The manganese transporter SLC39A8 links alkaline ceramidase 1 to inflammatory bowel disease

The metal ion transporter SLC39A8 is associated with physiological traits and diseases, including blood manganese (Mn) levels and inflammatory bowel diseases (IBD). The mechanisms by which SLC39A8 controls Mn homeostasis and epithelial integrity remain elusive. Here, we generate Slc39a8 intestinal epithelial cell-specific-knockout (Slc39a8-IEC KO) mice, which display markedly decreased Mn levels in blood and most organs. Radiotracer studies reveal impaired intestinal absorption of dietary Mn in Slc39a8-IEC KO mice. SLC39A8 is localized to the apical membrane and mediates 54Mn uptake in intestinal organoid monolayer cultures. Unbiased transcriptomic analysis identifies alkaline ceramidase 1 (ACER1), a key enzyme in sphingolipid metabolism, as a potential therapeutic target for SLC39A8-associated IBDs. Importantly, treatment with an ACER1 inhibitor attenuates colitis in Slc39a8-IEC KO mice by remedying barrier dysfunction. Our results highlight the essential roles of SLC39A8 in intestinal Mn absorption and epithelial integrity and offer a therapeutic target for IBD associated with impaired Mn homeostasis.

3. In Figure 4, the mRNA data on selective genes for TJ proteins are not sufficient to show the changes of TJs.The authors need provide western blots and immunostaining data to show the protein levels and locations of TJ proteins.Claudin-2 is a "leaky protein" known increased in human IBD and colitis models.Why the mRNA level of Claudin2 decreased more in the DSS treated KO mice, compared to the controls.4. The Fig. 5C-F, the authors need provide the data of Control and KO mice with or without DSS treatment.5. Fig. 6.The protein expression data of ACER1 is needed for validation of ACER1 expression.6. Fig. 7.The authors need provide data of protein levels of TJ proteins, e.g.ZO-1 and Claudins, in addition to the PCR data.7. Will ACER1 inhibitor rescue the IL-10 KO mice? 8.More discussion should be added on human IBD if the authors believe that Mn deficiency upregulates Acer1 expression in the DSS-colitis model.9.A working model of ACER1, Mn deficiency, and barrier in the development of chronic intestinal inflammation will help the readers.

Reviewer #2 (Remarks to the Author):
Choi et al. have submitted a solid manuscript that interrogates the role of the manganese transporter Slc39a8 in intestinal manganese absorption and inflammatory bowel disease using cell culture and mouse models.The work is highly original and will have a prominent impact on the field of manganese biology and intestinal disease.We do have several recommendations on how the manuscript can be improved.These are described here: -We highly recommended separating males and females for each panel instead of pooling them together.Even though no sex-specific differences are observed for Mn levels in Fig. 1, this doesn't necessarily mean such differences do not occur for other phenotypes.For instance, in Fig. S4, the authors show that female mice have lower sensitivity to effects of DSS.Additionally, the authors point out in the discussion that "inconsistent Mn concentrations among the two studies (#39,40) [of Slc39a8 A391T KI mice] complicate any interpretations of how SLC39A8 deficiency alters Mn homeostasis and exacerbates DSS-induced colitis"-could sex-specific differences contribute to the inconsistencies in these studies?-This study used a mixture of three genotypes as a control (Slc39a8-flox (Slc39a8fl/fl and Slc39a8fl/+), Slc39a8-WT (Slc39a8+/+), and Vil-Cre (Slc39a8+/+:Villin-Cre)).Authors should not use four different genotypes as controls.This is a key issue that needs to be addressed.
-For the 54Mn gavage experiments, the authors collected blood 15 minutes after gavage and tissues four hours after gavage.Four hours is more than enough time for 54Mn to have undergo biliary excretion and enterohepatic circulation.Can the authors comment on why tissues weren't harvested at the same time as blood and how this discrepancy would influence their results?-It appears that the confocal immunofluorescence images in Fig. 3b are misaligned-ZO-1 staining crosses through nuclei!-Were there any changes in expression of Dmt1 or other relevant metal transporters in the intestines of the Slc39a8 IEC KO mice?-Ceramide accumulation is known to mediate inflammation (https://www.nature.com/articles/nm1748).ACER1 inhibition will result in ceramide accumulation according to the pathway shown in Fig. 6c, so would ACER1 inhibition be expected to exacerbate colitis?(Also, alkaline ceramidase 3 deficiency aggravates colitis and colitis-associated tumorigenesis in mice by hyper-activating the innate immune system https://www.nature.com/articles/cddis201636).Please comment on this.
-Did the authors measure ceramide and sphingosine levels in cell culture and mouse models with and without Acer1 inhibition?This would help to confirm the physiologic relevance of altered Acer1 expression and the efficacy of Acer1 inhibition.
-The authors state that the increased spleen mass in Slc39a8 IEC KO mice implicates splenic macrophage infiltration.Was splenic macrophage infiltration observed by histology?-The histological images included in figures are quite small.The authors are recommended to include images of the same magnification and large enough to visualize differences between images.
-It is striking that only four genes were found to be differentially expressed in RNA seq analysis.Can the authors confirm that this is correct?Also, do the genes other than Acer1 have potential implications for the study at hand?What is known about these other genes?These issues need to be addressed.
-It appears that a legend is missing from Fig. 9B.
-The authors are recommended to include a model figure summarizing their results. 1

Response to reviewers
We thank the reviewers for the thoughtful comments and suggestions.As detailed in a pointby-point response below, we have addressed all of them with additional experiments, and we feel that the manuscript has been substantially strengthened as a result.

Reviewer #1
The authors presented comprehensive epithelial functions associated with the risk of inflamed intestine.This study is interesting and provides new insights into the involvements of the Mn homeostasis and epithelial integrity in the development and therapy of IBD.There are some comments for further improving the current manuscript.
We sincerely appreciate the reviewer's positive feedback and acknowledgment of the significance of our study.We value the reviewer's comments and suggestions for further improving our manuscript.

1.
Figure 1.Only female mice were used in Fig. 1A and only male mice were used in Fig. 1B.Please explain any gender differences observed in the mouse model?
We appreciate the reviewer's comment and the opportunity to address it.We conducted additional experiments using both female and male mice to investigate potential gender differences in Slc39a8 expression and localization in the intestines.We performed qPCR analysis of Slc39a8 expression in male mice, in addition to the female mice analyzed in Fig. 1a.The results from male mice showed similar patterns of Slc39a8 expression, with higher levels observed in the distal small intestine and colon (Fig. 1a).We also performed immunofluorescent Slc39a8 staining in female mice, in addition to the male mice analyzed in Fig. 1b.The results from female mice revealed a similar localization of Slc39a8 in the apical membrane of enterocytes (Fig. 1b).These results indicate no sex-specific differences in Slc39a8 expression and localization in the intestines.These new data have been added to Fig. 1a and Fig. 1b and are indicated in the text and Figure legends.

2.
For Figure 4F This is another important point.Our new analysis showed that, without DSS treatment, no significant differences were observed in H&E staining (Fig. 4f), total pathological score (Fig. 4g), 4 kDa FITC-dextran permeability (Fig. 4h), qPCR quantification of tight junction markers (Fig. 4i), or proinflammatory cytokine and chemokine expression (Fig. 4l) between control and Slc39a8-IEC KO mice.We have updated Fig. 4f, 4g, 4h, 4i, and 4l, and indicated these findings in the text and Figure legends.

3.
In Figure 4, the mRNA data on selective genes for TJ proteins are not sufficient to show the changes of TJs.The authors need provide western blots and immunostaining data to show the protein levels and locations of TJ proteins.Claudin-2 is a "leaky protein" known increased in human IBD and colitis models.Why the mRNA level of Claudin2 decreased more in the DSS treated KO mice, compared to the controls?
We agree with the reviewer and have now provided immunofluorescence and immunoblot blot analysis of tight junction proteins.The expression levels and localization of the tight junction proteins were examined by immunofluorescence.Specifically, Zo1, Zo2, Cldn2, Cldn5, and Ocln were observed at the apical side of the colonic epithelium, whereas Cldn3 and Cldn7 were found at the lateral membrane in control mice (Fig. 4j and Supplementary Fig. 3e).The localization and expression of these tight junction proteins were severely impaired by DSS treatment (Fig. 4j).Notably, the expression levels of Cldn3 and Cldn5 were significantly reduced in the Slc39a8-IEC KO mice after DSS treatment (Fig. 4j).Immunoblotting analysis confirmed the diminished levels of Cldn3 and Cldn5 (Fig. 4k).These data indicate a selective impact of Slc39a8 on the transcripts and proteins related to tight junctions.These new data have been included in Fig. 4j, 4k, and Supplementary Fig. 3e, and the relevant information has been added to the text and Figure legends.
As the reviewer points out, Claudin 2 protein levels generally increase in inflammatory states in human tissues 1,2 .By contrast, we observed a reduced mRNA level of Claudin 2 in DSS-treated Slc39a8-IEC KO mice compared to the controls (Fig. 4i).To address this question, we carefully performed additional experiments.In these new experiments, we specifically evaluated the Claudin 2 protein levels instead of mRNA expression to obtain more conclusive results.
We thought that the apparent Claudin 2 decrease was due to DSS-induced epithelial cell loss.To test this, we normalized the Claudin 2 western blot signal against the level of Keratin, an epithelial marker.This analysis indicated that the normalized Claudin 2 levels were still decreased in control mice after DSS treatment (Fig 4k and Supplementary Fig. 3f).The Slc39a8-IEC KO mice exhibited a more pronounced reduction of Claudin 2 levels following DSS treatment, although the difference did not reach statistical significance (Supplementary Fig. 3f).We also performed immunofluorescence labeling to visualize the localization and expression of Claudin 2. Our immunofluorescence analysis demonstrated impaired localization and reduced expression of Claudin 2 in the control mice following DSS treatment (Fig. 4j and Supplementary Fig. 3e), with more pronounced effects observed in Slc39a8-IEC KO mice, although the differences again did not reach statistical significance (Fig. 4j and Supplementary Fig. 3e).
While we cannot definitively explain the reasons for the reduced levels of Claudin 2 in both control and Slc39a8-IEC KO mice after DSS treatment, we have found supporting evidence from similar studies that demonstrate the downregulation of Claudin 2 in wild-type C57BL/6 mice following DSS treatment 3,4 .The discrepancy could stem from species differences or variations in experimental conditions, and this merits future investigation.Overall, these additional experiments and observations highlight the complex regulation of Claudin 2 in the context of DSS-induced inflammation.We have included these additional data in Fig. 4j, 4k, and Supplementary Fig. 3f, 3e and referenced them in the text and Figure legends accordingly.

4.
The Fig. 5C-F, the authors need provide the data of Control and KO mice with or without DSS treatment.
Our new analysis showed that, in the absence of DSS treatment, no significant differences were observed in colon length (Fig. 5c), survival percentage (Fig. 5d), H&E staining (Fig. 5e), or total pathological scores (Fig. 5f) between control and Slc39a8-IEC KO mice.We added these data to Fig. 5c, 5d, 5e, and 5f and indicated them in the text and Figure legends.

5.
Fig. 6.The protein expression data of ACER1 is needed for validation of ACER1 expression.
We conducted immunoblot analysis of ACER1 in the ileum and colon of both control and Slc39a8-IEC KO mice, as well as in the enteroid and colonoid monolayers derived from these mice.The immunoblotting results confirmed a significant increase in Acer1 protein levels in the ileum (~1.4 fold, P < 0.05) and colon (~1.3 fold, P < 0.05) of Slc39a8-IEC KO mice and the enteroid monolayers (~4.5 fold, P < 0.01) of Slc39a8-IEC KO mice.Although the increases in the levels of the Acer1 protein in the colonoid monolayers (1.3 fold) of Slc39a8-IEC KO mice did not reach statistical significance, the smaller changes in the colon were consistent with the RNA-seq results.We have added these new data to the manuscript as Fig. 6e and 6f and indicated the changes in the text and figure legends.

6.
Fig. 7.The authors need provide data of protein levels of TJ proteins, e.g.ZO-1 and Claudins, in addition to the PCR data.
In response to this comment, we conducted immunoblot analysis to assess the protein levels of tight junction proteins ZO-1, Claudin 3 (Cldn3), Claudin 5 (Cldn5), and Claudin 7 (Cldn7) in the enteroid monolayers derived from Slc39a8-IEC KO mice and control mice.The immunoblot analysis confirmed that D-e-MAPP treatment enhanced levels of tight junction proteins ZO-1, Cldn5, and Cldn7 in the enteroid monolayers derived from Slc39a8-IEC KO mice.A similar trend was observed for the Cldn3 protein levels, although the difference did not reach statistical significance.We have added these new data to the manuscript as Fig. 7e and indicated the changes in the text and figure legends.
We intended to do this experiment; however, for unknown reasons, IL-10 KO mice do not develop spontaneous colitis in our animal facility.Consequently, we were unable to directly test the effect of the ACER1 inhibitor in rescuing colitis in IL-10 KO mice.However, based on our findings that ACER1 is upregulated in the intestines of Slc39a8-IEC KO mice and that ACER1 inhibitor mitigates colitis in these mice, we speculate that if IL-10 KO mice had displayed a deficiency in Slc39a8 and Mn levels resulting in up-regulation of ACER1, that treatment with an ACER1 inhibitor could potentially attenuate colitis in IL-10 KO mice.Further studies would be required to investigate this hypothesis.

8.
More discussion should be added on human IBD if the authors believe that Mn deficiency upregulates Acer1 expression in the DSS-colitis model.
Our findings in mice align with epidemiological studies that have reported a positive relationship between Mn deficiency and the risk of IBD.For instance, a recent survey of pediatric patients newly diagnosed with Crohn's disease and ulcerative colitis showed significantly lower Mn levels in their hair samples 5 .Another human study reported that blood Mn levels are significantly lower in patients with active Crohn's disease and ulcerative colitis than in patients in remission 6 .These findings suggest that IBD patients may be at risk of Mn deficiency, leading to potential upregulation of Acer1 expression.Therefore, interventions aimed at restoring ACER1 levels may hold promise for improving human IBD conditions.We have further elaborated on this point in the Discussion section, as indicated by the yellow-highlighted text.

9.
A working model of ACER1, Mn deficiency, and barrier in the development of chronic intestinal inflammation will help the readers.
We appreciate the reviewer's suggestion and have included a working model in our revised manuscript to provide readers with a visual summary of the proposed mechanism involving ACER1, Mn deficiency, and barrier function in the development of chronic intestinal inflammation (Fig. 9i).

Choi et al. have submitted a solid manuscript that interrogates the role of the manganese transporter
Slc39a8 in intestinal manganese absorption and inflammatory bowel disease using cell culture and mouse models.The work is highly original and will have a prominent impact on the field of manganese biology and intestinal disease.We do have several recommendations on how the manuscript can be improved.These are described here: We appreciate the positive feedback from the reviewer regarding our manuscript.

1.
We highly recommended separating males and females for each panel instead of pooling them together.Even though no sex-specific differences are observed for Mn levels in Fig. 1, this doesn't necessarily mean such differences do not occur for other phenotypes.For instance, in Fig. S4, the authors show that female mice have lower sensitivity to effects of DSS.
We have now included separate data for both male and female mice for Slc39a8 expression (Fig. 1a and 1b), tissue Mn levels (Fig. 1d), and Mn radiotracer assays (Fig. 2).In these measurements, we did not find significant sex differences.Thus, the lower sensitivity of female mice to DSS-induced colitis is not caused by differences in Slc39a8 expression or Mn levels between sexes.Consequently, for further analysis, we focused on male mice, which resulted in the inability to separate males and females in other panels.

2.
Additionally, the authors point out in the discussion that "inconsistent Mn concentrations among the two studies (#39,40) [of Slc39a8 A391T KI mice] complicate any interpretations of how SLC39A8 deficiency alters Mn homeostasis and exacerbates DSS-induced colitis"could sex-specific differences contribute to the inconsistencies in these studies?
We appreciate the reviewer's question.As discussed in the manuscript, two recent studies independently reported altered Mn homeostasis and exacerbated DSS-induced colitis in Slc39a8 A391T KI mice 7,8 .Please see the summary table below comparing tissue Mn levels in Slc39a8 A391T knock-in (KI) mice from two studies.Sunuwar et al. 5 reported reduced Mn levels in whole blood, liver, and kidney of Slc39a8 A391T KI mice but no alterations in other tissues, including the distal small intestine, colon, lung, heart, and brain.Nakata et al. 8 reported reduced Mn levels in whole blood, liver, and colon of Slc39a8 A391T KI mice, but they did not examine Mn levels in other tissues 8 .The colon Mn data between the two studies were inconsistent.However, whole blood is the only cell type for which the two reports measured Mn levels in both sexes; Mn in other tissues was only measured in male animals.In whole blood, no sex difference was detected.Therefore, it is difficult to conclude whether the inconsistent colon Mn data reflect the sexes examined in the two studies.We have edited the Discussion section accordingly in the yellow-highlighted text.
This is an important point.We took great care to compare key experimental results, such as metal concentrations and experimentally induced colitis models, among all four control genotypes.We found no significant differences in Mn concentrations in the ileum and colon among the four control genotypes at 6-8 weeks of age (Supplementary Fig. 9a).We also observed that the four genotypes of controls did not display significant differences in DSS sensitivity at 6-8 weeks of age (Supplementary Fig. 9b).These data indicate that the flox sequence and Cre transgene did not impact Mn homeostasis or DSS sensitivity.We have added this information to Supplementary Fig. 9a and 9b, and the Animals in the Methods section.

4.
For the 54Mn gavage experiments, the authors collected blood 15 minutes after gavage and tissues four hours after gavage.Four hours is more than enough time for 54Mn to have undergo biliary excretion and enterohepatic circulation.Can the authors comment on why tissues weren't harvested at the same time as blood and how this discrepancy would influence their results?
We appreciate the reviewer's comment.We chose the two different time points for assaying the blood and other organs based on prior studies in which we monitored the relatively fast entry of Mn into blood circulation compared to the entry into tissues 9,10 .The observed 54 Mn levels in the tissues after a 4 h interval from gavage or intravenous injection may reflect enhanced Mn excretion.However, we found a significant reduction in transcript levels of intestinal Mn excretion proteins, Slc39a14/ZIP14 and Slc30a10/ZnT10, in Slc39a8-IEC KO-derived enteroids (Supplementary Fig. 3a), thereby limiting the possibility of enhanced excretion of Mn, at least in the intestine.Future studies are warranted to determine the contribution of absorption and excretion of Mn in each tissue.We have added this information to the Results section and in the yellow-highlighted text.

5.
It appears that the confocal immunofluorescence images in Fig. 3b are misaligned-ZO-1 staining crosses through nuclei!
We appreciate the reviewer's careful observation.We have provided new confocal immunofluorescence images in Fig. 3b to ensure an accurate representation of our data.

6.
Were there any changes in expression of Dmt1 or other relevant metal transporters in the intestines of the Slc39a8 IEC KO mice?
We did provide the data on these changes in Supplementary Fig. 3a, in the original manuscript, although this may not have been easily recognizable.Our results showed that the Slc39a8-IEC KO-derived enteroids exhibited significantly increased Slc11a2/DMT1 (divalent metal transporter 1) transcript levels, along with significantly decreased Slc39a14/ZIP14, Slc30a10/ZnT10, and Slc40a1/FPN (ferroportin-1) transcript levels.These findings suggest that DMT1 may compensate for the loss of SLC39A8 expression in the intestines of the Slc39a8-IEC KO mice.

7.
Ceramide accumulation is known to mediate inflammation (https://www.nature.com/articles/nm1748).ACER1 inhibition will result in ceramide accumulation according to the pathway shown in Fig. 6c, so would ACER1 inhibition be expected to exacerbate colitis?(Also, alkaline ceramidase 3 deficiency aggravates colitis and colitis-associated tumorigenesis in mice by hyper-activating the innate immune system https://www.nature.com/articles/cddis201636).Please comment on this.
Thank you for bringing up this important point.Given the similarity in content between your questions 7 and 8, I have provided a consolidated response that addresses the common points raised in both questions, thereby avoiding redundancy and ensuring clarity in the feedback provided.
To determine the mechanisms underlying the upregulation of ACER1 and the rescue effect of D-e-MAPP, we performed global, unbiased lipidomics profiling of the intestine from WT and mutant mice with or without the inhibitor using triple time of flight liquid chromatography-mass spectrometry (Triple-TOF LC-MS).While ACER1 converts ceramide into sphingosine, sphingomyelin synthase (SMS1) converts ceramide into sphingomyelin; therefore, these two enzymes represent the two major arms of ceramide metabolism (Fig. 8j).The lipidome profiling identified 40 differentially regulated lipids in Slc39a8-IEC KO intestines compared to control (Padj < 0.2, Supplementary Fig. 7a).Among the altered lipids, the heatmap (Fig. 8k) represents lipid species relevant to ceramide metabolism, including sphingomyelins, hexosylceramides, sphinganine, phosphatidylethanolamine ceramides, and ceramides.While we predicted a reduction in ceramides following ACER1 upregulation, the reduction was only observed for Cer 38:2;2O and Cer 43:1;2O, whereas other four ceramide species were upregulated in the Slc39a8-IEC KO intestines (Supplementary Fig. 7b-g).
Notably, sphingomyelins were all significantly downregulated in the mutant (Fig. 8m), which is consistent with the fact that Mn is necessary for the activity of ceramide phosphoethanolamine synthase (Cpes) 11 , the insect homolog of SMS1 12 .Indeed, SMS activity was lower in Slc39a8-IEC KO intestines compared to control intestines (Fig. 8l).Furthermore, all three sphingomyelin species showed partial or complete restoration of their abundance upon D-e-MAPP treatment; the inhibitor alone did not affect the sphingomyelin levels (Fig. 8m).By contrast, we did not find any ceramide species whose changes completely agreed with the phenotypic outcomes; i.e., dysregulation in Slc39a8-IEC KO, restoration by D-e-MAPP and no change by inhibitor alone (Supplementary Fig. 7b-g); the six KO-dysregulated ceramides largely satisfied the first two criteria, but D-e-MAPP alone also changed their levels in the control intestine, ruling them out as the mediator for gut phenotype rescue effects.

These results led us to propose the following mechanisms underlying Acer1 upregulation in
Slc39a8-IEC KO intestines and the phenotypic rescue by D-e-MAPP.First, the reduced SMS1 activity due to Mn deficiency led to reduced sphingomyelin and increased ceramide levels, thereby triggering the compensatory increase of Acer1 expression to offset the ceramide increase.The increased Acer1 expression indirectly led to alterations in lipid composition, including the upregulation of some ceramide species and dysregulation of other detected lipids.The phenotypic rescues by D-e-MAPP (Figs. 7, 8, and 9 and Supplementary Fig. 6) suggest that the upregulation of Acer1 and associated lipidome alterations contribute to the impaired gut function in Slc39a8-IEC KO mice.However, the rescue effect is unlikely to be mediated by ceramides; instead, the restoration of other lipid species, such as sphingomyelin, is the candidate mediator for the observed impact of D-e-MAPP on gut physiology.We have incorporated these new data into the manuscript as Fig. 8j, 8k, 8l, and 8m and Supplementary Fig. 7a, 7b, 7c, 7d, 7e, 7f, and 7g, and have indicated the corresponding changes in the text and figure legends.
Regarding ACER3, we conducted additional experiments to measure the mRNA levels of the other ceramidases, including acid ceramidase (Asah1), neutral ceramidase (Asah2), and alkaline ceramidases 1, 2, and 3 (Acer1, Acer2, and Acer3).Our qPCR analysis revealed no significant differences in the mRNA levels of Asah1, Asah2, Acer2, and Acer3 in the ileal and colonic mucosa from Slc39a8-IEC KO mice and control mice.However, we observed a significant upregulation of Acer1 expression in the ileal and colonic mucosa from Slc39a8-IEC KO mice compared to control mice, which is consistent with our findings shown in Fig. 6c and 6d.These new results have been included in the Supplementary Fig. 8, and the relevant information has been incorporated into the text and Supplemental Figure legends.

8.
Did the authors measure ceramide and sphingosine levels in cell culture and mouse models with and without Acer1 inhibition?This would help to confirm the physiologic relevance of altered Acer1 expression and the efficacy of Acer1 inhibition.
Please see the above response to your question 7. Regarding the lipid levels in the cell cultures, we indeed attempted the lipid measurements; however, we were not able to measure the lipid levels reliably due to the low amounts of materials.

9.
The authors state that the increased spleen mass in Slc39a8 IEC KO mice implicates splenic macrophage infiltration.Was splenic macrophage infiltration observed by histology?
To address this, we performed H&E staining of the spleen tissues from both control and Slc39a8 IEC KO mice.Indeed, our histological analysis revealed increased follicle destruction and infiltration of inflammatory cells in the spleens of Slc39a8 IEC KO mice compared to the control, as demonstrated in the newly added data in Supplementary Fig. 3c and indicated in the text and Supplementary Figure legends.

10.
The histological images included in figures are quite small.The authors are recommended to include images of the same magnification and large enough to visualize differences between images.
We appreciate the reviewer's feedback.We have made efforts to ensure that the images are clear and representative of the observed differences between the groups (Fig. 4f, Fig. 5e, Fig. 8e, Fig. 9g, and Supplementary Fig. 4d).We also revised the figure legends to clearly indicate the magnification of the images for better understanding.

11.
It is striking that only four genes were found to be differentially expressed in RNA seq analysis.Can the authors confirm that this is correct?Also, do the genes other than Acer1 have potential implications for the study at hand?What is known about these other genes?
These issues need to be addressed.
Yes, only four genes reached our statistical threshold (Padj < 0.1) with DEseq2 analysis.We now include the full results for the DEseq2 analysis as Supplementary Material.Among these four genes, Slc39a8 was the only downregulated gene in the Slc39a8-IEC KO intestines.The three upregulated genes are Ighv1-55 (ENSMUSG00000095589), Entpd4b, and Acer1.
Ighv1-55 has no known function, while Entpd4b encodes ectonucleoside triphosphate diphosphohydrolase 4 (ENTPD4), a member of the apyrase protein family involved in the hydrolysis of nucleotide diphosphates and triphosphates 13 .Acer1 encodes alkaline ceramidase 1 (ACER1), which catalyzes the hydrolysis of very long chain ceramides to generate sphingosine 14,15 .Given the significant association between sphingolipid metabolism and 16 , we selected to further investigate ACER1 in our study.
We have included a new table in the manuscript (Supplemental Table 1) that provides detailed information on these four genes, including gene description, p-value, direction of misregulation, known function, and references.This information is also incorporated in the revised Discussion section, with the relevant text highlighted in yellow.1.The four most dysregulated genes in Slc39a8-IEC KO intestines.

Acer1
Alkaline Ceramidase 1 4.65 X 10 -2 Hydrolysis of very long chain ceramides to generate sphingosine 14,15 12.It appears that a legend is missing from Fig. 9B.
We have added a legend to Fig. 9b, and the corresponding text has been highlighted in yellow.

13.
The authors are recommended to include a model figure summarizing their results.
We have included a model figure (Fig. 9i) in our revised manuscript to summarize our results concisely.

14.
The authors suggest that Mn deficiency reduces Cpes activity in the discussion about how Slc39a8 IEC KO mice develop misregulated Acer1 gene expression.Can the authors

3.
There are also studies using alkaline ceramidase 3 (ACER3) deficient mice in colitis and colitis associated cancer.This is not included in the discussion, but likely should be.
Thank you for your valuable input.We have added the following paragraph to the Discussion section as suggested.
In addition to Acer1, two other members of the alkaline ceramidase family, Acer2 and Acer3, are present in both mice and humans 15,21 .D-e-MAPP can inhibit ACER2 and ACER3 22,23 ; however, since we did not find any expression changes in Acer2 and Acer3 in Slc39a8-IEC KO mice (Supplementary Fig. 8), the rescue effect observed with D-e-MAPP was likely due to ACER1 inhibition.Nevertheless, the lack of expression changes still does not rule out a potential involvement of Acer2 and Acer3.A previous study has demonstrated that ACER3 plays a crucial role in mediating the immune response in innate immune cells and colonic epithelial cells 24 .Acer3 deficiency has been shown to increase the local and systemic production of proinflammatory cytokines and exacerbated colitis in the DSS-induced murine colitis model 24 .These observations indicate that Acer activities are tightly regulated to ensure normal gut functions.

1.
We appreciate the effort made by the authors to address sex-specific differences in phenotypes.However, in Fig 2b (and others) where authors separated sex of mice, it is not prudent to run statistics on data with N=2.Ideally authors should have included three or more mice per sex.
Because we do not have enough N in Fig. 2b, we decided not to present sex-separated statistical analysis.Investigating of the sex differences in intestinal absorption of 54 Mn is a potential subject for future research.

2.
Please include full uncropped blots for immunoblots as supplemental data.
In compliance with the journal's requirements, we have included the full, uncropped immunoblots in the source data file, not in supplemental data.

Reviewer #3:
The authors have significantly improved the quality of the manuscript and included several new and necessary experiments.

1.
However, their untargeted lipidomic studies are insufficient to draw the conclusions outlined in the manuscript.Untargeted lipidomics provides compositional data, not quantitative assessments of lipid levels.Several of the SM and Ceramide species that were identified in the untargeted method are not "canonical" sphingolipids and therefore may have been mis-identified.Moreover, untargeted lipidomics is often not sensitive enough to measure sphingosine -the produce of ACER1.Targeted lipidomic measurements for SM, CER, sphingoid bases and HexosylCer would significantly improve the validity of the conclusions drawn from the data.
We greatly appreciate your thoughtful comments and the opportunity to clarify our approach to untargeted lipidomics in response to your concerns.We would like to address each of your points as follows: In our current manuscript, lipids were identified using the MS/MS product ion fragmentation of molecular ions (m/z range 200-1200) resolved by RP-UPLC-ESI-QTOF-MS/MS.The method is highly sensitive and reproducible even for profiling low abundance lipids.We acknowledge the reviewer's concern that compound identification in untargeted lipidomics methods can be prone to error when MS/MS search results alone are used and trusted without further review.However, we do not find that identification in untargeted lipidomics is inherently flawed when an appropriate review of identifications is undertaken.
First, regarding the MS/MS spectral search strategy for initial compound identification, we used the "LipidBlast" library coupled with National Institute of Standards and Technology (NIST) spectral search tools, which matches compounds by observed mass and MS/MS fragmentation pattern.To facilitate accurate lipid identification, a more stringent mass error rate of 0.001 m/z for the positive mode and 0.005 m/z for the negative mode was determined based on the mass accuracy of internal standards to reduce the rate of false positives and identify likely correct candidates.The sphingolipid species considered for identification were limited to features with (%RSD < 20) in pooled samples.More details regarding the spectral matching procedure used for compound identification against the LipidBlast library are outlined in the original report 1 .
A manual review of spectral matches was performed in all potentially uncertain cases, including the sphingolipid species questioned by the reviewers.Examples of spectral matches are now included in Supplementary Figures 10a and 10b.The product ion scan of [M+Na] + species of sphingomyelin (SM 34:1, 2O) yields three fragments corresponding to NL of C3H9N (m/z 666.477,C36H70NO6PNa+), NL of C5H14NO4P (m/z 542.486,C34H65NO2Na + ), NL of C5H16NO5P (m/z 502.497,C34H64NO+), respectively.A product ion scan of m/z 300.2860, which corresponds to the [M+H] + ion for SPB 18:1;2O molecular species, was searched against the LipidBlast In-Silico library, revealing an exact match for the expected compound with high library scores (Supplementary Figure 10a).MS/MS spectra of ceramides in the positive mode were dominated by fragments from the sphingoid base (Supplementary Figure 10b).Collisional activation of ceramides in positive ion mode showed the characteristic product ions of m/z 264.2684 [C18H34N] + and 282.279 that were attributed to the loss of N-linked fatty acid and one and two water molecules, respectively.We also added the following data, according to the latest communication with the reviewer #3.The positive mode MSMS product ion spectrum of HexCer 18:0;3O/16:0;(2OH), which is likely one of the noncanonical ceramides questioned by the reviewer, is also illustrated in Supplementary Figure 10b.This compound shows an NL of C6H10O5, the hexose ring, results m/z 572.5232 (Supplementary Figure 10b), as well as other peaks that are m/z 554.5143 and m/z 264.2686 correspond to NL of C6H10O5 and H2O and SPB 18:0;O3 -3H2O respectively.
To provide additional evidence for the accuracy of our compound identifications, chromatographic retention time can be used in lipidomics performed with reversed-phase liquid chromatography.Lipids of a particular class (sphingolipids in this case) can be expected to follow a regular order of elution, wherein chromatographic retention time should increase with increasing alkyl chain length and decrease with increasing degree of desaturation (number of double bonds).Deviations from the observed elution pattern can highlight potentially misidentified compounds.To investigate these patterns with the sphingolipid species in question, we plotted sphingolipid RT against alkyl chain length and highlighted degree of desaturation using different color marker points.As seen in the scatter plot (Supplementary Figures 10c and 10d), observed retention time increases with increasing alkyl chain length, and RT decreases with increasing number of double bonds.All such plots follow a highly linear pattern for a given compound class and show no substantial RT deviation or any of the detected species, supporting the accuracy of our identifications.
Last, the reviewers raised the question regarding the validity of the quantitation of untargeted lipidomics data compared to targeted methods.We acknowledge that true absolute quantitation of lipids should only be performed using a targeted strategy with authentic standards and compoundmatching isotopically labeled internal standards.However, relative quantitative measurements are certainly valid from untargeted lipidomics methods, as has been demonstrated previously 2 .
Overall, it is our conclusion that with appropriate controls and a review of compound identification and method performance, the identification and relative quantitative analysis of lipids using a carefully managed untargeted workflow is not necessarily more prone to error than targeted methods.Each technique still has its advantages and disadvantages; targeted methods can achieve greater sensitivity under certain circumstances than untargeted ones, whereas untargeted ones enable the detection of un-predicted lipid species not included in a multiplexed selected reaction monitoring method.
Taken together, we included Supplementary Figures 10a-d in the revised version of the manuscript and incorporate the relevant information into the Methods section.

2.
In addition, ACER1 not only produces sphingosine, but SPH is the substrate for sphingosine kinases (SKs), which have also been extensively implicated in IBD.This should be added to the discussion and potentially mRNA levels/protein levels of SKs added to the data.
To address this point, we conducted additional experiments to assess the mRNA levels of sphingosine kinases (SphKs), specifically SphK1 and SphK2.Our qPCR analysis revealed a significant upregulation of Sphk1 in the ileum and colon of Slc39a8-IEC KO compared to control group (Supplementary Fig. 8f).However, no significant differences were observed in the mRNA levels of Sphk2 in the ileum and colon of both control and Slc39a8-IEC KO mice (Supplementary Fig. 8g).
We also conducted immunoblot analysis targeting SphK1 and SphK2 in the ileum and colon of both control and Slc39a8-IEC KO mice.We used two different antibodies for each protein-anti-SphK1 antibodies (Thermo Fisher, Cat.No. PA514068 and Abclonal, Cat.No. A0139) and anti-SphK2 antibody (Thermo Fisher, Cat.No. PA551064 and Abclonal, Cat.No. A6748).However, we were unable to detect SphK1 and SphK2 proteins at the predicted molecular weights from intestine tissues of both control and Slc39a8-IEC KO mice.The updated results of mRNA expression levels of Sphk1 and Sphk2 have been included in Supplementary Fig. 8f and 8g, and the relevant information has been incorporated into the revised text and Supplemental Figure legends.
As suggested, we have incorporated the following point into the Discussion accordingly.In addition to ACER1's role in sphingosine production, sphingosine serves as the substrate for sphingosine kinases (SphKs), extensively implicated in IBD.Three ceramidases, operating in distinct cellular compartments, convert ceramide into sphingosine, which is then phosphorylated by either sphingosine kinase 1 (Sphk1) or sphingosine kinase 2 (Sphk2) to form sphingosine-1-phosphate (S1P).Studies in the mouse small intestine have shown twofold higher SphK activity compared to the colon, with SphK1 being the predominant isoform in the colon 3 .In a rat model of intestinal inflammation, the 1-phosphorylated forms of ceramide (C1P) and sphingosine (S1P) increase with inflammation 4 .Elevated S1P levels are particularly relevant in IBD, observed not only in the intestine but also in the blood due to increased SphK1 expression 5 .Conversely, SphK2 deficiency has been shown to reduce the severity of IBD 6 .These studies collectively suggest the involvement of the SphK/S1P pathway in the development of IBD.In line with these observations, our Slc39a8-IEC KO mice exhibited increased Sphk1 mRNA expression in the ileum and colon, whereas Sphk2 expression levels remain unchanged (Supplementary Figure 8f and 8g), indicating a potential upregulation of the SphK1/S1P pathway.However, our lipidomics data indicate no changes in sphingosine, likely due to compensatory mechanisms.Future studies are needed to explore the intricate regulation of sphingosine metabolism in response to Slc39a8 deficiency.
We would like to thank our reviewers for all the thoughtful comments and suggestions, and we believe that these revisions have significantly improved the manuscript.
Sincerely yours, Young-Ah Seo , H&E data should include mice with and without DSS.The Fig. G-J should provide the data of Control and KO mice with or without DSS treatment.

Table .
Tissue Mn Levels of Slc39a8 A391T KI mice reported in the literature.