Axonal T3 uptake and transport can trigger thyroid hormone signaling in the brain

Article activity feed

1. 

   Version published to 10.7554/elife.82683 on eLife

   May 19, 2023
2. 

   eLife

   Mar 16, 2023

   Author Response

   Reviewer #1 (Public Review):

   Part 1: Type 2 deiodinase

   Table I is supposed to clarify and summarize the results but brings confusion. The text says that table I supports the claim that "in the cerebellum, Luc-mRNA was lower in the Ala92-Dio2 mice" whereas figure 1G does not show any difference. It is unclear whether Table I and figure 1 report the same data, and what the statistical tests are actually addressing (effect of genotype vs effect of treatment, whereas what matters here is only the interaction between genotype and treatment). Overall, it is not acceptable to present quantitative data without giving numbers, standard deviation, p-value, etc. as in Table I.

   Thank you. We agree with the reviewer. We intended to minimize the amount of data presented, which was already very large, and therefore only presented …

   Author Response

   Reviewer #1 (Public Review):

   Part 1: Type 2 deiodinase

   Table I is supposed to clarify and summarize the results but brings confusion. The text says that table I supports the claim that "in the cerebellum, Luc-mRNA was lower in the Ala92-Dio2 mice" whereas figure 1G does not show any difference. It is unclear whether Table I and figure 1 report the same data, and what the statistical tests are actually addressing (effect of genotype vs effect of treatment, whereas what matters here is only the interaction between genotype and treatment). Overall, it is not acceptable to present quantitative data without giving numbers, standard deviation, p-value, etc. as in Table I.

   Thank you. We agree with the reviewer. We intended to minimize the amount of data presented, which was already very large, and therefore only presented the ratios of thr/alaDio2 and which created confusion. This part was removed from the new version of the MS.

   Also, evaluating T3 signaling by only looking at the luc reporter and the Hprt housekeeping gene is not always sufficient (many T3 responsive genes can be found in the literature and more than one housekeeping gene should be used as a reference).

   Thank you. The advantage of using the THAI mouse is that the Luciferase reporter gene is driven by a promoter that is only sensitive to T3, which is not the case for any other T3-responsive responsive gene. The Hprt housekeeping signal was stable among the samples, and the differences observed were not caused by differences in the housekeeping gene expression. This part was removed from the new version of the MS.

   Another important weakness is that the wild-type mice have a proline at position 92. Why not include them? In absence of structural prediction, one wonders whether the mouse models are relevant to the human situation and whether the absence of the proline reduces the enzymatic activity when substituted for an Ala or Thr. This might have been addressed in previous work, but the authors should explain.

   The position 92 in DIO2 is occupied by Thr in humans. Its Km(T4) is indistinguishable from mouse Dio2 which has a Pro in the position 92 (4nM vs. 3.1nM) [PMID 8754756; PMID: 10655523]. Humans also carry an Ala in position 92. Comparing the two human alleles is the purpose of the study.

   Experiment 2: Ala92-Dio2 Astrocytes Have Limited Ability to Activate T4 to T3

   Here, the authors use primary cell cultures from different areas of the brain to measure the in vitro conversion of T4 to T3 by Dio2. They find that hippocampus astrocytes are less active, notably if they come from Ala92-Dio2 mice.

   This part has the following weaknesses:

   • This result correlates with the results from Fig 1F however the difference between Ala92-Dio2 and Thr92-Dio2 is significant in vitro, but not in vivo.

   From a deiodinase perspective, TH signaling in vivo depends on the presence of D2 (expressed in glial cells) and D3 (expressed in neurons), whereas in vitro it only depends on D2. In fact, D2 and D3 are known for a reciprocal regulation to preserve TH signaling [PMID: 33123655]. Thus, it is conceivable that the differences observed between the two models are explained by the intrinsic differences in the models.

   What matters is not the activity/astrocytes, but the total activity of the brain area, which depends on the number of astrocytes x individual activity. This is not measured.

   We respectfully disagree with the reviewer. The total D2 activity in a brain area depends fundamentally on the number of astrocytes in that area and on the intrinsic activity of the enzyme. The reviewer is suggesting that having an area denser in astrocytes expressing a catalytically less active D2 preserves a normal local T3 production. This is unlikely to be the case because we have no evidence that the density of astrocytes is different in Ala-DIo2 mice. Please keep in mind that the intimate relationship between astrocytes and neurons is what defines the microenvironment that surrounds the neuron. By separating astrocytes from neurons we are able to measure T3 production that is occurring in the neuronal microenvironment and show that cells obtained from AlaDio2 mouse produce less T3.

   • What the authors called 'primary astrocytes' is an undefined mixed population of glial cells, (including radial glial cells, stem cells, ependymal cells, progenitor cells, etc...) that proliferated differentially for more than a week in culture, among which an unknown ratio expresses Dio2. The cellular model is thus poorly characterized, and the interpretation must be prudent.

   • Again, wild-type mice are not included.

   Thank you. We now include a reference to illustrate the types and percentages of cells present in our cultures. Given that the study is to compare the Thr92 and the Ala92 alleles, which are both present in humans, we did not believe it was necessary to include them here. Please note (as explained above) the Km(T4) for Thr92 and Pro92-Dio2 is indistinguishable.

   Part 2: Neuronal response to T3 Involves MCT8 and Retrograde TH transport

   The authors next move to primary neuronal cultures, prepared from the fetal cortex which they grow in the microfluidic chamber to study axonal transport. This is a surprising move: the focus is not on Dio2 anymore, but on the MCT8 transporter, which is known in humans to play an important role to transfer TH into the brain. It is expressed mainly in glia, but also in neurons. They study the influence of endosomes and type 3 deiodinase on the trafficking and metabolism of TH.

   Thank you.

   It would be useful to perform an experiment, in which radioactive T3 is introduced in the "wrong" side of the chamber, in an attempt to detect a possible anterograde transport. This would address the possibility that Mct8 also promotes efflux and control so that the chamber is not leaking.

   Thank you. To satisfy the reviewer, we have conducted three new experiments adding 125IT3 in the MC-CS. The first experiment verified that the T3 transport in the cortical neurons also occurs anterogradely. The second experiment showed that the anterograde transport depends on mct8. The third experiment shows that D3 activity in the neuronal soma is limiting the amount of T3 transported along axons. We have included a new paragraph in the results section describing these experiments (Line 154 to 167), and a new supplementary figure (Figure 3—figure supplement 3). We have also discussed these new findings. Line 383 to 386. In every experiment, we have controlled for the possibility of leaking using one device without neurons that received radioactive T3. After 24 and 72h samples from the opposite side were obtained but did not contain any radioactive T3. We refer the reviewer to figure 1, where this is explained.

   The authors use sylichristin as an inhibitor of Mct8, to demonstrate that transport is Mct8 dependent. They do not provide indications or references that would clearly indicate that this drug is a fully selective antagonist of Mct8 (but not of Oatp1c1, Mct10, Lat1, Lat2, etc., the other TH transporters). A good alternative would be to use Mct8 KO mice as controls.

   Thank you. We refer the reviewer to reference 27 [J. Johannes et al., Silychristin, a Flavonolignan Derived from the Milk Thistle, Is a Potent Inhibitor of the Thyroid Hormone Transporter MCT8. Endocrinology 157, 1694-1701 (2016)] clearly indicating that Silychristin has a remarkable specificity toward MCT8. While using mct8 KO is interesting, it would have prevented us from testing some of our hypotheses. Being able to selectively inhibit Mct8 either in the MC-CS or in the MC-AS was a clear advantage. For example, pls see the experiment in which we add T3 in the MC-AS and the silychristin in the MC-CS (Fig. 3F). Here, we discovered new roles of mct8, such as its involvement in the release of T3 from the endosomes (line 228 to 231).

   The B27 used in primary neuronal culture might contain TH. This is not easy to know, but at least some batches do.

   Thank you. While the neurons were cultured in B27, all experiments were performed in cells incubated with neurobasal only (B27 was removed 24 earlier). This was not clear in the initial version, where there was only a vague reference in the legend of figure 3F. Now, this has been explained in the footnote of figure 3 and in line 207.

   The presence of astrocytes, probably expressing Mct8 and Dio2 is inevitable in primary neuronal cultures, and is not mentioned, but might interfere with TH metabolism.

   Thank you. We were aware that, under normal conditions, primary neuronal culture contains 25% of astrocytes. This was however minimized/eliminated by 2-day culture with the anti-mitotic cytosine arabinoside, which restricts astrocytes and microglia to <0.01 in this type of culture. This was explained in the initial version of the manuscript in the material and methods section (lines x to x) and supported with reference 53 (reference 57 in the previous version).

   Part 3: T3 Transport Triggers Localized TH Signaling in the Mouse Brain

   The authors return to in vivo experiments, implanting T3 crystals, labeled or not with radioactive iodine. They do so in the hypothalamus, where they address the retrograde transport of TH in TRH neurons, and in the cortex, looking for contralateral transport. These data are the most difficult to interpret. - First, T3 is hydrosoluble and would probably migrate without active transport.

   Thank you. Please note that at no point we characterized the T3 transport “active transport”, which by definition is an ATP-dependent process. Please note that to address the issue raised by the reviewer “migrate without active transport”, in both experimental approaches, we included controls to assess the random diffusion of T3.

   In hypothalamic studies, we used the (i) cerebral cortex and (ii) the lateral hypothalamus, a region that is immediately adjacent to the PVN. Neither region exhibit an axonal connection to the median emminence. The results, in both cases, show that the presence of radioactive T3 in the control areas was minimal when compared to the PVN (Fig. 5C).

   In the cerebral cortical studies, we included ipsi- and contra-lateral hypothalamic measurements that served as controls given the absence of a connection between the cortex and the hypothalamus. Accordingly, T3 signaling was not detected in any of the control regions (Fig. 6C previous version; now figure 5). Thus, these controls indicate that it is unlikely that the results could be explained by “migrate without active transport” of T3.

   • The authors do not demonstrate that these specific neuronal populations contain Mct8, and that these observations are connected to the previous in vitro observation (which used cortical neurons prepared from the fetus).

   Thank you. In the previous version, we did not make it abundantly clear that the EM pictures in Fig. 3D-G (previous version; now figure 2 D-G) were from neurons in the mouse motor cortex (this information is now explained in lines 149 to 151), which is where we inserted the T3 crystals. In addition, we have done more histological work on the brain M1 (cortex) of adult mice and found that many neurons in the M1 express D3 and Mct8—lines 433-434 and Figure 5 G-K (along with histological studies showing the specificity of the ab against D3 Fig S6).

   The possibility that astrocytes are involved, as reported in the literature, is not considered.

   • Here again, using Mct8KO mice would greatly help to interpret the data. In particular, the experiments with cold T3 involve a 48h delay which is very long in comparison to the 30 minutes required for long-distance transfer of radioactive T3.

   Thank you. We are unsure about the question posed by the reviewer. We are wondering how would astrocytes play a role in inter-hemispheric transport of T3? Given that astrocytes are not known to project across long distances, we have not considered this possibility. We agree that using the Mct8KO mouse could have provided supporting evidence of the role played by Mct8 in this process, but please keep in mind that the Mct8KO mouse does not have or exhibits a very mild brain phenotype, indicating that during development compensatory mechanisms have occurred that obviate the function of the transporter. This compensatory mechanism most likely involved Oatp1c1, given that only the double Mct8 and Oatp1c1 KO mouse develops a significant phenotype. This consideration directed us to the utilization of sylycristin, the highly selective Mct8 inhibitor, which disrupts the Mct8 pathway in a mouse that developed normally.

   The two approaches used to demonstrate neuronal T3 transport in vivo are fundamentally different. The hypothalamus experiments employed radioactive T3, whereas T3 crystals were used in the cerebral cortex. The first approach studied T3 transport whereas the second studied downstream T3 effects, logically requiring more time. The solid T3 implant requires time to release T3 and activate gene expression. In the original paper that utilized T3 implants in the rodent brain, samples were processed after 4 days. (Dyess et al. 1988 Endo; PMID 3139393)

   Discussion

   Considering the diversity of questions that are addressed in the study, it is not surprising that the discussion is not covering all aspects. The authors implicitly consider that their conclusions can be extended to all neurons, while they use in their experiments a variety of different populations coming from either the fetal cortex, hippocampus, adult cortex, or hypothalamus. The claim that they discovered a mechanism applying to all neurons is not supported by the data.

   Thank you. We agree with the reviewer: the high number of neuronal subtypes might include different mechanisms in T3 transport. Our studies involved cortical (central) and dorsal root ganglia (peripheral) neurons in vitro and cortical and hypothalamic neurons in vivo. Thus we think that the described mechanism is not confined to specific neuronal subtypes. The discussion has been modified accordingly (lines 402 to 411).

   Moreover, we have done immunofluorescence studies to characterize the neurons present in the MC-CS better. We have found that all the neurons residing in the MC-CS are excitatory, expressing the vesicular glutamate transporter 1 (Vglut1). But no neurons were expressing GAD67, a marker for inhibitory neurons Figure 5—figure supplement 5). This is supported by the fact that during the mouse's brain development, the embryonic days 14.5 to 17.5 is the birth date of layer 4 and 2/3 excitatory neurons (PMID: 34163074). These neurons are migrating and have not extended their cellular processes, making them more likely to survive the isolation protocol from the cortex. On the other hand, the neurons (mostly excitatory) already residing in the cortex may have expanded their processes and changed their morphology, making them less capable of surviving the isolation process.

   Some highly relevant literature is not cited. In particular:

   • Mct8 KO mice do not have marked brain hypothyroidism (PMID: 24691440) which at least suggests that the pathway discovered by the authors can be efficiently compensated by alternative pathways.

   We agree with the reviewer. As mentioned above, a compensatory mechanism triggered during development “compensates” for the inactivation of Mct8. That, however, does not mean that mct8 is not critically important. We have added that limitation to the discussion (lines 342); ref 46.

   • Dio3 KO only increases T3 signaling in a few brain areas and only in the long term (PMID: 20719855).

   Thank you. That is now included in the ms; ref 25.

   • Anterograde transport of T3 has been reported for some brainstem neurons (PMID: 10473259).

   Thank you. This was our mistake, indeed. We had worked on several versions of the manuscript that included references to her seminal work but unfortunately deleted it from the final version. This is now included in refs 48 and 49.

   Reviewer #2 (Public Review):

   Salas-Lucia et al. investigated two main questions: whether the Thr92Ala-DIO2 mutation impairs brain responsiveness to T4 therapy under hypothyroidism induction and the mechanisms of neuronal retrograde transport of T3. They find that the Thr92Ala-DIO2 mutation reduces T4-initiated T3 signaling in the hippocampus, but not in other brain regions. Using neurons cultured in microfluidic chambers, they further describe a novel mechanism for retrograde transport of T3 that depends on MCT8 and endosomal loading (possibly protecting T3 from D3-mediated cytosolic degradation) and microtubule retrotransport. Finally, they present evidence of retrograde transport of T3 through hypothalamic projections and interhemispheric connections in vivo. The main novelty of this study is the delineation of the mechanism of T3 retrograde transport in neurons. This is interesting from the cell biology perspective. The notion of impaired hippocampal T3 signaling is relevant for the cognitive outcomes of hypothyroidism and its associated therapy.

   Thank you.

   Although the data are exciting and relevant for the community, some issues need to be addressed so that conclusions are more clearly justified by data:

   1. The title and the abstract mean that dissecting this novel mechanism of T3 retrograde transport may help improve cognition or brain responsiveness in patients taking T4 or L-T3 therapy. However, how initial results (Figs 1 and 2) connect to later data is not essentially clear. For example, do Thr92Ala-DIO2 mice present altered retrograde transport of T3? Would stimulation of retrograde transport in Thr92Ala-DIO2 mice rescue neurological phenotypes? Can the authors address this experimentally?

   Thank you. These are all interesting points raised by the reviewer. However, the three reviewers felt that a connection between the studies in astrocytes and the studies in neurons was missing, and complained about the disjoint nature of the manuscript. To satisfy the reviewers we removed from the MS the experiments with astrocytes and DIO2 polymorphism, and focused on the neuronal transport of T3.

   1. Although the authors present in vivo evidence of retrograde T3 transport in the hypothalamus and motor cortex, given the select susceptibility of the hippocampus to hypothyroidism, it would be especially interesting to test whether this mechanism also happens in a hippocampal circuit (CA3-CA1 Schaffer collaterals, mossy fibers or perforant pathway).

   Thank you. We agree that this would be interesting, but technically challenging. Nonetheless, we intend to study this in the future.

   1. Table 1 should present the raw values for Ala92-DIO2 mice and treatments instead of only displaying the direction of change and statistical significance. From Panels 1E-J, it is unclear if Thr92Ala-DIO2 mice or treatments caused any real change in brain regions other than the hippocampus.

   Thank you. These experiments were removed from the new version of the MS.

   1. The authors put forward the notion that a rapid nondegradative endosome/lysosome incorporation protects T3 from D3 degradation in the cytosol. Their experiments with pharmacological modulation of MCT8, lysosomes, and microtubules are in this direction. However, they do not represent an unequivocal demonstration of this mechanism. Therefore, the authors should be more cautious in their interpretation and discuss the limitations of their approaches.

   Thank you. The manuscript was edited to reflect these important points.

   Reviewer #3 (Public Review):

   Initially, Salas-Lucia et al examined the effect of deiodinase polymorphism on thyroid hormone-medicated transcription using a transgenic animal model and found that the hippocampus may be the region responsible for altered behavior. Then, by changing to topic completely, they examined T3 transport through the axon using a compartmentalized microfluid device. By using various techniques including an electron microscope, they identified that T3 is uptaken into clathrin-dependent, endosomal/non-degradative lysosomes (NDLs), transported in the axon to reach the nucleus and activate thyroid hormone receptor-mediated transcription.

   Although both topics are interesting, it may not be appropriate to deal with two completely different topics in one paper. By deleting the topic shown in Table 1, Figure 1, and Figure 2, the scope of the manuscript can be more clear.

   Thank you. We did as suggested by the reviewer. These studies were removed from the present version of the ms.

   Their finding showing that triiodothyronine is retrogradely transported through axon without degradation by type 3 deiodinase provides a novel pathway of thyroid hormone transport to the cell nucleus and thus can contribute greatly to increasing our understanding of the mechanisms of thyroid hormone action in the brain.

   Thank you.

   Read the original source
3. 

   eLife

   Mar 14, 2023

   eLife assessment

   This valuable paper examines the effect of deiodinase polymorphism on thyroid hormone signaling in the brain by employing a transgenic animal model and then switching to studying T3 axonal transport using microfluid devices. Although methodologically extensive this paper has several claims that are not convincingly supported by the current experiments and furthermore some disjoint is observed between the two halves of the study. The therapeutic implications of understanding T3 signaling in the brain makes it a potentially important manuscript.

   Read the original source
4. 

   eLife

   Mar 14, 2023

   Reviewer #1 (Public Review):

   The article from Salas Lucia et al addresses the distribution and transport of thyroid hormones (TH, including T4 and T3) in the adult brain. This is a complex and important question. Overall, the manuscript is difficult to follow as it jumps from one question (Dio2 polymorphism) to another (Mct8 function in the uptake of TH in neurons, and then the connection between TRH neurons and tanycytes), without deepening any aspect. There are, however, interesting findings in the article, but they should be confirmed by additional experiments.

   Part 1: Type 2 deiodinase
   T4 entry is easier than T3 entry in the brain. However, type 2 deiodinase (Dio2 expressed mainly in glial cells) converts T4 to T3 and produces around 80% of the brain T3. In the introduction, the authors mention the controversial observation …

   Reviewer #1 (Public Review):

   The article from Salas Lucia et al addresses the distribution and transport of thyroid hormones (TH, including T4 and T3) in the adult brain. This is a complex and important question. Overall, the manuscript is difficult to follow as it jumps from one question (Dio2 polymorphism) to another (Mct8 function in the uptake of TH in neurons, and then the connection between TRH neurons and tanycytes), without deepening any aspect. There are, however, interesting findings in the article, but they should be confirmed by additional experiments.

   Part 1: Type 2 deiodinase
   T4 entry is easier than T3 entry in the brain. However, type 2 deiodinase (Dio2 expressed mainly in glial cells) converts T4 to T3 and produces around 80% of the brain T3. In the introduction, the authors mention the controversial observation according to which a polymorphism of type 2 deiodinase, Thr92Ala-DIO2, is detrimental to the entry of TH into the brain. One of the associated issues, mentioned by the authors, is that some patients treated with TH have normalized circulating levels of hormones but still complain of fatigue, a typical feature of brain hypothyroidism.
   Experiment 1: Hippocampal Responsiveness to L-T4 is Impaired in the Ala92-Dio2 Mouse
   This first part is a continuation of a previous study published by the same authors. Here, they use transgenic mice with Ala92-Dio2 and Thr92-Dio2 to address possible differences in the TH response of several areas of the brain. The readout is a reporter mRNA, coming from an additional reporter transgene.
   Table I is supposed to clarify and summarize the results but brings confusion. The text says that table I supports the claim that "in the cerebellum Luc-mRNA was lower in the Ala92-Dio2 mice" whereas figure 1G does not show any difference. It is unclear whether Table I and figure 1 report the same data, and what the statistical tests are actually addressing (effect of genotype vs effect of treatment, whereas what matters here is only the interaction between genotype and treatment). Overall, it is not acceptable to present quantitative data without giving numbers, standard deviation, p-value, etc. as in Table I. Also, evaluating T3 signaling by only looking at the luc reporter and the Hprt housekeeping gene is not always sufficient (many T3 responsive genes can be found in the literature and more than one housekeeping gene should be used as a reference).
   Another important weakness is that the wild-type mice have a proline at position 92. Why not include them? In absence of structural prediction, one wonders whether the mouse models are relevant to the human situation and whether the absence of the proline reduces the enzymatic activity when substituted for an Ala or Thr. This might have been addressed in previous work, but the authors should explain.
   Experiment 2: Ala92-Dio2 Astrocytes Have Limited Ability to Activate T4 to T3
   Here, the authors use primary cell cultures from different areas of the brain to measure the in vitro conversion of T4 to T3 by Dio2. They find that hippocampus astrocytes are less active, notably if they come from Ala92-Dio2 mice.
   This part has the following weaknesses:
   - This result correlates with the results from Fig 1F however the difference between Ala92-Dio2 and Thr92-Dio2 is significant in vitro, but not in vivo. What matters is not the activity/astrocytes, but the total activity of the brain area, which depends on the number of astrocytes x individual activity. This is not measured.
   - What the authors called 'primary astrocytes' is an undefined mixed population of glial cells, (including radial glial cells, stem cells, ependymal cells, progenitor cells, etc...) that proliferated differentially for more than a week in culture, among which an unknown ratio expresses Dio2. The cellular model is thus poorly characterized, and the interpretation must be prudent.
   - Again, wild-type mice are not included.

   Part 2: Neuronal response to T3 Involves MCT8 and Retrograde TH transport
   The authors next move to primary neuronal cultures, prepared from the fetal cortex which they grow in the microfluidic chamber to study axonal transport. This is a surprising move: the focus is not on Dio2 anymore, but on the MCT8 transporter, which is known in humans to play an important role to transfer TH into the brain. It is expressed mainly in glia, but also in neurons. They study the influence of endosomes and type 3 deiodinase on the trafficking and metabolism of TH.
   It would be useful to perform an experiment, in which radioactive T3 is introduced in the "wrong" side of the chamber, in an attempt to detect a possible anterograde transport. This would address the possibility that Mct8 also promotes efflux and control so that the chamber is not leaking.
   The authors use sylichristin as an inhibitor of Mct8, to demonstrate that transport is Mct8 dependent. They do not provide indications or references that would clearly indicate that this drug is a fully selective antagonist of Mct8 (but not of Oatp1c1, Mct10, Lat1, Lat2, etc., the other TH transporters). A good alternative would be to use Mct8 KO mice as controls.
   The B27 used in primary neuronal culture might contain TH. This is not easy to know, but at least some batches do.
   The presence of astrocytes, probably expressing Mct8 and Dio2 is inevitable in primary neuronal cultures, and is not mentioned, but might interfere with TH metabolism.

   Part 3: T3 Transport Triggers Localized TH Signaling in the Mouse Brain
   The authors return to in vivo experiments, implanting T3 crystals, labeled or not with radioactive iodine. They do so in the hypothalamus, where they address the retrograde transport of TH in TRH neurons, and in the cortex, looking for contralateral transport.
   These data are the most difficult to interpret.
   - First, T3 is hydrosoluble and would probably migrate without active transport.
   - The authors do not demonstrate that these specific neuronal populations contain Mct8, and that these observations are connected to the previous in vitro observation (which used cortical neurons prepared from the fetus). The possibility that astrocytes are involved, as reported in the literature, is not considered.
   - Here again, using Mct8KO mice would greatly help to interpret the data. In particular, the experiments with cold T3 involve a 48h delay which is very long in comparison to the 30 minutes required for long-distance transfer of radioactive T3.
   Discussion
   Considering the diversity of questions that are addressed in the study, it is not surprising that the discussion is not covering all aspects. The authors implicitly consider that their conclusions can be extended to all neurons, while they use in their experiments a variety of different populations coming from either the fetal cortex, hippocampus, adult cortex, or hypothalamus. The claim that they discovered a mechanism applying to all neurons is not supported by the data. Some highly relevant literature is not cited. In particular:
   - Mct8 KO mice do not have a marked brain hypothyroidism (PMID: 24691440) which at least suggests that the pathway discovered by the authors can be efficiently compensated by alternative pathways.
   - Dio3 KO only increases T3 signaling in a few areas of the brain and only in the long term (PMID: 20719855).
   - Anterograde transport of T3 has been reported for some brainstem neurons (PMID: 10473259)

   Read the original source
5. 

   eLife

   Mar 14, 2023

   Reviewer #2 (Public Review):

   Salas-Lucia et al. investigated two main questions: whether the Thr92Ala-DIO2 mutation impairs brain responsiveness to T4 therapy under hypothyroidism induction and the mechanisms of neuronal retrograde transport of T3. They find that the Thr92Ala-DIO2 mutation reduces T4-initiated T3 signaling in the hippocampus, but not in other brain regions. Using neurons cultured in microfluidic chambers, they further describe a novel mechanism for retrograde transport of T3 that depends on MCT8 and endosomal loading (possibly protecting T3 from D3-mediated cytosolic degradation) and microtubule retrotransport. Finally, they present evidence of retrograde transport of T3 through hypothalamic projections and interhemispheric connections in vivo. The main novelty of this study is the delineation of the mechanism of T3 …

   Reviewer #2 (Public Review):

   Salas-Lucia et al. investigated two main questions: whether the Thr92Ala-DIO2 mutation impairs brain responsiveness to T4 therapy under hypothyroidism induction and the mechanisms of neuronal retrograde transport of T3. They find that the Thr92Ala-DIO2 mutation reduces T4-initiated T3 signaling in the hippocampus, but not in other brain regions. Using neurons cultured in microfluidic chambers, they further describe a novel mechanism for retrograde transport of T3 that depends on MCT8 and endosomal loading (possibly protecting T3 from D3-mediated cytosolic degradation) and microtubule retrotransport. Finally, they present evidence of retrograde transport of T3 through hypothalamic projections and interhemispheric connections in vivo. The main novelty of this study is the delineation of the mechanism of T3 retrograde transport in neurons. This is interesting from the cell biology perspective. The notion of impaired hippocampal T3 signaling is relevant for the cognitive outcomes of hypothyroidism and its associated therapy. Although the data are exciting and relevant for the community, some issues need to be addressed so that conclusions are more clearly justified by data:

   1. The title and the abstract mean that dissecting this novel mechanism of T3 retrograde transport may help improve cognition or brain responsiveness in patients taking T4 or L-T3 therapy. However, how initial results (Figs 1 and 2) connect to later data is not essentially clear. For example, do Thr92Ala-DIO2 mice present altered retrograde transport of T3? Would stimulation of retrograde transport in Thr92Ala-DIO2 mice rescue neurological phenotypes? Can the authors address this experimentally?

   2. Although the authors present in vivo evidence of retrograde T3 transport in the hypothalamus and motor cortex, given the select susceptibility of the hippocampus to hypothyroidism, it would be especially interesting to test whether this mechanism also happens in a hippocampal circuit (CA3-CA1 Schaffer collaterals, mossy fibers or perforant pathway).

   3. Table 1 should present the raw values for Ala92-DIO2 mice and treatments instead of only displaying the direction of change and statistical significance. From Panels 1E-J, it is unclear if Thr92Ala-DIO2 mice or treatments caused any real change in brain regions other than the hippocampus.

   4. The authors put forward the notion that a rapid nondegradative endosome/lysosome incorporation protects T3 from D3 degradation in the cytosol. Their experiments with pharmacological modulation of MCT8, lysosomes, and microtubules are in this direction. However, they do not represent an unequivocal demonstration of this mechanism. Therefore, the authors should be more cautious in their interpretation and discuss the limitations of their approaches.

   Read the original source
6. 

   eLife

   Mar 14, 2023

   Reviewer #3 (Public Review):

   Initially, Salas-Lucia et al examined the effect of deiodinase polymorphism on thyroid hormone-medicated transcription using a transgenic animal model and found that the hippocampus may be the region responsible for altered behavior. Then, by changing to topic completely, they examined T3 transport through the axon using a compartmentalized microfluid device. By using various techniques including an electron microscope, they identified that T3 is uptaken into clathrin-dependent, endosomal/non-degradative lysosomes (NDLs), transported in the axon to reach the nucleus and activate thyroid hormone receptor-mediated transcription.

   Although both topics are interesting, it may not be appropriate to deal with two completely different topics in one paper. By deleting the topic shown in Table 1, Figure 1, and Figure …

   Reviewer #3 (Public Review):

   Initially, Salas-Lucia et al examined the effect of deiodinase polymorphism on thyroid hormone-medicated transcription using a transgenic animal model and found that the hippocampus may be the region responsible for altered behavior. Then, by changing to topic completely, they examined T3 transport through the axon using a compartmentalized microfluid device. By using various techniques including an electron microscope, they identified that T3 is uptaken into clathrin-dependent, endosomal/non-degradative lysosomes (NDLs), transported in the axon to reach the nucleus and activate thyroid hormone receptor-mediated transcription.

   Although both topics are interesting, it may not be appropriate to deal with two completely different topics in one paper. By deleting the topic shown in Table 1, Figure 1, and Figure 2, the scope of the manuscript can be more clear.

   Their finding showing that triiodothyronine is retrogradely transported through axon without degradation by type 3 deiodinase provides a novel pathway of thyroid hormone transport to the cell nucleus and thus can contribute greatly to increasing our understanding of the mechanisms of thyroid hormone action in the brain.

   Read the original source
7. 

   Version published to 10.1101/2022.08.17.504300v1 on bioRxiv

   Aug 18, 2022