanoparticles for gene delivery has documented similar aggregation issues, which the authors attributed to the lyophilization and reconstitution process used in the preparation of the nanoparticle. 103This is not a consideration since lyophilization was not used in our study.Another study attributed the aggregation to polyethyleneimine reaching its saturation point in the reaction mixture, and suggested optimization of the loading density of the polyethyleneimine on the gold nanoparticle surface (determined to be 1000 molecules of polyethyleneimine per gold nanoparticle in their system). 104This limit was not exceeded in our study.Extracellular aggregation of the nanoparticle may have contributed significantly to the seemingly poor uptake by lymphocytes.

Cytotoxicity and Antiviral Activity
The nanoparticles did not display significant toxicity (Table 1) or antiviral activity (Fig. 8) which was expected, given the poor uptake.

Conclusion
In this paper, a multifunctional nanoparticle, designed to facilitate transfection of a single-stranded RNA into MT4 lymphocytes, was synthesized and characterized.The loading of the RNA per nanoparticle (1 strand per ~4.8 nm 2 of nanoparticle surface area) exceeded that of Lee et al. 15 (1 strand per ~18 nm 2 and note: the RNA used in their study was double-stranded).The multifunctional nanoparticle successfully transfected MT4 lymphocytes, although at low efficiency.Measurement of fluorescence after lysis of treated cells showed uptake of ~45 000 strands of RNA (or 43 nanoparticles) per cell.This degree of uptake was below the limit of detection by flow cytometry and epifluorescent microscopy and was probably insufficient to exert a significant cytotoxic or antiviral effect.However, other reasons, such as quenching of fluorescent signal and lack of potency of RNA construct, may also explain the negative uptake and antiviral effect, respectively.
The limited uptake and antiviral effect was probably due to significant aggregation of the polyethyleneimine-coated nanoparticles in the cell culture medium.Similar aggregation issues with polyethyleneimine-coated nanoparticles have been encountered in previous studies. 42,103It has been shown that aggregation has an effect on uptake, and that such effects are dependent not only on the size of the aggregates, but also on cell type. 105Furthermore, aggregation may be associated with leeching of the polyethelenimine from the surface of nanoparticles, with subsequent loss of its protective effect and thus degradation of RNA by nucleases (although we did not confirm this experimentally).In our study, aggregation was quantified by photographic software and showed a dose response effect to RESEARCH ARTICLE R. Parboosing, T. Govender, G.E.M. Maguire and H.G. Kruger, 10 S. Afr.J. Chem., 2018, 71, 1-14, <http://journals.sabinet.co.za/content/journal/chem/>.increasing concentration of FBS, which suggests that the aggregation is induced by serum factors, most likely proteins.Attachment of the cyclic peptide ligand to the surface of the polyethyleneimine-coated gold nanoparticle exacerbated the aggregation.
It must be noted that we did not study whether particles were taken up singly, in clusters or as aggregates, nor did we study their fate upon entry into the intracellular environment (particularly with regard to whether aggregation increased or decreased over time, and how quickly, if at all, they exited the cell).Furthermore, we did not study the effect of aggregation on the release of RNA from the surface of the nanoparticle.If RNA release was hampered due to aggregation, this may have had led to quenching of fluorescent signals and subsequent underestimation of uptake.These are aspects for further study, probably by the use of ex vivo, real-time imaging techniques. 106he conjugation of RNA to the gold nanoparticle surface via the SPDP linker molecule was relatively inefficient (peaking at 1 strand per 4.8 nm 2 of nanoparticle surface area).This means that the RNA was conjugated to only ~8 % of SPDP binding sites available on the surface of the gold nanoparticle.Over 90 % of the RNA fails to conjugate and remains in the supernatant post-reaction.Another group using a similar strategy to link RNA onto gold nanoparticles revealed a similar low efficiency (by our calculation, only 6 % of available SPDP binding sites were conjugated to RNA).Increasing the input concentration of RNA did not lead to an increase in RNA loading per nanoparticle.This is a major limitation compared to transfection using dendrimers for example, where effectively all the RNA is complexed into the dendriplex. 31urthermore, the inefficient binding of RNA to the surface of the gold nanoparticle may be a mechanistic factor that contributes to the aggregation of the nanoparticle.The low density of RNA (as evident by relatively high zeta potential values) implies that the surface of the gold nanoparticle may not have sufficient negative charge to retain the positively charged polyethyleneimine upon its surface (which may have led to partial leaching of the polyethyleneimine and aggregation -see Fig. S3 in Supplementary Material).The negative charge of gold nanoparticles at the RNA conjugation stage (prior to polyethyleneimine coating) in our experiments did not exceed ~8 mV.By comparison, the zeta potential achieved in a layer-by-layer (non-covalent) approach by Elbakry et al. was at least double this value (-20 to -40 mV). 42However, other factors, such as the size of the gold nanoparticle (either as single entities or clusters) and the length of polyethyleneimine are likely to play a role since the wrapping process is governed by the length of the polymer relative to surface curvature of the nanoparticle. 42he reasons for the low conjugation of RNA to the linker molecule may include steric hindrance or charge repulsion, which prevent high-density packing of covalently bound RNA molecules.By contrast, a layer-by-layer approach facilitates higher density RNA packing due to the presence of the positively charged polyethyleneimine in the underlying layer (which may counteract charge repulsion).Furthermore, the density and orientation of RNA packing in a layer-by-layer approach are not limited by the availability of linker molecules for covalent bond formation.Possibly, for these reasons, there were 780 siRNA strands per 15 nm nanoparticle (i.e.~1.1 RNA per nm 2 ) in Elbakry's layer-by-layer approach, which is more than 5-fold greater loading than achieved by the approach in our paper.
Although the construct described in this paper had limited uptake and antiviral effect, details provided in terms of synthesis and characterization of the nanoparticle and quantification of aggregation may be useful for the assessments of nanoparticles with similar issues.Further studies are required to understand the factors that may explain and improve the loading of RNA onto the surface of gold nanoparticles using the approach adopted in this paper.In addition, it will be of interest to study whether aggregation may be prevented by increasing the loading of the RNA, or alternatively replicating this study with other cationic polymers (including low molecular weight and linear polyethyleneimine).Finally, it is not known whether this phenomenon is unique to single-stranded RNA and studies with double-stranded RNA and DNA may yield more favourable results.at the Microscopy & Microanalysis Unit of UKZN for use of the confocal microscope and Phillip Christopher for assistance with electron microscopy and the DST-NRF Internship program for intern (William Serumula) to assist with laboratory maintenance and assistance.Dr Parboosing received an Academic Fellowship award from the Discovery Foundation of South Africa and funding from the National Research Foundation Thuthuka Program.
Dr Louis Chonco, under the supervision of Dr R. Parboosing, performed the experiments to provide a positive control and assisted with the images for Fig. 6.To avoid side-reactions, cyclization was performed on resin with the N-terminal amino acid still protected with the FMoc group.A disulphide bond was formed between two the cysteine residues (C*) at positions 2 and 15, by iodine-oxidation.The details of the procedure are as follows: 1.The resin was swelled with DMF (3x).
3. The peptide was oxidized with I2 (10 eq) in DMF for 1.5 hours, with gentle stirring at room temperature.
5. Peptides were cleaved from the resin and the remaining side chain protecting groups removed with 1% TIS(tri-isopropylsilane) + 1% thioanisole + 1% 1,2 ethanedithiol + 95% TFA in DCM for 2 hours. 13e peptides were purified on a Younglin ACME 9000 instrument using an ACE C-18 reverse phase semi-preparative High Performance Liquid Chromatography (HPLC) column and mass spectra were obtained on a Bruker ESI-QTOF mass spectrometer and Shimadzu Prominence LC-MS System.The purified peptide was lyophilized using a VirTis benchtop K freeze dryer and stored at -20°C until use.

Linker Conjugation
SPDP is an amine-sulfhydryl crosslinking molecule that has an amine reactive portion (Nhydroxysuccinimide (NHS) ester) and a sulfhydryl-reactive portion (2-pyridyldithio group) The 2-pyridyldithio group reacts optimally with sulfhydryl-containing molecules, such as cysteine, between pH 7 and 8.The cysteine at the end of the "AAC" tail of the cyclic peptide was conjugated to SPDP as follows: 25µL of 20mM SPDP solution was added to 2 mg cyclic peptide dissolved in 1 mL of PBS.The mixture was incubated for 30 minutes at room temperature.Excess SPDP was then removed by Sephadex G-25 PD-10 desalting column (GE Healthcare Life Sciences, Pittsburgh, Pennsylvania, USA).Then the solution was added to 1 mL polyethyleneimine coated gold nanoparticle and incubated overnight to allow conjugation of the peptide-SPDP with the amine side chains of polyethyleneimine via the NHS ester of SPDP.The excess peptide-SPDP was removed by repeated centrifugation and washing with PBS.Germany).

Legend
5. The images were imported into Image J software in an 8-bit grayscale format.
6.The black:white threshold was adjusted by the maximum entropy method. 26The choice of threshold adjustment method was made empirically.The method that most accurately represented the percentage of black pixels that corresponded to visual approximation was chosen.The important point is that data using different methods cannot be compared.Therefore, the method, once chosen, was used for all experiments in which data was compared.
7. The number of black pixels (representing aggregated nanoparticles) was enumerated by Image J software and expressed as a percentage of the total number of pixels.
8. In order to generate a standard curve, serial, 2-fold dilutions of a highly aggregated gold nanoparticle solution were made with PBS, in the wells of a clear, flat bottom microtiter plate.
The solution was thoroughly mixed at each step to ensure that aggregated nanoparticles were likewise diluted.Dilutions were prepared in triplicate.Photographs were taken of each dilution and the percentage black pixels was determined as described above (steps 1-7).The percentage black pixels was plotted against serial dilutions of the aggregated gold nanoparticle (log transformed to base 2) (Figure S4).The method showed excellent correlation (r = 0.99) between percentage black pixels vs. actual aggregation of serially diluted aliquots in the range of 0. 8.The difference in absorbance was calculated as follows: Difference in absorbance = GNP-DTT -(GNP-PBS) Where GNP-DTT = mean absorbance of supernatant of GNP solution treated with DTT GNP-PBS = mean absorbance of supernatant of GNP solution treated with PBS PBS-TWEEN was read as "Blank".9.The molar concentration of the SPDP was then calculated as follows: Molar concentration (in mM) = difference in absorbance ÷ (8.08 mM -1 cm -1 x path-length in cm), where 8.08 x 10 -3 M -1 cm -1 is the"extinction coefficient" or molar absorptivity for pyridine-2thione at 343 nmand the path-length, when using the BioSpec Nano in the 0.7 mm mode.
In a typical experiment, the molar concentration of pyridine-2-thione in the final 1 mL aliquot was found to be 407 µM or 407 000 nM (based on mean absorbance difference of 0.23 units).The concentration of gold nanoparticle in the same aliquot was 20nM (as determined by absorbance measurements). 27erefore, each nanoparticle was decorated with an average of (407 000 ÷ 20 =) 20350 SPDP linker molecules (or ~2.6 SPDP molecules per nm 2 nanoparticle surface area).
number of nanomoles of RNA per million cells (e.g.7.5 x 10 -5 nmoles per million cells) which corresponds to 45 000 strands of RNA per cell.
The limit of detection of RNA uptake per cell was calculated based on the lower limit of detection of RNA of 1nM, corresponding to a Relative Fluorescence Units (RFU) = 5.3.Moreover, RFU values <7 (i.e.falling within 2 standard deviations away from the mean) are likely due to instrument "noise" rather than biological factors.Therefore, only readings > 7 RFU above the mean were regarded as significant.
The standard deviation of the fluorescence measurements was calculated by reading 30 wells of a "blank" plate.

Figure 1
Figure 1 Schematic representation of multifunctional, PEGylated gold nanoparticle construct, showing RNA with putative anti-HIV activity, attached via an SPDP linker.The particle was coated with polyethyleneimine and bears a cyclic anti-CD4 peptide on the surface.The attachment of the SPDP linker to the thiol-PEG-amine and RNA is shown in more detail on the right.Portion of this figure are based on Lee et al.15 a. Buffer control: 200 µL PBS b.Complete nanoparticle: 200 µL GNP-PEG-SPDP-RNA-polyethyleneimine nanoparticle (~3 nM) c.Uncoated nanoparticle (without polyethyleneimine) 200 µL AuNP-PEG-SPDP-RNA nanoparticle (~3 nM) d.Free RNA: 200 µL RNA (~10 µM)

Figure 2
Figure 2 UV-vis spectra of the gold nanoparticle construct at various stages of synthesis.The raw data were downloaded from the BioSpec Nano instrument and plotted in Excel™.Absorbance values were normalized for differences in concentration by setting the absorbance at 450 nm to 1 for all samples.(GNP = bare gold nanoparticle and PEG = PEGylated gold nanoparticle, i.e. gold nanoparticle conjugated to thiol-PEG-amine and SPDP = SPDP linker conjugated to PEGylated gold nanoparticle and RNA = RNA conjugated to PEGylated gold nanoparticle via SPDP linker and PEI = polyethyleneimine-coated nanoparticle following RNA conjugation they provide sufficient data for the calculation to be made.In their study, only 6 % of the RNA RESEARCH ARTICLE R. Parboosing, T. Govender, G.E.M. Maguire and H.G. Kruger, 7 S. Afr.J. Chem., 2018, 71, 1-14, <http://journals.sabinet.co.za/content/journal/chem/>.

Figure 3
Figure 3 Electron micrograph of ~30 nm diameter polyethyleneimine-coated gold nanoparticles.The distribution of the diameter of the particles is shown on the right.

Figure 4
Figure 4 Graph showing percentage aggregation (as determined by light microscopy and Image J software) at various concentrations of FBS.Data presented as mean ± 1 standard deviation of three independent experiments.

Figure 5
Figure 5 Black and white (8-bit) image of aggregated gold nanoparticles under light microscopy (A) and the transformed image used by ImageJ (B) to estimate the 2-D surface area (16.43 %) of the aggregated nanoparticles (C).

Figure 6 Figure 7
Figure 6 Epifluorescent microscopy (left) and flow cytometry results (right) showing untreated MT4 lymphocytes (A) and lymphocytes treated with polyethyleneimine-coated gold nanoparticles containing fluorescently tagged RNA (B).The nuclei of cells stained with DAPI are clearly visible on epifluorescent microscopy (blue).Visualization using the fluorescein filter showed no obvious uptake of the polyethyleneimine-coated gold nanoparticle.The corresponding flow cytometry results show no difference between the untreated control (A) and cells treated with the polyethyleneimine-coated gold nanoparticles (B) (3.8 vs. 3.5 %).By comparison, in an experiment with FITC labelled gold nanoparticles (C), there is obvious uptake (indicated by green fluorescence) and the corresponding flow cytometry results indicate that 96.6 % of treated cells had positive fluorescence (shown in blue, in this case).(See Acknowledgements section.)

Figure 8
Figure 8 Graphical representation of results of the antiviral MTT assay.(A) Results for the polyethyleneimine-coated nanoparticle, which show no activity or cytotoxicity at the concentrations tested.The AZT control (B) has potent antiviral activity (EC 50 = 0.05 µM) and is toxic only at high concentrations (IC 50 = 100.7 µM), with selectivity index (IC 50 /EC 50 ) = 2014.

Figure S2 :
Figure S2: Gel electrophoresis of gold nanoparticles showing migration of negatively charged bare gold nanoparticles towards the positive pole.PEGylated gold nanoparticles (left lane) and PEGylated gold nanoparticles with SPDP linker molecule (right lane) have a neutral charge and do not migrate.

Figure S3 : 2 . 3 . 4 .
Figure S3:Aggregation occurs when several nanoparticles are attracted to the same polyethyleneimine molecule. 20, 22-25A. Non-aggregated gold nanoparticles coated with polyethyleneimine.Exposure to serum proteinsis hypothesised to result in leeching of polyethyleneimine from the surface of nanoparticles (B, C).This results in the disorderly and random association of several (>1) nanoparticles with the same molecule of polyethyleneimine, resulting in the formation of aggregates (D, E).The polyethyleneimine used in this paper is branched but is shown here as a linear molecule for simplicity.The aggregatesshown here (D and E) consists of 2 nanoparticles (a doublet); the number of nanoparticles contained in aggregates described in this paper is unknown, but is most likely much larger and variable.

3 to 50 % 10 Figure S4 :Figure S5 :
Figure S4:Graph showing linear relationship between serially diluted aggregated gold nanoparticles and photographically determined aggregation (expressed as a percentage of black pixels)

Figure S6 :
Figure S6:Graph showing log-linear relationship between RNA concentration (in pM) and relative fluorescence units (dimensionless) over a wide range (from about 1 nM to 10 µM).Fluorescence readings of RNA below 1 nM concentration were close to zero and are not shown.The dotted line is the linear trend line.

Table 1
Size, zetapotential and IC 50 of gold nanoparticles.

Table 2
Fluorometric measurement of uptake of RNA and nanoparticles by MT4 lymphocytes.