Main text
A recent paper in Cell entitled “Soluble ACE2-mediated cell entry of SARS-CoV-2 via
interaction with proteins related to the renin-angiotensin system” by Yeung et al.
(2021) reported that the secretory form of soluble angiotensin converting enzyme 2
(ACE2), can foster SARS-CoV-2 infectivity (Yeung et al., 2021). ACE2 in its cell membrane-bound
form is the main receptor for SARS-CoV-2 cell entry. Yeung et al. (2021), however,
reported that, in a permissive human cell line (HK2-cells), exposure to very low concentrations
of soluble ACE2 promotes SARS-CoV-2 infectivity, whereas higher concentrations had
a neutralizing effect. The neutralizing effect of high concentrations of soluble ACE2
on SARS-CoV-2 infectivity indeed is confirmatory of previous reports in permissive
human cell lines and human kidney and vascular organoids (Monteil et al., 2020; Wysocki
et al., 2021). As a consequence, studies have taken advantage of this decoy action
of soluble ACE2 proteins to prevent and attenuate SARS-CoV-2 infectivity (Hassler
et al., 2022). If correct, the conclusion of Yeung et al. (2021) that low concentrations
of soluble ACE2 actually promote SARS-CoV-2 infectivity would have important implications
for the understanding of the role of soluble versus cell membrane-bound ACE2 as a
receptor essential for SARS-CoV-2 entry into the cell. Such conclusion could also
have a potential negative impact on therapeutic approaches based on administration
of soluble ACE2 proteins that exert a decoy effect and even for the design of vaccines
targeted to the receptor binding domain of the spike protein of SARS-CoV-2 that binds
to ACE2. We disagree with the conclusion of Yeung et al. (2021) that low soluble ACE2
concentrations increase SARS-CoV-2 infectivity for the theoretical reasons outlined
below and based on the data that we report here (Figure S1).
ACE2 exists in two forms: a full-length form that is cell membrane bound and a shorter
soluble form that is shed into body fluids and that normally circulates in the blood
in a small amount. Both forms contain the same sequence used by the receptor binding
domain of the SARS-CoV-2 spike protein, but soluble ACE2 lacks the transmembrane domain
necessary for anchoring in the cell membrane (Tipnis et al., 2000). The current understanding
of the SARS-CoV-2-ACE2 interaction is that, after binding of SARS-CoV-2 spike to the
membrane-bound full-length ACE2, activation via transmembrane serine protease 2 (TMPRSS2)
results in cell entry and viral replication. The full-length form of ACE2 is present
in epithelial cells in the upper respiratory tract and in type 2 pulmonary cells in
alveoli and also expressed in the kidney, heart, blood vessels, and intestine (Ziegler
et al., 2020). It is the interaction with full-length ACE2 in cells of the upper respiratory
tract that facilitates cell entry of SARS-CoV-2 particles (Ziegler et al., 2020).
The question we wished to address after the Yeung et al. (2021) paper is whether there
is any effect of low levels of soluble ACE2 that could promote SARS-CoV-2 infectivity.
ACE2 is a monocarboxypeptidase that cleaves angiotensin II to form angiotensin (1-7)
and several other peptides (Tipnis et al., 2000). Its physiologic role is to metabolize
these peptides and therefore prevent their accumulation, primarily at the local tissue
level. For instance, in experimental lung injury, angiotensin II levels increase and
the deficiency of ACE2 has a detrimental effect, while the administration of soluble
ACE2 protein is protective, respectively (Imai et al., 2005). Any loss of cell membrane-bound
full-length ACE2 as a result of SARS-CoV-2 infection would favor angiotensin II levels
to increase, which may cause inflammation and worsening of lung injury (Imai et al.,
2005). It should also be noted that there are reports that soluble ACE2 in plasma
is moderately increased in patients with cardiovascular disease and also in patients
with COVID-19 (Ramchand and Burrell, 2020; Kragstrup et al., 2021). The significance
of increased levels of soluble ACE2 in plasma in these conditions is not fully understood
but likely reflects shedding of membrane-bound ACE2 or cell-death-mediated ACE2 release
in these pathological conditions. The level of plasma-soluble ACE2 in humans and those
with cardiovascular disease or with COVID-19 could be roughly in the range of the
very low concentrations used in the study of Yeung et al. (2021). This could be erroneously
interpreted to signify that the virus could utilize soluble ACE2 for entry into host
cells. Binding of SARS-CoV-2 to soluble ACE2, however, should not foster infectivity
in vivo because soluble ACE2, unlike membrane-bound ACE2, lacks the transmembrane
domain essential for anchoring into the cell membrane (Tipnis et al., 2000). Therefore,
there is no sound rationale for the circulating form of soluble ACE2 to foster SARS-CoV-2
infectivity at any concentration because when a complex of soluble ACE2 and SARS-CoV-2
is formed it could not be internalized to initiate replication within the cell.
Notwithstanding the forgoing considerations, it is clearly important to validate or
refute the findings of Yeung et al. (2021) using low concentrations of soluble ACE2
protein. Their conclusions were derived from studies using a human kidney cell line
(HK-2) that is susceptible to SARS-CoV-2 infection. The data reported in Figure S3
of their paper showed that low concentrations of 0.1, 1, and 100 ng/mL soluble ACE2
increased SARS-CoV-2 RNA levels above control. We used the same HK-2 human cell line
in an effort to replicate the same protocol (details provided to us by the editors
of Cell, since they were lacking in the original paper [Yeung et al., 2021]). HK-2
cells were pre-treated with soluble ACE2 for 24 h (Figure S1A). This was followed
by infection of the HK-2 cells for 1 h with SARS-CoV-2 Wuhan strain isolated in Sweden
(GenBank: MT093571). Exposure to exactly the same low concentrations of soluble ACE2
and experimental conditions as in the Yeung et al. paper (Yeung et al., 2021) did
not result in any apparent change in infectivity assessed by qRT-PCR and expressed
as % of mock treated control (Figure S1A). Since there were more than two groups,
the data were analyzed by one-way ANOVA and, when significant, followed by post-hoc
Dunnett’s multiple comparisons test. This is the proper statistical analysis, and
we note that this was not done by Yeung et al. (2021) or previous work by Monteil
et al. (2020) who also used a paired t test. Yeung et al. (2021) used soluble ACE2
protein purchased from Sigma (Cat#SAE0064) which, like our protein (see Supplemental
methods), has amino acids 18–740 and dimerizes via its collectrin-like domain (Yan
et al., 2020). Thus, we could not replicate their findings using the same cell type,
protocol, and equally low concentrations of soluble ACE2 (Figure S1A).
Perhaps more relevant to human disease are our findings using human lung and kidney
organoids to assess the effect of soluble ACE2 concentrations in models that better
replicate organ physiology and pathophysiology. The experiments reported here in these
models were performed by different groups of investigators, independently of each
other. We used lung organoids of human embryonic origin that closely resemble the
target cells of SARS-CoV-2 entry into the body. These organoids express not only the
receptor ACE2 but also the protease TMPRSS2, needed for activation and subsequent
internalization (see Supplemental methods). In the lung organoids, we tested different
soluble ACE2 concentrations and assessed infectivity based not only on RNA levels
measured by qRT-PCR but also using plaque assay. None of the tested low soluble ACE2
concentrations, including 10 ng/mL as used by Yeung et al. (2021), had any significant
effect on SARS-CoV-2 infectivity as compared to PBS controls (Figures S1B and S1C).
By contrast, higher concentrations had the expected and previously reported strong
neutralizing effect (Monteil et al., 2020; Wysocki et al., 2021), which was also found
by Yeung et al. in HK-2 cells (Yeung et al., 2021).
In addition to these lung organoid experiments, we used different types of human kidney
organoids. One was derived from wild-type pluripotent stem cells and the others from
two different ACE2 knockout (KO) cell lines (Garreta et al., 2021). The kidney organoids
derived from the ACE2 KO cell lines are unique in that they do not express either
soluble or membrane-bound full-length ACE2. These ACE2 KO human kidney organoids,
therefore, cannot be infected by SARS-CoV-2, as shown by lack of any detectable nucleoprotein
(Garreta et al., 2021). Thus, this is an ideal genetic model system because there
is no endogenous ACE2 expressed that could bias the experimental data; hence the proposed
enhancement of SARS-CoV-2 by low-dose soluble ACE2 proteins would require the presence
of a receptor(s) other than ACE2 as proposed by Yeung et al. (2021). When ACE2 KO
human kidney organoids were exposed to SARS-CoV-2, however, we again did not detect
enhanced infectivity with the low concentrations of soluble ACE2 (including 10 and
100 ng/mL, Figures S1D and S1E). A higher concentration of soluble ACE2 also had no
effect on infectivity in these ACE2 KO human kidney organoids, whereas as expected,
it inhibited infection in orthologous wild-type kidney organoids (Figures S1D–S1F).
In wild-type kidney organoids, low concentrations of soluble ACE2 had no effect on
infectivity, whereas high concentrations had the expected suppressive effect (Figure S1F).
In the aggregate, our findings provide strong evidence for an essential requirement
of full-length membrane-bound ACE2 for SARS-CoV-2 infectivity and also show a lack
of effect of soluble ACE2 in the absence or presence of the membrane-bound full-length
ACE2 receptor.
In summary, the findings we report here show that soluble ACE2 when applied at low
concentrations does not promote a significant increase of SARS-CoV-2 infectivity in
any of the cell models studied. Importantly, in the absence of membrane-bound full-length
ACE2, kidney organoids cannot be infected with SARS-CoV-2. A faithful model of severe
SARS-CoV-2 infection in mice using ACE2 mutant animals, moreover, supports the evidence
for the essentiality of ACE2 as the SARS-CoV-2 receptor here reported (Gawish et al.,
2022). Based on our findings, including the use of the kidney organoid model of ACE2
deficiency to reduce experimental bias, we conclude that full-length membrane-bound
ACE2 is the essential determinant for infectivity by SARS-CoV-2 and that the low levels
of soluble ACE2 protein, that may correspond roughly to those present in vivo, cannot
promote infectivity.