Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This study investigated the role of SLC41A1 in OA imbalance of magnesium. They found IA injection of magnesium improved cartilage repair. Systemic in vitro assays demonstrated that SLC41A1 drived M2 macrophage to M1 macrophage phenotype transition which releases MMPs and inflammatory factor and aggravates OA phenotype. Knocking down SLC41A1protects chondrocytes and maintained M2 macrophage phenotype.  Overall, the study is well designed. The results are convincing and clearly presented. The writing is also good.  Following are some of the comments for authors to address.

 

1. No line numbers and page numbers in the manuscript.
2. Page 2, line 7 from top for secondary paragraph cGAS-STING should be spelled out before abbreviation. Same as SLC41A1 in the third paragraph.
3. Methods:

OA induction: Authors said animals were euthanized at 8 weeks. Is this 8 week after Mg injection of MIA injection.  MIA should be spelled out.

What decalcification method should be stated.

2.10, Authors said stain with DAPI, but their anti-fade medium has DAPI. Please clarify.

4. Results:

3.1. Authors said Safranin O staining stained histone polysaccharide, this is not correct. Safranin O stains cartilage extracellular matrix glycosaminoglycan (GAG).

Both H&E staining and SO staining should be made brighter.  Authors did not check white balance.

Figure 1B H&E staining showed OA group has more chondrocytes despite cartilage surface damage while in Figure 1D, this group showed no COL2.  Probably not correct. MIA is a chondrocyte toxin, it can cause entire cartilage loss without cells. TUNEL apoptosis staining revealed very few positive cells in OA.  The fonts on the quantification graph are blurry.

Figure 1 legend: panel label A,B,C and so on should have “.” after each. Whether bold or not bold should be consistent. It is better to label each graph with a panel letter. What is mean print area % quantified for between E and F panel?

Figure 2 DAPI staining is so weak, it is not easy to determine whether iNOS and SLC1A1 located in cytoplasm or membrane? Insets highlighted positive cells will be helpful in the merged images.

Figure 2 legend also has panel labeling the same issue as Figure 1. Same as the other figures.

Discussion: Thorough

Conclusion is supported by data.

Author contribution is missing.

 

 

Author Response

1. No line numbers and page numbers in the manuscript.

Reply. Thank you very much for your comments. This was our careless omission. We have adjusted the manuscript per the journal template, and no line numbers are retained in the updated version.

 

2. Page 2, line 7 from top for secondary paragraph cGAS-STING should be spelled out before abbreviation. Same as SLC41A1 in the third paragraph.

Reply. Thank you very much for your comments.We have now corrected the abbreviation formatting in the revised manuscript.

 

3. Methods:

OA induction: Authors said animals were euthanized at 8 weeks. Is this 8 week after Mg injection of MIA injection.  MIA should be spelled out.

What decalcification method should be stated.

Reply.Thank you for your reminder. MIA has been fully spelled out as monosodium iodoacetate on its first mention, and the 8-week euthanasia time point is explicitly defined as 8 weeks post MIA injection, with the detailed decalcification protocol added in the revised manuscript.

 

2.10, Authors said stain with DAPI, but their anti-fade medium has DAPI. Please clarify.

Reply.Thank you for your reminder. We sincerely apologize for the previous misleading description. In this study, we did not use any DAPI-containing anti-fade medium. Instead, we performed an independent 5-minute DAPI nuclear staining step, and immediately proceeded to fluorescence microscope observation and image acquisition.

 

1. Results:

3.1. Authors said Safranin O staining stained histone polysaccharide, this is not correct. Safranin O stains cartilage extracellular matrix glycosaminoglycan (GAG).

Both H&E staining and SO staining should be made brighter.  Authors did not check white balance.

Figure 1B H&E staining showed OA group has more chondrocytes despite cartilage surface damage while in Figure 1D, this group showed no COL2.  Probably not correct. MIA is a chondrocyte toxin, it can cause entire cartilage loss without cells. TUNEL apoptosis staining revealed very few positive cells in OA.  The fonts on the quantification graph are blurry.

Figure 1 legend: panel label A,B,C and so on should have “.” after each. Whether bold or not bold should be consistent. It is better to label each graph with a panel letter. What is mean print area % quantified for between E and F panel?

Reply.Thank you for your careful and constructive comments on Figure 1. We sincerely apologize for the errors, ambiguous descriptions and presentation defects in the original figure and its legend, and have fully revised Figure 1 accordingly, with point-by-point responses below:

1. We have corrected the wrong description of Safranin O staining target.
2. We have re-processed all H&E and Safranin O staining images.
3. Histological result clarification‌:Our 8-week MIA-induced OA model represents the early-to-mid OA stage rather than end-stage full-thickness cartilage loss. At this time point, residual chondrocytes still exist in the middle and deep cartilage layer.These residual chondrocytes exhibit significantly elevated intracellular oxidative stress and mitochondrial dysfunction, and have already lost their normal capacity to synthesize and secrete type II collagen, leading to the near-complete depletion of COL2 in the surrounding extracellular matrix. This histological feature is fully consistent with the actual performance of our sampled articular cartilage sections.As a well-recognized chondrocyte metabolic toxin, MIA triggers rapid chondrocyte apoptosis within the first 1–2 weeks after intra-articular injection. By the 8-week sampling point, nearly all those early-stage apoptotic chondrocytes have been completely cleared by synovial resident macrophages and local tissue phagocytes, leaving only a tiny fraction of newly initiated apoptotic chondrocytes that can be captured by TUNEL staining.
4. All quantification graphs in Figure 1 have been completely redrawn.The full Figure 1 legend has been standardized: every panel label (A., B., C., etc.) now has a period after the letter, and bold formatting for all panel labels is kept completely consistent.
5. We have supplemented a clear, standardized description for both quantification indicators in the revised Figure 1 legend, and the full revised text is as follows:
6. Quantitative analysis of the mean print area (cm²) of Safranin O-positive glycosaminoglycan (GAG) regions in articular cartilage.
7. Quantitative analysis of the apoptosis rate, calculated as the percentage of TUNEL-positive chondrocyte area relative to the total DAPI-stained nuclear area across the full cartilage field of view.

 

Figure 2 DAPI staining is so weak, it is not easy to determine whether iNOS and SLC1A1 located in cytoplasm or membrane? Insets highlighted positive cells will be helpful in the merged images.

Figure 2 legend also has panel labeling the same issue as Figure 1. Same as the other figures.

Reply. Thank you very much for your comments.We have fully revised Figure 2 (optimized DAPI brightness) and its corresponding legend. All revised high-resolution figures have been uploaded in the revised manuscript.

Author contribution is missing.

 Reply. Thank you very much for your comments. We have corrected it in the revised

Manuscript.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript proposes that dysregulated SLC41A1 expression in synovial macrophages disrupts intracellular Mg²⁺ balance, drives pro-inflammatory M1 polarization, and, through this mechanism, damages chondrocytes and accelerates cartilage degradation in osteoarthritis (OA). The authors support this hypothesis with an in vivo MIA-induced rat OA model, where intra-articular magnesium sulfate attenuates cartilage damage and lowers synovial SLC41A1/iNOS expression, and with in vitro macrophage-chondrocyte co-culture systems, where SLC41A1 knockdown in M1 and M2 macrophages is used to test downstream effects on chondrocyte oxidative stress, apoptosis, mitochondrial function, and matrix turnover. The experimental design has real strengths. It combines an in vivo disease model with two co-culture setups run in opposite directions, M1 acting on chondrocytes and M2 acting on IL-1β-stimulated chondrocytes, which lets the authors examine both the pro-inflammatory and the protective sides of the mechanism they propose. That said, the manuscript as it stands has several problems that weaken confidence in its central claims:

1. The case for SLC41A1 as the central target is thin. The cited evidence (refs. 11, 12, 14, 15) comes from contexts unrelated to OA and doesn't establish a direct link between SLC41A1 and either OA pathology or macrophage-chondrocyte crosstalk. The authors need preliminary data from this specific system, for instance, differential SLC41A1 expression in OA versus healthy synovium or cartilage, to justify choosing this target. Ref. 14 already reported SLC41A1-mediated Mg²⁺ dysregulation driving mitochondrial damage in an LPS setting, so the authors also need to state explicitly what this study adds beyond that prior work.

2. Every functional result comes from knockdown experiments. Without overexpression or rescue experiments, and without controls ruling out shRNA off-target effects, the causal role claimed for SLC41A1 is not demonstrated, only correlated. This is a substantial gap and likely requires additional experiments to close.

3. Several terms need correction throughout the manuscript:

• "SLC41A1 knockout" (Section 2.6) should read "knockdown"; shRNA doesn't produce a knockout.
• "OA histone polysaccharides" (Section 3.1) isn't a real term; Safranin O staining detects glycosaminoglycans (GAGs) and proteoglycans.
• SLC41A1 is called a generic "magnesium transporter" in the Figure 2 legend but a "magnesium exporter" in the Figure 3H legend, and the text never states the transport direction explicitly or cites support for it. Since Mg²⁺ efflux is central to the whole hypothesis, this needs to be stated clearly and consistently from the Introduction onward, backed by the relevant literature.

4. There are also figure and data-reporting errors. The text cites "Figure 1C and 1D" for synovial SLC41A1/iNOS immunofluorescence, but those panels are actually Figure 2C and 2D. Figure 4C shows no significant difference between RNAi constructs #2 and #3, yet construct #3 was chosen without any stated reason; the authors should explain the selection criterion. And the MOI used for lentiviral transduction is only described as "predetermined"; the actual number needs to be reported for reproducibility.

5. The statistics need more detail. Figures report only "*P<0.05" without saying which test was used in each panel (t-test versus ANOVA with post hoc), which matters here since some figures compare three groups (Figs. 1–2) and others compare two (Figs. 3–5). Sample sizes aren't given per panel either, and some panels show as few as three data points. On top of that, the figures only compare baseline to treated groups; without a direct comparison to the M2 or sham reference, the claims of "reversal" or "restoration" to a normal phenotype aren't statistically backed.

6. The Discussion's opening paragraph repeats itself: two consecutive sentences say almost the same thing about the M1 macrophage-synovium-cartilage cascade, and the first one is even missing a word ("degradation"). This needs to be cut down to one sentence. The manuscript's language should also be toned down wherever it overstates what the data show. "Can fully reverse" should become something like "significantly attenuated" or "ameliorated," and "SLC41A1 drives magnesium efflux, promoting M1 polarization" should be framed as an inference rather than a fact, given that there's no rescue data to back the causal claim.

Comments on the Quality of English Language

.

Author Response

1. The case for SLC41A1 as the central target is thin. The cited evidence (refs. 11, 12, 14, 15) comes from contexts unrelated to OA and doesn't establish a direct link between SLC41A1 and either OA pathology or macrophage-chondrocyte crosstalk. The authors need preliminary data from this specific system, for instance, differential SLC41A1 expression in OA versus healthy synovium or cartilage, to justify choosing this target. Ref. 14 already reported SLC41A1-mediated Mg²⁺ dysregulation driving mitochondrial damage in an LPS setting, so the authors also need to state explicitly what this study adds beyond that prior work.

Reply. Thank you for this critical and constructive comment, which greatly strengthens our study’s target rationale and novelty.

We acknowledge that the original submission lacked sufficient direct OA-relevant evidence to justify SLC41A1 as our central target, and the previously cited references (refs. 11, 12, 14, 15) were mostly from non-OA contexts. To address this, we have added new transcriptomic data in the Results section: reanalysis of GSE271903 confirms significantly elevated SLC41A1 mRNA in the ACRL-induced OA group versus sham group (Figure 2F), and independent reanalysis of GSE305061 and GSE290919 further verifies marked SLC41A1 upregulation in M1-polarized RAW264.7 macrophages (Figure 2F). These new data directly link SLC41A1 dysregulation to OA pathology and M1 macrophage polarization, providing explicit rationale for our target selection. We have also supplemented the Discussion section with additional SLC41A1-related literature (refs. 24, 25, 26) .

For the novelty beyond Ref. 14, which reported SLC41A1-mediated Mg²⁺ dysregulation drives LPS-triggered mitochondrial damage: our work uniquely extends this mechanism to the OA context, and further defines that aberrant SLC41A1 upregulation in synovial M1 macrophages disrupts magnesium’s physiological anti-inflammatory effect, and subsequently drives pathological abnormal crosstalk between synovial macrophages and chondrocytes, forming a previously unreported magnesium-homeostasis-centered pathological amplifier axis in OA progression.

We hope these revisions fully address your concern, and we are ready for further adjustments.

 

2. Every functional result comes from knockdown experiments. Without overexpression or rescue experiments, and without controls ruling out shRNA off-target effects, the causal role claimed for SLC41A1 is not demonstrated, only correlated. This is a substantial gap and likely requires additional experiments to close.

Reply. Thank you for this rigorous and professional comment. We acknowledge that the original study only presented knockdown-based correlation evidence, and we have made targeted revisions and supplements in the revised manuscript:

We have used two independent validated shRNA sequences against SLC41A1 in all functional experiments, and the highly consistent phenotypic changes across both sequences strongly minimize potential off-target effects. We have systematically revised all causal claims to strictly align with our current evidence, replacing overstated descriptions with precise statements such as "Knockdown of SLC41A1 attenuates the pro-oxidative and pro-apoptotic effects of M1 macrophages on co-cultured chondrocytes" to avoid overinterpreting correlation results. We have also added a dedicated section in the Discussion to clearly state this experimental limitation, and explicitly note that SLC41A1 overexpression and in vivo rescue experiments will be performed in follow-up studies to further define its definitive causal role in OA progression. All revisions are clearly marked in red in the revised manuscript.

 

3. Several terms need correction throughout the manuscript:
   • "SLC41A1 knockout" (Section 2.6) should read "knockdown"; shRNA doesn't produce a knockout.
   • "OA histone polysaccharides" (Section 3.1) isn't a real term; Safranin O staining detects glycosaminoglycans (GAGs) and proteoglycans.
   • SLC41A1 is called a generic "magnesium transporter" in the Figure 2 legend but a "magnesium exporter" in the Figure 3H legend, and the text never states the transport direction explicitly or cites support for it. Since Mg²⁺ efflux is central to the whole hypothesis, this needs to be stated clearly and consistently from the Introduction onward, backed by the relevant literature.

Reply. Thank you for these precise, constructive corrections that have helped us eliminate inaccurate statements and conceptual inconsistencies throughout the manuscript. We have addressed all three issues point by point in the revised version:

1. Every previously misused instance of "SLC41A1 knockout" has been systematically corrected to "SLC41A1 knockdown" across the full text, to accurately reflect that shRNA-mediated gene silencing produces only partial, transient target suppression rather than complete, permanent genetic ablation.
2. The non-standard, incorrect term "OA histone polysaccharides" has been fully removed. We now explicitly and accurately state that Safranin O staining detects cartilage glycosaminoglycans (GAGs) and proteoglycans.
3. All inconsistent descriptions of SLC41A1’s transport direction have been fully unified. Starting from the Introduction, we consistently define SLC41A1 as a magnesium efflux transporter, with peer-reviewed functional literature properly cited to support this statement. For Figure 3, which profiles a panel of multiple magnesium regulatory genes in addition to SLC41A1, we have adopted the accurate, general term "magnesium transport-related genes" for the corresponding description, resolving the prior conflicting functional labeling across different figure legends.

All revisions outlined above are clearly marked in red in the revised manuscript.

 

4. There are also figure and data-reporting errors. The text cites "Figure 1C and 1D" for synovial SLC41A1/iNOS immunofluorescence, but those panels are actually Figure 2C and 2D. Figure 4C shows no significant difference between RNAi constructs #2 and #3, yet construct #3 was chosen without any stated reason; the authors should explain the selection criterion. And the MOI used for lentiviral transduction is only described as "predetermined"; the actual number needs to be reported for reproducibility.

Reply.Thank you for identifying these critical errors in figure labeling, data reporting and methodological reproducibility. We have fully addressed all three concerns point by point in the revised manuscript:

1.We corrected the mislabeled cross-reference for synovial SLC41A1/iNOS immunofluorescence results, updating the incorrect citation "Figure 1C and 1D" to the correct panel numbering "Figure 2C and 2D".

2. We added a clear explicit explanation for shRNA selection in the Results section: construct #3 was chosen for its higher SLC41A1 knockdown performance verified in preliminary qPCR and Western blot assays, which outperformed construct #2 despite no statistically significant difference between the two groups.

3.We replaced the vague "predetermined MOI" description and reported the exact, experimentally optimized lentiviral transduction MOI value of 20 for RAW264.7 macrophages.

All these corrections are clearly marked in red in the revised manuscript.

 

5. The statistics need more detail. Figures report only "*P<0.05" without saying which test was used in each panel (t-test versus ANOVA with post hoc), which matters here since some figures compare three groups (Figs. 1–2) and others compare two (Figs. 3–5). Sample sizes aren't given per panel either, and some panels show as few as three data points. On top of that, the figures only compare baseline to treated groups; without a direct comparison to the M2 or sham reference, the claims of "reversal" or "restoration" to a normal phenotype aren't statistically backed.

Reply.We greatly appreciate these critical comments that enhance our statistical rigor. All concerns have been fully addressed in the revised manuscript:

Explicit statistical test annotations are added to each figure legend: one-way ANOVA with LSD post hoc for three-group comparisons in Figs.1–2, and unpaired two-tailed Student’s t-test for two-group comparisons in Figs.3–5.

Exact sample sizes for every panel are supplemented, with 3 data points clearly defined as independent biological replicates.

Direct statistical comparisons between the treatment group and the sham reference group are added to all relevant panels, making the "reversal/normal phenotype restoration" claims statistically validated.

All revisions are marked in red in the revised manuscript.

 

6. The Discussion's opening paragraph repeats itself: two consecutive sentences say almost the same thing about the M1 macrophage-synovium-cartilage cascade, and the first one is even missing a word ("degradation"). This needs to be cut down to one sentence. The manuscript's language should also be toned down wherever it overstates what the data show. "Can fully reverse" should become something like "significantly attenuated" or "ameliorated," and "SLC41A1 drives magnesium efflux, promoting M1 polarization" should be framed as an inference rather than a fact, given that there's no rescue data to back the causal claim.

Reply.Thank you for these precise and constructive suggestions. We have fully revised the Discussion section accordingly in the revised manuscript:

1.We have merged the two repetitive sentences about the M1 macrophage-synovium-cartilage cascade in the opening paragraph into one concise, complete statement.

2.We have systematically toned down all overstated claims throughout the manuscript: the expression "can fully reverse" has been replaced with "significantly attenuated" and "ameliorated" to accurately reflect the observed phenotypic improvement.We have rephrased the statement "SLC41A1 drives magnesium efflux, promoting M1 polarization" from a definitive factual claim to a reasonable, clearly labeled inference, explicitly noting that this regulatory relationship is supported by our current correlative data and will be further validated in subsequent rescue experiments, which fully avoids overstating the causal relationship without corresponding rescue data.

All these language adjustments are clearly marked in red in the revised manuscript.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The paper entitled SLC41A1-mediated magnesium homeostasis imbalance promotes osteoarthritis progression by inducing abnormal interaction between synovial macrophages and chondrocytes s presented.  The authors showed that as an essential modulator of magnesium homeostasis in synovial macrophages, SLC41A1 fuels pathological macrophage-chondrocyte communication and facilitates OA pathogenesis, suggesting it as an attractive therapeutic target for magnesium-centered OA treatment strategies. This paper is  interesting. The methods used were appropriate and conclusions were justified.

 

Some possible issues

Figure 1. Intraarticular injection of magnesium exerts a protective effect. It is suggested to include the sample size of analyses? N=?  in the figure legend.

Figure 2. Magnesium ion intervention significantly downregulates the aberrant expression of

SLC41A1 in synovial tissue and suppresses M1 pro-inflammatory polarization of macrophages. It is suggested to include the sample size of analyses? N=?  in the figure legend.

 

Figure 3. M1 macrophage polarization links SLC41A1 upregulation to reduced intracellular mag

nesium, reversed by magnesium supplementation. It is suggested to include the sample size of analyses? N=?  in the figure legend.

 

Figure 4. Knockdown of SLC41A1 attenuates the pro-oxidative and pro-apoptotic effects of M1

macrophages on co-cultured chondrocytes. It is suggested to include the sample size of analyses? N=?  in the figure legend.

 

Figure 5. SLC41A1 knockdown preserves M2 macrophage phenotype and magnesium homeostasis

in IL-1β-activated chondrocyte co-culture. It is suggested to include the sample size of analyses? N=?  in the figure legend.

 

The cellular mechanism of SLC41A1 in OA seems not fully explored or discussed. Recent studies have found OA is multiple factorial with many cell types and various pathological changes involved  (for example, PMID: 30937213).  It would be informative to discuss how SLC41A1 might modulate these factors and cells with these reports.

There are several subtypes of SLC family proteins such as Slc37a2 which regulate bone and joint pathology, it would be relevant to discuss the comparative role of SLC41A1 with other SLC molecules such as Slc37a2.

Author Response

Figure 1. Intraarticular injection of magnesium exerts a protective effect. It is suggested to include the sample size of analyses? N=?  in the figure legend.

Figure 2. Magnesium ion intervention significantly downregulates the aberrant expression of

SLC41A1 in synovial tissue and suppresses M1 pro-inflammatory polarization of macrophages. It is suggested to include the sample size of analyses? N=?  in the figure legend.

 

Figure 3. M1 macrophage polarization links SLC41A1 upregulation to reduced intracellular mag

nesium, reversed by magnesium supplementation. It is suggested to include the sample size of analyses? N=?  in the figure legend.

 

Figure 4. Knockdown of SLC41A1 attenuates the pro-oxidative and pro-apoptotic effects of M1

macrophages on co-cultured chondrocytes. It is suggested to include the sample size of analyses? N=?  in the figure legend.

 

Figure 5. SLC41A1 knockdown preserves M2 macrophage phenotype and magnesium homeostasis

in IL-1β-activated chondrocyte co-culture. It is suggested to include the sample size of analyses? N=?  in the figure legend.

 Reply. Thank you very much for your comments. We have corrected it in the revised

Manuscript.

 

The cellular mechanism of SLC41A1 in OA seems not fully explored or discussed. Recent studies have found OA is multiple factorial with many cell types and various pathological changes involved  (for example, PMID: 30937213).  It would be informative to discuss how SLC41A1 might modulate these factors and cells with these reports.

There are several subtypes of SLC family proteins such as Slc37a2 which regulate bone and joint pathology, it would be relevant to discuss the comparative role of SLC41A1 with other SLC molecules such as Slc37a2.

 Reply. Thank you very much for these insightful suggestions, which have significantly strengthened the contextual depth and logical completeness of our Discussion section.

1. We fully agree that the cellular mechanism of SLC41A1 in OA remains largely underexplored. We have cited PMID: 30937213 to support that OA is a multifactorial disease involving multiple cell types and complex pathological alterations. In the revised manuscript, we have expanded the discussion section to elaborate how SLC41A1-mediated magnesium efflux dysregulation in synovial M1 macrophages disrupts the physiological anti-inflammatory function of magnesium, and further drives aberrant crosstalk between macrophages and chondrocytes during OA progression.
2. For the comparative discussion of other SLC family members: Slc37a2 acts as a glucose-6-phosphate transporter that regulates osteoclastic lysosomal metabolism and bone resorption, which is mechanistically distinct from the Mg²⁺ efflux regulatory function of SLC41A1 in synovial macrophages, with no confirmed overlapping functional axis in current experimental settings.

We hope this arrangement addresses your valuable concern, and we are ready for further adjustments.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Thank you for the reply and the revised manuscript. Several round 1 points are resolved, and the added public transcriptomic data and declarations are welcome. However, the main concerns remain:

1. The new transcriptomic data (Fig. 2F-G) show that SLC41A1 is upregulated, but they don’t fully justify the choice of target. The “ACRL” model, species, and tissue of GSE271903 are not defined, and no methods describe the reanalysis. The macrophage datasets compare M1 with M0, not M2, with n=3, so “specifically upregulated in M1” is not supported. The novelty relative to ref. 14 is stated in the reply but not in the manuscript. Please add it to the introduction.
2. All functional data come from partial knockdown (about 50% at protein and mRNA level for construct #3, n=3), with no overexpression or rescue. The reply says two independent shRNA sequences were used in all functional experiments, yet Sections 2.6 and 3.4 indicate that only construct #3 was used downstream. Please clarify. Provide a rescue or a second independent shRNA/siRNA, show the data behind the choice of #3 over #2 (not significantly different in Fig. 4C-D), or restrict the conclusions accordingly. I could not find the limitations section mentioned in the reply.
3. Several statements are unchanged despite the reply: “SLC41A1 drives magnesium efflux, promoting M1 polarization” (Conclusions), “key regulator” and “promising therapeutic target” (Abstract), “can fully reverse” (Discussion), and “reversed” in Sections 3.1 and 3.3. The new comparisons against the reference groups also show that Mg²⁺ does not normalize the phenotype: sham vs. Mg²⁺ remains significant in Figs. 2B and 2D, and so does M2 vs. M1+Mg²⁺ in Fig. 3. The text should say “partial attenuation”.
4. Fig. 3H calls SLC41A1 an “exporter”, but the introduction gives no direction or reference. The efflux claim rests on the steady-state Mag-Fluo-4 signal alone. Please state the direction with references in the introduction and acknowledge that efflux was not measured directly.

Comments on the Quality of English Language

.

Author Response

1. The new transcriptomic data (Fig. 2F-G) show that SLC41A1 is upregulated, but they don’t fully justify the choice of target. The “ACRL” model, species, and tissue of GSE271903 are not defined, and no methods describe the reanalysis. The macrophage datasets compare M1 with M0, not M2, with n=3, so “specifically upregulated in M1” is not supported. The novelty relative to ref. 14 is stated in the reply but not in the manuscript. Please add it to the introduction.

Reply. We greatly appreciate your careful and insightful feedback, which has helped us substantially improve the transparency and rigor of this study.

We have now made corresponding revisions throughout the manuscript: we added the complete description of GSE271903 (ACLR model, C57BL/6J mice synovial tissue) and the full bioinformatic reanalysis protocol in the Methods section, supplemented the M1 vs M2 comparison data from GSE290919 in the Results, and incorporated the novelty relative to ref. 14 into the Introduction. All these changes are clearly marked in the revised version.

 

2. All functional data come from partial knockdown (about 50% at protein and mRNA level for construct #3, n=3), with no overexpression or rescue. The reply says two independent shRNA sequences were used in all functional experiments, yet Sections 2.6 and 3.4 indicate that only construct #3 was used downstream. Please clarify. Provide a rescue or a second independent shRNA/siRNA, show the data behind the choice of #3 over #2 (not significantly different in Fig. 4C-D), or restrict the conclusions accordingly. I could not find the limitations section mentioned in the reply.

Reply. We greatly appreciate the reviewer’s precise and critical observation, which has helped us further standardize our data presentation and eliminate ambiguous descriptions.

We sincerely apologize for the inconsistent statement in our previous reply. We have now updated the corresponding Results section to read: “The second and third transfection systems produced comparable significant SLC41A1 silencing with no statistical difference. The third system was selected for subsequent experiments, as it showed a trend toward slightly lower residual SLC41A1 expression across all preliminary validation tests.”

Given the current project timeline does not allow additional overexpression, rescue or second-shRNA functional validation to be completed in this revision, we have fully restricted all relevant conclusions accordingly: all previously implied causal statements have been replaced with association-based descriptions, and we no longer claim a definitive causal role for SLC41A1 in driving the observed phenotypes. We have also added an explicit Limitations subsection in the Discussion, clearly acknowledging that all functional results are derived from a single validated SLC41A1 partial knockdown sequence, and that independent shRNA/siRNA replication and rescue experiments will be prioritized in future work to further confirm our current observations. All revised sections are marked in red in the manuscript for easy tracking.

 

3. Several statements are unchanged despite the reply: “SLC41A1 drives magnesium efflux, promoting M1 polarization” (Conclusions), “key regulator” and “promising therapeutic target” (Abstract), “can fully reverse” (Discussion), and “reversed” in Sections 3.1 and 3.3. The new comparisons against the reference groups also show that Mg²⁺ does not normalize the phenotype: sham vs. Mg²⁺ remains significant in Figs. 2B and 2D, and so does M2 vs. M1+Mg²⁺ in Fig. 3. The text should say “partial attenuation”.

Reply. We thank the reviewer for this careful and critical observation. We have now revised the manuscript systematically to remove all overstated claims:

1.Rewrote “SLC41A1 drives magnesium efflux, promoting M1 polarization” as “Elevated SLC41A1 expression in M1 macrophages leads to magnesium efflux”.

2.Replaced “key regulator” and “promising therapeutic target” with “potential modulator” and “represents a potential target” in Abstract.

3.Changed “can fully reverse” to “partially attenuated” in Discussion

4.Updated “reversed” to “partially attenuated” in Sections 3.1 and 3.3.

 

 

4. 3H calls SLC41A1 an “exporter”, but the introduction gives no direction or reference. The efflux claim rests on the steady-state Mag-Fluo-4 signal alone. Please state the direction with references in the introduction and acknowledge that efflux was not measured directly.

Reply. We thank the reviewer for this important and constructive comment. “Consistent with previous functional evidence, SLC41A1 is generally recognized as a plasma membrane Na⁺/Mg²⁺ exchanger that mediates Mg²⁺ efflux (refs. 22). In the present study, we inferred altered Mg²⁺ homeostasis primarily from steady-state Mag-Fluo-4 fluorescence measurements in macrophages, and direct transport assays were not performed to confirm unidirectional Mg²⁺ efflux.”

Accordingly, we have updated the label in Fig. 3H from “SLC41A1 exporter” to “SLC41A1, putative Mg²⁺ efflux-related transporter” to avoid overstatement.

Round 3

Reviewer 2 Report

Comments and Suggestions for Authors

Thank you for the revised manuscript and the point-by-point reply, but four points remain.

1. The reply says the novelty relative to ref. 14 was added to the introduction, but I could not find it. The introduction only says that the role of SLC41A1 in this process is undefined. State what ref. 14 showed (LPS, dental stem cells) and what this study adds in OA. The discussion still says “specifically upregulated in M1” on the basis of GSE305061, which compares M1 only with M0.
2. I accept that a rescue and a second shRNA are not feasible now, as long as the conclusions are restricted to match. The new limitations paragraph mentions only the missing overexpression/rescue. It should also say that a single partial knockdown (about 50%) from one construct was used, so off-target effects cannot be excluded. Section 2.6 still lists three sequences as one construct and does not say which is #3. Clarify this, and either show the validation data behind the choice of #3 or report the choice as a limitation.
3. The reply says all causal statements were replaced, but several remain: the title (“promotes... by inducing”), the abstract methods (“upregulation mediates... drives”), “reversed by magnesium supplementation” in the abstract, “can reverse this trend” in Section 3.3, “effectively protects the cells from M2-to-M1 transition” in the discussion (this transition was not measured), and the conclusions (“leads to magnesium efflux, which subsequently impairs chondrocyte mitochondrial function”). The abstract also says “modulator”, while the reply says “potential modulator”. The text should match the reply.
4. The passages quoted in the reply are not in the manuscript. The introduction does not state the direction of transport, and neither the introduction nor the limitations say that efflux was not measured directly. The Fig. 3H legend still says “magnesium exporter” instead of the new label given in the reply. Add these changes.

Comments on the Quality of English Language

.

Author Response

Thank you for the revised manuscript and the point-by-point reply, but four points remain.

1. The reply says the novelty relative to ref. 14 was added to the introduction, but I could not find it. The introduction only says that the role of SLC41A1 in this process is undefined. State what ref. 14 showed (LPS, dental stem cells) and what this study adds in OA. The discussion still says “specifically upregulated in M1” on the basis of GSE305061, which compares M1 only with M0.

Reply. Thank you for your careful and constructive comments. We have now clearly specified the core findings of ref. 14 and the distinct, OA-specific novelty of the present study in the revised introduction, with the updated text reading as follows:

"That study was conducted in dental stem cells and did not explore SLC41A1 function in osteoarthritis. Though Mg²⁺ is known to promote macrophage M2 polarization and alleviate OA injury, how SLC41A1 governs this magnesium-dependent process remains unknown. We will investigate the unreported mechanism by which aberrant SLC41A1 upregulation in synovial M1 macrophages blunts magnesium’s anti-inflammatory action, disrupts macrophage-chondrocyte homeostasis, and drives OA progression."

We have also corrected the overstatement in discussion to "significantly upregulated in M1 versus M0 macrophages ", to strictly match the dataset design.

 

2. I accept that a rescue and a second shRNA are not feasible now, as long as the conclusions are restricted to match. The new limitations paragraph mentions only the missing overexpression/rescue. It should also say that a single partial knockdown (about 50%) from one construct was used, so off-target effects cannot be excluded. Section 2.6 still lists three sequences as one construct and does not say which is #3. Clarify this, and either show the validation data behind the choice of #3 or report the choice as a limitation.

Reply. Thank you very much for your pragmatic and critical guidance. We fully agree that all conclusions should be strictly constrained to match the available experimental evidence, and have made targeted, transparent revisions across the manuscript to address every point you raised.

First, we acknowledge that under the current experimental conditions, the best achievable silencing efficiency among all three screened shRNA sequences was approximately 50% partial SLC41A1 knockdown, and we were not able to obtain a second independent shRNA with comparable or higher efficiency for parallel functional validation. We have therefore expanded the limitations paragraph to fully document this constraint, with the revised text reading as follows:

"Nevertheless, several key limitations of this study should be explicitly noted. First, all functional results are derived from a single shRNA construct that produced approximately 50% partial SLC41A1 knockdown, without parallel overexpression or rescue validation. This design means potential off-target effects cannot be fully excluded, and restricts the causal interpretation of SLC41A1 as a core driver in this OA-related phenotype. Second, SLC41A1 is annotated as a bidirectional magnesium exchanger, and the magnesium efflux inference in this study is indirectly derived from intracellular Mg²⁺ concentration changes, rather than verified by direct transport assay. Third, the exact downstream effectors that mediate the function of this SLC41A1-Mg²⁺ regulatory axis in synovial macrophage-chondrocyte crosstalk remain uncharacterized. Future work will validate these preliminary findings using cell-type-specific in vivo genetic models, multiple independent silencing constructs and direct magnesium transport measurement, to further support the translational development of SLC41A1-related magnesium-targeted OA interventions."

Second, in revised Section 2.6, we have now clearly listed all three independent shRNA sequences with explicit labeling as Sequence #1, Sequence #2 and Sequence #3, to resolve the previous ambiguity that treated the three sequences as a single unseparated construct.

All related claims in the results and discussion sections have been further moderated to strictly align with this level of experimental evidence, and all revisions have been clearly marked in red in the revised manuscript.

 

3. The reply says all causal statements were replaced, but several remain: the title (“promotes... by inducing”), the abstract methods (“upregulation mediates... drives”), “reversed by magnesium supplementation” in the abstract, “can reverse this trend” in Section 3.3, “effectively protects the cells from M2-to-M1 transition” in the discussion (this transition was not measured), and the conclusions (“leads to magnesium efflux, which subsequently impairs chondrocyte mitochondrial function”). The abstract also says “modulator”, while the reply says “potential modulator”. The text should match the reply.

Reply. We thank the reviewer for this rigorous check. All remaining overstated causal and inconsistent statements in the manuscript have been fully revised to strictly align with our previous reply, with all changes marked in red in the revised version. Details are as follows:

1.‌Title‌: Changed the original strong causal phrasing "promotes... by inducing" to the cautious formulation "SLC41A1-associated magnesium dyshomeostasis correlates with osteoarthritis progression: a preliminary investigation into abnormal synovial macrophage-chondrocyte crosstalk".

2.‌Abstract Methods‌: Revised "upregulation mediates... drives" to "SLC41A1 expression in M1/M2 macrophages and magnesium’s regulatory effect on M1 were analyzed. A macrophage-chondrocyte co-culture system was used to evaluate how SLC41A1 knockdown in M1 affects chondrocyte apoptosis, reactive oxygen species (ROS) level, mitochondrial membrane potential, and cartilage matrix anabolism/catabolism.".

3.‌Abstract ‌: The original statement "reversed by magnesium supplementation" has been removed.

4.‌Section 3.3‌: Rewrote "can reverse this trend" to "Mg²⁺ treatment in M1 macrophages attenuated this trend".

5.‌Discussion‌: Deleted the unmeasured claim "effectively protects the cells from M2-to-M1 transition", and replaced it with "and CD206 expression, along with reduced iNOS levels, compared with controls".

6.‌Conclusions‌: Revised the causal chain "leads to magnesium efflux, which subsequently impairs chondrocyte mitochondrial function and contributes to cartilage degradation" to "Elevated SLC41A1 expression in M1 macrophages correlates with increased magnesium efflux, which is associated with impaired chondrocyte mitochondrial function and altered cartilage homeostasis".

7.‌Abstract ‌: Corrected "modulator" to "potential modulator" to match our previous reply.

 

 

4. The passages quoted in the reply are not in the manuscript. The introduction does not state the direction of transport, and neither the introduction nor the limitations say that efflux was not measured directly. The Fig. 3H legend still says “magnesium exporter” instead of the new label given in the reply. Add these changes.

Reply. We thank the reviewer for this careful check. All required adjustments have been completed and marked in red in the revised manuscript:

1. We have relocated all SLC41A1 Mg²⁺ efflux-related descriptions to the Discussion section, rather than the Introduction, with the two literature-supported statements added exactly as specified: "SLC41A1 is a major Mg²⁺ efflux system that functions as a Na⁺/Mg²⁺ exchanger, and its dysregulation is associated with diseases such as Parkinson's disease, pulmonary arterial hypertension and various cancers[22]" and "Available literature further confirms the plausibility of this mechanism: SLC41A1 knockdown restricts excessive Mg²⁺ efflux, which not only alleviates pathological cardiac fibrosis[25]".
2. We have supplemented the explicit note in the Limitations section that direct Mg²⁺ efflux measurement was not performed in this study, and all related inferences are derived from detected intracellular Mg²⁺ changes.
3. The phrase "magnesium exporter" in the Fig. 3H legend has been completely removed.

All content now fully matches the commitments in our previous reply.

Author Response File: Author Response.pdf