Mechanism of stepwise electron transfer in six-transmembrane epithelial antigen of the prostate (STEAP) 1 and 2

Reviewer #1 (Public Review):

This important study reveals the structure of human STEAP2 for the first time and suggests the electron transport pathway, but some questions remain regarding the interpretation of the in vitro electron transport experiments, the lack of available redox couples, and potential alternative hypotheses that would if addressed, strengthen the claims in the manuscript.

Strengths

One of the clear strengths of the manuscript that stands out is the determination of the structure of human STEAP2. The structures of some other homologs are known, but STEAP2's structure was not, and STEAP2 seems to have an unusually low activity towards certain metal chelates. The approach of producing the human STEAP2 in insect cells with the supplementation of cofactor biogenesis components nicely results in cofactor-replete protein. The structure of STEAP2 reveals a domain-swapped trimer, with the NADPH-binding domain of the neighboring protomer within electron-transport distance of the FAD-heme axis. The FAD has an interesting and somewhat unusual extended conformation and abuts a Leu residue that may regulate electron transport. Another strength of the manuscript is the demonstration that STEAP1, which does not have the internal NADPH binding domain, can interact modestly and shuttle electrons to the heme in STEAP1 through FAD. These experiments nicely expand information on the function of STEAP1 and provide a structural basis for electron transport in STEAP2.

Weaknesses

A major weakness in the manuscript lies with the kinetics data and how the data, as presented, are unclear to the reader regarding their impact and their connection to the purported electron transport scheme. While multiple sets of data are reported, the analysis in all cases is simply the reduction of a hexacoordinate heme and its related spectra and kinetic parameters. In most cases, it's unclear to the reader which part of the electron pathway is being tested in which experiment. Simple diagrams would be helpful in each case. However, it's also unclear if all of the potential order of addition experiments were actually performed; i.e., flavin but no NADPH; NADPH but no flavin; flavin before NADPH; flavin after NADPH, etc. As there are multiple permutations that should be tested to demonstrate the electron transport pathway, presenting the data in a way that makes it clear to the reader is challenging. Particularly missing are the determined redox potentials of the hemes in both STEAP1 and STEAP2. Could differences in these heme redox potentials be drivers of the difference in metal reduction rates? Also, the text indicates that STEAP2 does not show a reduction rate dependence on the [Fe3+-NTA], but Figure 1A shows a difference in rates dependent on [Fe3+-NTA], and the shape of the reduction curve also changes based on [Fe3+-NTA]. This discrepancy should be rectified.

A second major weakness is the lack of any verification of the relevance of the STEAP2 oligomerization to its in vivo function. Is the same domain-swapped trimer known to exist in vivo? If the protein were prepared in other detergents, is the oligomerization preserved? It is alluded to in the text that another STEAP protein is also a trimer. Was this oligomerization verified in vivo? Could this oligomerization be disrupted to impede or abrogate electron transport to underscore the oligomerization relevance? This point is germane, as it would further suggest that the domain-swapped trimer observed in the STEAP2 cryo-EM structure is physiologically relevant, especially given how far the distance between the NADPH and the FAD would otherwise be to support electron transport.

Beyond these two areas in which the manuscript could be improved there are a couple of minor considerations. First, the modest resolution of the STEAP2 structure prevents assigning the states of NADP+/NADPH and FAD/FADH2 with confidence. An orthogonal measure would be useful for modeling the accurate states in the structure. Finally, the BLI b5R/STEAP1 binding/unbinding fits seem somewhat poor, especially at high concentrations of b5R in the dissociation regime, which likely influences the derived value of Kd. A different fitting equilibrium might yield better agreement between the experimental and theoretical results. Moreover, whether this binding strength is influenced by the reduction state of the NADPH would be helpful in understanding and contextualizing the weak binding affinity.

Reviewer #2 (Public Review):

The manuscript provides new insight into a family of human enzymes. It demonstrates that STEAP2 can reduce iron and STEAP1 can be promiscuous regarding the source of electron donors that it can use. The quality of the kinetics experiment and the structural analysis is excellent. I am less enthusiastic about the interpretation of data and the take-home message that the manuscript intends to deliver. Above all, the work combines data on STEAP2 and STEAP1 and it remains unclear which questions are ultimately addressed. A second critical point is about the interpretation of the experiment demonstrating that STEAP1 can be reduced by cytochrome b5 reductase. The abstract states that "We show that STEAP1 can form an electron transfer chain with cytochrome b5 reductase" whereas the main text discusses the data by suggesting that "we speculate that FAD on b5R may partially dissociate to straddle between the two proteins". The two statements seem to be partly contradictory. Cytochrome b5 reductases do not easily release FAD but rather directly donate electrons to heme-protein acceptors (PMID: 36441026). According to the methods section, no FAD was added to the reaction mix used for the cytochrome b5 reductase experiment. Overall, the data seem to indicate that the reductase might reduce the heme of STEAP1 directly. Would it be possible to check whether FAD addition affects the kinetics of the process? And to perform a control experiment to check that NAD(P)H does not directly reduce the heme of STEAP1 (though unlikely)? A final point concerns the "slow Fe3+-NTA reduction by STEAP2". The reaction is not that slow as the initial phase is 2 per second. The reaction does not show dependence on the substrate concentration suggesting "poor substrate binding". I am not convinced by this interpretation. Poor substrate binding would give rise to substrate dependency as saturation would not be achieved. A possible interpretation could be that substrate binding is instead tight and the enzyme is saturated by the substrate. Can it be that the reaction is limited by non-productive substrate-binding and/or by interconversions between active and non-active conformations?

Reviewer #3 (Public Review):

The six-transmembrane epithelial antigen of the prostate (STEAP) family comprises four members in metazoans. STEAP1 was identified as integral membrane protein highly upregulated on the plasma membrane of prostate cancer cells (PMID: 10588738), and it later became evident that other STEAP proteins are also over expressed in cancers, making STEAPs potential therapeutic targets (PMID: 22804687). Functionally, STEAP2-4 are ferric and cupric reductases that are important for maintaining cellular metal uptake (PMIDs: 16227996, 16609065). The cellular function of STEAP1 remains unknown, as it cannot function as an independent metalloreductase. In the last years, structural and functional data have revealed that STEAPs form trimeric assemblies and that they transport electrons from intracellular NADPH, through membrane bound FAD and heme cofactors, to extracellular metal ions (PMIDs: 23733181, 26205815, 30337524). In addition, numerous studies (including a previous study from the senior authors) have provided strong implications for a potential metalloreductase function of STEAP1 (PMIDs: 27792302, 32409586).

This new study by Chen et al. aims to further characterize the previously established electron transport chain in STEAPs in high molecular detail through a variety of assays. This is a well-performed, highly specialized study that provides some useful extra insights into the established mechanism of electron transport in STEAP proteins. The authors first perform a detailed spectroscopic analysis of Fe3+-NTA reduction by STEAP2 and STEAP1, confirming that both purified proteins are capable of reducing metal ions. A cryo-EM structure of STEAP2 is also presented. It is then established that STEAP1 can receive electrons from cytochrome b5 reductase, and the authors provide experimental evidence that the flavin in STEAP proteins becomes diffusible.

The specific aims of the study are clear, but it is not always obvious why certain experiments are performed only on STEAP2, on STEAP1, or on both isoforms. A better justification of the performed experiments through connecting paragraphs and proper referencing of the literature would improve readability of the manuscript. Experimentally, the conclusions are appropriate and mostly consistent with the experimental data, although one important finding can benefit from further clarification. Namely, the observation that STEAP1 can form an electron transfer chain with cytochrome b5 reductase in vitro is an exciting finding, but its physiological relevance remains to be validated. The metalloreductase activity of STEAP1 in vitro has been described previously by the authors and by others (PMIDs: 27792302, 32409586). However, when over expressed in HEK cells, STEAP1 by itself does not display metal ion reductase activity (PMID: 16609065), and it was also found that STEAP1 over expression does not impact iron uptake and reduction in Ewing's sarcoma (cancer) cells (PMID: 22080479). Therefore, the physiological relevance of metal ion reduction by STEAP1 remains controversial. The current work establishes an electron transfer chain between STEAP1 and cytochrome b5 reductase in vitro with purified proteins. However, the conformation of this metalloreductase activity of the STEAP1-cytochrome b5 complex will be required in a cell line to prove that the two proteins indeed form a physiological relevant complex and that the results are not just an in vitro artefact from using high concentrations of purified proteins.

The work will be interesting for scientists working within the STEAP field. However, some of the presented data are redundant with previous findings, moderating the study's impact. For instance, the new structural insights into STEAP2 are limited because the structure is virtually identical to the published structures of STEAP4 and STEAP1 (PMIDs: 30337524, 32409586), which is not surprising because of the high sequence similarity between the STEAP isoforms. Moreover, the authors provide experimental evidence to prove the previous hypothesis (PMID: 30337524) that the flavin in STEAP proteins becomes diffusible, but the molecular arrangement of a STEAP protein, in which the flavin interacts with NADPH, remains unknown. Based on the manuscript title, I would also expect the in-depth characterization of STEAP1/STEAP2 hetero trimers (first identified by the authors), but this is only briefly mentioned. When taken together, this study by Chen et al. strengthens and supports previously published biochemical and structural data on STEAP proteins, without revealing many prominent conceptual advances.