Targeted delivery system for cancer cells consist of multiple ligands conjugated genetically modified CCMV capsid on doxorubicin GNPs complex

Targeted nano-delivery vehicles were developed from genetically modified Cowpea chlorotic mottle virus (CCMV) capsid by ligands bioconjugation for efficient drug delivery in cancer cells. RNA binding (N 1-25aa) and β-hexamer forming (N 27-41aa) domain of capsid was selectively deleted by genetic engineering to achieve the efficient in vitro assembly without natural cargo. Two variants of capsids were generated by truncating 41 and 26 amino acid from N terminus (NΔ41 and NΔ26) designated as F1 and F2 respectively. These capsid were optimally self-assembled in 1:2 molar ratio (F1:F2) to form a monodisperse nano-scaffold of size 28 nm along with chemically conjugated modalities for visualization (fluorescent dye), targeting (folic acid, FA) and anticancer drug (doxorubicin). The cavity of the nano-scaffold was packed with doxorubicin conjugated gold nanoparticles (10 nm) to enhance the stability, drug loading and sustained release of drug. The chimeric system was stable at pH range of 4–8. This chimeric nano-scaffold system showed highly specific receptor mediated internalization (targeting) and ~300% more cytotoxicity (with respect to FA− delivery system) to folate receptor positive Michigan Cancer Foundation-7 (MCF7) cell lines. The present system may offer a programmable nano-scaffold based platform for developing chemotherapeutics for cancer.

. These sequences were submitted to I-TASSER and GalaxyGemini tool for the structural conformation (structure function relationship) and assembling potential respectively. Based on the outcome of these analysis, the NΔ41 and NΔ26 truncated capsid protein were selected for development of drug delivery vehicles. These NΔ41 and NΔ26 capsid protein variants were designated as F 1 and F 2 respectively.

Results and Discussion
: CCMV capsid is well-studied and first in vitro assembled capsid from individual components. Protein-protein and RNA interaction play an important role in capsid formation. The N-terminus sequence was employed for identifying and binding the negatively charged RNA strands while the C terminus sequence takes part in intersubunit linking. Hence the role of N-terminus was mainly to recognize and load RNA molecule in icosahedral capsid. However, these sequences were not required for in vitro assembly. Hence, these N-terminus sequences could be deleted without affecting the assembly in vitro 4 . Two variants of N-terminal truncated sequences were reported by deleting first 26 and 41 amino acid of capsid protein 4 . The deletion of these amino acid sequences completely blocked the RNA mediated assembly but had no effect on normal assembly 5 . The 27-35 N-terminal amino acid sequences form β-hexamer responsible for providing the stability to hexameric capsomers. This enhances the in vitro assembly through trimer formation in suitable condition 3 . However, amino acid sequence 44-51 forms the clamps that hold the C-terminal sequences of other protein. This is responsible for formation of non-covalent dimer, which finally lead to formation of aggregates/assembly depending upon solvent composition 3, 6,7 . Thus to achieve efficient virion formation in assembly condition only and also to avoid assembly during overexpression/ purification, two variants of N-terminal truncated capsid protein NΔ41 (F 1 ) and NΔ26 (F 2 ) were proposed to synthesize by recombinant DNA technology based expression. To generate NΔ41 (F 1 ) and NΔ26 (F 2 ), the capsid protein gene were PCR amplified using specific primers having extra sequence for introduction of BamHI (in both forward primers) and XhoI (reverse primer) (Supplementary Table 1). Forward primer 2 GTC GGA TCC GTG GTC CAG CCA GTA ATT GTG F 1 and F 2 Reverse primer ATC CTC GAG GTA AAC CGG GGT GAA GGA GTC PCR amplification using the full-length capsid protein gene as template and specific primers leads to generation of N-terminus deletion variants of capsid proteins (Supplementary Figure 2a). The N terminus deletion mutants were modelled and analyzed in silico for their ability to oligomerize and to form the capsid. The analysis involves the generation of 3D atomic models from iterative structural assembly simulation and multiple threading alignments followed by homology modeling 8 .

Supplementary
Supplementary Figure 2b: Conformational structure of the modified CCMV capsid protein modeled by I-TASSER bioinformatics tool. (i) CCMV capsid protein from PDB (PDB id 1cwp) (ii) for F 1 (NΔ41) and (iii) F 2 (NΔ26). The comparative structural study revealed that both the modified protein has nearly same structure as their original pdb structure. The cloning based additional sequence does not affect the structure function relationship in silico. The additional N-terminal and C-terminal protein sequence were represented in green and blue color respectively.

Supplementary Figure 2c:
In silico analysis of assembling ability using GalaxyGEMINI tool of (i) Native (ii) F 1 (NΔ41) and (iii) F 2 (NΔ26) capsid proteins. The N terminus additional 14 amino acid sequences (by cloning strategies) were shown in red while the C terminus his-tag was shown in blue color.  Figure 3a).

MALDI TOF of Intact protein
Mass spectrometry (MS) was used to determine the molecular mass of F1 and F 2 capsid protein just after the affinity purification. The proteins solutions were extensively dialyzed against Millipore water and 0.8 µl of the sample was spotted on a MALDI plate. Then 1 µl of sinapinic acid was used as matrix after proteins samples were dried. Multiple standard were used for the calibration of the equipment. The data were obtained in MALDI-TOF TOF 4800 using linear spectrum. sequences of F 1 and F 2 protein gene were PCR amplified using primers given in the Table   1 ( Supplementary Figure 3b-ii). These primers added the desired restriction sites at 5ʹ and 3ʹ end for the new genes. The PCR amplified genes of interest (F 1 and F 2 ) was cloned in to an intermediary vector pGEM-T using TA cloning. The cloned pGEM-T vectors were further amplified and the cloning of these genes were confirmed by restriction digestion using BamHI and XhoI. The pGEM-T constructs of 459 bp (F 1 ) and 504 bp (F 2 ) were sequenced to confirm the desired sequence of F 1 and F 2 genes.

Results and Discussion
This cloning helps to improve the cloning efficiency than direct cloning into expression vector pET21a from PCR product. The current cloning strategies was used to add minimum amino acids on N-and C-terminals of each capsid proteins (Supplementary Figure 3a). This is because the capsids proteins need to be in vitro assembled to form the capsid by oligomerization and these additional sequences may have adverse effect on formation of stable drug delivery vehicles. The pGEM-T constructs were cloned into pET-21a for bulk expression in E. coli. The computational analysis of protein demonstrated that the addition of sequences at N-terminus and his-tag at C-terminal derived from pET-21a vector on each capsid protein had minimal effect with respect to folding and structure function relationship 10 . Thus, these gene sequences were cloned into pET-21a vector that characteristically introduced his-tag at C-terminus for ease in purification 11 . pET-21a vector is based on promoter from T7 phage RNA polymerase and requires a host strain with DE3 phage fragment as lysogen, which encodes the T7 RNA polymerase (bacteriophage T7 gene 1) under the control of the IPTG inducible lacUV5 promoter for protein expression (Supplementary Figure 3a). The expression of the recombinant protein using these plasmids was tightly regulated by repression of the lacUV5 promoter by LacI and produces high levels of transcripts and recombinant proteins on induction by IPTG 12 .
T7 RNA polymerase is transcribed when IPTG binds and triggers the release of repressor from the lac operator. The pET-21a construct of modified capsid proteins (F 1 and F 2 ) were confirmed by double restriction digestion (Figure 3b-iv).
The F 1 and F 2 gene construct were cloned into expression vector pET-21a with BamHI and XhoI sites. Usage of BamHI and XhoI restriction enzymes added 14 amino acids at N terminus (MASMTGGQQMGRGS) and eight amino acid (LEHHHHHH) at Cterminus. These additions did not affect the folding and assembly of the capsid. The effect of the additional amino acid residues on folding and possible effect on structure was evaluated using I-TASSER 8 . This prediction tool was used for analyzing the effect of modified sequence to structure-function paradigm. Its prediction are highly reliable because it first generates 3D atomic models from iterative structural assembly simulation and multiple threading alignments followed by homology modeling 8 . The property of Nterminal truncated capsid proteins F 1 and F 2 to form trimeric T=3 symmetry based assembly remained intact after sequence modification. This was also confirmed by Galaxy Web computational tool (http://galaxy.seoklab.org/cgi-bin/submit.cgi?type=GEMINI) (Supplementary figure 2). This tool predicts the oligomerization pattern of proteins at molecular level based on the tertiary structure 9 .

Overexpression and purification of F 1 and F 2 capsid protein
The solubility predictions upon overexpression in E.coli were analyzed using SOLpro tool (http://scratch.proteomics.ics.uci.edu) for both F 1 and F 2 capsid protein of CCMV.
This tool indicated nearly 52% insoluble probability for F 1 and very high probability (~60%) for solubility of F 2 protein 13  HIS-select cobalt affinity gel (Sigma Aldrich, USA) was used for immobilized metal ion affinity chromatography (IMAC). The affinity column was washed with 5-column volume of buffer and equilibrated with chilled 1X PBS buffer. The overexpressed capsid proteins (F 1 and F 2 ) were allowed to bind pre-equilibrated cobalt-NTA beads separately and washed using 1XPBS and 20mM imidazole to remove non-specific bound proteins. Histagged capsid protein was eluted using 250mM imidazole prepared in 1XPBS buffer.

Supplementary Figure 4: Western blot
The affinity purified capsid proteins were visualized in 15% SDS-PAGE gel 14 . His-tagged CCMV capsid protein variants were confirmed by western blotting using anti-his antibody (from mouse). The blots were transferred to nitrocellulose membrane at constant voltage (100V) for an hour.
The membrane was blocked using 5% dried milk in 1X PBS for 1h at 4°C with continuous shaking. This membrane was washed thrice with 1X PBS with 0.1% Tween20.
The washed membrane was incubated with primary antibody (diluted 1:1000) containing 1% BSA in 1X PBS for 1 hour at room temperature. This membrane was again washed thrice to remove the excess antibodies and incubated with secondary antibody (Horse radish peroxidase (HRP)-conjugated rabbit anti-mouse IgG) at 1:8000 dilutions for 1 hour at room temperature. The membrane was washed five times and the blot was developed in dark on X-ray film by chemiluminescence produced due to action of HRP on luminol (dissolved in DMSO) with p-coumaric acid (dissolved in DMSO) and hydrogen peroxide.
These antibodies bind to the C-terminal his-tag of each capsid proteins (F 1 and F 2 ). The For ITF measurements the 0.01mg/ml of F 1 and F 2 capsid proteins were excited at 292 nm and emission was recorded from 300 to 400 nm. The emission spectra for the capsid proteins were found near 340nm, which confirmed the folded state of proteins. The Far UV CD spectra of F 1 and F 2 capsid proteins (0.10mg/ml) were taken from 190 to 260nm at ambient temperature. The CD spectra with the negative ellipticity at 208 nm and positive peak below 200nm suggested the folded structure of the proteins. capsid protein (F 1 and F 2 ) was measured in different conditions e.g. in presence of trifluoroethanol (30% TFE) and denaturing conditions (heat 95°C, 10 min) to see the effect on folding of both proteins.

Far UV circular dichroism (CD) spectrum
The capsid protein sample was diluted in low ionic strength phosphate buffer, pH 7 (50mM) to a final concentration of 0.1mg/ml. The protein was analyzed in a CD spectrometer (JASCO J-815 spectropolarimeter, Japan) using a quartz cell with a path length 1 mm. Spectra was recorded with 1.0 nm bandwidth over the wavelength range of 190 nm to 260 nm. The final spectra was averaged from 3 scans and subtracted from the solvent spectra. Ellipticity values of CD spectra were expressed as the mean residue ellipticity (ϴ).

Supplementary Figure 6: In vitro assembly of CCMV capsid
The assembly of individual F 1 and F 2 capsid protein were analyzed under purification overnight at 4°C. The appropriate condition for assembly was found to be 1:2 molar ratios of F 1 and F 2 and assembly buffer (sodium acetate 0.1M, sodium chloride 0.1M, pH 4.8).
Supplementary Figure 6a: Morphological characterization after purification of (i) F 1 and (ii) F 2 as well as after two months of (iii) F 1 and (iv) F 2 capsid protein. DLS measurement and TEM image of both the capsid protein were taken after the affinity purification (containing 0.25M Imidazole) and after 2 months storage at 4°C in 1x PBS (without in vitro assembling conditions. Both the capsid proteins did not show any assembly or aggregation just after the purification.

Zeta size analysis
The average particle size of CCMV capsid was determined using a zeta particle size analyzer based on the DLS principle (Nano ZS 90, Malvern, UK). The measurement of average particle mean diameter of CCMV capsid (1.2 µM) was performed on zeta sizer equipped with a 5 mW helium /neon laser at 4°C in triplicate. The results were expressed as average mean based on the number as variables for particle size distribution (PSD) in dynamic light scattering (DLS).

Transmission electron microscopy (TEM)
The size and morphology of CCMV capsid nanoparticles was observed under TEM. The in vitro assembled capsids were diluted to 1.2 µM and about 10 µl of dispersion was placed on carbon-coated copper grid. The samples were stained with 0.1% phosphotungstic acid (PTA) solution for 30 sec. The copper grid was allowed to dry at ambient temperature and observed for TEM (FEI Tecnai G20, Germany). The mean diameter of assembled particles was calculated using Image J software. dyes were generated to quantify the dye bioconjugation efficiency. Absorbance spectra of CFSE and AF610 showed characteristic peaks at 492 nm and 610nm respectively. The capsid protein (F 1 , nonconjugated) does not showed absorption at either 492nm or 610nm. However, dye conjugated capsid protein (DF 1 ) showed the characteristic absorption peak of corresponding dyes. These peaks remain while the absorbance was taken using equimolar of non-conjugated capsid proteins. For fluorescence analysis, diluted dye solution, equimolar concentration of native F 1 and completely dialyzed dye conjugated F 1 capsid protein were excited (492 nm for CFSE; & 610nm for AF610) and emission spectra were recorded (CFSE: 500 to 650 nm; AF610: 612 to 700 nm). The absence of fluorescence peak in native protein confirms the bioconjugation and it was used to quantify the conjugation efficiency of both the dyes on capsid protein.

Results and Discussion
HPLC analysis was performed for the determination of conjugation efficiency for Dye, FA and Doxorubicin on capsid protein. The results were given as suppl. Figure 7b (for dye conjugation), 8d (for FA conjugation), and 9d (for doxorubicin conjugation). The HPLC absorbance chromatogram peaks of initial dye and after bioconjugation were averaged as 1609737 and 1200420 respectively. This confirms that nearly ~26% dye was conjugated on capsid proteins. This is equivalent to 5. Calibration curve using standard concentrations of folic acid were generated to quantify the folic acid bioconjugation efficiency. Absorbance spectra of folic acid conjugated (ii) F 1 and (iii) F 2 capsid protein.
These spectra were recorded for folic acid (FA), non-conjugated protein (F 1 or F 2 ) and FA conjugated CCMV capsid proteins ( FA+ F 1 or FA+ F 2 ) using PBS as well as equimolar concentration of F 1 or F 2 as blank.
The non-conjugated protein does not showed absorption at 363nm. However, folic acid conjugated capsid Fluorescence Spectra proteins ( FA+ F 1 or FA+ F 2 ) showed the characteristic absorption peak of folic acid. These peaks remain while the absorbance was taken using equimolar of non-conjugated capsid proteins. The protein was allowed to react with EDC/NHS activated folic acid; FA+ F 1 and FA+ F 2 showed peak at 363 nm due to folic acid conjugation same is absent from the non-conjugated proteins.
The FA conjugation was performed using 10:1 molar ratio of dye and F The HPLC absorbance chromatogram peaks of initial FA and after bioconjugation (first peak) were averaged 45544 and 34720 respectively. This confirms that nearly ~23% FA was conjugated on capsid proteins. This is equivalent to 2.3 molar FA in experimental conditions. Thus, one molecule capsid (used for the experiment) protein has conjugated with approximately 2 molecule of FA. Based on these and work curve model experiment, it was reported that average 2 molecules were individually bio-conjugated with CP (F 2 ).
Supplementary Figure 8e: SDS PAGE Gel of FA conjugated proteins ( FA+ F 1 and FA+ F 2 ). Lanes: 1 and 3 were F 1 and F 2 capsid proteins; lanes 2 and 4 were FA conjugated FA+ F 1 and FA+ F 2 capsid proteins respectively. The conjugated proteins were found intact after FA conjugation.

Supplementary Figure 8f: Biophysical characterization by (i) intrinsic tryptophan fluorescence (ITF) and (ii) Far UV CD of folic acid conjugated capsid proteins.
For ITF measurements the 0.01mg/ml of F 1 , F 2 , FA+ F 1 and FA+ F 2 (Folic acid conjugated) capsid proteins were excited at 292 nm and emission was recorded from 300 to 400 nm. The protein spectra for the capsid proteins were found near 340nm, which confirmed the folded state of proteins. The Far UV CD spectra of F 1 and F 2 as well as folic acid conjugated ( FA+ F 1 and FA+ F 2 ) capsid proteins (0.10mg/ml) were taken from 190 to 260nm at ambient temperature. The CD spectra with the negative ellipticity at 208 nm and positive peak below 200nm suggested the folded structure of the proteins.

Results and Discussion:
The HPLC absorbance chromatogram peaks of initial doxorubicin and after bioconjugation were averaged as 80692 and 44369 respectively. This confirms that nearly ~45% doxorubicin was conjugated on capsid proteins. This is equivalent to 4.5 molar doxorubicin in experimental conditions. Thus, one molecule capsid (used for the experiment) protein has conjugated with approximately 4 to 5 molecule of doxorubicin.
Based on these and work curve model experiment, it was reported that average 4 doxorubicin molecules were individually bio-conjugated with CP (F 2 ). For ITF measurements the 0.01mg/ml of F 1 , F 2 , F 1 dox+ and F 2 dox+ (doxorubicin conjugated) capsid proteins were excited at 292 nm and emission was recorded from 300 to 400 nm. The protein spectra for the capsid proteins were found near 340nm, which confirmed the folded state of proteins. The Far UV CD spectra of F 1 and F 2 as well as folic acid conjugated (F 1 dox+ and F 2 dox+ ) capsid proteins (0.10mg/ml) were taken from 190 to 260nm at ambient temperature. The CD spectra with the negative ellipticity at 208 nm and positive peak below 200nm suggested the folded structure of the proteins.