Human Gene Therapy - April 2023 - 282

282
WESTHAUS ET AL.
Figure 4. AAV-treated NHP serum analysis. (a) Schematic of serum collection before apheresis (antibody depletion), before AAV injection, 1 and 24 h after
injection as well as at killing 1 week after injection. (b) NAb titers in collected serum at indicated time points for the indicated AAV variants. (c) AAV copy
number per microliter and AAV variant from collected serum at indicated time points. NAb, neutralizing antibody.
est, AAV5 was the serotype that entered cells most efficiently
in the gallbladder and liver capsule (Fig. 3b). AAV2
and AAV8 were the least efficient serotypes at cell entry
(Fig. 3b). In terms of transgene expression, AAV-LK03
was the best performing variant, followed by AAV3b and
AAV-NP59. AAV5, AAV2, and AAV8 performed relatively
poorly (Fig. 3c). Although these observations require
replication in additional NHPs, the hierarchies of capsid
performance observed were sufficiently distinctive to offer
preliminary insights worthy of discussion.
As the NHP experiment was terminated just 1 week
after vector infusion, it is possible that expression from
some of the AAV variants may not have reached the peak
level.58 It is therefore possible that AAVs that entered well
(such as AAV5) could have, with time, led to higher
mRNA expression compared with AAVs that entered less
efficiently, but appear to have higher kinetics of expression
(such as AAV-NP59). Although the authors appreciated
the fact that the timing of the study could directly
affect the study outcome, the main reason for the shorter
timeline was to ensure consistency between ex vivo (1
week or less) and xenograft in vivo (1 week) experiments.
Based on previously published data, we were not surprised
by the high performance of AAV-LK03.55 It was
also interesting to see how much better AAV-NP59 performed
compared with the prototypical AAV2, which is
highly homologous at the protein level, but potentially
tissue culture adapted.43 This indicates that the previously
published advantage of lower HSPG binding could be
beneficial in in vivo xenograft mouse models as well as
NHP models.31 As AAV-NP59 has not been evaluated in
human studies, we can only rely on the hFRG, Rh/CyFRG
and in vivo NHP data to infer clinical efficiency of this
variant. Our data strongly suggest that AAV-NP59 would
most likely perform substantially better than AAV2, but
less efficiently than AAV3b/AAV-LK03, as we can see
that the performance of AAV-NP59 in the CyFRG mice
was overestimated compared with performance in Cynomolgus
macaque in vivo. Finally, the most surprising
finding from the NHP experiment was the very low AAV8
performance,55 which warranted further investigation.
NHP serum analysis
Driven by the fact that AAV8 performance in the NHP
liver was lower than anticipated based on published data,55
we analyzed the levels of anti AAV NAbs in the serum
harvested before AAV infusion. In addition, the clearance
of AAVs from the serum was quantified. Both seroreactivity
and AAV clearance were evaluated at five time
points (Fig. 4a).
The NAb titers showed that the NHP had pre-existing
NAbs against AAV3b, AAV8, and AAV-NP59 before
apheresis. These NAb titers were reduced by the apheresis
in all cases. Although the anti-AAV3 NAbs were effectively
removed, antibodies against AAV8 and AAV-NP59
were not fully eliminated (Fig. 4b).
Seven days after vector administration we detected high
NAb titers against all six AAV variants (Fig. 4b). Previous
publications showed that even low NAb titers against
AAV8 capsids could have a strong neutralization effect

Human Gene Therapy - April 2023

Table of Contents for the Digital Edition of Human Gene Therapy - April 2023