Response by Dichek to Letter Regarding Article, “Jugular Vein Injection of HighTiter Lentiviral Vectors Does Not Transduce the Aorta”

To the Editor:

We understand and empathize with Dr. Redondo’s disappointment with our report.¹ We share his disappointment because we so much wanted this to work! Regarding specific critiques, we reaffirm that the titers of the lentivirus (LV) preparations that we used are within the range of those reported by Dr. Redondo and his collaborators, who also reported success with an LV preparation of 1×10⁸ transducing units/mL,² 90% below the 1×10⁹ transducing units/mL that Dr. Redondo recently suggested as essential,³ and well below the titers of the LV preparations that we used. All studies, including ours, have limitations. However, positive controls included in our report confirm our ability to detect high-level GFP expression in mouse tissue if present, our ability to extract and detect small amounts of LV DNA if present in mouse arteries transduced either in vitro (aortic segment) or in vivo (several carotid arteries), and our ability to inject sufficient intravascular LV to achieve in vivo gene transfer to tissues susceptible to LV transduction (liver and spleen) with 100% reproducibility. Our standard curves confirm our ability to detect the GFP DNA sequence with high sensitivity, even if it is contained in a small DNA fragment such as a plasmid. Fragmentation of genomic DNA in aortic extracts cannot explain our negative results.

Dr. Redondo criticizes our immunohistochemistry as inconsistent with our LV genome quantitation and lacking in sensitivity. Importantly, our immunohistochemistry was not designed to quantify LV genomes, if that could even be achieved by immunostaining. There are simple explanations for why results of LV genome quantitation and immunostaining of GFP in the same tissue would not match. A single GFP-expressing cell may contain several LV genomes. Alternatively, an unstained cell may contain LV genomes that are not expressing GFP, for example, because they are integrated in a silent area of the genome. As concerns sensitivity, our staining technique can clearly detect low-level LV transduction as shown by positive staining of spleen and liver (0.8 and 0.1 LV genomes/cell, respectively) and of right carotid arteries (~1 LV genome/cell). Supporting the specificity of our immunostaining, it is expected that GFP staining would be confined to the outer surfaces of the in vivo transduced carotids and the ex vivo transduced aorta, and that is exactly where we find it. Finally, Dr. Redondo misrepresents the methods we used to generate the ex vivo positive controls. The aortic segment used for DNA extraction (Figure 1B) was transduced at 5×10⁷ transducing units/mL, whereas the segment used for immunostaining (Figure 2B) was transduced at 1×10⁷ transducing units/mL. It is unsurprising that GFP staining of the aortic segment is less abundant or intense than might be expected in tissue with ~12 LV genomes per cell.

It is curious that Dr. Redondo questions our technical experience and training. Our laboratory has worked on vascular cell gene transfer for over 30 years.⁴ We have paid particular attention to quantitative aspects of arterial gene transfer⁵ and to defining the cell-type specificity of arterial gene transfer.⁶ We optimized methods for in vivo gene transfer to the artery wall,⁷ and we defined the role of endothelial barrier function in determining the distribution of viral vectors and the location of transduced cells in the artery wall.⁸ We were the first group to report in vivo transduction of mouse arteries.⁹ All of this work has withstood the test of time. The operator in the present study (LB) is a skilled microsurgeon who has performed scores of arteriotomies and vector injections in rabbits.^{10, 11}

If, as suggested by Dr. Redondo, our negative results are due to insufficient technical experience and training, this will become apparent when multiple laboratories independently report convincing molecular evidence of efficient intravenous LV-mediated gene transfer to the aorta. This evidence would be expected to include immunohistochemical, histochemical, and fluorescence microscopy detection of cell-localized expression of multiple LV-expressed reporter genes.^12–14 It would be expected that these methods would not reveal—as portrayed in Dr. Redondo’s publications—diffusely positive staining of ~100% of aortic mural cells and surrounding tissues.^{15, 16} Rather, it would be expected that some cells would be distinctly positive for reporter gene expression and others would be negative, as reported by others in efficiently LV-transduced mouse tissues.^12–14 Evidence of successful aortic transduction would also need to include robust detection of LV DNA, in an amount sufficient to explain phenotypic effects. To our knowledge, detection of aortic LV DNA in LV-injected mice has never been reported by any group. As explained in more detail in our report,¹ presence of LV DNA is a critical factor in confirming LV-mediated gene transfer. Detection of aortic LV DNA is not reported in any of Dr. Redondo’s own or collaborative publications^{2, 3, 15–17} nor is it reported in any of the other 4 papers he cites as supportive of his work.^18–21 None of these papers provide convincing evidence of cell-localized recombinant gene expression in the aorta. Indeed one of these papers does not even include a control group for detection of aortic GFP fluorescence (Xu et al; Fig. S2A).²⁰ All 4 papers use tail vein injections of low-titer LV preparations (1×10⁷–2×10⁸ transducing units/mL), an injection site and titers that Dr. Redondo has associated with failure to achieve efficient aortic transduction.³ Rather than being supportive of Dr. Redondo’s claims, these papers present yet another challenge to his conceptualization of how intravenous LV-mediated aortic gene transfer might work.

It is at least 10 years since in vivo LV-mediated aortic transduction was first reported.¹⁵ In 2021, there are still only a few reports of success. Most of these reports are co-authored by Dr. Redondo,^{2, 3, 15–17} and all of these reports lack critical data that would support efficacy. We suggest a pathway forward for Dr. Redondo, including steps that could help him to establish broad confidence in his results. (1) Publish methods for LV injection and detection in a publicly accessible format such as the Journal of Visualized Experiments. We have reported our own detailed vascular gene transfer methods in this format.^{11, 22} (2) Measure LV DNA in aortas, livers, spleens, and bone marrow of LV-infused mice and report his results. If abundant aortic LV DNA is found, we would be glad to learn how we missed it. If abundant aortic LV DNA is not found, Occam’s razor says that the aorta is not transduced and that diffuse GFP immunostaining suggesting otherwise^{15, 16} is a falsely positive indicator of aortic transduction. (3) Measure LV DNA in tissue adjacent to the jugular vein injection site. This approach is likely to provide a far more sensitive assessment of local LV release than is provided by the acute visual assessment of Evans blue dye leakage that Dr. Redondo suggests in his letter. (4) Provide a plausible biological explanation (as well as supportive experimental data) for how injection of 10⁷–10⁸ LV transducing units into a mouse jugular vein could yield ~100% transduction of aortic cells. This level of aortic cell transduction is implausible because it conflicts with numerous quantitative studies from leading gene-therapy groups showing that the mouse liver (~3.5×10⁸ cells),²³ spleen (~2×10⁸ cells),²⁴ and bone marrow (~5×10⁸ cells)²⁵ are all highly transduced by intravenously injected LV vectors whereas organs with large amounts of arterial and other vascular tissue (lung, heart, kidney) are essentially not transduced.^{12–14, 26–28} (5) Explain why injection of LV by his collaborators seems to transduce only the aortic luminal endothelium (de Yebenes et al.,¹⁷ Figure 5B, in which tissue and antibody controls are also absent), whereas LV injections in his own laboratory with the same LV pseudotype and backbone seem to transduce 100% of cells throughout the full thickness of the aortic wall.^{15, 16} (6) Finally, we encourage Dr. Redondo to accept my invitation, offered on 12/11/20, to engage in precommitment that would lead to a registered report²⁹ of results of experiments aimed at reproducing highly efficient aortic transduction after intravenous injection of LV vectors. Following these 6 suggestions would allow the field to move beyond recrimination and instead refocus community attention on experimental data and on methods that are both detailed and easy to reproduce.