Reduced blood lipid levels with in vivo CRISPR-Cas9 base editing of ANGPTL3

Naturally occurring loss-of-function mutations in ANGPTL3 (angiopoietin-like 3) are associated with reduced blood triglycerides, LDL cholesterol, and risk of coronary heart disease, with no apparent adverse health consequences, making ANGPTL3 a compelling therapeutic target.¹ We assessed whether base editing—a variation on CRISPR-Cas9 genome editing that does not require DNA double-strand breaks—could be used in vivo to introduce loss-of-function mutations into ANGPTL3 and reduce blood lipid levels. Base editor 3 (BE3) can introduce cytosine-to-thymine changes at desired sites in the genome,² e.g., nonsense mutations into Pcsk9 (proprotein convertase subtilisin/kexin type 9) in mice.³ Animal studies and other procedures were performed as previously described³ in accordance with University of Pennsylvania guidelines.

We screened potential sites of base-edited nonsense mutations in Angptl3 in Neuro-2a cells, identifying robust BE3 activity at the codon Gln-135 site with protospacer sequence AGCCCTTCAACACAAGGTCA (codon underlined). We produced adenoviral vectors encoding BE3 with no guide RNA (BE3-control), guide RNA targeting Angptl3 Gln-135 (BE3-Angptl3), and guide RNA targeting Pcsk9 Trp-159 (BE3-Pcsk9). First, we injected BE3-control or BE3-Angptl3 into 5-week-old male C57BL/6J mice. At day 7, deep sequencing of the Angptl3 target site in liver samples from euthanized BE3-Angptl3-treated mice revealed a median editing rate of 35%; deep sequencing of 10 top predicted sites of off-target mutagenesis (from CRISPOR website) showed no evidence of editing (Figure, A–B). Pre-treatment plasma ANGPTL3, triglycerides, and cholesterol (R&D Systems #MANL30; Thermo Fisher #TR22421, #TR13421) were similar between the groups (2.0%, 0.1%, 0.5% differences), whereas at day 7 the mean analyte levels were all significantly lower in BE3-Angptl3-treated mice (49%, 31%, 19%) (Figure, A).

Figure. In Vivo Base Editing of Angptl3.

(A) Post-treatment editing of Angptl3 alleles in liver (1 BE3-control-treated and 9 BE3-Angptl3-treated mice); post-treatment plasma analytes (n=9 mice/group) with medians indicated. (B) Frequencies of specific edited alleles in most-edited mouse and control mouse; targeted Angptl3 Gln-135 codon underlined in wild-type sequence, edited bases in bold. Base-editing frequencies at 10 top predicted off-target sites also indicated. (C) Post-treatment analytes (n=5 mice/group). *P<0.05; **P<0.01; ***P<0.001 (compared to BE3-control if not indicated). (D) Post-treatment analytes, bone marrow hematopoietic stem cells, and lipoprotein profiles (n=6 Ldlr-knockout mice/group). All mice obtained directly from The Jackson Laboratory. P-values from two-tailed Mann–Whitney U tests (A, D) or Conover–Iman post-hoc tests adjusted by the Holm FWER method following Kruskal–Wallis tests (C).

Second, we assessed 5-week-old male C57BL/6J mice injected with BE3-control, BE3-Angptl3, BE3-Pcsk9, or a half/half mix of BE3-Angptl3 and BE3-Pcsk9 for plasma ANGPTL3, PCSK9 (R&D Systems #MPC900), triglycerides, and cholesterol. At post-treatment day 7, Angptl3 targeting alone caused a greater decline in triglycerides than targeting Pcsk9 alone, and neither additivity nor synergism was observed when both genes were targeted; comparable declines in cholesterol occurred with all 3 interventions (Figure, C).

Finally, we assessed BE3-Angptl3 in hyperlipidemic Ldlr-knockout mice (5-week-old male B6.129S7-Ldlr^tm1Her/J mice), which phenocopy homozygous familial hypercholesterolemia patients and in which Pcsk9 knockout has little effect.⁴ At post-treatment day 14, BE3-Angptl3 substantially reduced triglycerides (56%) and cholesterol (51%) compared to BE3-control (Figure, D). Previous studies of Angptl3-knockout mice documented decreased bone marrow hematopoietic stem cells;⁵ we observed no decrease in BE3-Angptl3-treated mice (BD Biosciences #560492) (Figure, D).

In conclusion, we established in vivo base editing of ANGPTL3 as a potential strategy to treat patients with atherogenic dyslipidemia. Further refinements of base-editing technology could improve the efficiency of ANGPTL3 targeting beyond the already substantial ≈35% observed in this study.