Human Gene Therapy - April 2023 - 279

COMPARING AAV FUNCTIONAL TRANSDUCTION IN LIVER MODELS
279
to be most effective at entering cells (29% ofNGS reads at
the DNA level) (Fig. 1b). However, the efficient entry of
AAV5 did not lead to efficient transgene expression,
where only 8.9% of the NGS reads from RNA/cDNA
could be attributed to this variant. Instead, AAV-LK03
(38.8%) and AAV3b (34%) were most efficient at transgene
expression in the organoids (Fig. 1c), despite less
efficient entry (20.3% and 19.9% DNA NSG reads for
AAV-LK03 and AAV3b, respectively). The observed
decrease in the performance of AAV2 might indicate reduced
importance of binding to heparan sulfate proteoglycan
(HSPG) as a critical step for cell entry.43
To understand how the previous findings translate to
primary cells ex vivo, human and rhesus monkey (Macaca
mulatta) hepatocytes were seeded in 2D culture as well as
3D-printed in a hydrogel.51 The 2D cultured hepatocytes
from human and NHP were exposed to three different
AAV doses (1,000, 500, and 200 vg/cell) and harvested
3 days after exposure to AAV. The hepatocytes of human
origin were most efficiently entered by AAV-LK03 and
AAV3b, followed by AAV2, AAV-NP59, AAV5, and
AAV8 (Fig. 1b). In contrast, rhesus monkey primary hepatocytes
were most efficiently entered by AAV2, followed
by AAV5, AAV-LK03, AAV-NP59, AAV3b, and
AAV8 (Fig. 1b), showing a marked difference compared
with transduction observed in the human hepatocytes.
However, at the level of transgene expression, AAVLK03
was the most efficient variant in both species. In
human hepatocytes, it was followed by AAV3b, AAVNP59,
AAV2, AAV8, and AAV5, whereas in simian hepatocytes
it was followed by AAV2, AAV3b, AAV-NP59,
AAV5, and AAV8 (Fig. 1c).
As the next step, we wanted to transduce the same cells,
but 3D printed in a hydrogel substrate.38 However, before
doing so, we evaluate the potential effect of the 3D culturing
system on primary hepatocytes by performing
RNAseq using cells before and after 3D culture (Supplementary
Fig. S3). The results showed that although expression
of most genes did not change substantially,
several genes underwent upregulation and downregulation
after 3 days of the 3D culture (Supplementary Fig. S3b and
Supplementary Table S3). For the human hepatocytes, the
most upregulated gene was a long noncoding RNA
(PCAT1) with the proposed function of regulating genes
implicated in increased cell proliferation, migration, and
invasion.52,53 Owing to a lack of comprehensive annotation
coverage of the rhesus monkey genome, the RNAseq
data recovered for this species was less informative than
the human data.
Hepatocytes in 3D hydrogel cultures were exposed to a
dose of 500 vg/cell and cells were harvested and processed
for analysis 3 days later. Results obtained showed number
of differences compared with results obtained with conventional
2D cultures. Consistent with the 2D cultures,
AAV-LK03 entered 3D-printed human hepatocytes with
the highest efficiency among the vectors tested but was
more closely followed by AAV2 and AAV3b. AAVNP59,
AAV5, and AAV8 performing substantially less
efficiently. In the 3D-printed simian hepatocytes, AAV2
was found to be the most effective variant at cell entry,
followed by AAV5, AAV3b, and AAV-LK03, with AAVNP59
and AAV8 being the weakest performers (Fig. 1b).
At the transgene expression level (RNA), AAV2 gained in
function in both human and rhesus cells compared with 2D
cultures.
The data from 2D versus 3D culture systems as well as
human versus NHP hepatocytes led to interesting observations.
Although AAV2 worked in 2D and 3D cultures of
both species, there was a substantial increase in AAV2
performance in the 3D culture system over the 2D culture.
This observation could be explained with strong reliance
on HSPGs in the 3D-printed hepatocytes, a question we
explored further in subsequent studies. Furthermore, the
data showed that AAV5 had a markedly higher performance
in NHP-cultured hepatocytes than in human hepatocytes,
which might explain why clinical trials using
AAV5 need very high vector doses and have a lower efficacy
than NHP data would suggest.24,25
Another upregulated gene that is potentially relevant
for AAV biology was glypican proteoglycan 6, a cell
surface protein known to harbor HSPGs.54 This could
explain the increased transduction by AAV2 and decreased
transduction by AAV-NP59, which has lower affinity
to HSPG (Fig. 1),31 in the 3D printed human
hepatocytes compared with the conventionally cultured
human hepatocytes.
In summary, in the simplest models of human hepatocytes,
the immortalized cancer cell lines, AAV2 performed
substantially better than all other variants tested.
This performance decreased in stem cell-derived models
and primary cells, where the performance of the bioengineered
variant AAV-LK03 improved in both human and
simian cells. The bioengineered AAV-NP59 also seemed
to have improved in primary cells of human origin compared
with its performance in immortalized cells and stem
cell-derived models. One of the most interesting observations
was the relatively variable performance of AAV5
across the models tested. The performance was low in
HuH-7 cells and primary human hepatocytes, whereas a
relatively high performance was observed in HepG2 cells,
stem cell-derived iHeps, and ductal organoids, as well as
primary NHP hepatocytes. This might indicate that AAV5
(the most distantly related AAV capsid of the ones chosen
for this study) might utilize distinct cell entry and transduction
mechanism.
Xenograft in vivo models of human
and NHPs livers
Having studied the six vectors in several in vitro and ex
vivo models, we next wanted to evaluate one of the com
Human Gene Therapy - April 2023

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