Karsten Schmidt

Triantennary N-acetyl galactosamine (GalNAc, GN3: ), a high-affinity ligand for the hepatocyte-specific asialoglycoprotein receptor (ASGPR), enhances the potency of second-generation gapmer antisense oligonucleotides (ASOs) 6-10-fold in mouse liver. When combined with next-generation ASO designs comprised of short S-cEt (S-2'-O-Et-2',4'-bridged nucleic acid) gapmer ASOs, ∼ 60-fold enhancement in potency relative to the parent MOE (2'-O-methoxyethyl RNA) ASO was observed. GN3: -conjugated ASOs showed high affinity for mouse ASGPR, which results in enhanced ASO delivery to hepatocytes versus non-parenchymal cells. After internalization into cells, the GN3: -ASO conjugate is metabolized to liberate the parent ASO in the liver. No metabolism of the GN3: -ASO conjugate was detected in plasma suggesting that GN3: acts as a hepatocyte targeting prodrug that is detached from the ASO by metabolism after internalization into the liver. GalNAc conjugation also enhanced potency and duration of the effect of two ASOs targeting human apolipoprotein C-III and human transthyretin (TTR) in transgenic mice. The unconjugated ASOs are currently in late stage clinical trials for the treatment of familial chylomicronemia and TTR-mediated polyneuropathy. The ability to translate these observations in humans offers the potential to improve therapeutic index, reduce cost of therapy and support a monthly dosing schedule for therapeutic suppression of gene expression in the liver using ASOs.

We have purified contulakin-G, a 16-amino acid O-linked glycopeptide (pGlu-Ser-Glu-Glu-Gly-Gly-Ser-Asn-Ala-Thr-Lys-Lys-Pro-Tyr-Ile-Leu-OH, pGlu is pyroglutamate) from Conus geographus venom. The major glycosylated form of contulakin-G was found to incorporate the disaccharide beta-D-Galp-(1-->3)-alpha-D-GalpNAc-(1-->) attached to Thr10. The C-terminal sequence of contulakin-G shows a high degree of similarity to the neurotensin family of peptides. Synthetic peptide replicates of Gal(beta-->3) GalNAc(alpha-->)Thr10 contulakin-G and its nonglycosylated analog were prepared using an Fmoc (9-fluorenylmethoxycarbonyl) protected solid phase synthesis strategy. The synthetic glycosylated con- tulakin-G, when administered intracerebroventricular into mice, was found to result in motor control-associated dysfunction observed for the native peptide. Contulakín-G was found to be active at 10-fold lower doses than the nonglycosylated Thr10 contulakin-G analog. The binding affinities of contulakin-G and the nonglycosylated Thr10 contulakin-G for a number of neurotensin receptor types including the human neurotensin type 1 receptor (hNTR1), the rat neurotensin type 1 and type 2 receptors, and the mouse neurotensin type 3 receptor were determined. The binding affinity of the nonglycosylated Thr10 contulakin-G was approximately an order of magnitude lower than that of neurotensin1-13 for all the receptor types tested. In contrast, the glycosylated form of contulakin-G exhibited significantly weaker binding affinity for all of the receptors tested. However, both contulakin-G and nonglycosylated Thr10 contulakin-G were found to be potent agonists of rat neurotensin receptor type 1. Based on these results, we conclude that O-linked glycosylation appears to be a highly unusual strategy for increasing the efficacy of toxins directed against neurotransmitter receptors.

The comprehensive structure-activity relationships of triantennary GalNAc conjugated ASOs for enhancing potency via ASGR mediated delivery to hepatocytes is reported. Seventeen GalNAc clusters were assembled from six distinct scaffolds and attached to ASOs. The resulting ASO conjugates were evaluated in ASGR binding assays, in primary hepatocytes, and in mice. Five structurally distinct GalNAc clusters were chosen for more extensive evaluation using ASOs targeting SRB-1, A1AT, FXI, TTR, and ApoC III mRNAs. GalNAc-ASO conjugates exhibited excellent potencies (ED50 0.5-2 mg/kg) for reducing the targeted mRNAs and proteins. This work culminated in the identification of a simplified tris-based GalNAc cluster (THA-GN3), which can be efficiently assembled using readily available starting materials and conjugated to ASOs using a solution phase conjugation strategy. GalNAc-ASO conjugates thus represent a viable approach for enhancing potency of ASO drugs in the clinic without adding significant complexity or cost to existing protocols for manufacturing oligonucleotide drugs.

Follistatin is a transcriptional target and a modulator of activin action. Through an autocrine/paracrine loop, activin controls follistatin levels and thus regulates its own bioavailability. In gonadotropic alphaT3-1 cells, activin induces follistatin transcription primarily through the action of Smad3 at an intronic Smad-binding element (SBE1). Using a proteomics approach, we searched for endogenous alphaT3-1 proteins that participate in SBE1-mediated transcription. We identified FoxL2, a member of the forkhead family, as a candidate modulator of SBE1 function. Mutations of FoxL2 are associated with the blepharophimosis/ptosis/epicanthus inversus syndrome characterized with craniofacial defects and premature ovarian failure. FoxL2 localizes to alpha-glycoprotein subunit- and follicle-stimulating hormone beta-positive cells of the adult mouse pituitary and is present in alphaT3-1 and LbetaT2 cells, but its pituitary actions remain largely unknown. We have determined that FoxL2 binds to a forkhead-binding element (FKHB) located just downstream of the SBE1 site of the follistatin gene and functions as a Smad3 partner to drive SBE1-mediated transcription in alphaT3-1 cells treated with activin. Chromatin immunoprecipitation assays confirm that endogenous FoxL2 and Smad3 are recruited to the intronic enhancer of the follistatin gene where the SBE1 and FKHB sites are located. Exogenous FoxL2 enhances SBE1-mediated transcription, and short hairpin RNA-mediated knockdown of endogenous FoxL2 protein compromises this effect in alphaT3-1 cells. FoxL2 directly associates with Smad3 but not Smad2 or Smad4. This association between Smad3 and FoxL2 is mediated by the MH2 domain of Smad3 and is dependent on an intact forkhead domain in FoxL2. Altogether, these observations highlight a novel role for FoxL2 and suggest that it may function as a transcriptional regulator and a coordinator of Smad3 targets.

Targeted delivery of antisense oligonucleotides (ASO) to hepatocytes via the asialoglycoprotein receptor (ASGR) has improved the potency of ASO drugs ∼30-fold in the clinic (1). In order to fully characterize the effect of GalNAc valency, oligonucleotide length, flexibility and chemical composition on ASGR binding, we tested and validated a fluorescence polarization competition binding assay. The ASGR binding, and in vitro and in vivo activities of 1, 2 and 3 GalNAc conjugated single stranded and duplexed ASOs were studied. Two and three GalNAc conjugated single stranded ASOs bind the ASGR with the strongest affinity and display optimal in vitro and in vivo activities. 1 GalNAc conjugated ASOs showed 10-fold reduced ASGR binding affinity relative to three GalNAc ASOs but only 2-fold reduced activity in mice. An unexpected observation was that the ASGR also appears to play a role in the uptake of unconjugated phosphorothioate modified ASOs in the liver as evidenced by the loss of activity of GalNAc conjugated and unconjugated ASOs in ASGR knockout mice. Our results provide insights into how backbone charge and chemical composition assist in the binding and internalization of highly polar anionic single stranded oligonucleotides into cells and tissues.