Reviewer comments

SECTION 1 - AOP identifier/TitlePR: Strictly speaking the AOP does not conform to the naming convention in that the title of the MIE (Event 1391) does not include «chronic » in the title. Generally speaking, if chronic activation of the MIE is required to cause this particular AO, it would be preferable to specify that in the KER, as the event of Cyp2E1 activation could generalize to other pathways. Unless the difference between chronic and acute can be precisely defined, I think it is best that « chronic » is not included in the MIE title. That said, the guidance does not explicitly preclude including « chronic » in the title and it does provide some important and immediate information to the user. Consequently, while the authors might consider dropping « chronic » from the title, I would view that as optional and at their discretion.

I think the title adequately describes the AOPs content and domain.

We wanted to emphasize that the AO is not expected to occur following an acute exposure. Acute exposures will lead to ROS production and even, perhaps, hepatocytotoxicity (e.g., acetaminophen), but chronic exposures are needed to cause regenerative proliferation. We wanted to be clear about this in the title. Does the PR have advice?

Overall, we are concerned with this matter and would like expert advice on how to proceed. We need to be clear somehow that it is only chronic activation of Cyp2E1 that would lead to liver cancer. Having said that, we have not quantified ‘how much’ activation is necessary, nor have we defined how long ‘chronic’ is. We are not able to do this given the existing empirical data.

Overall, we’d prefer to retain in the title at the very minimum and will need to ensure that this is captured in the overall description of the AOP.

SR1: - --SR2: Not quite and I think this AOP highlights two issues for the EAGMST to consider in its guidance.

(1) Firstly, the guidance doesn’t mention how to handle a requirement for time-dependence in an AOP title – and in my view AOPs don’t handle time very well in general. So although the use of the word “chronic” in the title isnt excluded by the Guidance, I think it is implicit that temporal requirements are handled in the temporal concordance details of the AOP (ie in the KERs) rather than in its title. This is debateable. But if that was

(1) See response to ‘PR’ above.

(2) Yes, “activation” = “substrate-dependent stabilization”. We have

the intent of the guidance perhaps “chronic” should be removed; however, if the sustained nature of the MIE is essential to this AOP argueably it should remain.

(2) Secondly, when MIEs have used activation or inhibition of a protein in the past I think they meant that the intrinsic functional activity of the protein is changed due to the effects of some stressor. In this case, I think the authors intent is not quite this so perhaps the use of activation isnt appropriate? Are you saying that:

- an increase in the catalytic activity of Cyp2E1 is the MIE (perhaps due to substrate-dependent stabilisation of the enzyme), or

- that the sustained production of reactive metabolites due to a chemical reaction catalysed by Cyp2E1 is the MIE (that is chronic substrate metabolic acitvation ie the MIE), or is it both of these?

Perhaps it would be better described as (chronic) Cyp2E1-catalysed production of reactive oxygen and electrophilic metabolites leading to liver cancer.

added further explanatory text that we hope clarifies this issue.

See AOP Section 1.3 How the MIE Works for the explanation (suggested modification to the text is in red):

“Cyp2E1 is a membrane-bound monooxygenase that is primarily located in zone 3 hepatocytes (Ingelman-Sundberg, et al. 1988, Tsutsumi, et al. 1989). CYP2E1 is primarily located in the endoplasmic reticulum, but can also be present in the mitochondria. It is a phase I metabolism enzyme that catalyzes the oxidation of low molecular weight substrates. Unlike most cytochrome P450 enzymes, Cyp2E1 is constitutively expressed (i.e., its expression is not transcriptionally controlled by substrate-bound nuclear receptors). Alternatively, exposure to a substrate increases its activity through post-translational stabilization of the molecule. Thus, the presence of substrate significantly increases the half-life of the Cyp2E1 enzyme thereby allowing it to be active for a longer period of time (Gonzalez 2007, Song, et al. 1989).”

The explanation for why Cyp2E1 stabilization is a bad thing is in the KERs (KER1 and iKER1) as per the guidance (page 29; https://one.oecd.org/document/ENV/JM/MONO(2016)12/en/pdf ).

We believe the above is the best description of the MIE. Note that ‘Cyp2E1-catalyzed production of reactive oxygen’ does not describe how the stressor interacts with the protein, which is the necessary information for the MIE.

Section 2: AuthorsPR: Yes. The authors, point of contact, and contributors (to the ---

Wiki) are clearly defined.

SR1: - ---

SR2: yes ---

Section 3: Date of updatingPR: It is clear when the snapshot was created (11/17/2017). The history logs indicate the content was last upated around June, 2017.

---

SR1: - ---

SR2: 2017-11-17 10:02 ---

SECTION 4 : AbstractPR: Yes. The abstract includes a concise summary and touches on all the recommended information suggested in the handbook.

---

SR1: Graphical representations (p22): It is not explained, why some lines are dotted. Dot they reflect indirect relationships?

Yes, the dotted lines at the indirect (now called non-adjacent in new Users’ Handbook) relationships. As per the guidance, page 35: “By convention (and for clarity), KERs linking adjacent KEs in an AOP are represented using solid arrows, while KERs that link KEs that are not adjacent to one another in sequence are linked via dashed arrows” (https://one.oecd.org/document/ENV/JM/MONO(2016)12/en/pdf)

Note: We didn’t create this picture (the website auto-generates this).

SR2: there are some differences. In the abstract the authors mention Cyp2e1-dependent generation of reactive metabolites and decoupling produced reactive oxygen species, however in the detailed MIE discription they also mention substrate-dependent stabilisation of cyp2e1. Is this the “activation” of cyp2e1 catalytic activity they mention.

Please see “SR2” in Section 1—above for response.

We that the abstract is consistent with the above. The description of “…Cyp2e1-dependent generation of reactive metabolites and decoupling produced reactive oxygen species” refers to the KERs, not the MIE.

As per the guidance, page 24, the KERs are not explicitly included in the AOP abstract:

“…typically be 200-400 words in length (similar to an abstract for a journal article). Suggested content for the abstract includes the following:(1) the background/purpose for initiation of the AOP’s development (if there was a specific intent);(2) a brief description of the MIE, AO, and/or major KEs that define the pathway;(3) a short summation of the overall WoE supporting the AOP and identification of major knowledge gaps (if any);(4) if a brief statement about how the AOP may be applied (optional).”

(https://one.oecd.org/document/ENV/JM/MONO(2016)12/en/pdf)

The clarifying changes below (in red) would not be consistent with the guidance. Further, it would result in inconsistencies in the rest of the abstract because the rest of the KERs are not mentioned.

Briefly, the MIE occurs when Cyp2E1 binds a substrate. The Cyp2E1 catalytic cycle is prone to decoupling (KER1), which produces oxidative stress (KE1), and mono-oxidation of substrates produces reactive metabolites. Both reactive oxygen species and metabolites cause cytotoxicity (KE2).

SECTION 5 : Molecular Initiating Event

PR: Yes, a MIE is described.

1. The MIE description does not really define what constitutes “activation” of cyp2E1. Does activation indicate an increase in enzyme activity, and increase in cyp2E1 protein abundance, or an increase in anticipated monooxygenated metabolites (and lack-thereof in a knock-out) which could occur without an increase in abundance or activity. Any of these could be inferred based on the different methods proposed for measuring or detecting this MIE. Consider being more specific. Each of the methods described provide different types of information that can be relevant for either measuring the KE, establishing the essentiality of this KE relative to downstream events, establishing a particular stressor as a substrate for CYP2E1, etc., but I view those as different things, not necessarily measures of the same biological change. Consider making the description precise and only including the methods that specifically address what is meant by “activation”.

2. Restricting the applicability domain by specifying “in liver” in the title of the MIE may restrict use, but since primary expression and its role in xenobiotic metabolism seems predominantly localized in liver, this may not be a big deal for this particular event.

(1) Please see “SR2” in Section 1—above for response.

- Basically, Cyp2E1 is constitutively expressed and has a certain half life. But, when there is substrate around, the enzyme becomes stabilized, which increases the half life. The impact of an increased half life is that it will be in the cell for a longer period of time, which means it will be catalytically active for a longer period of time, which means that more metabolites and ROS will be produced. The mechanism of action is explained in detail in the two references and in KER1 and what is currently called iKER1.

- We agree that they all measure different things, but we disagree that that this is a negative attribute. We think that there is variety in both the research questions asked, and the resource availability in labs. Therefore, each of these assays has a place/context for determining if a substance interacts with Cyp2E1. However, if the reviewers would prefer them out, we are happy to delete.

(2). Agree. We will change the name of the MIE to remove ‘in the liver’.

The AOP is focused on the liver, thus we had narrowed to this originally. The research provided has been done in vivo in the liver and in vitro in liver cell cultures. As nearly all of the empirical evidence for this AOP is in the liver, we were uncomfortable extending to other tissues.

We have now noted that Cyp2E1 is expressed in other tissues (e.g., http://www.genecards.org/cgi-bin/carddisp.pl?gene=CYP2E1)

However, whether its chronic activation leads to the next event in

3. Referencing of the methods cited is good.

4. Support for applicability to rodents and primates is provided. Tissue context is defined. Life stage and sex applicability is not defined in either structured or free-text fields. Consider providing information on potential life stage and sex applicability and completing the structured fields as much as feasible.

5. A number of prototypical chemicals are identified in the free text. It would be advantageous if they could also be identified in the structured “stressors” field – at least for the four chemicals upon which most of the assembled evidence are based. Also, in reference to the CYP2E1 knock-out mouse description the authors identify a list of chemicals tested, but don’t say whether or not they were positive in terms of their interaction with CYP2E1.

tissues other than liver is not documented.

(3) --

(4) We have included the information that we have available as suggested (focus on CYP2E1 has been on adult hepatocytes). More taxonomic applicability has been added.

(5) The field ‘stressors’ was not there when we originally developed this AOP. We have now added stressors to this section. We have indicated that these chemicals are positive in their interaction as suggested by the reviewer.

SR1:

1. A closer look at the description reveals that two MIEs are described. 1) Formation of reactive metabolites by CYP2E and 2) Generation of ROS by CYP2E uncoupling.

2. In the description of these MIEs it could be pointed out that:

- CYP2E is expressed particularly in the liver, and- that activation is required or has an impact on the

quantitative manifestation of the MIE.

1. Please see responses above. The MIE is substrate dependent stabilization of the Cyp2E1 enzyme (as described in the Section 1.3 ‘How this MIE works’). Yes, the uncoupling does lead to ROS, which is why oxidative stress is the second KE. The inclusion of a potential KE ‘ROS levels, increase’ is something we discuss below in the KER section where it is relevant.

2. Done. Please see responses above. We have clarified all of this.

3. It should be highlighted what activation means, i.e. stabilization of CYP2E leading to a higher level of protein abundance.

4. What does not become clear from the description is whether the stabilization is really essential or whether constitutive levels of CYP2E would also lead to generation of reactive metabolites and ROS, but potentially at a lower level.

3. Done. Please see “SR2” in Section 1—above for response.

4. Yes, constitutive levels of CYP2E would also lead to generation of reactive metabolites and ROS, but at a lower level. Please note that we feel that this mechanism is well described in KER1.

(1) Because Cyp2E1 metabolizes O2, it basically always has substrate. Ref.: Mechanism of oxidation reactions catalyzed by cytochrome p450 enzymes. Chem Rev. 2004 Sep;104(9):3947-80. https://www.ncbi.nlm.nih.gov/pubmed/15352783

(2) Cyp2E1 is capable of ‘futile cycling’, which is substrate independent. Ref.: Rat liver microsomal NADPH-supported oxidase activity and lipid peroxidation dependent on ethanol-inducible cytochrome P-450 (P-450IIE1). Biochem Pharmacol. 1989 Apr 15;38(8):1313-9. https://www.ncbi.nlm.nih.gov/pubmed/2495801

(3) Cells and animals that over-express Cyp2E1 do have elevated oxidative stress and increased expression of anti-oxidant genes (the opposite is true for knock-downs).

E.g., CYP2E1-dependent toxicity and oxidative stress in HepG2 cells. Free Radic Biol Med. 2001 Dec 15;31(12):1539-43. https://www.ncbi.nlm.nih.gov/pubmed/11744327

Cytochrome P450 2E1-dependent oxidant stress and upregulation of anti-oxidant defense in liver cells. J Gastroenterol Hepatol. 2006

Oct;21 Suppl 3:S22-5. https://www.ncbi.nlm.nih.gov/pubmed/16958665

Expression of CYP2E1 increases oxidative stress and induces apoptosis of cardiomyocytes in transgenic mice. FEBS J. 2011 May;278(9):1484-92. https://www.ncbi.nlm.nih.gov/pubmed/21352494

Note that (1) the ‘lower level’ would not produce adverse outcomes as they are physiologically normal; and (2) Cyp2E1 is only catalytically active when it is bound to substrate, and thus only producing high levels of ROS when substrate is available.

SR2: Is a MIE described? YES If yes, then:

1. Is the MIE description clear and is it biologically plausible? yes

2. Is the MIE described in a way that allows its use in other AOPs? yes

3. Are measurement/prediction methods specified and adequately described/referenced? Somewhat superficial in detail but adequately referenced.

4. Is the biological context (inc. taxonomic applicability/relevance, level of biological organisation) specified and explained sufficiently? Not really, they only include information on presence in primates and rodents. It isnt clear what wider taxa orthologous genes exist.

Have chemical initiators (prototypical chemicals or chemical features) been identified? A few examples are named and

1. ---

2. ---

3. ---

4. We have now indicated other taxa as well (frogs and fish, with references).

many more referenced in papers.

SECTION 6 : Key EventsPR:

General comments on KEs:

1. The measurement methods are adequately described/referenced.

2. The taxonomic applicability is explained in free text, but sex and life stage applicability is not addressed. Consider adding those in both the free text and structured fields if possible (optional).

KE-specific comments:

3. Event 1392: The AOP-Wiki already contains at least six other KEs pertaining to oxidative stress. While most appear to lack content, Event 1088 (Increased, Oxidative Stress) is already linked to 5 other AOPs. Consider adding your content from Event 1392 to Event 1088 and linking to that KE instead in order to increase overall connectivity in the AOP-Wiki.

4. Event 1393: Although the title is specific to liver as are some of the in vivo methods cited, neither the event description nor the other methods cited are specific to cytotoxicity in liver. Consider whether specification as hepatocytotoxicty is really necessary, or whether this can be described as a more generalizable cytotoxicity key event with the specificity of the consequences added via the KER descriptions through which this event is linked to others. There is currently at least one

(1) ---(2) ---

(3) Event 1392. Given that KE 1088 is blank and ours is written, we suggest that they add their content to ours. Alternatively, the AOP curators could remove event 1088 and connect ours to the 5 other AOPs (might speed things along a bit?).

(3) Event 1393. We can see PR’s point and have considered it and discussed it extensively with colleagues. We are happy to discuss further, but the outcomes of our consultations have led us to conclude that cytotoxicity is perhaps a bit too broad. We want to focus on hepatic cytoxoxicity, which we feel is also very broad; we believe that a KE about hepatic cytotoxicity will be widely re-used KE, given that the liver is a primary target of chemicals and drugs. We also want to fully capture the hepatic cytotoxicity measurements herein

general cytotoxicity KE (Event 768, currently linked to 2 AOPs) and one hepatocyte specific KE (Event 786; currently linked to 1 AOP) present in the wiki. Consider whether your content could be added to either of those existing pages, rather than a new page in order to increase connectivity within the AOP-Wiki.

5. Event 1394: Ok. The description is specific to liver. There are a whole series of key events in the wiki related to regenerative cell proliferation, at least one of which (Event 787; linked to one AOP) is specific to hepatocytes. Consider whether the content from Event 1394 could be added to Event 787 and the redundant event eliminated from the AOP-Wiki.

(e.g., biomarkers that we have listed – AST, ALT, mir-122 that are specific to liver – listed in KE methods) but do not want to provide extensive text and methods for all of the other biomarkers of organ-specific cytotoxicity. Overall, we think that re-usability by focusing on hepatocytotoxicity is more effective. Also note that KE 786 indicated by the PR is also entirely blank.

(5) Event 1394. As above, given that KE 787 is blank, we suggest that they add their content to ours.

SR1: KEs:

1. Both the formation of ROS and KEs trigger various parallel KE. All of them may finally lead to liver cancer, but probably the joint activation is increasing the likelihood that cancer is developed. However, in the AOP only “Oxidative stress” is mentioned as a KE. I think in compatibility to other AOPs that describe development of cancer or cytotoxicity it might be a good idea to replace oxidative stress by different AOPs or to define oxidative stress more precisely. What is actually described as oxidative stress KE may be replaced by three independent or parallel KEs: lipid peroxidation; glutathione peroxidation; DNA mutation.

(1) In general, we disagree with the suggestion to split the KE—we feel that doing this may needlessly complicate, which is advised against in the Users’ Handbook supplement.

In general, what we have described are biomarkers of increased levels of ROS in the cells (e.g., through measurement of lipid peroxidation and glutathione depletion). All of these things are happening at the same time and are hard to parse apart. Since we use them as biomarkers, we feel appropriate as written. Finally, please note that this AOP was in development from 2016-2017. At this time, branching of AOPs was discouraged and we thus simplified as much as possible.

If pressed, we could envision an AOP that leads from Cyp2e1 activation -> increased ROS levels (measured by GSH depletion) ->

2. Lipid peroxidation and glutathione peroxidation would lead to the cytotoxicity KE, subsequently the proliferation KE and finally the AO liver cancer. DNA mutation could lead directly to the AO liver cancer but the parallel increase in proliferation would greatly enhance the proliferation.

Role of signaling pathways:3. The role of signaling pathways and pathway-specific

upregulation of transcription factors may outreach the role of only a biomarker associated with a certain KE. It may indeed represent a KE itself. For instance it is stated that “…NRF2 is required for hepatocellular carcinoma precursor cells (HcPC) to progress to become hepatocellular carcinoma cells; in fact, without NRF2 excess ROS accumulate and kill the HcPC cells.” This rather resembles an essential KE than a biomarker. The same applies for NF kappa signaling pathway. However, it potentially requires more data to decide whether these event represent a real KE, but it should be at least indicated in the description of the KE that there may be more KEs or that one KE could reflect a chain of KEs.

lipid peroxidation -> hepatocytotoxicity. We can explore this option if the internal reviewers deem this to be essential, but this will mean the development of multiple KERs and thus we’d need additional support and much longer to revise the document. Note that increased ROS levels are not generally measured (they are inferred by biomarkers such as GSH depletion), so we’re not sure this would be very easy to do (lack of empirical support would lead to some very weak KER). We also believe that these additional KEs/branches could be the subject of future work by ‘the crowd’ if it turns out that they are needed.

We agree that ROS damage to DNA may also lead to mutations and perhaps cancer, which would be occurring in parallel with this AOP. However, we feel that this is outside the scope of our work. Our hope is to network into that AOP once it is available. Please note the Health and Environmental Sciences Institute’s Genetic Toxicology Technical Committee (GTTC) currently has an AOP development working group that plans to develop two AOPs in this area: (1) ROS damage to DNA leading to double strand breaks (chromosome aberrations and micronuclei); and (2) ROS damage to DNA leading to mutations. SR1 has suggested that perhaps we could define oxidative stress more precisely. We will explore this option.

Overall, we are not in support of having a ROS-> DNA damage->mutation branch of the AOP as it is beyond the scope of our project and could be a separate linear AOP that networks with ours in future.

We agree that there is more than one way for ROS to cause liver cancer (or other AOs in the AOP Wiki). This AOP is meant to capture and describe a single route to the AO, as per the User

Species specificity (page 3):4. It is indicated that CYP2E1 is present at least in

mammals. What about other mammals, such as fish. Could it be briefly outlined if there is any evidence that CYP2E1 could activate chemicals or lead to ROS also in non-mammals (or if a different CYP is providing this function)? On page 7 it is mentioned that zebrafish shows liver regeneration as well. Hence, at least partially the AOP is conserved across species/vertebrates.

5. Page 23: “…example, acetaminophen is a Cyp2E1 substrate that does not cause cancer.” This indicates that CYP2E1 activation per se may not lead to liver tumors. It is required that a chemical is metabolized by the cytochrome P450. This could be reflected by changing the MIE name(s) appropriately. See figure below in the section on AO.

Handbook instructions. The AOP networks will help to visualize the additional ‘KE chains’.

(2) Agree. But please see comment above.

(3) Agree. These relationships are already well described in:

- NRF2: Primarily in iKER2; but also in KER1 and KE1.

- NF kappa B: Primarily in iKER3; but also in KER3 and KE3.

(4) Agree. We have added fish and frogs with appropriate references.

(5) Agree. All of these events occur in a specific context. It is accepted that the activation of the MIE is essential, but may not be sufficient to cause the AO.>Changing the MIE:

Disagree. The definition of an MIE is the point at which the substance interacts with the biological system. In the case of this AOP, the chemical substance directly interacts with Cyp2E1. So, this is the only possible MIE for this AOP.

AOP Guidance: https://one.oecd.org/document/ENV/JM/MONO(2016)12/en/pdf

SR2:

Key event 1: oxidative stress

Are the KE descriptions clear on how the events work and are (1) We see SR2’s point, but please note that Bax, ASK1, MAPKs

they biologically plausible? Yes. (reuse of an existing KE)

Are the KEs described in a way that allows their reuse in other AOPs?

Yes

1. Are measurement methods specified and adequately described/referenced?

A bit superficial and only based on NRF2. In this case they might have considered adding to the KE “how it is measured” section the regulation of thioredoxin-ASK1/ Jnk/p38 pathway as an additional option for the CYP2e1 mediated mechanism (see Wu D, Cederbaum A. Activation of ASK-1 and downstream MAP kinases in cytochrome P4502E1 potentiated tumor necrosis factor alpha liver injury. Free Radic Biol Med. 2010 Aug 1;49(3):348-60)?

2. Is the biological context (inc. taxonomic applicability/relevance, level of biological organisation) specified and explained sufficiently?

Yes

Key event 2: hepatocytotoxicity

3. Are the KE descriptions clear on how the events work and are they biologically plausible? Yes

and TNF are discussed in “KE3: ‘hepatocytotoxicity” because they occur downstream of oxidative stress, which is KE2 (as mentioned in the article title). We feel that they are more suitable in that section as they are mediators of cell death.

(2) --

(3) –

(4) We see SR2’s point, but in keeping with the conventions of AOP

4. Are the KEs described in a way that allows their reuse in other AOPs?

No, this Key event is actually a group that describes 3 separate processes and each is qualitatively different in terms of how it is detected/diagnosed by pathologists and what causes it. Used as described here, it prevents people specifically proposing only one of these as a key event in another AOP. I would recommend these be split out.

5. Are measurement methods specified and adequately described/referenced?

Partly. A bit superficial in the detail and lacks any details on how the ASK1 pathway would be measured. The authors make the comment that H&E sections examined for the presence of cytotoxicity, which is true by heptocyte toxicities are typically described as apoptosis or degeneration – the term cytotoxicity is not used. Therefore this could be confusing.

6. Is the biological context (inc. taxonomic applicability/relevance, level of biological organisation) specified and explained sufficiently?yes

Key event 3: regenerative proliferation

7. Are the KE descriptions clear on how the events work and are they biologically plausible? AND Are the KEs described in a way that allows their reuse in other AOPs?

development, which indicate that not every single mechanism should be described in detail, we have used a high-level approach herein. We suggest that we retain as written as these events generally occur together. It would be very hard (if not impossible) to collect the empirical data to support the subsequent KERs if all of these KEs were independently in the AOP.

If cases arise where these are required to be split out, then since this is a living document we propose that this happen in future iterations of the AOP. We welcome collaboration on this AOP.

(5) We feel that the current list provides a sufficient amount of choice and variety to measure cell death. However, because this is a living document, additional assays can always be added later if there is desire.

(6) –

(7) and (8)

Agree. KE3 will be merged with the existing KE: “Increase, Cell

Proliferation; Event: 870”

- Information on regeneration has been moved to KER3.

7.1: I don’t think they are clear or allow reuse. Regenerative proliferation isnt a description of a pathology observation made in hepatocytes: ie it isnt an acutal thing that you can measure and therefore isnt a key event. Regenerative proliferation is typically used as a summary description of the process by which necrosis/apoptosis induced loss of hepatocytes is counterbalanced by proliferation of hepatocyte- i.e. a description of the key event relationship. I havent seen any evidence that the key event you measure: the proliferation induced by a cytotoxic stimulus is qualitatively different from the proliferation induced in hepatocytes by other stimuli such as hepatocyte mitogens. If the proliferation is qualitatively different included the evidence for it, if it isnt just call this key event hepatocyte proliferation and move the regeneration process description to the key event relationship. This will allow proliferation to be a reusable key event.

7.2: The description is focused on liver regeneration after partial hepatectomy and is highly superficial for a complex process (see for instance recent reviews Alvarez-Sola et al: Bile acids, FGF15/19 and liver regeneration: From mechanisms to clinical applications. Biochim Biophys Acta. 2017 Jul 12. pii:S0925-4439(17)30222-3 or Michalopoulos: Hepatostat: Liver regeneration and normal liver tissue maintenance. Hepatology. 2017 Apr;65(4):1384-1392).

8. Are measurement methods specified and adequately described/referenced?

For proliferation yes. For the processes stimulating that proliferation, no. Nor is there any proposed measure for

Note, however, that there are many other regenerative proliferation KEs in the AOP Wiki across many other tissues; thus, we think it’s worth hearing other people’s opinions on this before we make the change.

the proliferation and transdifferentiation of stem cells that could occur.

9. Is the biological context (inc. taxonomic applicability/relevance, level of biological organisation) specified and explained sufficiently?

Section 5:Adverse OutcomePR: Yes, an AO is described.

1. The measurement methods are adequately described and referenced.

2. Use in other AOPs would be reasonable, although in describing risk factors it suggests specific KEs that may lie upstream. I don’t see this as problematic for maintaining the modularity of the description though. There were no other well connected or well described KEs already present in the Wiki that was an obvious synonym for the present AO.

3. Some basic information on applicability to vertebrates and greater prevalence in males is provided, but life stage applicability is not described and the structured fields have not been filled out. Consider adding information to the structured fields if possible.

4. The regulatory relevance of the AO is not explicitly described.

1--

2--

3. This comment has been raised and addressed above.

4: Agree. Text inserted:

Any cancer endpoint is considered to be adverse from a regulatory perspective. Substances causing cancer are regulated such that the general population is not exposed to levels that exceed the carcinogenic dose. The standard assay for carcinogens is the two-

year rodent bioassay, which is conducted by the National Toxicology Program in the U.S.A. (https://ntp.niehs.nih.gov/). The International Agency on Research on Cancer (IARC; https://www.iarc.fr/) categorizes substances based on available evidence pointing to their ability to cause cancer in humans and/or animals.

SR1: Suggestion for modified AO

1. Based on the comments above (see section on MIE and KE), I propose to change the AOP as given in the graph below:

- The suggestion that we produce an additional 5 KEs and 2 KERs is not feasible.

- Also, the MIEs do not conform with the OECD’s definition of an MIE (discussed above).

- A portion of this could be produced as separate AOP, and this would become a network. We feel that the suggestion is beyond the scope of what we are capable of doing. Please also see responses above.

SR2:

1. Is an AO described? YES2. Is the AO description clear and is it biologically

plausible? YES3. Is the AO described in a way that allows its use in other

AOPs? NO, other AOPs have used liver adenoma or carcinomas to describe their liver cancers. Perhaps reuse of these terms would be a better way to enable reuse?

4. Are measurement methods specified and adequately described/referenced? Lacking in specific detail but referenced yes

5. Is the biological context (inc. taxonomic applicability/relevance, level of biological organisation) specified and explained sufficiently? yes

Has the regulatory relevance of the AO been described?no

1--

2--

3: Other similar AOs include: KE378, which is not relevant because it is written specifically for AFB1-dependent mutation of p53. KE1395, which is blank (as above, we suggest they integrate into our AO, since ours is already written).

4--

5--

6--

Section 6:

Key Event RelationshipsPR

Y/N responses to each question are summarized for each KER – specific comments follow

Relationship Q1-modular

Q2-plausible

Q3-emp. support

Q4 – uncertainties

Q5-quant.

1512 Y [1] Y Y [2] Y Y 1513 Y Y [3] Y Y Y 1514 Y Y [3] Y [4] Y Y 1515 Y N [5] Y Y Y 1516 Generally

[7] Y/N [6] Y Y [6] Y

1517 Y [8] Y Y Y Y 1518 Y Y Y Y Y

Relationship 1512: [1] The final sentence of the empirical support section seems more aligned with relationship 1516 than 1512. Otherwise, described in a way that allows use with other AOPs. [2] Consider bulletizing the list of empirical support for the linkage to make it a bit easier to read. Doesn’t necessarily need to be presented as narrative text. Often bulleted text under some sub-headings or a table can make easier for a user to digest than a long narrative stringing together many different lines of evidence. That’s just a suggestion though – nothing inconsistent with the guidance here.

Relationship 1513: [3] Most of the plausibility is described in the “How this key event relationship works section” That is ok though. The distinction between those two sections can be a bit fuzzy.

Relationship 1514: [4] In their empirical support section the authors allude to specific example results the address temporal concordance (for example) between the KEs. The new version of the handbook suggests organizing empirical support under sub-headings for results that support temporal concordance, dose-response concordance, etc. While this is optional, it can be helpful as readers try to evaluate the evidence in the context of the different Bradford-Hill considerations. Most of the evidence seems to be focused on establishing temporal

1512-1514: We thank PR for these suggestions. We will consider these during our revisions to the AOP.

1515 (iKER1): Agree. Revised text in biological plausibility

concordance.

Relationship 1515: [5] The “how does this key event relationship work” section provides useful information regarding the biology that plausibly links cyp2E1 activation to cytotoxicity. However, the biological plausibility statement just talks about metabolic activation of substrates – it doesn’t really address the downstream event of cytotoxicity, only a general reference to “metabolite-dependent” toxicity.

Relationship 1516: “The focus of this iKER is on (b) and (c), as the details of (a) are mapped out elsewhere”. I think “i” could be deleteted from “iKER”.

Second, if you’re going to suggest that (a) is detailed elsewhere, it would be helpful to indicate where one would find that (e.g., AOP:220).

Since it seems that you’re basically proposing three different paths through which oxidative stress could lead to cancer, do you plan to develop these other branches (b) and (c) in separate AOPs (which may share the KEs leading up to oxidative stress and converge on the AO of liver cancer)? It appears that’s what you are setting up for. Essentially three, potentially concurrent AOPs that fall under the umbrella of biology that this KER spans. However, you also suggest the understanding of (b) and (c) may not be sufficient to support development of more detailed description of those paths at this time. In the end, I think this is an appropriate use of this indirect KER.

[6] This is another case where the “how does this KER work” section really seems to address the plausibility. In this case, the

section:

Metabolite-dependent toxicity and adduct formation are well known side-effects of cytochrome P450 mono-oxygenase metabolism of xenobiotics in the liver. Because primary metabolites are more reactive than the parent compound, they often create adducts to cellular proteins or DNA. In both cases, this prevents the normal functioning of the molecules. In extreme cases this will lead to hepatocytoxicity due to: (1) the large number of adducts, (2) the loss of function of important cellular proteins and the related cellular processes, and (3) loss of function of important genes due to DNA damage and mutation.

1516:

- Noted - we will delete the i.

-

- Agreed. We have now noted that this is AOP:220.

- The reviewer is correct that this was our intent. We believe these are different AOPs that bridge in with this one, and we were setting this up within this KER, and we felt important to note this. We, personally, do not have resources at this time to develop the additional AOPs mentioned by PR. We are happy that PR thinks this is a useful approach for this non-adjacent, broadly encompassing KER.

plausibility reads more like uncertainty.

[7] The chloroform and carbon tetrachloride paragraph of the empirical support section seems to deviate somewhat far afield from the relationship between oxidative stress and liver cancer. A lot of different endpoints are discussed, but the connection or correlation with a cancer outcome is not made strongly in the paragraph. Likewise, the paragraph regarding chronic Nrf2 activation raises the question of whether this should be represented as a separate KE and a KER linking to liver cancer. There seems to be quite a bit of evidence for an association.

Relationship 1517: The description of this KER contains a setence or two on cytotoxicity, but a bunch of information regarding inflammation and NF-kappaB. While this may be biology that underlies the connection between cytotoxicity and cancer, it is not laid out in a way that is particularly easy to follow. The plausibility lays out the connection more clearly and simply. One could consider adding the more detailed description of inflammation and NF-kappaB as a sub-heading under plasuibility.

[8] It may be worth considering how this AOP ties with

AOP:118 (chronic cytotoxicity leading to hepatocellular adenomas and carcinomas).

Likewise, AOP:32 includes several similar KEs (hepatotoxicity, regenerative proliferation, liver tumors).

AOP:41 shares a number of similarities as well.

- [6] and [7] We will review and consider revisions based on these suggestions.

1517 (iKER3): We agree with PR’s comments. We swapped the text from the “how does this work” and the “plausibility” sections, which we think addresses this comment.

[8] This comment points out the challenges of having multiple authors developing AOPs at the same time and in their own silos. It is worth discussion by the EAGMST.

- AOP118 and 32 are blank; therefore, suggest that they use ours during their AOP development.

It would appear the present AOP is part of a closely connected AOP network in which a variety of initiating events leading to oxidative stress and/or inflammation can lead to downstream liver tumor formation via cytotoxity and regenerative proliferation. It would be great to see these developed in a way that promoted the connectivity of shared elements within the AOP-Wiki. Likewise, there is new proposed guidance from a recent workshop on how to represent the process of inflammation as KEs in an AOP. If you view inflammation as an important part of the process connecting cytotoxicity to liver cancer, you could consider including one or more of those key events. (feel free to contact me for more details if you are interested).

Relationship 1518: While the inclusion of this relationship as an indirect KER is consistent with current guidance, the use of “indirect KERs” to indicate a lack of understanding of the details connecting the two KEs in question. Under the new guidance, we would encourage this to be designated as an “adjacent KER” – that is, the two KEs in question are adjacent in the sequence you’ve described in your AOP. This does not preclude the possibility that additional KEs might be inserted in the future, turning this into a non-adjacent KER, just a recognition that there are always finer and finer levels of biological resolution that could be inserted between any two KEs. The confidence scores are used to indicate the gaps or weakness in our understanding, while the line drawn between to KERs just indicates the pair for which evidence of an upstream-downstream association is being assembled. With this in mind, consider changing the “directness”, soon to be termed “adjacency” from indirect (non-adjacent) to direct

- AOP 41… we’re not really sure what to do about this comment. We agree in seeing the flow-chart that there is a lot of overlap, although there are also notable differences… This comment also relates back to what we noted above. We note that this AOP merges even more information into the KEs and is comprised of many non-adjacent events. We also note that this AOP requires revision, as the KEs and KERs as described refer to the entire AOP and other modules of the AOP. We think it might be worthwhile bringing both of these AOPs through and having other select from the KE(R)s.

In terms of promoting connectivity… Is there a database curator who has a bird’s eye view of everything coming down the pipeline who can merge and prune AOPs as they are incorporated into the database?

1518: This AOP was written in 2016, before the new guidance was available. Therefore, it was written according to those rules. We will do a search and change mentions of ‘indirect’ relationships noted by PR to ‘non-adjacent’ as per new guidance.

(adjacent).

SR1: - ---

SR2:

(1) I think the descriptions of the KERs were generally well described. I would say that there is lot of detail hidden in the KERs that could be used to describe additional key events in this AOP but for some reason the authors have chosen to chunk and not split KE and KERs. As the mechanisms of the hepatostat continue to be uncovered it is likely that there will be more MIEs identified and they will need KE and KERs at a highly grnaular detail to enable reuse in AOP networks- so I would encourage you to split KE and KERs as much as you can given the available data.

(2) When describing the quantative nature of the regenerative proliferation KER (cytotoxicity to proliferation) they use the term “significant amounts of hepatotoxcity” but do not define what this means.

(1) We feel that we have addressed this comment above.

(2) We will refine the text to address this comment.

SECTION 7: Overall Assessment of the AOPPR: The authors do a nice job of summarizing the overall weight of evidence supporting their AOP and provides brief justifications for all their weight of evidence calls, referring directly to the guiding questions provided in Annex 1 of the handbook. The domain of applicability is reasonably defined

Thank you!

as being applicable to both sexes, all life stages, with the strongest evidence available in rodent models, but likely translating to humans as well. Applicability to non-mammalian taxa is not addressed. Overall, the authors have done a very nice job on this section and have clearly followed the guidance.

SR1: See general observations and conclusions below ---

SR2: - ---

Section 8:

PR: Yes, the authors provide some context for the development and intended use.

---

SR1: - ---SR2: - ---

Section 9:

General Observations and Conclusions of the ReviewerPR: Generally speaking, a well developed AOP description that is consistent with the guidance provided in the current version of the handbook. Given strong similarity between a number of the KEs included in the current AOP and others in the AOP-Wiki, as well as synonymous events found in other AOPs, I would encourage the authors to try to coordinate/integrate their KE/KER descriptions with similar ones by other authors in order to produce greater connectivity in the overall AOP network, and perhaps strengthen the current descriptions further. Inclusion of structured terms regarding life-stage, sex, and taxonomic applicability in the KE

We thank PR for the positive comments and support.

We have not made any changes at this time to use other people’s KE(R)s. We feel that the comments of PR are really important and warrant discussion within the EAGMST to determine what approach should be used, as this will likely occur fairly routinely in AOP development.

Please note: Because our KEs and KERs are already developed, we suggest that the authors of the overlapping (but currently blank) KEs

descriptions would also aid later network applications of this AOP. However, overall I felt this was a strong AOP description which should be suitable for external review with only minor revisions.

and KERs (or curators of the AOP-wiki) use ours.

SR1:

1. The relation of the activation and subsequent KE to the final adverse effects is well described, concise and supported by appropriate evidence from the scientific literature. Based on this description a simple and straight forward AOP chain was developed. However, if the AOP description is carefully read, it becomes evident that the provided AOP is highly branched, involves at least two MIE and various KEs, partially branched or parallel, but finally conflated at the AO. Hence, it requires discussion, whether the provided AOP does indeed not reflect several and/or partially branched AOPs.

2. Furthermore, it requires discussion whether the activation of CYP2E does represent the MIE instead of reactivity to lipids or DNA. And the CYP2E activation would rather represent a toxicokinetic constraint that drives and explains the specificity to liver. Indeed, one may argue that if an organism is exposed to the reactive metabolite directly, this would also trigger the KE – but probably not in a liver-specific manner. A potential solution to describe the AOP could be to use a different terminology for the MIE(s) with focus on the formation of reactive metabolites.

1. We agree that the biology is more complex than the AOP. As described above, we purposefully simplified, and believe some of these additional details belong in separate AOPs. We are open to providing this AOP to the community for further development, but have no resources or plans to add all of the additional KEs/KERs proposed.

2. Disagree. The definition of an MIE is the point at which the stressor interacts with the biological system. In the case of this AOP, the chemical substance directly interacts with Cyp2E1 to activate/stabilize it. So, as we argue above, we strongly believe that this is the only possible MIE for this AOP, with the downstream consequences (e.g., ROS production, lipid peroxidation) being downstream KE.

SR2: I think this is a well written AOP that reviews and integrates a complex and well studied area of science. I think our goal to enable reuse of key events would benefit from the authors splitting their propsed KE into their component parts to capture some of this complexity. Methods are often described in a superficial way. Their use of the term CYP2e1 activation is ambiguous in meaning and should be clarified.

Once the AOP becomes an evergreen document, there will be the capacity to add to it and modify it.

As described above, we have clarified the cyp2e1 activation MIE.