Identification of efficient prokaryotic cell-penetrating peptides with applications in bacterial biotechnology

Abstract

In bacterial biotechnology, instead of producing functional proteins from plasmids, it is often necessary to deliver functional proteins directly into live cells for genetic manipulation or physiological modification. We constructed a library of cell-penetrating peptides (CPPs) capable of delivering protein cargo into bacteria and developed an efficient delivery method for CPP-conjugated proteins. We screened the library for highly efficient CPPs with no significant cytotoxicity in Escherichia coli and developed a model for predicting the penetration efficiency of a query peptide, enabling the design of new and efficient CPPs. As a proof-of-concept, we used the CPPs for plasmid curing in E. coli and marker gene excision in Methylomonas sp. DH-1. In summary, we demonstrated the utility of CPPs in bacterial engineering. The use of CPPs would facilitate bacterial biotechnology such as genetic engineering, synthetic biology, metabolic engineering, and physiology studies.

Fig. 1. The schematic illustration of our strategy for finding efficient prokaryotic CPPs with applications in microbial biotechnology.

a We constructed a library of 98 CPPs from the literature and online databases. b For screening and the development of an efficient CPP delivery method, TAMRA-labeled CPPs were synthesized. c A new method for the improved delivery of CPP-conjugated proteins was developed that was suitable for bacterial engineering applications. d The library of CPPs was screened to identify the most efficient CPPs in E. coli. e The five most efficient CPPs in E. coli in terms of penetration efficiency and cytotoxicity were selected, and their efficiencies for delivering GFP cargo were measured. f The final two selected CPPs were used for bacterial engineering applications including plasmid removal from live E. coli cells using CPP-conjugated I-SceI and marker gene excision in Methylomonas sp. DH-1 using CPP-conjugated Cre recombinase.

Fig. 2. Electroporation-based method improved the penetration efficiencies of TAMRA-labeled CPPs.

a The list of the five selected CPP sequences used to evaluate the efficiencies of different treatment methods and incubation buffers. b The cell penetration efficiencies of the five CPPs measured by flow cytometry (in arbitrary units; A.U.). The CPPs were treated with either chemicals or electroporation with two different buffers (distilled water or Tris-Cl buffer at pH = 7.5). The mean and standard error were calculated from three independent experiments (n = 3–5). The dataset used to draw the graph is compiled in Supplementary Data 1.

Fig. 3. Penetration efficiencies and cytotoxic effects of the CPPs in the library and the distinct features of highly efficient CPP sequences.

a Since the CPPs were conjugated with TAMRA, the TAMRA fluorescence intensities within the cells indicate the cell membrane penetration efficiencies of the CPPs (n = 3–5). b–e Property comparison between the top 15 CPPs (efficient CPPs) and bottom 15 CPPs (inefficient CPPs). In the t test, the p values of the four physicochemical properties were lower than 0.0001. b Average disorder propensity value^†. c Average pI value^‡. d Average net charge at pH = 7.0^¶. e Average hydrophobicity value of CPP sequences^§. f The effect of the CPPs on E. coli cell growth (indicating cytotoxicity). E. coli cells treated with CPPs were grown to the stationary phase. Max OD₆₀₀ denotes the maximum optical density (OD) of CPP-treated cells divided by the maximum OD of control cells. The mean and standard error were calculated from three independent experiments (n = 3). g The list of selected efficient and safe CPPs. The datasets used to draw the graphs are compiled in Supplementary Data 1. Dagger symbol, https://iupred2a.elte.hu/plot Double dagger symbol, https://www.biosyn.com/peptidepropertycalculator/peptidepropertycalculator.aspx Paragraph symbol, https://www.biosyn.com/peptidepropertycalculator/peptidepropertycalculator.aspx Section sign, https://www.thermofisher.com/kr/ko/home/life-science/protein-biology/peptides-proteins/custom-peptide-synthesis-services/peptide-analyzing-tool.html.

Fig. 4. Electromicroscopic pictures of E. coli cells treated with the top five highly efficient CPPs.

Each image set shows a fluorescence image (TAMRA-labeled CPP, top right), an electron microscope image (EM, bottom right), and a merged CLEM image. a–e CPPs that were efficient in penetration. f CPP 22 was inefficient in penetration and was selected as a negative control. Black arrowheads indicate the bacteria in which the cytoplasm was intact and TAMRA-labeled CPPs emitted a red color. White arrowheads indicate bacteria in which the cytoplasm was removed during sectioning and that therefore lacked CPPs.

Fig. 5. The nine designed CPP sequences and their penetration efficiency and cytotoxicity.

a The list of nine CPP sequences that were predicted by our developed computational model to have highly efficient cell membrane penetration. b, c Evaluated penetration efficiency and cytotoxicity of the TAMRA-labeled CPPs, respectively. Biological replicates (n): n = 3–5 for penetration efficiency, n = 5 for cytotoxicity. Max OD (%) denotes the maximal OD of CPP-treated cells divided by the maximal OD of non-treated cells, multiplied by 100. The datasets used to draw the graphs are compiled in Supplementary Data 1.

Fig. 6. Penetration efficiencies of the five selected CPP-conjugated GFP proteins.

a The genetic structure designed to express CPP-conjugated GFP proteins in pET21a(+) plasmids, which contained T7 promoter-lac operator-RBS-CPP-(Gly–Ala linker)-GFP-(6 × His tag)-terminator. b Cell penetration assay results of the five purified CPP-conjugated GFP proteins. GFP intensities of cells were measured by flow cytometry after washing out extracellular CPP-conjugated GFP proteins. The protein cargo (GFP) delivery efficiencies of the CPPs were normalized by the efficiency of a control GFP protein (with no CPP). The mean and standard error were calculated from three independent experiments (n = 3). The dataset used to draw the graph is compiled in Supplementary Data 1.

Fig. 7. Plasmid removal using CPP-conjugated I-SceI in E. coli and marker gene excision from the genome of Methylomonas sp. DH-1 using CPP-conjugated Cre recombinase.

a, b Plasmid removal in E. coli. c–e Marker gene excision from the genome of Methylomonas sp. DH-1. a The strategy for CPP-based plasmid removal. The target plasmid to be removed contained an I-SceI recognition site, the tetracycline resistance gene for selection and counterselection, and three different replication origins (pSC101 origin, ~5 copies; ColE1 origin, 15–20 copies; RSF1030 origin, >100 copies). b Measured plasmid removal efficiencies using CPP-conjugated I-SceI proteins (n = 3). CPP 45 and CPP 70 had significantly increased efficiencies compared with that of the control, but there were no statistically significant differences between CPP 45 and CPP 70. c The DNA structure used for genomic integration into the genome of Methylomonas sp. DH-1. d The strategy for CPP-based marker gene excision from the genome of Methylomonas sp. DH-1. e Marker gene excision efficiencies of CPP-conjugated Cre recombinases (n = 3). Using Tris-Cl buffer did not result in significant efficiency differences among the control, CPP 45 and CPP 70. However, nitrate mineral salts (NMS) growth media resulted in a statistically different efficiency only for CPP 70. The datasets used to draw the graphs are compiled in Supplementary Data 1.

Fig. 8. Penetration efficiencies and cytotoxic effects of the CPPs in HEK293 and CHO cells, and comparison of CPP effects in different cell types.

a, b Ten highly efficient CPPs and ten highly inefficient CPPs in E. coli were selected, and their cell penetration efficiency was measured in HEK293 and CHO cells (n = 9). c–e Linear correlations of the penetration efficiencies among the three cell types. f, g Ten highly toxic CPPs and ten highly safe CPPs in E. coli were selected, and their cytotoxicity was measured in HEK293 and CHO cells (n = 3). h–j Linear correlations of the cytotoxicity among the three cell types. The datasets used to draw the graphs are compiled in Supplementary Data 1.