Disease-associated programming of cell memory in glycogen storage disorder type 1a

  1. U Sprecher
  2. J D’Souza
  3. K Mishra
  4. N Muchtar
  5. O Shalev
  6. A Eliassaf
  7. A Morshina
  8. A Canella Miliano
  9. G Mithieux
  10. F Rajas
  11. S Avraham
  12. Moses F Castellani
  13. H Kauffman
  14. Y Bergman
  15. N Garti
  16. S Garti
  17. M Linial
  18. Y Anikster
  19. O Kakhlon
  20. M Weil

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Glycogen storage disorder type 1a (GSD1a) is caused by loss-of-function mutations in the catalytic subunit of glucose-6-phosphatase enzyme ( G6PC1 ) in the liver, kidney and intestine exclusively. Here we show the surprising results that while not expressing G6PC1 , primary skin fibroblasts isolated from GSD1a patients’ skin biopsies preserve a distinctive disease phenotype irrespective of the different culture conditions under which they grow. This discovery was initially made by phenotypic image-based high content analysis (HCA). Deeper analysis into this disease phenotype, revealed impaired lysosomal and mitochondrial functions in GSD1a cells, which were driven by a transcriptional dysregulation of the NAD + /NADH-Sirt-1-TFEB regulatory axis. This dysregulation impacts the normal balance between mitochondrial biogenesis and mitophagy in the patients’ cells. The distinctive GSD1a fibroblasts phenotype involves elevated H3 histone acetylation, global DNA hypomethylation, differences in the chromatin accessibility and different RNA-seq and metabolomic profiles, all of which suggesting that in some way a distinctive disease cell phenotype is programmed in these cells in vivo and that this phenotype is maintained in vitro . Supporting this notion, reversing H3 acetylation in these cells erased the original cellular phenotype in GSD1a cells. Remarkably, GHF201, an established glycogen reducing molecule, which ameliorated GSD1a pathology in a liver-targeted inducible L.G6pc - knockout mouse model, also reversed impaired cellular functions in GSD1a patients’ fibroblasts. Altogether, this experimental evidence strongly suggests that GSD1a fibroblasts express a strong and reversible disease phenotype without expressing the causal G6PC1 gene.

Article activity feed

  1. Review Commons

    Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    Reviewer #1

    __Evidence, reproducibility and clarity __

    The manuscript explores mild physiological and metabolic disturbances in patient-derived fibroblasts lacking G6Pase expression, suggesting that these cells retain a "distinctive disease phenotype" of GSD1a. The manuscript is well written with well-designed experiments. However, it remains unclear whether these phenotypes genuinely reflect the pathology of GSD1a-relevant tissues. The authors did not validate these findings in a liver-specific G6pc knockout mouse model, raising concerns about the study's relevance to GSD1a. Additionally, the lack of sufficient in vivo evidence undermines the therapeutic potential of GHF201 for this disease. Overall, the study lacks a few key pieces of evidence to completely justify its conclusions at both fundamental and experimental levels.

    __Reply:__We thank the reviewer for this general comment which gives us the opportunity to better explain the scope of our work. The purpose and focus of this work are not to test the pathological relevance of skin fibroblasts to GSD1a pathology. We do not claim that skin fibroblasts are involved in GSD1a pathogenesis. It is also not a developmental work claiming to uncover GSD1a pathogenic axis throughout embryonic development. As a matter of fact, since skin fibroblasts originate from the mesoderm embryonic germ layer and hepatocytes develop from the endoderm embryonic germ layer, it would even be unlikely that the pathological phenotype found in skin fibroblasts directly contributes to GSD1a pathology in model mice or in patients. Indeed, we are not aware of any dermatological contribution to GSD1a pathology in patients. However, our results suggest that in addition to the established and mutated organ (liver in the liver-specific G6pc knockout mouse model), other, relatively less studied, patho-mechanisms in distal tissues may also contribute to GSD1a pathology. Notably, this work is also not testing a therapeutic modality for GSD1a. Our work uses GSD1a disease models as a tool for demonstrating, or reviving, the concept of epigenomic landscape (Waddington, 1957): Different cell phenotypes, such as healthy and diseased, are established by innate metabolic differences between their respective cell environments, which impose epigenetic changes generating these different phenotypes. In this respect, our manuscript has a similar message to the one in the recently published paper Korenfeld et al (2024) Nucleic Acids Res 53:gkae1161. doi: 10.1093/nar/gkae1161: The Kornfeld et al paper shows that intermittent fasting generates an epigenetic footprint in PPARα-binding enhancers that is "remembered" by hepatocytes leading to stronger transcriptional response to imposed fasting by up-regulation of ketogenic pathways. In the same way, the diseased GSD1a status imposes metabolic changes, as detailed here, leading to permanent epigenetic changes, also described here, which are "remembered" by GSD1a fibroblasts and play a major role in the transcription of pathogenic genes in these patient's cells. This in turn is how the diseased state is preserved, even in cells not expressing the G6Pase mutant, which is the direct cause of the disease. We added this perspective to the Discussion to better highlight the key takeaway from our manuscript.Naturally, research such as ours with a claim on biological memory would involve ex vivo experiments where tissues are isolated from their in-situ environments and tested for preservation of the original in situ phenotype. The few in vivo experiments we performed (Fig. 5) are mainly aimed at demonstrating that not only the phenotype, but also therapy response is "remembered" ex vivo: In the same way that the G6PC-loss-of-function liver responded positively to GHF201 therapy in situ, ex vivo cells not expressing G6PC also responded positively to the same therapy. This observation only demonstrates further support for "memorization" of the disease phenotype by cell types not expressing the mutant: Both the diseased phenotype and response to therapy were preserved ex vivo.Lastly, while interesting, validation of our findings in vivo (as suggested by the reviewer) is not related to the scope of this manuscript. Such experiments, using the liver-targeted G6pc knockout mouse model, are the follow-up story, which is related to the origin of inductive signals that cause the curious and novel phenotype mechanism in GSD1a fibroblasts described in this manuscript. The scope and volume of such research constitute a novel manuscript.

    Since dietary restriction is the only management strategy for GSD1a, the authors should clarify whether the patient fibroblast donors were on a dietary regimen and for how long. Given that fibroblasts do not express G6Pase, it is possible that the observed phenotype could be influenced by the patient's diet history.

    __Reply:__We thank the reviewer for this important comment, we agree that it is important to note the dietary regimen assigned to the cohort of patients described in this study. We added an explanation to the manuscript on patient's diets as shown below.Briefly, all patients besides patient 6894 were treated with the recommended dietary regimen for GSD1a as explained in Genereviews (Bali *et al *(2021)). This dietary treatment (now added to the Methods section in the manuscript) allows to maintain normal blood glucose levels, prevent secondary metabolic derangements, and prevent long-term complications. Specifically, this dietary treatment includes- nocturnal nasogastric infusion of a high glucose formula in addition to usual frequent meals during. By constantly maintaining a nearly normal level of blood glucose, this treatment causes a remarkable decrease, although not normalization, of blood lactate, urate and triglyceride levels, as well as bleeding time values. A second layer in the treatment includes the use of uncooked starch in the dietary regimen to allow maintenance of a normal blood glucose levels for long periods of time. Patient 6894 did not tolerate well the uncooked cornstarch and therefore was treated with a tailored dietary treatment planned by metabolic disease specialists and dedicated certified dieticians highly experienced with the management of pediatric and adult patients with GSDs and other inborn errors of metabolism. The biopsies of patients were taken in the range of 3 month to several years from receiving the aforementioned dietary regimen.Importantly, the strict metabolic diet imposed on GSD1a patients might influence the observed phenotype described throughout the manuscript. This concept aligns with our claim that the GSD1a skin cells are affected by the dysregulated metabolism in patients in comparison to healthy individuals. Interestingly, while patient 0762 harbors a mutation in the SI gene in addition to the G6PC mutation and patient 6894 did not receive the same dietary regimen as other patients (as explained above), all patients do show similar disease related phenotypes, perhaps highlighting the role of an early programing process that affected these cells due to the severe metabolic aberrations presented in this disease from birth.One of the main pathological features of GSD1a is glycogen buildup. The authors should compare glycogen levels between healthy controls and GSD1a fibroblasts and provide a dot plot analysis.

    One of the main pathological features of GSD1a is glycogen buildup. The authors should compare glycogen levels between healthy controls and GSD1a fibroblasts and provide a dot plot analysis.

    __Reply:__We thank the reviewer for this important comment. We added glycogen levels of HC to Figure S2A and accordingly also edited the relevant text in the Results section.

    Figure S2A - As mentioned above, the authors should present healthy control vs. patient fibroblast glycogen data. Without this, the rationale for using GHF201 is questionable.

    __Reply:__We thank the reviewer for this important comment. We added glycogen levels of HC to Figure S2A as mentioned above.

    Figure S2B-C - If the authors propose that GHF201 reduces glycogen and increases intracellular glucose in GSD1a fibroblasts, they need direct evidence. Either directly quantifying glycogen levels or even better would be a labeling experiment to confirm that the free intracellular glucose originates from glycogen. Additionally, the reduction in sample size from N=24 in glycogen analysis to N=3 in the glucose assay needs justification.

    __Reply:__We thank the reviewer for this comment. To clarify, the results shown in Figure S2A left are based on PAS assay, directly quantifying glycogen in cells with and without GHF201 treatment. We have now added HC glycogen levels as requested above. Regarding N, this is explained in Methods: In imaging experiments N was determined based on wells from the experiments done in three independent plates following the rationale that each well is independent from the others and reflects a population of hundreds of cells as previously described in (Lazic SE, Clarke-Williams CJ, Munafò MR (2018) What exactly is 'N' in cell culture and animal experiments?. PLOS Biology 16(4):e2005282. https://doi.org/10.1371/journal.pbio.2005282, Gharaba S, Sprecher U, Baransi A, Muchtar N, Weil M. Characterization of fission and fusion mitochondrial dynamics in HD fibroblasts according to patient's severity status. Neurobiol Dis. 2024 Oct 15;201:106667. doi: 10.1016/j.nbd.2024.106667. Epub 2024 Sep 14. PMID: 39284371.). Figure S2A right shows the glucose quantification experiment that we think the reviewer is referring to. Glucose increase is normally concomitant with glycogen reduction and we therefore show these results in support of the glycogen reduction results. These glucose results are part of our metabolomics results done on the same cells (Figure 6), where glucose is one of the metabolites analyzed. This metabolomics analysis was repeated three times; therefore, N is 3. In summary, these results show that GHF201 directly contributes to glycogen reduction in GSD1a fibroblasts and concomitantly increases glucose levels.

    Figure S2B-C- It is not shown how GHF201 increases intracellular glucose? If glycophagy is a possibility, the authors should do an experiment to confirm this.

    __Reply:__Assuming the reviewer's comment is related to Figure S2A right, glucose levels are only shown to validate the glycogen reduction results (also see point 4): When glycogen levels are reduced, especially by inhibition of glycogen synthesis, glucose levels are supposed to concomitantly rise, being spared as an indirect substrate of glycogen synthesis. There is no proof, and as a matter of fact we also do not assume, that the GHF201-mediated reduction in glycogen levels is a result of increased glycophagy: Glycophagy has been described in cell types with high glycogen turnover, e.g., muscle and liver cells, not fibroblasts. Additionally, glycophagy is a glycogen-selective process implicating STBD1 whose expression in skin fibroblasts is negligible (https://www.proteinatlas.org/ENSG00000118804-STBD1/tissue).On the other hand, glycogen in GSD1a does not accumulate in lysosomes. It is built up in the cytoplasm (Hicks et al (2011) Ultrastr Pathol 35: 183-196; Hannah et al (2023) Nat Rev Dis Primers DOI: 10.1038/s41572-023-00456-z). Therefore, we do not believe that GHF201 reduced glycogen by enhancing glycophagy. As we show, GHF201 activated several key catabolic pathways. It is more likely that activation of one of these pathways, the AMPK pathway, inhibited glycogen synthesis via phosphorylation and ensuing inhibition of glycogen synthase. Alternatively, excessive cytoplasmic glycogen might enter lysosomes by bulk autophagy, or microautophagy (not by glycophagy) and GHF201 might induce lysosomal glycogenolysis by alpha glucosidase as an established lysosomal activator (Kakhlon et al (2021)). However, since, as explained, the mechanism of action of GHF201 is not the topic of this manuscript and therefore we did not dwell more into that.

    Figure 2- How can GSD1a fibroblasts have significantly reduced OCR (Fig. 2B) but increased mitochondrial ATP production (Fig. 2H)?

    __Reply:__We thank the reviewer for highlighting this important topic. OCR, measured in Fig. 2B, is an indirect measure of ATP production. Therefore, changes in OCR only measure the capacity of the mitochondria to produce ATP, and not the direct quantity of ATP. Other factors might influence ATP production, *e.g., *substrate availability and the activity of other metabolic pathways. On the other hand, the ATP Rate Assay (Figure 2h), provides a real-time direct measurement of ATP levels incorporating coupling efficiency and P/O ratio assumptions. Therefore, these two measurements do not necessarily match. We will add this information to the relevant segment in the text to clarify why OCR is reduced and mitochondrial ATP production increased in GSD1a cells.

    Why do GSD1a fibroblasts show reduced glycolytic ATP (Figure 2h) despite increased glycolysis and glycolytic capacity (Fig 2J-K)?

    __Reply:__We thank the reviewer for highlighting this important topic. ECAR measures medium acidification and thus reflects the production of lactic acid, which is a byproduct of glycolysis. However, medium acidification is also influenced by other factors that can acidify the extracellular environment, especially CO2 production which can originate from the intramitochondrial Krebs cycle which produces reductive substrates for mitochondrial respiration, or OCR. Moreover, the buffering capacity of the Seahorse mito stress assay medium might mask changes in lactic acid production, leading to an underestimation of glycolytic activity. On the other hand, glycolytic ATP production measured by the ATP rate assay directly quantifies the rate of ATP production from glycolysis. Notably, there is a major difference between ECAR and the ATP rate assay: The ATP rate assay is less sensitive to variations in buffering capacity than ECAR measurements. This is because the ATP rate assay relies on inhibitor-driven changes in OCR and ECAR, rather than absolute pH values.Teleologically, as indicated, the increased ECAR in GSD1a cells represents a known compensatory response to deficient ATP production which is stimulation of glycolysis (Figure 2i). To test the success of this known compensatory attempt, we applied the real-time ATP rate assay, but as explained they do not report the same entities. We will add this information to the relevant segment in the text to clarify how reduced glycolytic ATP can be co-observed with increased glycolytic capacity.

    The authors should clarify how many healthy control and patient fibroblast lines were compared per experiment. Given the wide age range, the unexpectedly small error bars raise concerns about variability and statistical robustness.

    __Reply:__We thank the reviewer for raising this topic. Number of samples per experiment is reported in the Methods section. As for the age range, patients age was matched to healthy controls to account for age differences and experiments were performed under similar passages range. This procedure allowed us to control for technical differences between samples that might arise due to different passages and ages. Importantly, the cohort of samples used in this manuscript included GSD1a patients with different ages further implying the strength of the observed disease phenotype found in patients' cells which exists regardless of the different age and gender of patients. The HC samples were chosen to match age and gender and passages were used in the recommended range (L. Hayflick,The limited in vitro lifetime of human diploid cell strains,Experimental Cell Research,Volume 37, Issue 3,1965,Pages 614-636, änzelmann S, Beier F, Gusmao EG, Koch CM, Hummel S, Charapitsa I, Joussen S, Benes V, Brümmendorf TH, Reid G, Costa IG, Wagner W. Replicative senescence is associated with nuclear reorganization and with DNA methylation at specific transcription factor binding sites. Clin Epigenetics. 2015 Mar 4;7(1):19. doi: 10.1186/s13148-015-0057-5. PMID: 25763115; PMCID: PMC4356053., Magalhães, S.; Almeida, I.; Pereira, C.D.; Rebelo, S.; Goodfellow, B.J.; Nunes, A. The Long-Term Culture of Human Fibroblasts Reveals a Spectroscopic Signature of Senescence. Int. J. Mol. Sci. 2022, 23, 5830. https://doi.org/10.3390/ijms23105830). Finally, for the error bars, assuming the reviewer is addressing this for all experiments, this means that results are consistent across each compared group and reflects robustness of the results. Further, to ensure statistical robustness we used bootstrapping, 95% confidence intervals and other statistical methodologies that were designed to increase the validity of the conclusions drawn from different experiments.

    Figure 5- The study should include Tamoxifen-untreated mice as a control to properly assess the efficacy of GHF201 in regulating glucose-6-P and glycogen levels.

    __Reply:__GHF201 reduced liver glucose-6-phosphate (G6P) with p-/-* mice livers and their normalization by GHF201.

    Reviewer #2 (Evidence, reproducibility and clarity (Required)):

    General comments: the authors propose a very intriguing concept, that metabolic abnormalities trigger epigenetic changes in tissues distal from the disease site, even in cells in which the affected gene is not expressed. This is demonstrated in primary fibroblasts from patients with Glycogen Storage Disease type 1a (GSD1a). The authors provide a large amount of data to support the compelling concept of "Disease-Associated Programming", a term that they have coined to describe this effect. The level of novelty is very high and so is the impact of the study, since the above may apply to many different pathological conditions. Although, the study is well performed and employs multiple approaches and analyses to address the raised hypothesis, there are some limitations and concerns that need to be addressed by the authors.

    __Reply:__We thank the reviewer for this comment and will address each comment raised.

    The different phenotypic characteristics are only demonstrated in skin fibroblasts which is not sufficient to support the conclusions made in the Discussion about the general applicability of the proposed disease-induced, metabolite-driven epigenetic programming to all cells and tissues. The authors should discuss this as a limitation of the study and general conclusions should be formulated with more caution.

    __Reply:__We concur with this comment and accept that this is a general limitation of the study. We added a reservation clause at the beginning of the Discussion section.

    The authors describe a range of alterations in patients' fibroblasts as compared to healthy control fibroblasts. However, they draw parallels to the liver which is the organ primarily affected by GSD1a, stating that tissues other than the liver such as skin fibroblasts phenocopy the liver pathology (Discussion). Extrapolation of the findings to the liver is also made in the section "ATAC-seq, RNA-seq and EPIC methylation data integration". Here, the authors comment on the finding that identified genes are associated with tumour formation and draw parallels to hepatocellular carcinoma which is an important co-morbidity of GSD1a. These correlations, although interesting, should be presented as indications and not as "strong links". A major difference between fibroblasts and liver cells in the case of GSD1a is the massive accumulation of glycogen in the liver. This is a major metabolic feature which largely defines the disease's pathology. In addition to the similarities in the pathological features between the liver and other tissues such as fibroblasts, the authors should highlight this major difference and discuss their findings within this context.

    __Reply:__We thank the reviewer for this important comment. We have toned down the language correlating the regulation of gene expression between fibroblasts and liver in GSD1a. We have also alluded to the key metabolic difference between fibroblasts and liver - glycogen levels and turnover - in the second paragraph of the Discussion. We are aware that if our deep analyses were conducted on a different tissue with different basal metabolism the results might have been different. However, the GSD1a-pathogenic findings in fibroblasts suggest that they might also contribute to pathology in situ, perhaps by modulating the expression of functionally redundant genes.

    For basically all experiments performed in the study the authors follow the approach of culturing cells for 48 hours under serum and glucose starvation, followed be 24-hour cultivation in complete medium. This was practiced in a previous study by the authors (PMID: 34486811) to enhance the levels of glycogen in skin fibroblasts of patients with Adult Polyglucosan Body Disease. For the current study the selection of this treatment protocol is not sufficiently justified. Although, differences are described between patients' fibroblasts and controls under these conditions, it would have been interesting to address the reported parameters also at standard culturing conditions. This might be too much to ask for the purposes of this revision, but the authors may provide a better justification for the selection of the above treatment protocol and discuss whether the described phenotypic features are constitutive abnormalities present at all times or are induced by the metabolic stress imposed to the cells through this treatment.

    __Reply:__We thank the reviewer for pointing this important topic. Previously, we used the 72 h condition (48 h starvation followed by 24 h glucose supplementation) to attain two goals: generation of glycogen burden by excessive glucose re-uptake after glucose starvation and induction of basal autophagy by serum starvation so as to sensitize detection of the action of the autophagic activator GHF201 on a background of already induced autophagy. As stated, this 72 h condition was used previously in other GSD cell models (Kakhlon et al (2021) - GSDIV, Mishra et al (2024) - GSDIII, GSDII - in preparation), so we decided to use it in this work as well to enable cross-GSD comparison of GHF201 efficacy in GSD cell models. Moreover, as shown in Figure 1, the largest differences between HC and GSD1a fibroblasts, especially in lysosomal and mitochondrial features, were observed at the 72 h time condition. We therefore used this condition in all other fibroblasts experiments presented in this manuscript. Our ultimate aim was to test whether the metabolic reprograming induced in situ by the patients' diseased state before culturing generates stable epigenetic modifications withstanding seclusion from the original in situ environment. Thus, using the non-physiological 72 h condition, after the fibroblasts were cultured in full media remote from the in situ environment, can only confirm the stability and environment-independence of these metabolically-driven epigenetic modulations. We now provide this justification at the beginning of the Results section.

    In the Figures, the authors provide comparisons between controls and patient fibroblasts (+/- GHF201). Although the authors provide the respective p values in all figures, it is not clear which differences are considered significant and which are not. Since some of the indicated p values are > 0.0. The authors should indicate which of these changes are significant or non-significant and these should be presented and discussed accordingly in the text.

    __Reply:__We thank the reviewer for highlighting this important topic. We will add this information to the methods segment. Throughout the manuscript, p https://doi.org/10.1080/00031305.2018.1529624, Cumming, G. (2013). The New Statistics: Why and How. Psychological Science, 25(1), 7 29. https://doi.org/10.1177/0956797613504966 (Original work published 2014)). Along with the p values we presented all data points in each comparison and added bootstrap mediated 95 % confidence intervals as well. Since our sample size was small, we chose to focus on effect sizes, to use a higher p value threshold and to implement various advanced methodologies that allowed us to find important biological patterns.

    In Figure S2A, the authors show a reduction of glycogen levels in GSD1a fibroblasts following treatment with GHF201. Glycogen accumulation is central to this study, since a) is considered by the authors "a disease marker which is reversed by GHF201" - this is demonstrated in the liver of L.G6pc-/- mice and, according to the authors, replicated in the fibroblasts, b) as suggested by the authors it is the biochemical aberration that drives epigenetic modifications generating "disease memory". It is therefore important to appreciate whether GSD1a cells display pathologically increased levels of glycogen. This is also pertinent to the lack of G6PC expression in fibroblasts. The authors should include in Fig. S2A glycogen measurements of HC control fibroblasts cultured under the same conditions to compare with the levels present in GSD1a cells.

    __Reply:__We thank the reviewer for highlighting this issue. We added glycogen levels of HC to Figure 2SA as requested. Expectedly, glycogen levels are similar between HC and GSD1a fibroblasts because neither wild type G6PC1 in HC, or mutated G6PC1 in GSD1a fibroblasts is expressed. We have now corrected the manuscript text suggesting that glycogen is accumulated in GSD1a fibroblasts and rephrased the text to express the more versatile state where epigenetic modulation could be mediated by different metabolic perturbations according to the expression profile: G6PC1 mutant expressers (notably liver and kidney cells) could inhibit p-AMPK by glycogen accumulation, while non-expressers could inhibit p-AMPK by lowering NAD+. Text changes related to this new concept are found in the Results section "Exploring epigenetics as a phenotypic driver in GSD1a fibroblasts by ATAC-seq analysis" and in the Discussion section "Metabolic-driven, disease-associated programming of cell memory."

    Comparisons between protein levels (AMPK/pAMPK, Sirt1, TFEB, p62 ane PGC1a) are made on the basis of fluorescence intensity in immunostained cells. These results need to be supported by relevant western blot images to exclude that binding of the antibodies to unspecific sites contributes to the measured fluorescence.

    __Reply:__We thank the reviewer for this comment allowing us to clarify the reasoning behind the selected methods for the main markers identification. Throughout the manuscript we employed both Western blot and immunofluorescence experiments. We believe that immunofluorescence present as a more robust and efficient method for the following reasons: i. It allows to focus on proteins in their native state; ii. Immunofluorescence allows to observe proteins in relation to their location in the cells (for example TFs in nuclei area); iii. Immunofluorescence allows to focus on each cell and exclude cells which are dead, stressed or with a low viability characteristic; iv. Immunofluorescence allows to generate much more data. For the following reasons, the main proteins explored in this work we used immunofluorescence, in each immunofluorescence experiment we added a control for the secondary antibody alone, verifying the signal is related to the antibodies only. This information can be added if requested. Importantly, some of the antibodies used were recommended for immunofluorescence and not for Western blot. As the reviewer requested, we now provide western blot results for proteins that produced a signal with the antibodies in Western blots, all markers mentioned except TFEB were added to Figure S3 d.

    The authors demonstrate that treatment of GSD1a fibroblasts with histone deacetylase inhibitors reverses some of the phenotypic alterations. Given that GHF201 also improves these phenotypic differences it would be interesting to address whether GHF201 has any effect on histone acetylation.

    Reply: We strongly agree with this comment and have therfore tested for the effect of GHF201 on H3K27 acetylation levels as shown in Fiugre 3f and on the deacetylase -SIRT-1 as shown in Figure 3e, Figure S3d and representative images in Figure S2b.

    The authors report reduced levels of the transcription factors PGC1α and TFEB in GSD1a fibroblasts. Does this correlate with lower levels of expression of PGC1α and TFEB target genes in the RNA-seq experiments?

    Reply:

    We thank the reviewer for raising this topic, since there were thousands of differentially expressed genes and we cannot mention all we focused on the most important ones that comprise key pathways we wanted to highlight as described in the Results section. We have now linked in the Results section examples of PGC1α and TFEB target genes that were reduced due to lower levels of these transcription factors in GSD1a, as compared to HC cells. Importantly, a full list of the genes from the RNA-seq experiment can be found in Table S3. Genes regulated by TFEB contain the CLEAR (Coordinated Lysosomal Expression and Regulation) motif. Two notable genes regulated by CLEAR binding TFs such as TFEB, which are very important biologically, are cathepsin L and S (Figure 6A right) both of which were reduced in GSD1a and are now elaborated in the Results section referring to Figure 6a right. Additionally, Table S3 shows differentially expressed genes in GSD1a cells where there are many other lysosomal related genes that are downmodulated in GSD1a, we now added another important example, ATP6V0D2 to the Discussion as the reviewer suggested. As for PGC1alpha, a notable gene whose expression is up-modulated by PGC1alpha, which is down-modulated in GSD1a, is ALDH1A1 (Figure 6a right). In addition, we have now added PPARG and its coactivators alpha and beta to the discussion as requested by the reviewer, these genes are shown in Table S3 and are downmodulated in GSD1a. Finally, the transcriptional effect of PGC1alpha and TFEB is also mentioned in the Discussion within the cell phenotyping section, where we describe the deep impact of dysregulation of NAD+/NADH-Sirt-1-TFEB regulatory axis on the cell phenotype at all the levels described in the manuscript.

    Please revise the following sentences as the statements made are not adequately supported by the provided data a. "This NAD+/NADH increase correlated with reduced cytotoxicity and increased cell confluence (Figure 3d) suggesting that NAD+ availability prevails over ATP availability as an effector of cell thriving in GSD1a cells."

    __Reply:__If one ranks treatments according to NAD+/NADH (Figure 3c) and according to cytotoxicity (Figure 3d left) and cell confluence (Figure 3d right), then the mentioned correlation can be supported. ATP availability is compromised by gramicidin, yet gramicidin, which also increased NAD+/NADH, reduced cytotoxicity and enhanced cell confluence.

    b. "....in further support that respiration-dependent NAD+ availability mediate GHF201's corrective effect in GSD1a cells."

    __Reply:__Our data (Figure 3c) show that GHF201 increased NAD+/NADH both alone and with gramicidin.

    Please indicate on the densitometry graph of Fig. 10b the treatment (HDACi), for better visibility.

    __Reply:__We agree and have corrected the Figure as requested.

    The reference list (n=160) is probably too long for a research article.

    __Reply:__The number of references reflect the length and depth of the manuscript and we believe that each reference merits its place. We agree that the number of references is large but we are not sure which criteria to use to exclude some references and to reduce them to a more acceptable number that we assume would be determined by the publishing journal.

    The study is of high novelty and impact, as it proposes a so far undescribed biological mechanism contributing to disease pathology that could apply for general pathological conditions. Although this is a compelling concept, it is only demonstrated in skin fibroblasts which limits its applicability at an organismal level.

    __Reply:__We thank the reviewer for this comment and for raising the important comments that allowed us to improve our manuscript, please see our reply to point 1.

  2. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    General comments: the authors propose a vey intriguing concept, that metabolic abnormalities trigger epigenetic changes in tissues distal from the disease site, even in cells in which the affected gene is not expressed. This is demonstrated in primary fibroblasts from patients with Glycogen Storage Disease type 1a (GSD1a). The authors provide a large amount of data to support the compelling concept of "Disease-Associated Programming", a term that they have coined to describe this effect. The level of novelty is very high and so is the impact of the study, since the above may apply to many different pathological conditions. Although, the study is well performed and employs multiple approaches and analyses to address the raised hypothesis, there are some limitations and concerns that need to be addressed by the authors.

    1. The different phenotypic characteristics are only demonstrated in skin fibroblasts which is not sufficient to support the conclusions made in the Discussion about the general applicability of the proposed disease-induced, metabolite-driven epigenetic programming to all cells and tissues. The authors should discuss this as a limitation of the study and general conclusions should be formulated with more caution.
    2. The authors describe a range of alterations in patients' fibroblasts as compared to healthy control fibroblasts. However, they draw parallels to the liver which is the organ primarily affected by GSD1a, stating that tissues other than the liver such as skin fibroblasts phenocopy the liver pathology (Discussion). Extrapolation of the findings to the liver is also made in the section "ATAC-seq, RNA-seq and EPIC methylation data integration". Here, the authors comment on the finding that identified genes are associated with tumour formation and draw parallels to hepatocellular carcinoma which is an important co-morbidity of GSD1a. These correlations, although interesting, should be presented as indications and not as "strong links". A major difference between fibroblasts and liver cells in the case of GSD1a is the massive accumulation of glycogen in the liver. This is a major metabolic feature which largely defines the disease's pathology. In addition to the similarities in the pathological features between the liver and other tissues such as fibroblasts, the authors should highlight this major difference and discuss their findings within this context.
    3. For basically all experiments performed in the study the authors follow the approach of culturing cells for 48 hours under serum and glucose starvation, followed be 24-hour cultivation in complete medium. This was practiced in a previous study by the authors (PMID: 34486811) to enhance the levels of glycogen in skin fibroblasts of patients with Adult Polyglucosan Body Disease. For the current study the selection of this treatment protocol is not sufficiently justified. Although, differences are described between patients' fibroblasts and controls under these conditions, it would have been interesting to address the reported parameters also at standard culturing conditions. This might be too much to ask for the purposes of this revision, but the authors may provide a better justification for the selection of the above treatment protocol and discuss whether the described phenotypic features are constitutive abnormalities present at all times or are induced by the metabolic stress imposed to the cells through this treatment.
    4. In the Figures, the authors provide comparisons between controls and patient fibroblasts (+/- GHF201). Although the authors provide the respective p values in all figures, it is not clear which differences are considered significant and which are not. Since some of the indicated p values are > 0.0. The authors should indicate which of these changes are significant or non-significant and these should be presented and discussed accordingly in the text.
    5. In Figure S2A, the authors show a reduction of glycogen levels in GSD1a fibroblasts following treatment with GHF201. Glycogen accumulation is central to this study, since a) is considered by the authors "a disease marker which is reversed by GHF201" - this is demonstrated in the liver of L.G6pc-/- mice and, according to the authors, replicated in the fibroblasts, b) as suggested by the authors it is the biochemical aberration that drives epigenetic modifications generating "disease memory". It is therefore important to appreciate whether GSD1a cells display pathologically increased levels of glycogen. This is also pertinent to the lack of G6PC expression in fibroblasts. The authors should include in Fig. S2A glycogen measurements of HC control fibroblasts cultured under the same conditions to compare with the levels present in GSD1a cells.
    6. Comparisons between protein levels (AMPK/pAMPK, Sirt1, TFEB, p62 ane PGC1a) are made on the basis of fluorescence intensity in immunostained cells. These results need to be supported by relevant western blot images to exclude that binding of the antibodies to unspecific sites contributes to the measured fluorescence.
    7. The authors demonstrate that treatment of GSD1a fibroblasts with histone deacetylase inhibitors reverses some of the phenotypic alterations. Given that GHF201 also improves these phenotypic differences it would be interesting to address whether GHF201 has any effect on histone acetylation.
    8. The authors report reduced levels of the transcription factors PGC1α and TFEB in GSD1a fibroblasts. Does this correlate with lower levels of expression of PGC1α and TFEB target genes in the RNA-seq experiments?

    Minor points

    1. Please revise the following sentences as the statements made are not adequately supported by the provided data

    a. "This NAD+/NADH increase correlated with reduced cytotoxicity and increased cell confluence (Figure 3d) suggesting that NAD+ availability prevails over ATP availability as an effector of cell thriving in GSD1a cells."

    b. "....in further support that respiration-dependent NAD+ availability mediate GHF201's corrective effect in GSD1a cells."

    1. Please indicate on the densitometry graph of Fig. 10b the treatment (HDACi), for better visibility.
    2. The reference list (n=160) is probably too long for a research article.

    Significance

    The study is of high novelty and impact, as it proposes a so far undescribed biological mechanism contributing to disease pathology that could apply for general pathological conditions.

    Although this is a compelling concept, it is only demonstrated in skin fibroblasts which limits its applicability at an organismal level.

  3. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    Major Comments:

    1. Since dietary restriction is the only management strategy for GSD1a, the authors should clarify whether the patient fibroblast donors were on a dietary regimen and for how long. Given that fibroblasts do not express G6Pase, it is possible that the observed phenotype could be influenced by the patient's diet history.
    2. One of the main pathological features of GSD1a is glycogen buildup. The authors should compare glycogen levels between healthy controls and GSD1a fibroblasts and provide a dot plot analysis.
    3. Figure S2A - As mentioned above, the authors should present healthy control vs. patient fibroblast glycogen data. Without this, the rationale for using GHF201 is questionable.
    4. Figure S2B-C - If the authors propose that GHF201 reduces glycogen and increases intracellular glucose in GSD1a fibroblasts, they need direct evidence. Either directly quantifying glycogen levels or even better would be a labeling experiment to confirm that the free intracellular glucose originates from glycogen. Additionally, the reduction in sample size from N=24 in glycogen analysis to N=3 in the glucose assay needs justification.
    5. Figure S2B-C- It is not shown how GHF201 increases intracellular glucose? If glycophagy is a possibility, the authors should do an experiemnt to confirm this.
    6. Figure 2- How can GSD1a fibroblasts have significantly reduced OCR (Fig. 2B) but increased mitochondrial ATP production (Fig. 2H)?
    7. Why do GSD1a fibroblasts show reduced glycolytic ATP (Fig. 2H) despite increased glycolysis and glycolytic capacity (Fig. 2J-K)? The authors should clarify how many healthy control and patient fibroblast lines were compared per experiment. Given the wide age range, the unexpectedly small error bars raise concerns about variability and statistical robustness.
    8. Figure 5- The study should include Tamoxifen-untreated mice as a control to properly assess the efficacy of GHF201 in regulating glucose-6-P and glycogen levels.
    9. Fig. 5B-C - The authors should explain how GHF201 reduces glucose-6-P levels. Additionally, they should demonstrate whether GHF201 activates lysosomal pathways and induces autophagy in the liver of G6pc knockout mice, as claimed in the fibroblast experiments.

    Significance

    The manuscript explores mild physiological and metabolic disturbances in patient-derived fibroblasts lacking G6Pase expression, suggesting that these cells retain a "distinctive disease phenotype" of GSD1a. The manuscript is well written with well designed experiments. However, it remains unclear whether these phenotypes genuinely reflect the pathology of GSD1a-relevant tissues. The authors did not validate these findings in a liver-specific G6pc knockout mouse model, raising concerns about the study's relevance to GSD1a. Additionally, the lack of sufficient in vivo evidence undermines the therapeutic potential of GHF201 for this disease. Overall, the study lacks a few key pieces of evidence to completely justify its conclusions at both fundamental and experimental levels.

  4. Version published to 10.1101/2023.02.20.529109v2 on bioRxiv
  5. Version published to 10.1101/2023.02.20.529109v1 on bioRxiv