*2.2. His\_eGFP\_C3botE174Q Was Selectively Internalized into Human Macrophages Ex Vivo*

After confirming that transport of cargo into macrophages can be enhanced by C3bot or C3botE174Q, we investigated whether His\_eGFP\_C3botE174Q still possessed the cell-type selectivity of wild-type C3bot. Hence, cellular uptake into human-monocyte-derived macrophages was compared with uptake into lymphocytes of the same blood donor. Initially, the different cells were treated separately with His\_eGFP\_C3botE174Q, and internalization of eGFP was quantified in flow cytometry (Figure 2a). Since only internalized cargo should be analyzed, cell-surface-bound eGFP signals were quenched by using an established trypan-blue-based assay [38–40]. Uptake of His\_eGFP\_C3botE174Q into macrophages was strongly enhanced in comparison with lymphocytes, which internalized only a minor portion of the fusion protein (Figure 2a). Notably, even with 800 nM of His\_eGFP\_C3botE174Q, only a minor portion was internalized by the lymphocytes (Figure 2a; for the experiment with a lower concentration, see Figure S1).

**Figure 2.** Cell-type-selective uptake of His\_eGFP\_C3botE174Q into primary human macrophages compared to lymphocytes ex vivo. (**a**) Human-monocyte-derived macrophages and lymphocytes of the same donor were separately treated with 800 nM His\_eGFP\_C3botE174Q for 20 min at 37 ◦C. The cells were washed with PBS, and extracellular or membrane-bound eGFP signals were quenched with trypan blue directly before flow cytometry measurement. The averaged relative median fluorescence intensity (normalized to untreated cells (NC)) is depicted in the column diagrams (mean ± SD, n=5). (**b**) Monocyte-derived macrophages and lymphocytes of the same human blood donor were co-cultured (1:5 ratio) and treated with 400 nM His\_eGFP\_C3botE174Q or His\_eGFP at 37 ◦C for 1 h. The cells were fixed, and the cell nuclei (blue) were stained with DAPI. MHC class II (red) was stained as marker for macrophages. All in confocal microscopy detected fluorescence signals (eGFP, DAPI, and MHC class II) were merged, and representative images are depicted with magnification as indicated by the white squares. Scale bars on the right apply to the complete row of images (25 μm for full-size images, and 10 μm for magnifications). The figure is modified from [37] under the authors' rights.

On the basis of these findings, a co-culture of macrophages and lymphocytes of the same blood donor was treated with His\_eGFP\_C3botE174Q. For the control, the co-culture was treated with His-eGFP alone. The macrophage population was marked by staining of the major histocompatibility complex (MHC) class II on the cell surface (Figure 2b). In this co-culture, His\_eGFP\_C3botE174Q was cell-type selectively internalized into human macrophages, as indicated by the green signals surrounded by red MHC class II staining. Moreover, about 70–90% of the total macrophage population internalized His\_eGFP\_C3botE174Q (see Figure S2). Hence, direct coupling of cargo molecules to the transporter system did not affect the cell-type selectivity, and C3botE174Q can be used to enhance the uptake of cargo molecules.

#### *2.3. Functional eGFP Was Delivered into the Cytosol of DCs and Macrophages via C3bot and C3botE174Q*

Next, we investigated whether the cargo model eGFP was released into the cytosol of target cells. Since eGFP is hardly detectable in the cytosol, even with state-of-the-art fluorescence microscopic techniques, at first, an indirect approach based on the cytotoxic effect of C3bot was used. We compared the morphological changes induced by C3bot with those induced by cargo-labeled His\_eGFP\_C3bot to evaluate whether coupling of cargo interfered with the uptake process of the toxin. Therefore, cells of a mouse macrophage cell line (J774A.1) were treated with increasing concentrations of C3bot or His\_eGFP\_C3bot (Figure 3). Treatment with His\_eGFP\_C3bot induced characteristic changes in cell macrophage morphology, i.e., formation of long cell protrusions, comparable to wildtype C3bot (Figure 3a). In contrast, His\_eGFP\_C3botE174Q did not induce such characteristic changes in cell morphology, confirming that enzymatic activity of C3bot is required for this effect. Since the fusion proteins were produced in *Escherichia coli* (*E. coli*), the effect of lipopolysaccharides (LPS) was also tested. LPS endotoxins are common impurities in recombinant protein expression in Gram-negative bacteria that can activate immune cells, e.g., macrophages [41,42]. Despite treatment with LPS slightly influencing J774A.1 cell morphology and more intracellular inclusions being visible compared to the negative control (NC), the clear and striking formation of protrusions cannot be explained by a potential LPS contamination (Figure 3a). Hence, the results indicate that active C3bot is required for induction of the morphological change and, most importantly, eGFP labeling did not affect this effect. This becomes even more obvious by quantifying the cells with C3 morphology and comparing different concentrations in a time course (Figure 3b) or at defined time points (Figure 3c). Importantly, no significant differences were detected by comparing same concentrations of wild-type C3bot with His\_eGFP\_C3bot.

These results were furthermore confirmed by analyzing the Rho ADP-ribosylation status inside the cells. A sequential Rho ADP-ribosylation assay was used for detection of non-ADP-ribosylated Rho in intact cells (Rhonon-ADP-rib.) (Figure 4). Notably, in this assay, weak signals indicate strong ADP-ribosylation of Rho by the toxin inside the target cells. Equal protein loading was controlled by detection of heat shock protein 90 (HSP90). The detected Rhonon-ADP-rib. decreased, i.e., Rho-ADP-ribosylation in the intact cells increased with the concentration of C3bot or His\_eGFP\_C3bot. In accordance with the morphological assays, eGFP labeling inhibited neither the Rho ADP-ribosyltransferase activity of C3bot in J774A.1 macrophages (Figure 4a) nor the activity in human DCs derived from a sarcoma cell line (U-DCS cells, see Figure 4b). Taken together, the results indicate that cytosolic release of C3bot into target cells (macrophages and DCs) is not inhibited by fusion with the cargo model eGFP.

Since this approach is quite indirect for the detection of cytosolic cargo release, alongside the fact that intracellular degradation of the fusion constructs cannot be excluded, we directly analyzed the presence of cytosolic eGFP with an established digitoninbased cell fractionation [34,36]. After treatment with the eGFP-labeled proteins (His\_eGFP, His\_eGFP\_C3bot, and His\_eGFP\_C3botE174Q), the cells were separated into cytosolic and membrane fractions.

For isolation of the cytosol, digitonin was used to form small pores into the cell membrane that are smaller than endosomes or other organelles. Hence, only the cytosolic proteins were able to leave the cells. Successful separation of the cytosolic fraction from the membrane fraction (containing endosomes) was confirmed by Western blot detection of EEA1, which was only present in the membrane fractions (Figure 5a). As a loading control for cytosolic proteins, HSP90 was detected.

**Figure 3.** Fusion of eGFP to C3bot did not affect the cytosolic release of C3bot in J774A.1 macrophages. (**a**) J774A.1 cells were treated with C3bot, His\_eGFP\_C3bot, His\_eGFP\_C3botE174Q (200 nM each), or 1 μg/mL LPS, or were left untreated (NC). The images show representative phase contrast images after an incubation time of 12 h with 50 μm scale bar. (**b**,**c**) J774A.1 macrophages were treated with increasing concentrations (25, 50, 100, 200 nM) of C3bot or His\_eGFP\_C3bot, with 200 nM His\_eGFP\_C3botE174Q, with 1 μg/mL LPS, or were left untreated (NC). Phase contrast images were taken after incubation times of 2, 4, 6, 8, 10, 12, and 24 h at 37 ◦C. The cell portion with characteristic stellate C3 morphology was quantified and divided by the total amount of cells per image. The ratios (mean ± SD, n = 3) are depicted in time course diagrams (**b**) and column diagrams after 6 or 12 h (**c**). (**c**) Compared to the NC, statistical significance was tested via Student's *t*-test, and columns are labeled according to following significance levels: ns *p* > 0.05, \* *p* < 0.05, \*\* *p* < 0.01, \*\*\*\* *p* < 0.0001. Additionally, the same concentrations of C3bot and His\_eGFP\_C3bot were compared via Student's *t*-test, as indicated by the lines and significance levels above the representative columns. The figure is modified from [37] under the authors' rights.

**Figure 4.** Fusion of eGFP to C3bot did not affect the cytosolic ADP-ribosylation of Rho in J774A.1 macrophages or U-DCS cells. J774A.1 cells (**a**) and U-DCS cells (**b**) were treated with increasing concentrations (50, 100, 200 nM) of C3bot or His\_eGFP\_C3bot, respectively, with 200 nM His\_eGFP\_C3botE174Q, with 1 μg/mL LPS, or they were left untreated (NC). After 7 h of incubation at 37 ◦C, the cells were washed and lysed, and then a sequential ADP-ribosylation assay was performed for detection of non-ADP-ribosylated Rho in intact cells (lower band in left lower panels). The relative integrated density values for Rhonon-ADP-rib. were normalized to the HSP90 loading control (left upper panels) and depicted in a column diagram (mean ± SD, right panels). The detected bands above 26 kDa (upper band in left lower panels) were probably unspecific binding sites of the used streptavidin–peroxidase conjugate, which did not correlate with any treatment conditions. For J774A.1 macrophages in (**a**), five replicates (n = 5), and for U-DCS cells in (**b**), four replicates (n = 4) were averaged. Notably, weak signals in the sequential ADP-ribosylation assay indicated strong toxin activity in intact cells. Compared to the NC, statistical significance was tested via Student's *t*-test, and columns are labeled according to following significance levels: ns *p* > 0.05, \* *p* < 0.05, \*\* *p* < 0.01, \*\*\*\* *p* < 0.0001. Additionally, the same concentrations of C3bot and His\_eGFP\_C3bot were compared via Student's *t*-test, as indicated by the lines and significance levels above the representative columns. The figure is modified from [37] under the authors' rights.

Already after 1 h of incubation with the constructs, signals for eGFP were detected in both fractions, indicating cytosolic release of eGFP (Figure 5a). For His\_eGFP\_C3bot and His\_eGFP\_C3botE174Q, the eGFP signals were detected at a height of about 55 kDa, corresponding to the size of full-length fusion proteins (calculated at 52 kDa). For His\_eGFP, the signals were detected at about 34 kDa, which also fitted into the calculated molecular weight of about 28.4 kDa. These results indicate that eGFP-labeled proteins were not degraded during cellular uptake. Due to the nature of this assay, it is not possible to compare the protein amounts in one fraction with that in the other fraction, but rather the signals within one fraction should be compared. In both fractions, the detected eGFP signals were much stronger for the fusion constructs His\_eGFP\_C3bot and His\_eGFP\_C3botE174Q compared to His\_eGFP alone (Figure 5a), indicating that uptake into intracellular vesicles

(membrane fraction) as well as cytosolic release of the cargo model eGFP were significantly enhanced by coupling to C3bot or C3botE174Q (for Western blot quantification and statistical analysis, see Figure 5b).

**Figure 5.** C3bot and C3botE174Q increased the cytosolic release of the functional cargo model eGFP. (**a**,**b**) U-DCS cells were treated with His\_eGFP\_C3bot, His\_eGFP\_C3botE174Q, or His\_eGFP (250 nM each), or were left untreated (NC) for 1 h at 37 ◦C. Subsequently, the cytosol was separated from the membrane fraction (containing endosomes) in a digitonin-based assay. (**a**) In a Western blot, the separation was confirmed by detection of EEA1 as a maker for endosomes that should not be present in the cytosolic fraction. HSP90 was detected as a cytosolic marker, present in both fractions, since the cytosol did not completely leave the cells (see Section 5.14). The presence of eGFP in the cytosol or membrane fraction was analyzed by using a primary anti-GFP antibody. (**b**) The Western blots were quantified, and the normalized integrated density values were plotted in a column diagram as mean ± SD (n = 7). For the cytosolic fraction, eGFP was normalized to the detected HSP90 signals, and for the membrane fraction, it was normalized to the EEA1 signals. The individual values for each repetition are available in the Supplementary Materials (Table S3 for the cytosolic fraction and Table S4 for the membrane fraction). (**c**) U-DCS cells were treated with His\_eGFP\_C3bot, His\_eGFP\_C3botE174Q, or His\_eGFP (250 and 500 nM each), or left untreated (NC) for 5 h at 37 ◦C. Digitonin-based cell fractionation was performed as described above and controlled by Western blotting (data not shown). The functionality of eGFP in each fraction (cytosol in the left panel and membrane in the right panel) was analyzed by fluorescence detection at 488 nm and an emission wavelength of 510 nm in a microplate reader. (**b**,**c**) Compared to the respective His\_eGFP treatment, statistical significance was tested via Student's *t*-test, and the following significance levels were defined: ns *p* > 0.05, \* *p* < 0.05, \*\* *p* < 0.01. (**b**) Additionally, the same concentrations of C3bot and His\_eGFP\_C3bot were compared via Student's *t*-test, as indicated by the lines and significance levels above the representative columns.

For assay control, it was confirmed that the fusion proteins His\_eGFP\_C3bot and His\_eGFP\_C3botE174Q were equally detected compared to His\_eGFP by the GFP antibody (see Figure S3b,c). In this additional Western blot (Figure S3b,c), the detected signals for His\_eGFP\_C3bot were slightly weaker compared to His\_eGFP and His\_eGFP\_C3botE174Q; however, this minor difference was neglected since it did not object to the final conclusion in the digitonin-based cell fractionation assay.

After confirming that cargo delivery can be enhanced by coupling to C3bot and C3botE174Q, we investigated whether the cytosolic eGFP was still functional. Therefore, fluorescence of the cytosolic and membrane fraction was analyzed with a microplate reader after treatment of cells with His\_eGFP, His\_eGFP\_C3bot, or His\_eGFP\_C3botE174Q and digitonin-based separation. Fluorescence signals thereby increased in a concentrationdependent manner, and cytosolic fluorescence was significantly enhanced by coupling eGFP to the C3bot transporters (Figure 5c, left panel). Taken together, the results indicated that the endosomal uptake and cytosolic release of the eGFP cargo model into macrophages and DCs can be strongly enhanced by using C3bot or C3botE174Q as delivery tools and that the cargo remains functional inside these target cells.

#### *2.4. Characterization of a Modular System for Fast Attachment of Cargo to C3botE174Q*

After proofing the concept of using C3botE174Q as a shuttle system for the cargo model eGFP, the next step was to generate a modular delivery system for fast attachment of various cargos, i.e., small molecules and proteins. Since the earlier published streptavidinbiotin system comes with the cost of reduced cell type selectivity [34,35], direct coupling of C3botE174Q with cargo via thiol–maleimide click chemistry was investigated to possibly overcome this problem. By nature, C3botE174Q does not contain any thiol group (no cysteine), and therefore this reaction group can simply be inserted side specifically by a point mutation. A single cysteine amino acid was inserted at the N-terminal end of C3botE174Q (A1C), and this mutant will be further referred to as Cys-C3botE174Q. After protein purification, the thiol group can be loaded with different maleimide-labelled cargo molecules (see the cartoon in Figure S4a and example cargo in Figure S4b) by incubation for 2 h on ice. Successful protein loading was indicated by a shift to higher molecular weight, as observed by SDS-PAGE. As proof of concept, we loaded different maleimide-labeled cargos, i.e., a small molecule fluorophore (maleimide\_Dylight 488 (mDL488), 0.8 kDa), larger fluorescein-isothiocyanate-labeled polyethylene glycol (mPEG\_FITC, 5 kDa), or the reporter enzyme C2I (mC2I, 50 kDa). For all cargo molecules, a shift in molecular weight was detected as expected for successful coupling (Figure S4c).

After confirming the cargo coupling to the Cys\_C3botE174Q-system, successful cellular delivery was tested in different assays. First, the uptake of the small molecule fluorophore mDL488 into macrophages (Figure 6a) and DCs (Figure 6b) was investigated. Internalization of mDL488 was only facilitated when the cargo was directly coupled to Cys\_C3botE174Q. In contrast, neither for the treatment with cargo alone (mDL488) nor in combination with free uncoupled C3botE174Q green internalization signals were detected. These results indicate that the generated modular system enhanced cellular uptake directly and not indirectly by simply enhancing endocytosis. Comparable results were obtained for the lager mPEG\_FITC cargo model, which was also more efficiently internalized into target cells after coupling to Cys\_C3botE174Q (Figure S5).

Finally, we investigated whether the cytosolic release of cargo was also enhanced by the modular thiol–maleimide system. Therefore, the established reporter enzyme C2I was labeled with maleimide (mC2I) via the bifunctional linker m-maleimidobenzoyl-Nhydroxysuccinimide ester (MBS) and coupled to Cys\_C3botE174Q. When C2I reached the cytosol, it ADP-ribosylated G-actin, leading to cell rounding without affecting the cell viability within 2 days of incubation [43,44]. Cell rounding also occurred when U-DCS cells were treated with C2I in combination with its activated B-component C2IIa, providing a robust and established control for successful cytosolic release of cargo (Figure 7a). In

accordance with the literature [44,45], only minimal effects on cell viability/proliferation were detected for the delivered C2I (Figure S6).

**Figure 6.** The small-molecule fluorophore was delivered into macrophages and DCs via Cys\_C3botE174Q. Human-monocyte-derived macrophages (**a**) and U-DCS cells (**b**) were treated with free mDL488, mDL488-Cys\_C3botE174Q, or uncoupled mDL488 together with C3botE174Q (250 nM each), or were left untreated (NC) for 30 min at 37 ◦C. The cells were fixed, and cell nuclei (DAPI) and MHC class II were stained as markers for cell shape and size. Confocal microscopy was used to detect fluorescence signals, and the signals for mDL488 (green), DAPI (blue), and MHC class II (red) were merged. Representative images are depicted with magnification as indicated by the white squares. Scale bars on the right are applied for the complete row of images (25 μm for full-size images, and 10 μm for magnifications).

Although mC2I alone applied in high concentrations (220 or 880 nM) for 27 or 50 h also induced some cell rounding of the dendritic sarcoma cells, this effect was significantly enhanced by coupling mC2I to Cys\_C3botE174Q (mC2I-Cys\_C3botE174Q), as shown in the quantification (Figure 7b). For proper control, 4.4-fold higher concentrations of mC2I were used in comparison to mC2I-Cys\_C3botE174Q in order to compensate for free mC2I present in the coupling product (Figure S4b). Nevertheless, the Cys\_C3botE174Q transporter significantly increased the cytosolic release of mC2I, proving that the modular C3bot system can serve as selective and specific tool for delivery of cargo molecules into the cytosol of human monocytic cells.

**Figure 7.** Cys\_C3botE174Q enhanced the cytosolic release of the reporter enzyme C2I. (**a**) U-DCS cells were treated with 200 nM mC2I-Cys\_C3botE174Q, 880 nM mC2I, or a combination of 1 nM C2I with 1.66 nM C2IIa, or were left untreated (NC). Phase contrast images were taken after 27 and 50 h of incubation, and representative micrographs are shown with a 50 μm scale bar. (**b**) U-DCS cells were treated with 50 or 200 nM mC2I-Cys\_C3botE174Q, 220 or 880 nM mC2I, or a combination of 1 nM C2I with 1.66 nM C2IIa, or were left untreated (NC). Phase contrast images were taken after 27 and 50 h of incubation. The number of rounded cells was counted and divided by the total cell number, and the resulting ratios are depicted in column diagrams as mean ± SD (n = 3) for the indicated time points. The values for mC2I-Cys\_C3botE174Q were compared with the corresponding mC2I concentrations, and statistical significance was tested via Student's *t*-test (indicated with a line above columns) according to the following significance levels: ns *p* > 0.05, \* *p* < 0.05, \*\* *p* < 0.01. The figure is modified from [37] under the authors' rights.

#### **3. Discussion**

The protein toxin C3bot selectively enters the cytosol of monocytic cells, including macrophages and DCs, by endocytosis, and inhibits Rho-dependent processes in such cells. Thereby, the toxin down-modulates essential functions of these important immune cells, which should be detrimental in the context of an infection with C3-toxin-producing bacteria. On the other hand, the cell-type selectivity of the C3 toxin can be exploited for the targeted pharmacological down-modulation of excessive pro-inflammatory activity of macrophages in the context of traumatic diseases [21]. Since hyperactive macrophages and DCs play crucial roles in several inflammatory diseases, they are important drug targets, and the C3 toxin should be a promising compound to selectively and specifically suppress excessive reactions of these innate immune cells. However, the cell-type selectivity of C3bot towards monocytic cells can be exploited for pharmacological purposes in a second

way, i.e., for targeted delivery of therapeutic (macro-) molecules into these target cells and their controlled release into the cytosol. Various bacterial AB-type protein toxins have been used as drug delivery systems because of their unique mode of action, i.e., endocytic uptake and endosomal release of the therapeutic cargo molecules to reach cytosolic drug targets [46–48]. Since C3bot by nature enters monocyte-derived cells, a non-toxic variant of C3bot should represent an ideal molecule for targeted drug delivery into macrophages and DCs. Potentially, the cell penetration of novel therapeutics, e.g., peptides, proteins, nucleic acids, or cell-membrane-impermeable small molecules, can be facilitated, and/or selective targeting of macrophages and DCs as important immune cells can be improved by using C3botE174Q as a drug delivery tool. The basic concepts for such a transport system were investigated in this study.

As proof of concept, the cargo model eGFP was genetically fused to both C3bot and C3botE174Q, and cellular uptake of the resulting fusion proteins into target cells was investigated in comparison to non-target human blood cells. The fusion protein His\_eGFP\_C3botE174Q was selectively internalized into the human-monocyte-derived macrophages much stronger compared to lymphocytes of the same blood donor. Hence, the cell-type selectivity of the wild-type C3bot was maintained, despite the attachment of eGFP. The fluorescent protein eGFP is commonly used to follow uptake processes or to localize intracellular proteins. The generated fusion proteins (His\_eGFP\_C3bot and His\_eGFP\_C3botE174Q) were specifically internalized into early endosomes of humanmonocyte-derived macrophages, as unveiled by STED super-resolution microscopy. These results confirm earlier results obtained for the J774A.1 cell line [18] and provide more details about the localization within early endosomes. The eGFP signals are not evenly distributed over the complete vesicles, but rather are condensed at specific sites where potential translocation into the cytosol could take place. Earlier, our group observed a similar C3bot condensation in endosomes for human-monocyte-derived mature or immature DCs [23]. Only neglectable amounts of His\_eGFP alone were internalized into human-monocyte-derived cells, indicating that the C3bot transporters strongly enhance the cellular uptake of this cargo model. After endocytosis, wild-type C3bot is released into the cytosol of macrophages or DCs [18,23]. However, due to the stable β-barrel structure, eGFP can potentially hinder the translocation process, resulting in lower toxin activity, as shown for the diphtheria toxin [49] or for the anthrax pore [50]. In the present study, we investigated whether this was also the case for C3bot translocation by comparing wild-type C3bot with His\_eGFP\_C3bot in intoxication assays. In contrast to the diphtheria toxin or the anthrax pore, not even the slightest reduction in toxin activity was detected for the fusion construct His\_eGFP\_C3bot. Hence, this could be the result of (i) a very effective delivery of the cargo together with C3bot into the cytosol, or (ii) very effective cleavage of the fusion construct in the early endosomes to only release C3bot into the cytosol. This second possibility, however, was ruled out by detection of cytosolic full-length His\_eGFP\_C3bot and His\_eGFP\_C3botE174Q in a digitonin-based cell fractionation assay. Notably, no degradation products were detected in either the cytosol or the membrane fraction, excluding endosomal degradation. Importantly, cytosolic eGFP delivered by C3bot or C3botE174 was also proven to be functional by fluorescence detection. Since hypothesis (ii) could be rejected, the possibility of (i) being correct seems plausible, indicating that even relatively stable proteins such as eGFP are efficiently delivered into the cytosol of target cells via the toxins' uptake mechanism. It is possible that C3bot uses another translocation mechanism compared to the diphtheria or anthrax toxin. This could explain the preserved toxin activity of the eGFP fusion construct (His\_eGFP\_C3bot). This seems reasonable since C3bot has no B-component that remains inside the endosomes [2,18], while this is the case for the classical AB-type protein toxins. Moreover, it was shown earlier that C3bot does not form pores but rather induces membrane flickering upon acidification [18]. By using C3bot or C3botE174Q as a drug delivery tool, this special mechanism could potentially be utilized to deliver stable cargo molecules into the cytosol of C3bot-target cells. In the digitonin-based cell-fractionation assay, low amounts of cytosolic His\_eGFP were detected when the cells

were treated with cargo alone. This minor portion can be explained by unspecific uptake and release. Alternatively, the His-tag could potentially function as a very inefficient cellpenetrating peptide, as shown for cationic elastin-like polypeptide tags that enhance the cellular uptake of GFP in dependency on the number of positive charges [51]. However, in contrast to cells treated with His\_eGFP\_C3botE174Q or His\_eGFP\_C3bot, this portion is neglectable, indicating that the C3bot-based transporters significantly enhanced the cytosolic release of cargo.

To evaluate drug delivery via C3bot in more detail, a modular system for fast attachment of cargo was generated on the basis of thiol–maleimide click chemistry. Three different cargo molecules were efficiently loaded onto the Cys\_C3botE174Q-transporter, and uptake into DCs and macrophages was evaluated. Labeling with mDL488 was used to simulate small molecule cargo (0.8 kDa). A polyethylene glycol cargo model (mPEG\_FITC) was used to simulate more bulky cargo molecules (5 kDa), and the reporter enzyme C2I was used as a model for larger protein cargo (50 kDa). For mDL488 and mPEG\_FITC, uptake into macrophages and DCs was strongly enhanced by coupling to Cys\_C3botE174Q. Increased cytosolic release into DCs was detected for mC2I. Thereby, the modular system was proven to be functional and can be used to deliver a wide range of different drugs into these target cells.

Besides the action on macrophages and DCs, earlier publications have shown that C3bot can independently of its enzymatic activity promote the outgrowth of neuronal cells after spinal cord injury [22]. Coupling C3botE174Q with novel drugs could potentially enhance this effect. Moreover, C3botE174Q could potentially also be used as a drug delivery tool for targeting neuronal cells upon local application. The thiol–maleimide system could serve as a platform for fast attachment of different drug candidates to C3botE174Q and thereby accelerate such a drug development process.

By comparing this thiol–maleimide system with the earlier published biotin–streptavidin coupling system [19,34] for coupling of cargo molecules to C3botE174Q, this novel system provides some benefits but also comes with some costs. A disadvantage of the thiol– maleimide system is that the cargo molecules themselves should not contain thiol groups to ensure a specific and easy coupling process. Otherwise, maleimide-labeled cargo molecules could form multimers. This could be prevented by using protection groups, but this complicates the coupling process. In contrast, the main advantages of the thiol–maleimide system over the biotin–streptavidin coupling system are that streptavidin-induced multimerization can be excluded and that streptavidin cannot influence the natural cell-type selectivity of C3bot, which was reduced/lost, as shown in earlier publications for the biotin– streptavidin coupling system [34,35]. Internalization of the novel system only depends on Cys\_C3botE174Q and the attached cargo molecule.

Overall, the fully modular thiol–maleimide system will widely extend the applications of C3botE174Q as a cell-type-selective drug delivery system for monocyte-derived cells, as well as allowing for fast and easy attachment of various therapeutic cargo molecules including macromolecules such as enzymes, peptides, or nucleic acids.

#### **4. Conclusions**

Taken together, we demonstrated that C3bot and its non-toxic variant C3botE174Q can be used to deliver different cargo molecules into the cytosol of human macrophages and DCs. Direct fusion of eGFP as a cargo model to C3botE174Q neither influenced the cell-type selectivity as evaluated for human blood cells, nor interfered with cytosolic release of the C3 toxin. C3bot and C3botE174Q can therefore serve in the delivery of relatively stable functional proteins into the cytosol of their target cells. Additionally, thiol–maleimide technology was used for rapid and easy attachment of various maleimide-labeled cargo molecules to Cys\_C3botE174Q in order to generate a fully modular transport system for the targeted delivery and cytosolic release of molecules into macrophages and DCs.

#### **5. Materials and Methods**

#### *5.1. Cell Culture*

All cells were cultured at 37 ◦C at constant saturated humidity and 5% CO2 up to passage number 25. For U-DCS cells, a medium mix of IMDM (Lonza, Basel, Switzerland) and RPMI 1640 (Gibco-Life Technologies, Carlsbad, CA, USA) in 4:1 ratio was used, and it was supplemented with 10% fetal calf serum (Gibco-Life Technologies, Carlsbad, CA, USA), 0.1% insulin–transferrin–sodium selenite supplement (Roche Diagnostics, Basel, Switzerland), 1% L-glutamine (Thermo Fisher Scientific, Waltham, MA, USA), and 100 U/mL (1%) penicillin–streptomycin (Gibco-Life Technologies, Carlsbad, CA, USA). Subconfluent U-DCS cells were passaged after trypsin detachment (Roche Diagnostics, Basel, SUI) every 2 to 3 days and split into ratios of 1:2 or 1:3, respectively. J774A.1 macrophages were cultivated in DMEM with 10% FCS, 1 nM sodium pyruvate, 1% penicillin–streptomycin, and 1% non-essential amino acids (all Gibco-Life Technologies, Carlsbad, CA, USA). Subconfluent J774A.1 cells were passaged after mechanical detachment with a cell scraper (Sarstedt, Nümbrecht, Germany) every 2 to 3 days and split into ratios of 1:4 or 1:10, respectively. During passaging, cells were seeded in microtiter plates and used for the experiments on the following or next but one day, respectively.

#### *5.2. Differentiation of Human Monocytes into Macrophages and Co-Cultivation with Lymphocytes*

Density gradient centrifugation (Ficoll-Paque™ Plus; GE Healthcare, Chicago, IL, USA) was used to isolate human peripheral blood mononuclear cells (PBMCs) from buffy coat preparations of anonymous healthy donors (Institute of Transfusion Medicine, Ulm University). PBMCs were allowed to adhere in a cell culture flask for 90 min in AIM V cell culture medium (Gibco-Life Technologies, Carlsbad, CA, USA). Non-adherent autologous cells (lymphocytes) were transferred to a fresh flask and kept in AIM V until their use in the incubator. Plastic-adherent monocytes were incubated with granulocyte-macrophagecolony-stimulating factor (10 ng/mL; Miltenyi Biotec, Bergisch Gladbach, Germany) in macrophage serum-free medium (Gibco-Life Technologies, Carlsbad, CA, USA) to generate primary human macrophages. After 6 days, macrophages were harvested using 1 mM EDTA/PBS (Sigma-Aldrich, St. Louis, MO, USA). If indicated, macrophages and autologous PBMCs were co-cultured in a 1:5 ratio in a poly-L-lysine-coated 8-chamber slide.

#### *5.3. Protein Expression and Cell Lysis*

The plasmids coding for His\_eGFP\_C3bot, His\_eGFP\_C3botE174Q, His\_eGFP (for its origin, see [23]), C3bot, C3botE174Q, Cys\_C3botE174Q (for origin see [52]), C2I, or C2IIa were heat-shock transformed into competent *Escherichia coli* BL21. For the first preculture, 5 mL LB medium (1% tryptone, 0.5% yeast extract, 1% NaCl, 100 μg/mL ampicillin) was inoculated with a single colony for 5–8 h at 37 ◦C and at 180 rpm in a shaking incubator. A second 150 mL overnight culture in an Erlenmeyer flask was inoculated with the preculture. The main culture of 4 L LB medium was inoculated with the overnight preculture (30 mL per L) and incubated at 37 ◦C and 180 rpm until the OD600 reached 0.6–0.8. Protein expression was induced by adding 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG, Carl Roth, Karlsruhe, Germany). Subsequently, the incubation temperature was decreased to 16 ◦C (for eGFP-labeled proteins) or to 29 ◦C (GST-tagged proteins) for the main culture incubated at 180 rpm overnight. For harvesting, the cells were centrifuged at 5500 rcf and 4 ◦C for 10 min, and the pellet was resuspended in 40 mL buffer. GST-Lysis buffer (10 mM NaCl, 20 mM Tris, 1% Triton X-100, 1% phenylmethylsulfonyl fluoride (PMSF); pH 7.4) was used for GST-tagged proteins and NPI-20 buffer (50 mM NaH2PO4, 300 mM NaCl, 20 mM imidazole, 1% PMSF; pH 8.0) for His-tagged proteins. The cells were lysed by sonication (10 pulses each for 20 s with intermediate pauses of 30 s). Cell fragments were removed by centrifugation at 13,000 rcf and 4 ◦C for 30 min. The supernatant was filtered through 0.45 μm and 0.2 μm syringe filters.

#### *5.4. Purification of GST-Tagged Proteins*

C3bot, C3botE174Q, C2I, and Cys\_C3botE174Q were purified as GST-tagged proteins, as described previously [23]. The filtered cell lysates were incubated overnight at 4 ◦C with 1.2 mL Protino Glutathione Agarose 4B-beads (Macherey-Nagel., Düren, Germany), which were preequilibrated in PBS (137 mM NaCl, 2.7 mM KCl, 8 mM Na2HPO4, and 1.8 mM KH2PO4; pH 7.4). After centrifugation at 3000 rcf for 5 min, the beads were washed three times (twice with washing buffer (150 mM NaCl, 20 mM Tris–HCl; pH 7,4) and once with PBS). The proteins were eluted by removing the GST-tag with thrombin (80 NIH units, Amersham Biosciences, Little Chalfont, GBR) for 1 h at RT. The beads were removed by centrifugation for 30 s at 10,000 rcf and 4 ◦C. Thrombin was removed by incubating the supernatant with 120 μL Benzamidine–Sepharose 6B-beads (GE Healthcare, Chicago, IL, USA) for 10 min at RT. Centrifugation at 10,000 rcf and 4 ◦C for 30 s was used to remove the benzamidine beads, and the concentration of purified proteins was determined in SDS-PAGE by comparison to a BSA standard.

#### *5.5. Purification of His-Tagged Proteins*

The proteins His\_eGFP\_C3bot, His\_eGFP\_C3botE174Q, and His\_eGFP were produced as described previously [23]. C2II was produced as a His-tagged protein, as described in [45]. In general, filtered cell lysate containing the His-tagged proteins was incubated with preequilibrated (in NPI-20, see Section 5.3) PureCube 100 INDIGO Ni-agarose (Cube Biotech, Monheim am Rhein, Germany) overnight at 4 ◦C. PureCube 1-step batch Midi Plus Columns (Cube Biotech, Monheim am Rhein, Germany) was used to collect the beads, and they were washed three times with 20 mL NPI-20. For elution of the His-tagged proteins, NPI-250 (50 mM NaH2PO4, 300 mM NaCl, 250 mM imidazole; pH 8.0) was used. Protein fractions were analyzed in SDS-PAGE, and the ones with the highest amount of target protein and the lowest content of impurities were collected. The buffer was exchanged with PBS by ultrafiltration (Vivaspin20 with 10 kDa molecular weight cutoff, Merk, Darmstadt, Germany). The protein solutions were stored at −80 ◦C, and the concentration of purified proteins was determined via SDS-PAGE by comparison to a BSA standard.
