Author response:
Response to Comments from reviewer #1
Many thanks for appreciating that SZN-043 can promote hepatocyte proliferation via the Wnt-signaling pathway.
(1) The reviewer is concerned with using only CYP1A2 expression as an endpoint to make a conclusion about the effect of SZN-043 on Wnt activity in human ALD samples. The reviewer raises a good point as the more commonly used Wnt target gene, AXIN2, is not consistantly changed in both cohorts. We were at first also surprised by this finding. However, upon closer analysis we found that the expression of hepatocyte-specific target genes such as CYP1A2 (Figure 2), CYP2E1, OAT, LGR5, GLUL (Table 1) and ZNRF3 were mostly expressed in hepatocytes and ductal cells were all down-regulated in ALD samples. Others Wnt target genes expressed in epithelial and mesenchymal liver cell populations, such as AXIN2, CCND1 and NOTUM are indeed not consistently and significantly changed. Given that SZN-043 is not active on mesenchymal cells, this discrepancy could be best explained by the large increase in mesenchymal cells in ALD tissue samples, thereby confounding the results. We have now clarified this in the discussion. Another method to assess Wnt activity is to measure b-catenin phosphorylation and nuclear transfer. In our hands, this method was found to be better suited for tissue culture than histological sections from in vivo studies. We have also amended the manuscript title to refer to expression of Wnt target genes, rather than Wnt activity.
(2) We have now added a supplemental figure to show the lack of Ki-67+ human hepatocytes in the cirrhotic tissue samples to confirm the absence of hepatocyte proliferation (Figure S1).
(3) The differences in amino acid sequence between SZN-043 and its precursor, αASGR1-RSPO2-RAIgG, can be found in the material and method section. These changes in amino acid sequences improved the biophysical properties of the final clinical candidate, such as oxidation and nonspecific binding. The biochemical analysis of those differences exceeds the scope of the current manuscript. We present here the pharmacokinetic properties of SZN-043 only, as this was the only molecule advanced to clinical trial and used in the studies presented here.
(4) The reviewer suggests to assess the effect of SZN-043 in Ctnnb1-KO mice to confirm that SZN043 acts via a canonical Wnt pathway. Indeed, there were several reports on the ability of Rspondin to act on other pathways besides the Wnt signaling pathway (for recent review, Niehrs et al, 2024, Bioessays). However, while an interesting suggestion, this line of investigation belongs to MOA studies and exceeds the scope of the current manuscript. An additional manuscript presenting MOA studies for SZN-043 was recently submitted elsewhere. Still, we have added this possibility in the discussion section.
(5) The reviewer is asking how SZN-043 is affecting liver functions in general. Indeed, we have observed a consistent reduction in the international normalized ratio of prothrombin time using the thioacetamide (TAA)-induced fibrosis model and previously published those findings (Zhang, 2020). In our hands, the TAA is the only liver injury model that significantly increases INR. This increase is modest compared to that observed in clinical patients. Therefore, we do not report INR findings for other models. We have not seen any effects of SZN-043 on hepatocyte differentiation markers such as HNF4A (data not shown) and the hepatocyte specific ASGR1/2 as shown in Figure 5. Rather we focused on proliferation as the main potentially beneficial endpoint, to restore the parenchymal mass in injured livers. Finally, consistent with what was reported in the literature, we have observed a transient and reciprocal effect on albumin and alfa-fetoprotein expression during the proliferative phase of liver regeneration. These results are detailed in an additional manuscript presenting MOA studies for SZN-043, which was recently submitted elsewhere.
(6) We have used females only in the ethanol-induced injury models because there are numerous reports in the literature stating that males are not as susceptible to those injuries.
(7) The reviewer questions the relevance of the ethanol-induced injury model used to evaluate SZN043 efficacy. Indeed, none of the disease model developed to date reproduce the severity and complexity of alcohol-associated liver diseases, although some, such as the ethanol supplemented Lieber DeCarli diet, are more commonly used than others – which is the reason why this model was selected.
(8) The reviewer questions the relevance of the fibrosis model used to evaluate SZN-043 efficacy. Indeed, none of the fibrosis models developed to date reproduce the severity and complexity of cirrhosis in human livers. While combining ethanol with CCl4 would lead to more severe fibrotic livers, CCl4 itself is not involved in ALD in humans. Both models are likely to result in similar pericentral fibrosis with central-to-central bridging. In this study, we were mostly interested in addressing the effects of SZN-043 in a tissue affected by fibrotic scars.
(9) The sex of CCl4-treated mice is male. We added this information in the methods section.
(10) A summary of histology and fibrosis assessment data for alcohol-fed mice was added in supplemental Table S3. In our hands, the use of aging mice did not induce the presence of fibrosis, in contrast to published results.
(11) The rationale for using 13.5-month-old mice in the alcohol studies and scid mice in the CCl4 studies has been clarified in the results and discussion sections.
a. Briefly, aging mice were reported to be more susceptible to ethanol-induced injury than young mice and to include induction of fibrosis. However, we were unable to reproduce the presence of fibrosis reported in the literature.
b. Scid mice were used in the CCl4 studies to test whether a stronger response could be observed in the absence of a potential anti-drug antibodies response. While a modest reduction in fibrosis was observed in both B6 and scid mice following the SZN-043 treatment, the effect size did not seem affected by the mouse strain.
Response to Comments from reviewer #2
Many thanks for appreciating that the use of multiple disease models to identify SZN-043 as a potential novel drug for liver regeneration.
(1) The importance of restoring liver regeneration capacity to reduce the need for liver transplantation had been emphasized in the introduction.
(2) There is continuous damage to the mouse hepatocytes in the FRG mice, due to the Fah mutation. They undergo repair mechanisms favoring the proliferation of human hepatocytes during the production period. Injury models that affect the human hepatocytes population have been developed in these mice. However, the primary goal of this study was to confirm that SZN043 was efficacious in inducing human hepatocytes proliferation, a feature difficult to reproduce in primary hepatocyte cultures. Given the artefactual nature of the chimeric liver in FRG mice and the high cost of these mice, further studies were not judged to be necessary.
(3) Corrected
(4) A figure including DAPI staining has now been included in supplemental Figure S2.
(5) Clarification that the 8 weeks alcohol feeding used in our study design is a modification of the NIAAA model. While some ASGR1 has been reported on the surface of macrophages, additional data from MOA studies strongly suggest that the effect of SZN-043 is mediated via a hepatocytespecific mechanism (submitted manuscript).
(6) The reviewer inquired about the potential role of macrophages in promoting an antiinflammatory state in response to SZN-043. While a direct effect is unlikely, a potential effect of macrophages in response to SZN-043 is plausible. Wnt activation is known to induce the secretion of hepatokines, such as LECT2, which in turn can influence macrophage activity. This possibility is discussed in the discussion section.
(7) The potential off-target effects of SZN-043 such as stellate cell activation is discussed in the discussion section.
(8) The discussion of the limitations of current models has been included in the discussion section of the manuscript.
(9) We have now included a discussion of prior RSPO-based therapies, such as OMP-131R10. We explain why the hepatocyte-targeting of RSPO activity minimizes undesired effects.
