Electrostatic anti-CD33-antibody–protamine nanocarriers as platform for a targeted treatment of acute myeloid leukemia

Development of an AML-specific siRNA carrier system

In order to develop an siRNA-based targeted therapy to treat leukemia, we chose the humanized anti-CD33-monoclonal antibody (αCD33-mAB) gemtuzumab as a carrier. We expressed the αCD33-mAB in CHO-S-cells (Fig. 1A) and isolated it via affinity chromatography (Fig. 1A and B). Then, the αCD33-mAB was conjugated via the linker sulfo-SMCC to commercially available salmon protamine (Fig. 1B). As depicted here, we found out that antibodies coupled with SMCC-protamine form vesicular structures in the presence of free SMCC-protamine when they were complexed with anionic cargo molecules such as siRNA (see below in this manuscript and [10]) or anionic ibrutinib-Cy3.5 [6]. Therefore, we abbreviate the empty carrier components as αCD33-mAB-P/P. Cysteine residues within the αCD33-mAB were conjugated to SMCC-protamine, detectable by a molecular weight shift (Fig. 1B), which resulted in the antibody–protamine complex (Fig. 1C). As revealed by electrophoretic band-shift assays, the αCD33-mAB-P/P can complex around 16 molecules of siRNA, while a higher molar excess of siRNA leads to an overflow of unbound siRNA (Fig. 1D). Specific CD33 internalization induced by αCD33-mAB-protamine complex incubation of CD33-positive cells was confirmed by flow cytometry (Additional file 1: Figure S1 A-D). Moreover, αCD33-mAB-P/P-siRNA complexes internalized efficiently and specifically into CD33-positive OCI-AML2, OCI-AML3 and KG1 AML cells (Fig. 1L-N and P-R), whereas CD33-negative OCI-Ly19 DLBCL cells did not internalize αCD33-mAB-P/P-Alexa488-control-siRNA complexes (Fig. 1O and S). As anticipated when αCD33-mAB-P/P-Alexa488-control-siRNA complexes were incubated on coated slides overnight, micellar-like Alexa488-positive structures appeared (Fig. 1E-K) depending on the αCD33-mAB-P/P:siRNA ratio, illustrating the formation of a supramolecular complex that we call αCD33-mAB-P/P-siRNA nanocarrier.

Fig. 1figure 1

The αCD33-monoclonal antibody (mAB) gemtuzumab-protamine (αCD33-mAB-P/P) conjugates bind and transport siRNA only into CD33-positive cells. A We cloned a plasmid expressing the heavy (HC) and light chain (LC) of the full IgG1 monoclonal antibody gemtuzumab. After transfection of the production cell line CHO-S, the mAB was purified via FPLC. Shown here is a representative elution profile (right panel). IRES, internal ribosomal entry side; GFP, green fluorescent protein. B First, protamine was conjugated chemically to the linker sulfo-SMCC, then this linker was coupled to the αCD33-mAB. The naturally anionic siRNA could then complex by electrostatic binding to the protamine moieties. In complex with siRNA, the αCD33-mAB-P and free SMCC-protamine spontaneously form vesicular structures, which we call αCD33-mAB-P/P-siRNA nanocarrier. C Coomassie-stained SDS–PAGE showing uncoupled αCD33-mAB and αCD33-mAB coupled with SMCC-protamine (αCD33-mAB-P/P). The shift of molecular weight after protamine conjugation via sulfo-SMCC to the αCD33-mAB can be seen for the heavy chain (HC-P) and for the light chain (LC-P). SMCC-P, unbound SMCC-protamine; M, molecular weight marker. D Band-shift assay. Agarose gel-electrophoretic analysis of the binding capacity of siRNA to the αCD33-mAB-P/P complex. Up to 16 mol siRNA per mol of mAB-P can bind to the αCD33-mAB-P/P complex. E–K Fluorescence microscopy of nanocarriers. Vesicles were formed by self-assembly upon incubation of 60 nM αCD33-mAB-P/P with 600 nM Alexa488-control-siRNA for 2 h at RT and documented by fluorescence microscopy after immobilization of the nanocarriers on coated slides without cells o/n at 37 °C. L–S Fluorescence microscopy analysis of siRNA internalization mediated by αCD33-mAB-P/P-siRNA complex into CD33-positive cells (L-N and P-R) but not into negative control cells (O and S). Upper panels (L-O): Green fluorescence depicting Alexa488-control siRNA in vesicular structures within the cells. Lower panels (P-S): Overlay of blue Hoechst staining and green fluorescence from upper panels. α, anti

The αCD33-mAB-P/P-nanocarrier directed RNAi inhibits target gene expression and colony growth in DNMT3A-mutated CD33-expressing AML cell lines

The DNA-methyltransferase DNMT3A is mutated in a high number of AML patients (20–30%, [12, 15, 16]), however, the consequences of this mutation event are not fully understood today. According to its prominent abundance even preceding leukemogenesis, we hypothesized an oncogenic function of DNMT3A mutated protein driving AML and intended to inhibit it via RNAi. We verified by Western blot analysis that treatment with αCD33-mAB-P/P-DNMT3A-siRNA led to a significant knockdown of DNMT3A in DNMT3A-mutant OCI-AML2 cells (Fig. 2A). Moreover, we observed that the colony forming capacity of DNMT3A-mutant OCI-AML2 (Fig. 2B) and OCI-AML3 (Fig. 2C) cells was significantly decreased, when treated with αCD33-mAB-P/P-DNMT3A-siRNA compared to αCD33-mAB-P/P-scrambled control-siRNA and PBS treatment, while colony formation in DNMT3A-wild type KG1 cells was unchanged (Fig. 2D). In order to control the cell surface molecule specificity of the αCD33-mAB-P/P carrier antibody for the nanocarrier efficacy, we applied αCD33-mAB-P/P-siRNA nanocarriers to cell lines that do not express CD33, for instance A549 non-small cell lung cancer (NSCLC) cells as well as OCI-Ly19 lymphoma cells (Additional file 1: Figure S1 E–F). In KRAS-dependent A549 cells, KRAS-specific siRNA was effective in colony reduction when the respective nanocarrier was decorated with αEGFR mAB cetuximab, but ineffective when decorated with αCD33-mAB gemtuzumab (Additional file 1: Figure S1 E). In CD33-negative, CD20-positive and DNMT3A-independent OCI-Ly19 cells, αCD33-mAB-P/P-decorated DNMT3A nanocarriers had no effect on colony formation (Additional file 1: Figure S1 F). Vice versa, we applied αCD20-mAB-P/P-decorated DNMT3A-siRNA nanocarrier on CD20-negative and DNMT3A-dependent OCI-AML2 cells, which again exhibited no significant effect on colony formation (Additional file 1: Figure S1 G). Moreover, cell proliferation was compromised after DNMT3A-knockdown in OCI-AML2 and OCI-AML3 cells, while KG1 cells were unaffected (data not shown). Hence, we conclude that the efficacy of the αCD33-mAB-P/P-DNMT3A-nanocarrier is indeed dependent on the presence of both, the expression of CD33 on the cell surface and the addiction of the respective cell to the action of the mutated DNA methylase DNMT3A.

Fig. 2figure 2

αCD33-mAB-P/P-siRNA-mediated RNAi inhibits target gene expression and decreased growth of DNMT3A-mutant AML cells in vitro and in vivo. A Western blot analysis of treated OCI-AML2 cells. B–D Colony formation assay of DNMT3A mutant cell lines OCI-AML2 (B, n = 6) OCI-AML3 (C, n = 6) and DNMT3A-wild type KG1 cells (D, n = 3). There is a significant decrease in colony growth due to αCD33-mAB-P/P-DNMT3A-siRNA treatment in OCI-AML2 and OCI-AML3 (B-C, in contrast to control-siRNA), but not in KG1 (D). Significance: *, p < 0.05, 2-tailed T-test. Means plus SD of three independent experiments. E. Schematic overview about in vivo treatments after subcutaneous (s.c.) injection of 1 × 107 AML cells in CD1-nude mice. Mice were treated intraperitoneally (i.p.) with PBS, αCD33-mAB-P/P-control (cntr)-siRNA or αCD33-mAB-P/P-DNMT3A-siRNA three times weekly. F,G Tumor growth curves of OCI-AML2 (F) and KG1 (G). Growth of OCI-AML2 tumors was significantly delayed due to systemic αCD33-mAB-P/P-DNMT3A-siRNA treatment in contrast to control-siRNA treatment and PBS treatment (G), whereas KG1 tumors (DNMT3A wild type) did not show significant differences in tumor growth (G). H,I Tumor weight of isolated OCI-AML2 (H) and KG1 (I) tumors. Shown are means plus SD: T-test *, p < 0.05. α, anti

DNMT3A-mutant xenograft tumors are sensitive to DNMT3A-knockdown via αCD33-mAB-P/P-DNMT3A-siRNA treatment

We then performed in vivo evaluation of anti-CD33-mAB-P/P-siRNA delivery in the DNMT3A-dependent cellular system OCI-AML2 cells in a CD1 nude mouse xenograft model (see Fig. 2E for schematic overview). Besides the systemic transplantation of leukemia cells, the ectopic subcutaneous transplantation of AML cells in immunodeprived mice has a long history in preclinical tests and is widely accepted as a screening experiment for drug evaluation [36,37,38]. The subcutaneous tumor growth was significantly reduced (p < 0.01) upon αCD33-mAB-P/P-DNMT3A-siRNA application, but not upon application of carriers with control siRNA or PBS (Fig. 2F).

Moreover, the knockdown of DNMT3A in wild type KG1 subcutaneous tumors was therapeutically ineffective (Fig. 2G). These results were substantiated by ex vivo tumor weight differences in each treatment cohort (Fig. 2H and I).

Immunohistochemical analysis of OCI-AML2 xenograft tumors at the end of observation revealed diminished DNMT3A-expression (Fig. 3E–F, compared to A-D), decreased numbers of proliferating Ki67-positive cells (Additional file 1: Figure S2 E–F, compared to A-D) and increased numbers of apoptotic TUNEL-stain positive cells (Additional file 1: Figure S2 Q-R, compared to M–N and O-P) of αCD33-mAB-P/P-DNMT3A-siRNA treated mice in comparison to PBS and to αCD33-mAB-P/P-scr (control)-siRNA treated mice. DNMT3A-expression was also decreased in KG1 xenograft tumors of αCD33-mAB-P/P-DNMT3A-siRNA treated mice (Fig. 3K-L, compared to G-J), this knockdown did not lead to decreased Ki67 (Additional file 1: Figure S2 G-L) or increased TUNEL positive cells (not shown). These results underline that specifically DNMT3A-mutated cells, but not DNMT3A-wild type cells were vulnerable to DNMT3A-knockdown in vivo.

Fig. 3figure 3

Expression analysis and downstream factors in xenograft tumor samples. A–L DNMT3A staining of OCI-AML2 and KG1 tumor sections. DNMT3A was detected in nuclei of OCI-AML2 tumor sections treated with PBS and αCD33-mAB-P/P-control-siRNA; scr, scrambled (= control) (A–D), but DNMT3A staining was almost completely lost after treatment with αCD33-mAB-P/P-DNMT3A-siRNA (E–F). Reduced DNMT3A staining was also observed in KG1 tumor sections from the αCD33-mAB-P/P-DNMT3A-siRNA treatment group (K–L compared to G–J). Nuclear counterstain was performed using Hoechst33342 (B, D, F, H, J, L). M–Q Relative expression (RT-PCR) of DNMT3A and downstream genes in OCI-AML2 tumors ex vivo upon previous in vivo exposure to PBS, control carriers (αCD33-mAB-P/P-scr-siRNA) and αCD33-mAB-P/P-DNMT3A-siRNA. D3A, DNMT3A siRNA; α, anti. R,S Gene set enrichment analysis (GSEA) of RNA sequencing data in αCD33-mAB-P/P-DNMT3A-siRNA vs. scr-siRNA carrier-treated OCI-AML2 tumors, ex vivo. GSEA analysis of hallmark gene sets from the molecular signature database identified significant enrichment for hallmark_oxidative phosphorylation (R) and hallmark_c-Myc targets (T) in αCD33-mAB-P/P-scr-siRNA carrier-treated OCI-AML2 tumors. Shown are representative gene set enrichment plots. FDR, false discovery rate q-value and NES, normalized enrichment score. S, U Graphical presentation of central genes within the biological networks identified in the GSEA analysis. Top 15 hub genes identified by String/cytoscape/cytoHubba interface were presented as a circular layout (color coded: red and yellow indicating a higher and lower rank, respectively) using degree based algorithm of cytoHubba plug-in. Rank, gene and the scores were presented in the adjacent table. Furthermore, protein–protein interaction network/subnetworks were analyzed using 12 different algorithms of cytoHubba plug-in and hub genes representing intersection of at least nine algorithms (75% overlapping signature) potentially indicate key DNMT3A related hub genes in inducing c-Myc targets and oxidative phosphorylation. Detailed information on these hub genes and their importance in different cancers and therapy outcome is depicted in Additional file 1: Figure S4

To get a deeper insight in downstream effects of DNMT3A-knockdown in vivo, we further analyzed RNA from OCI-AML2 tumors samples. Real-Time RT-PCR for different formerly identified downstream target factors of DNMT3A function revealed – besides the significant downregulation of DNMT3A itself (Fig. 3M)—that MEIS1, HOXA11, c-Myc and BCL2 were significantly downregulated upon DNMT3A-knockdown (Fig. 3N-Q). Interestingly, GSEA analysis of RNA sequencing data in scr- but not DNMT3A-siRNA targeted OCI-AML2 tumors demonstrated significant enrichment for hallmark gene sets of oxidative phosphorylation (Fig. 3R) and c-Myc targets (Fig. 3T). To provide a better overview of these gene sets and to underscore possible interactions between genes, we performed STRING analysis followed by topological analysis using the CytoHubba plugin in the analysis platform Cytoscape (Fig. 3S and U), which results in a graphic visualization of the 15 highest ranking genes in both biological networks (oxidative phosphorylation: Fig. 3S; c-Myc targets: Fig. 3U; Additional file 1: Figure S4). Our data thus underscore that mutated DNMT3A-dependent regulation of genes is responsible for oxidative phosphorylation and c-Myc transcriptional programs as potential targets of DNMT3A and their contribution in leukemogenesis in these cells.

αCD33-mAB-P/P-FLT3-siRNA treatment leads to significantly reduced tumor growth in a FLT3-ITD positive MV4-11 xenograft mouse model

One of the most frequent mutations in AML is FLT3-ITD [23]. Moreover, DNMT3A-mutations and FLT3-ITD most likely cooperate as oncogenes in AML development [17, 39]. Therefore, a targeted therapy using siRNA may also include this important oncogene. We treated FLT3-ITD-bearing MV4-11 cells with our αCD33-mAB-P/P in complex with a mixture of two siRNAs and found that the complex was effectively internalized (Fig. 4A) into the MV4-11 cells and downregulated FLT3 (Fig. 4B). Colony formation was significantly decreased upon αCD33-mAB-P/P-FLT3-siRNA treatment of MV4-11 cells compared to αCD33-mAB-P/P-control-siRNA treated cells (Fig. 4C). This was not the case when DNMT3A-siRNA was delivered (Fig. 4C). We performed systemic in vivo treatment with αCD33-mAB-P/P-siRNA delivery in the FLT3-ITD-dependent MV4-11 xenograft in a CD1 nude mouse model (see Fig. 4D for schematic overview). The subcutaneous tumor growth was significantly reduced (p < 0.05) upon i.p. αCD33-mAB-P/P-FLT3-siRNA application, but not upon control siRNA and PBS application, which again underlines the effective transport and activity of the nanocarrier (Fig. 4E).

Fig. 4figure 4

Knockdown of FLT3 via αCD33-mAB-P/P-nanocarrier leads to significantly decreased colony and tumor growth of MV4-11 cells and of colony formation of primary AML blasts. A Internalization of αCD33-mAB-P/P-Alexa488-control-siRNA into MV4-11 cells. B Western blot for FLT3 of FLT3-ITD-mutant AML cell line MV4-11. Expression of FLT3 was suppressed upon αCD33-mAB-P/P-FLT3-siRNA treatment in contrast to control-siRNA in MV4-11 with β-actin as control. C Colony formation assay of FLT3-ITD mutated cell lines MV4-11 (n = 3). There is a significant decrease in colony growth due to αCD33-mAB-P/P-FLT3-siRNA treatment in contrast to control (cntr)- or DNMT3A-siRNA carriers, respectively. Means plus SD of three independent experiments. 2-sided T-test: *p < 0.02. D Schematic overview of in vivo i.p. treatment after s.c. injection of 1 × 107 MV4-11 cells in CD1-nude mice. Mice were treated with PBS, αCD33-mAB-P/P-control (cntr)-siRNA or αCD33-mAB-P/P-FLT3-ITD-siRNA three times weekly. E Tumor growth curves of MV4-11 transplants. Growth of MV4-11 tumors was significantly inhibited due to αCD33-mAB-P/P-FLT3-siRNA treatment in contrast to control-siRNA carrier treatment or PBS treatment. F Flow cytometric analysis of CD33-expression on the surface of AML patient #751 (driven by DNMT3A-R882H and FLT3-ITD mutations) cells. G The αCD33-mAB-P/P nanocarrier is able to internalize Alexa488-siRNA into patient blasts. Left: Fluorescence microscopy; right: Quantification via flow cytometry of non-treated (upper panels) and αCD33-mAB-P/P-Alexa488-control-siRNA internalized AML patient cells (lower panels). H Colony formation capacity of primary AML blasts. Cells were pre-incubated with the antibody-siRNA complex, resuspended in methylcellulose and cultivated for 8—12 days. Colonies were stained with INT, counted and photographed. There is a significant decrease in colony growth upon αCD33-mAB-P/DNMT3A-siRNA and αCD33-mAB-P/P-FLT3-siRNA treatment in contrast to control-siRNA carrier. Significance: *p < 0.05, 2-tailed T-test. Means plus SD of 3 independent replicates. α, anti

Colony growth of primary AML blasts is reduced by treatment with αCD33-mAB-P/P and functional siRNAs

An in vitro standard to test a novel therapeutic strategy preclinically is the analysis of anchorage-independent colony growth of treated vs. non-treated primary patient cells. We chose an AML patient sample (patient #751), which harbored both, a FLT3-ITD and a DNMT3A-R882H driver mutation (Fig. 4). Around 86% blasts of patient #751 expressed CD33 on their surface as detected by flow cytometric analysis (Fig. 4F). To investigate internalization efficiency of αCD33-mAB-P/P-siRNA carriers, primary blasts were incubated o/n at 37 °C with αCD33-mAB-P/P complexed to Alexa488-labeled control-siRNA, and internalization was verified by fluorescence microscopy (Fig. 4G, lower panel).

Next, blasts were treated with αCD33-mAB-P/P with “payloads” of DNMT3A-siRNA or FLT3-siRNA, respectively. In contrast to αCD33-mAB-P/P-cntr (control)-siRNA, a significant reduction of colony formation could be detected after treatment with αCD33-mAB-P/P-DNMT3A-siRNA or αCD33-mAB-P/P-FLT3-siRNA (Fig. 4H). We substantiated the results of the in vitro-treatment of primary patient cells with three additional patient leukemia cells, patient #805 (DNTM3A- and FLT3-ITD-mutant), patient #770 (only DNMT3A-mutant) and #719 (FLT3-ITD) (Additional file 1: Figure S3). These results imply that 1) primary leukemic cells internalize the antibody–protamine-siRNA carriers, 2) the siRNA-mediated knockdown was efficient and 3) FLT3-ITD and DNMT3A-mutated blasts depend on both oncogenic factors to survive.

Safety

To evaluate the effect of the RNAi in normal hematopoietic cells, CD34+/CD19− peripheral blood mononuclear cells (PBMCs) of healthy persons were analyzed in colony formation assays (Additional file 1: Figure S5 G). A slight inhibition of colony growth caused by treatment with αCD33-mAB-P/P-siRNA for both, the cntr-siRNA and the functional siRNA carriers, could be detected (Additional file 1: Figure S5 G) probably representing an antibody-only effect. On the other hand, the differences between control-siRNA and FLT3-siRNA or DNMT3A-siRNA carrier treatment were not significant, indicating that normal PBMC are not specifically sensitive to FLT3 or DNMT3A knockdown in contrast to AML blasts (Additional file 1: Figure S5 G). Treatment with equimolar antibody concentrations of gemtuzumab-ozogamicin (GO, MylotargR), however, led to a significant reduction in colony growth of normal PBMC (Additional file 1: Figure S5 G). This indicated that treatment with αCD33-mAB-P/P-siRNA is less toxic for normal human hematopoietic cells than gemtuzumab-ozogamicin.

In addition, blood urea nitrogen (BUN) and serum creatinine (CR) values of mice treated with αCD33-mAB-P/P-siRNA carriers were determined from blood samples drawn from mice treated as described (see Additional file 1: Figure S5). The treatment with αCD33-mAB-P/P-siRNA carriers had no effect on BUN (Additional file 1: Figure S5 A) or CR (Additional file 1: Figure S5 B) values compared to values of mice treated with PBS. Moreover, mouse weight was unchanged under treatment (Additional file 1: Figure S5 E–F). In an independent OCI-AML2 xenograft transplant and treatment experiment, also liver enzymes AST (aspartate aminotransferase; other name: glutamic oxaloacetic transaminase (GOT)) and ALT (alanine aminotransferase; other name: glutamic pyruvic transaminase (GPT)) in mouse serum were unchanged when compared to values from saline control treated animals (Additional file 1: Figure S5 C-D), BUN and CR were reproduced as unchanged (not shown). Together, this shows that anti-CD33-mAB-P/P-siRNA carriers are not causing major clinical or laboratory toxicity in mice.

As some siRNAs can activate TLR as adverse effects, we further substantiated safety with TLR activation tests by in vitro assays that allow examination of NF-κB pathway activation via TLR2-signaling and of IRF pathway activation via TLR3 signaling (Additional file 1: Figure S9) in the monocytic reporter cell line THP1-DUAL designed for this purpose. While recommended positive controls such as poly(I-C) and Pam3CSK4 robustly induced TLR2 and TLR3 signaling, respectively, no nanocarrier combination treatment except αCD33-mAB-P/P-NRAS-siRNA induced a significantly elevated signal for induction. The latter is probably due to a cellular stress reaction, because THP-1 cells are NRAS mutated and dependent from this oncogenic signaling pathway. Therefore, a general TLR activation effect of αCD33-mAB-P/P-siRNA nanocarriers was not observed.

Physical characterization of the nanocarriers

Having obtained these therapeutic and safety results, we further analyzed the properties of nanocarrier formation and function for different conjugation ratios (Fig. 5) and in absence of free (SMCC-)protamine (Fig. 6 and Additional file 1: Figure S7). To this end, we conjugated molar ratios from 1:1 to a 1:100 (Fig. 5A) excess of SMCC-protamine over αCD33-mAB, which showed different coupling efficiencies as checked by the gel-electrophoretic properties of the resulting conjugates (Fig. 5B). Efficient coupling of heavy (HC) and light (LC) chain of the αCD33-mAB only appeared at ratios of  > / = 1:10 (Fig. 5B), which was in line with siRNA binding (Fig. 5C–H, left side). Interestingly, at ratios of 1:50 or higher, precipitation was visible (Fig. 5G, H, right side, compared to C-F, right side). Internalization of fluorescence-tagged siRNA into CD33-expressing OCI-AML2 cells (Fig. 5I–N) was seen with conjugates with a molar excess of 32 or 50 mol SMCC-protamine over αCD33-mAB.

Fig. 5figure 5

Attributes of effective αCD33-mAB:protamine conjugation ratios. A Concentrations tested and resulting molar ratios of αCD33 antibody (αCD33-mAB) to SMCC-protamine for the effective conjugation of both components. B Coomassie-stained SDS–PAGE showing uncoupled αCD33-mAB compared to the conjugation products that were coupled as depicted in A. The formation of a protamine-conjugated heavy chain (HC-P) and light chain (LC-P) showed an optimum at a 1:32 conjugation ratio with no further increase at higher ratios. C–H Left: Band-shift assays exhibiting siRNA binding capacity. Right: αCD33-mAB-P/P stored at 4 °C for several days shows precipitation at the bottom of the tube only at ratio 1:50 and 1:120 (blue triangles). I-N Internalization of Alexa488-control-siRNA complexed αCD33-mAB-P/P into OCI-AML2 cells. Complexes of αCD33-mAB-P/P transport Alexa488-siRNA into cells (left panel rectangles), with detailed magnifications (right panels). At conjugation ratios of 1:1–1:10 (I-K) only diffuse greenish background can be detected. O-T Left: Colony formation assays of the different conjugations in OCI-AML2 cells. Significance: *, p < 0.05, 2-tailed T-test. Means plus SD of three independent experiments. D3A, DNMT3A siRNA. S-P, SMCC-protamine. Right: αCD33-mAB-P/P-Alexa488-siRNA complexes form vesicles with different size and efficacy in presence of rising amounts of free SMCC-protamine. No vesicle formation at ratios of 1:1–10 (right: O-Q), apart from some unspecific aggregates (right side in P). α, anti

Fig. 6figure 6

The αCD33-mAB-P/P nanocarriers only transport siRNA in presence of free SMCC-protamine (SMCC-P). A Coomassie-stained SDS–PAGE showing αCD33-mAB, αCD33-mAB coupled with SMCC-P and HPLC-fractions 29–30 of αCD33-mAB-P/P upon effective depletion of unbound SMCC-P; HC = heavy chain, LC = light chain, -P = SMCC-protamine. B Antibody–protamine conjugates with fluorescent Alexa488-siRNA in cell-free incubation overnight on chamber slides. αCD33-mAB-P/P with free SMCC-P forms visible vesicular structures (left panel), while αCD33-mAB-P after depletion of free SMCC-P (fraction 30, see A) do not form visible structures (right panel). C. DLS and zeta-potential measurement of αCD33-mAB-P/free protamine-scr-siRNA carriers. D. αCD33-mAB-P/P-scr-siRNA nanoparticles were left to form for 2 h and subjected to electron microscopy on copper grids by phosphotungstate negative staining. E–H. Immunostaining of nanocarriers with an anti-human IgG antibody to illustrate the accessibility and location of the αCD33-mAB in the outside rim and the siRNA in the lumen of the αCD33-mAB-P/P-nanocarriers. I. Protamine was chemically coupled to Cy3 and then incubated with αCD33-mAB-P that was depleted from free protamine and with non-fluorescent control-siRNA to form nanocarriers. This complexation was performed for 2 h at RT and nanocarriers were then immobilized o/n on slides for immunostaining as in G. J. αCD33-mAB-P/P-Cy3-control-siRNA show homogeneous Cy3 (blue) micelles. K. The same vesicles as in L show anti-human IgG-Alexa647 (red) fluorescence in ring-like structure around each vesicle (staining as depicted in E). L. Overlay of panels J and K. α, anti

As a functional analysis of the different conjugation ratios, we treated OCI-AML2 cells with different αCD33-mAB-P/P with control (scrambled, scr) versus anti-DNMT3A (D3A)-siRNA, respectively, and cultured equal numbers of cells in methylcellulose for anchorage-independent colony growth indicative for tumorigenicity. Of note, only the 1:32 (Fig. 5R) and 1:50 conjugates (Fig. 5S) showed a specific functional impact of DNMT3A-siRNA versus control siRNA with 50% less colony formation, compared to non-functional control siRNA. In contrast, compositions with lower excess such as 1:1 up to 1:10 showed no impact of DNMT3A-knockdown on colony growth (Fig. 5O–Q). Furthermore, concentrations of free SMCC-protamine yielding ratios higher than 1:50 led to cellular toxicity unrelated to the nature of the applied siRNA (Fig. 5T).

Further, when the αCD33-mAB-SMCC-protamine (αCD33-mAB-P) was conjugated with free SMCC-protamine at ratios of 1:1 up to 1:10, no efficient cell-free vesicle formation could be observed (Fig. 5O–Q, right side). At the ratios of 1:32 to 1:100, the vesicle formation was abundant (Fig. 5R–T, right side). Taken together, results shown in Fig. 5 suggest an optimal nanoparticle formation of αCD33-mAB-P to free SMCC-protamine ratio of 1:32 or 1:50, corresponding to the efficient siRNA delivery and internalization into cells without or only with little (1:50) precipitation or non-specific toxicities.

To confirm 1) that free (SMCC-)protamine is necessary for the proper formation of the nanocarrier and 2) that it is not the free (SMCC-)protamine, which mediates the cargo internalization but the antibody-receptor binding, as we saw in our recently published [6] and unpublished studies, we removed all excess free SMCC-protamine (“SMCC-P”) from the reaction mixture by preparative size exclusion chromatography or by protein G interaction chromatography (Fig. 6 and Additional file 1: Figure S7). Protamine-conjugated αCD33-mAB-P without free SMCC-protamine (Fig. 6A, fraction 29/30) was neither able to bind siRNA in a band-shift assay (Additional file 1: Figure S7 A), nor to internalize into CD33-positive OCI-AML2 cells (Additional file 1: Figure S7 D-E compared to B-C). Also, free SMCC-protamine alone could not transport Alexa488-siRNA into OCI-AML2 cells (Additional file 1: Figure S7 F-G). Vesicular nanocarrier formation was not detectable with αCD33-mAB-P without free SMCC-protamine (Fig. 6B). Moreover, neither colony formation of OCI-AML2 cells treated with αCD33-mAB-P after depletion of free SMCC-P and added to DNMT3A-siRNA was inhibited (Additional file 1: Figure S7 I), nor when treated with free SMCC-protamine with DNMT3A-siRNA (Additional file 1: Figure S7 J) when compared to added scr-siRNA or PBS, respectively. This confirmed the necessity of an intact αCD33-mAB-P/P nanoparticle in complex with siRNA to form what we now call αCD33-mAB-P/P-nanocarrier (Additional file 1: Figure S7 H).

To elucidate whether siRNA is needed to form vesicles, we incubated αCD33-mAB-P/P with rising amounts of Alexa488-control-siRNA (Additional file 1: Figure S7 K). Here, nanocarriers are efficiently formed with a molar excess of siRNA of 5–10 times over the antibody (Additional file 1: Figure S7 K).

Once formed properly by auto-assembly, the nanocarriers demonstrated a high stability toward pH shifts between pH 4.8 and 8.0 (Additional file 1: Figure S7 L) and different serum concentrations (Additional file 1: Figure S7 M). The αCD33-mAB-P/P-nanocarrier possesses a distinct size range of 359 ± 82 nm as determined by dynamic light scattering spectroscopy DLS (Fig. 6C), which was confirmed by electron microscopy (Fig. 6D). Here, spheroid, electron dense structures were determined by phosphotungstate negative staining (Fig. 6D).

We subsequently characterized the nanocarrier by immunostaining the human IgG proportion of the carrier complex (schematic overview in Fig. 6E) and observed a strong staining of the nanostructure rim (Fig. 6G), which revealed a position of the targeting antibody facing outward, while the Alexa488-tagged siRNA fills the lumen of the structure (Fig. 6F and H).

To further characterize the nanocarrier structure we also investigated the actual position of the protamine as the main electrostatic connector. Theoretically, free protamine could also form 1) a shell that is filled with siRNA or 2) siRNA and protamine could be distributed equally, as it was hypothesized before [40]. We therefore conjugated free protamine with Cy3 as a traceable chromophore (Additional file 1: Figure S8) and combined it with the αCD33-mAB-P depleted from free SMCC-protamine (as described in Fig. 6A and B) as well as with non-fluorescent siRNA to establish nanocarriers (Fig. 6I). We identified a luminal fluorescence staining of protamine-Cy3 (blue fluorescence in Fig. 6J), which closely resembled the pattern of the siRNA-Alexa488 (Fig. 6F and H compared to Fig. 6J and L). Moreover, immunostaining again with an anti-human IgG-Alexa647 antibody showed a ring-like staining surrounding the protamine-Cy3 containing vesicle (Fig. 6K and L). We conclude that the auto-assembly process forms a spheroid structure with the targeting IgG facing outward and a balanced composition of siRNA and free protamine filling the lumen of the nanocarrier sphere (Fig. 6H and L).

In summary, these characterization experiments show a sufficient corridor for nanocarrier production according to Good Manufacturing Practice (GMP) guidelines necessary for translation into early clinical trials.

Treatment of patient-derived xenotransplanted (PDX) AML with αCD33-mAB-P/P-siRNA nanocarriers

The analysis of primary cells directly obtained from patients and xenotransplanted into immune-deficient mice represents a preclinical tumor model close to the patient’s situation and is a standard in the development of therapy options. To this end, we established a patient-derived AML xenograft (PDX) transplantation of blood AML mononuclear cells in NSG mice using the same patient sample as analyzed in vitro before (Fig. 

留言 (0)

沒有登入
gif