Deep Learning Enables Discovery of a Short Nuclear Targeting Peptide for Efficient Delivery of Antisense Oligomers

Abstract

Therapeutic macromolecules such as proteins and oligonucleotides can be highly efficacious but are often limited to extracellular targets due to the cell’s impermeable membrane. Cell-penetrating peptides (CPPs) are able to deliver such macromolecules into cells, but limited structure–activity relationships and inconsistent literature reports make it difficult to design effective CPPs for a given cargo. For example, polyarginine motifs are common in CPPs, promoting cell uptake at the expense of systemic toxicity. Machine learning may be able to address this challenge by bridging gaps between experimental data in order to discern sequence–activity relationships that evade our intuition. Our earlier data set and deep learning model led to the design of miniproteins (>40 amino acids) for antisense delivery. Here, we leveraged and expanded our model with data augmentation in the short CPP sequence space of the data set to extrapolate and discover short, low-arginine-content CPPs that would be easier to synthesize and amenable to rapid conjugation to desired cargo, and with minimal in vivo toxicity. The lead predicted peptide, termed P6, is as active as a polyarginine CPP for the delivery of an antisense oligomer, while having only one arginine side chain and 18 total residues. We determined the pentalysine motif and the C-terminal cysteine of P6 to be the main drivers of activity. The antisense conjugate was able to enhance corrective splicing in an animal model to produce functional eGFP in heart tissue in vivo while remaining nontoxic up to a dose of 60 mg/kg. In addition, P6 was able to deliver an enzyme to the cytosol of cells. Our findings suggest that, given a data set of long CPPs, we can discover by extrapolation short, active sequences that deliver antisense oligomers.

In-cell penetration selection—mass spectrometry produces noncanonical peptides for antisense delivery

“…The PMO delivery activity of the library is likely energy-dependent, similar to PMO-CPP conjugates previously investigated. 18,19,46 A 1,000-member portion of the PMO-library was incubated with cells at 4 °C, conditions that arrest energy-dependent uptake. After incubation with the PMO-CPP conjugates, each well was washed extensively with PBS and heparin in order to disrupt and remove membrane-bound conjugates.…”

“…This trend of positive charge correlating with PMO activity has been observed consistently in our lab. [18][19][20]46 Although the hit peptides do not show higher activity than the parent peptide penetratin in its native L-form, they do show less toxicity and membrane disruption than penetratin at 25 µM in an LDH release assay (Fig. 5D), demonstrating their potential utility as PMO delivery vehicles.…”

“…Previous PMO-CPPs discovered in our laboratory are hypothesized to enter cells via clathrin-mediated endocytosis, as demonstrated through incubation with a panel of chemical endocytosis inhibitors, suggesting this could be a likely mechanism of energy-dependent uptake for a PMO-CPP library. 18,19,46 To confirm this, a 1,000 member library was tested with a series of chemical endocytosis inhibitors in a pulse-chase format EGFP assay, in which HeLa 654 cells were pre-incubated with inhibitors to arrest various endocytosis pathways before PMO-CPPs were added. Following 3 h co-incubation, cells were washed extensively with heparin to dissociate membrane-bound constructs.…”

“…Following 3 h co-incubation, cells were washed extensively with heparin to dissociate membrane-bound constructs. 18,19,46 Activity of the library sample was reduced by 10 µM chlorpromazine (Fig. 3E), a known inhibitor of clathrin-mediated endocytosis.…”

“…[12][13][14][15][16][17] The latter include our own recent work that leverages machine learning to design new sequences using a model trained with a library tested for the desired activity: nuclear localization. [18][19][20] Finally, another common strategy involves screening platforms employing libraries from phage or mRNA display. [21][22][23][24] For example, a screening platform identified several "phylomer" CPPs from bacterial and viral genomes that were then shown to deliver antisense cargo in vivo.…”