A new mRNA structure prediction based approach to identifying improved signal peptides for bone morphogenetic protein 2

Abstract

Background: Signal peptide (SP) engineering has proven able to improve production of many proteins yet is a laborious process that still relies on trial and error. mRNA structure around the translational start site is important in translation initiation and has rarely been considered in this context, with recent improvements in in silico mRNA structure potentially rendering it a useful predictive tool for SP selection. Here we attempt to create a method to systematically screen candidate signal peptide sequences in silico based on both their nucleotide and amino acid sequences. Several recently released computational tools were used to predict signal peptide activity (SignalP), localization target (DeepLoc) and predicted mRNA structure (MXFold2). The method was tested with Bone Morphogenetic Protein 2 (BMP2), an osteogenic growth factor used clinically for bone regeneration. It was hoped more effective BMP2 SPs could improve BMP2-based gene therapies and reduce the cost of recombinant BMP2 production.

Results: Amino acid sequence analysis indicated 2,611 SPs from the TGF-β superfamily were predicted to function when attached to BMP2. mRNA structure prediction indicated structures at the translational start site were likely highly variable. The five sequences with the most accessible translational start sites, a codon optimized BMP2 SP variant and the well-established hIL2 SP sequence were taken forward to in vitro testing. The top five candidates showed non-significant improvements in BMP2 secretion in HEK293T cells. All showed reductions in secretion versus the native sequence in C2C12 cells, with several showing large and significant decreases. None of the tested sequences were able to increase alkaline phosphatase activity above background in C2C12s. The codon optimized control sequence and hIL2 SP showed reasonable activity in HEK293T but very poor activity in C2C12.

Conclusions: These results support the use of peptide sequence based in silico tools for basic predictions around signal peptide activity in a synthetic biology context. However, mRNA structure prediction requires improvement before it can produce reliable predictions for this application. The poor activity of the codon optimized BMP2 SP variant in C2C12 emphasizes the importance of codon choice, mRNA structure, and cellular context for SP activity.

Keywords: Bone morphogenetic protein2; Gene therapy6; Protein engineering4; RNA structure3; Regenerative medicine7; Signal peptides1; Synthetic biology5.

Fig. 1

Overview of the filtering process employed in the in silico pipeline. Initially ~ 35,000 TGF-β superfamily sequences were retrieved, though only ~ 19,000 were predicted to contain SPs. Of these, ~ 10,000 were found to be valid records and to have unique nucleotide sequences. Existing computational methods indicated ~ 7,000 were predicted to still function and to target the extracellular space when attached to BMP2. ~2,500 were found to have strong Kozak sequences and were taken forward to mRNA structure prediction. Sequences with the least structure at the translational start site but high global stability were thought to be preferable. The top 5 sequences from the pipeline and two manually selected alternatives were taken forward to in vitro work

Fig. 2

Further nucleotide sequence analysis (A) Nucleotide sequence alignment of the final 7 SP sequences. Note the moderate variety with few positions being strongly conserved across the final set. A ~ 25 bp insertion can be seen in the SPs from Artemin related genes (SPs 3–5), making them noticeably longer than the other sequences. B Predicted mRNA secondary structures at the ribosomal attachment site. The ± 15 bp window is highlighted in blue, with start codon found at positions 151–153

Fig. 3

Further protein sequence analysis (A) Protein sequence alignment of the final 7 SP sequence. Note the series of conserved leucine residues, the only conserved element across the final panel. Unsurprisingly the three SPs from Artemin related genes (SPs 3–5) were highly similar. B Detailed SP region boundary predictions from the SignalP long model. The Artermin related SPs displayed elongated N-regions in comparison to the rest of the set, corresponding to the position of the insertion in the nucleotide sequences (see Fig. 2A). Also note the leucine motif from the alignment is contained within the SP H-region, thought to be important for maintaining hydrophobicity and ensuring the alpha helical conformation required for transmembrane function

Fig. 4

In vitro validation data (A) hBMP2 ELISA data from HEK293T cells. All computationally selected sequences showed non-significant increases in secretion. N = 7 ± SD, **** = p < 0.0001 (B) hBMP2 ELISA data from C2C12 cells. All computationally selected sequences showed decreased secretion versus the positive control, with several showing significant decreases. Particularly notable were the sequences that showed major decreases in the C2C12 results compared to the HEK293T experiment. These were SP1, SP2, SP hIL2 and SP hBMP2 TIS. N = 3 ± SD. * = p < 0.05, ** = p < 0.01, *** = p < 0.001 C) ALP assay data from C2C12 cells. All groups except the positive control showed background levels of ALP activity, indicating the novel SPs were not sufficient to induce osteogenesis in this context. N = 3 ± SD, *** = p < 0.001

Fig. 5

Linear regression analyses of the SP score values and ELISA data (A) ELISA data from the HEK239T experiment. Correlation was found to be significant (gradient = 0.0509, R2 = 0.096, F [1, 44] = 4.976, p = 0.0305) (B) ELISA data from the C2C12 experiment. No significant correlation was found (gradient = 0.0427, R2 = 0.138, F [1, 19] = 3.038, p = 0.0975)