Delivery: CRISPR Nanoblades Target HIV in CD4 Cells

Initial experiments used mCherry-loaded particles to track uptake. In primary human peripheral blood mononuclear cells, the particles bound and entered CD4⁺ T cells, CD4⁺ dendritic cells and CD4⁺ monocytes far more strongly than CD4-negative subsets. In humanised mice, the particles accumulated mainly in kidney, liver and spleen, but in blood they still showed enrichment for human CD4⁺ immune cells. This distinction is important: the delivery system showed clear CD4 preference in circulation, but tissue targeting was less complete.

The team next asked whether delivery translated into editing. To test this, they used mixed CD4⁺ and CD4⁻ Sup-T1 reporter cells, both of which had been engineered beforehand to carry randomly integrated eGFP. Nanoblades carrying an eGFP-targeting guide disrupted eGFP expression in CD4⁺ cells but not in CD4⁻ cells, showing that the CRISPR cargo was delivered selectively through CD4 recognition.

In primary CD4⁺ cells from six donors, CD4-NBs achieved knockout efficiencies comparable to RNP electroporation while using roughly 250-fold less Cas9 per cell, and did so without the viability loss associated with electroporation.

For the HIV experiments, the researchers loaded CD4-NBs with two guide RNAs targeting conserved regions of the viral tat locus. Because tat overlaps functionally with rev and env coding regions, cutting this area can interfere with several viral functions at once. In latently infected J.Lat cells, the tat-targeting nanoblades produced the expected excised viral DNA fragment and reduced HIV p24 production. In actively infected Sup-T1 cells and primary CD4⁺ cells, treatment lowered viral gene expression and reduced production of infectious virus, although the effect varied across primary donors.

The in vivo experiments were more restrained but informative (see Figure 2). Humanised mice were infected with HIV, treated briefly with antiretroviral therapy, then given either control nanoblades or tat-targeting CD4-NBs before treatment interruption. Viral rebound occurred in all mice, but plasma viraemia was lower over time in animals receiving tat-targeting nanoblades. A second experiment using two doses again lowered plasma viraemia, and indels at the tat guide site were detected in bone marrow and lung tissue in some treated mice. However, viral RNA and proviral DNA were not consistently or significantly cleared from tissues.

The work shows that CD4-directed nanoblades can target CRISPR editing towards HIV-relevant immune cells and suppress viral replication, but it does not yet demonstrate sterilising reservoir clearance. The platform therefore looks most valuable as a proof of concept for targeted, transient CRISPR delivery to HIV-susceptible cells, rather than as a finished cure strategy.

The authors suggest that a limitation of the study is the low in vivo efficacy observed, and they attribute this partly to sequestration by liver and kidney and to inefficient extravasation into lymphoid tissue where HIV reservoirs reside. They suggest that incorporating capillary-targeting ligands or stealth modifications such as PEGylation or CD47 "don't-eat-me" signals could improve tissue delivery in future iterations. They also propose combining CD4-directed editing with CCR5 disruption, or pairing the platform with shock-and-kill strategies using latency-reversing agents and anti-HIV CAR T cells, rather than expecting a single dosing regimen to clear the reservoir outright.

The study was led by Jolien Van Cleemput and Linos Vandekerckhove at Ghent University in Belgium. It was published in EMBO Molecular Medicine on 6 July 2026.