Genome wide analysis reveals heparan sulfate epimerase modulates TDP-43 proteinopathy

Pathological phosphorylated TDP-43 protein (pTDP) deposition drives neurodegeneration in amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration (FTLD-TDP). However, the cellular and genetic mechanisms at work in pathological TDP-43 toxicity are not fully elucidated. To identify genetic modifiers of TDP-43 neurotoxicity, we utilized a Caenorhabditis elegans model of TDP-43 proteinopathy expressing human mutant TDP-43 pan-neuronally (TDP-43 tg). In TDP-43 tg C. elegans, we conducted a genome-wide RNAi screen covering 16,767 C. elegans genes for loss of function genetic suppressors of TDP-43-driven motor dysfunction. We identified 46 candidate genes that when knocked down partially ameliorate TDP-43 related phenotypes; 24 of these candidate genes have conserved homologs in the human genome. To rigorously validate the RNAi findings, we crossed the TDP-43 transgene into the background of homozygous strong genetic loss of function mutations. We have confirmed 9 of the 24 candidate genes significantly modulate TDP-43 transgenic phenotypes. Among the validated genes we focused on, one of the most consistent genetic modifier genes protecting against pTDP accumulation and motor deficits was the heparan sulfate-modifying enzyme hse-5, the C. elegans homolog of glucuronic acid epimerase (GLCE). We found that knockdown of human GLCE in cultured human cells protects against oxidative stress induced pTDP accumulation. Furthermore, expression of glucuronic acid epimerase is significantly decreased in the brains of FTLD-TDP cases relative to normal controls, demonstrating the potential disease relevance of the candidate genes identified. Taken together these findings nominate glucuronic acid epimerase as a novel candidate therapeutic target for TDP-43 proteinopathies including ALS and FTLD-TDP.


4) The screen was performed in a mutant TDP-43 line, however the cell culture and FTLD-TDP would presumably express wild-type TDP-43. Therefore, it would be interesting to see if the hse-5 RNAi or mutant is protective against the WT TDP-43.
We now include data showing that hse-5(-) protects against wild-type TDP-43 as well (Supplemental Figure 2B). Table 2 is referred to in the text, but was not included in the manuscript for review. We thank reviewers for pointing out this absence. We apologize for this oversight and have included Table 2 with the resubmission materials.

5) Also,
Reviewer #2: Major points Unbiased genetic screens are powerful ways to advance understanding. However, sometimes it is not evident by which mechanism a modifier achieves its phenotypic effects. Here the authors sequentially limit their candidate genes to a select few before settling on hse-5/GLCE. There is no doubt that this gene has an effect on phosopho TDP-43 phenotypes. But the mechanism for the suppression is not sufficiently accounted for. p. 10 "Preliminary characterization of GLCE suggests it can act to detoxify pTDP and is a therapeutic target worth further examination." The choice of focusing on one gene, hse-5 is reasonable, but I have several questions. How does a potential ECM protein affect the levels of TDP-43 phosphorylation in worms and tissues? What is the working model? The gap in understanding should be addressed, at least as speculation in the discussion. We appreciate this suggestion and we have expanded our discussion of the potential protective mechanisms at work in hse-5/ GLCE loss of function . This includes discussion of the ability of ECM to interact and influence neuronal function, and in support of this, we have included new data showing that hse-5(-) partially restores aberrant pre-synaptic function in TDP-43 tg C. elegans (Fig 3G). We have added discussion of potential feedback mechanisms whereby an ECM protein can signal and affect cell phosphatases and kinases. We have also included a statement of our working model for the mechanism by which hse-5/ GLCE protects against TDP-43.
Has GLCE showed up as a modifier of ALS in GWAS studies? GLCE has not previously been identified as a modifier of ALS via GWAS. However, few of the previous ALS GWAS studies have been consistently replicated suggesting that many of these studies may be underpowered. We have added this information into the discussion.

Where is hse-5 expressed in C. elegans and how does this relate to the expression of mutant TDP-43?
A previous publication reported hse-5 expression predominantly in the hypodermis and intestine [1], and this is consistent with our own observation in an Phse-5::GFP reporter strain (data not shown). We have added these details to the text and expanded discussion of interactions between the hypodermis, extracellular matrix, and neurons.
Because this allele has additional unexpected deleterious phenotypes associated with it, we feel its use is inappropriate. We believe that precise gene deletions would be the best approach for further exploring these findings using null mutations.
While such mutations do not currently exist, we plan to generate such precise deletions using CRISPR mediated genome editing. However, we believe this time consuming work beyond the scope of the current manuscript. Regardless, the hse-5 data in the manuscript includes complimentary RNAi loss of function and authentic loss of function mutation validation in C.
elegans. In addition we show translational validation using siRNA assays of the hse-5 homolog GLCE in mammalian cultured cells, and IHC of GLCE in human pathological tissue. Taken together, we believe the preponderance of evidence suggests hse-5 plays a role in TDP-43 neurotoxicity.
The fact that hse-5 does not suppress neurodegeneration is unfortunate, perhaps this is the key mechanism, but at the same time maybe it is non-specific. Does hse-5 result in alternative neuronal connections that simply compensate for impaired movement in TDP-43 animals? While we do hypothesize that alternative neuronal connections are supporting function of remaining neurons in TDP-43 tg; hse-5(-), we have also added data showing hse-5(-) restores pre-synaptic function in TDP-43 tg ( Figure 3). We have expanded our treatment of these data in the results and discussion.
Do hse-5 animals have generally improved motility? hse-5(-) animals have similar motility to N2 (non-Tg). We have added these data as Supplemental Figure 2A.

How do they respond to aldicarb and/or levamisole?
We have tested hse-5(-) sensitivity to aldicarb and interestingly found a modest but significant defect in its responsiveness to aldicarb. We also tested the TDP-43 transgenic animals sensitivity to aldicarb and found a strong defect in aldicarb responsiveness. Excitingly, hse-5(-) was able to restore TDP-43 tg sensitivity, indicating second * that loss of hse-5 improves TDP-43 tg synaptic transmission. We have included this new data as Figure 3G and added discussion of this data in the manuscript text.
They say that RNAi candidates working in non-transgenic worms were excluded as part of the selection process, but it is possible that retesting of hse-5 RNAi in N2 worms would miss subtle effects. Thus, do hse-5 mutants have improved motility compared to N2 worms? We have added locomotion data for hse-5(-) as Supplemental Figure 2A showing that it has similar motility to N2 (non-Tg).
Furthermore, a control that is missing is tests versus wild type TDP-43 transgenics. The 2010 Liachko et al. manuscript describes the WT control strains, thus these are likely the floor for the amount of suppression possible with this approach and hse-5 should be tested in these WT TDP-43 strains. We now include data showing that hse-5(-) protects against wild-type TDP-43 as well (Supplemental Figure 2B).
Suppression of phenotypes associated with human disease proteins is of wide interest. Thus, it would be good to know if suppression of TDP-43 toxicity by hse-5 is specific. Does hse-5 protect against other forms of toxicity, perhaps Tau, given the role of phosphorylation in tau toxicity. This is a good point, and to address it we have tested whether hse-5(-) can protect against the motility defects in C. elegans expressing human tau. Interestingly, we find that hse-5(-) does not significantly protect against either wild-type or mutant V337M tau toxicity. In fact, motility defects in these animals are slightly enhanced in the hse-5(-) background. We have added these data to Supplemental Figure 2.

Does hse-5 affect the expression of the TDP-43 transgenes? This would be good to know as it could change interpretation of the potential mechanism for reduced pTDP-43. Perhaps for the other 4 candidate genes as well.
We agree this is an important point, and as such, we have performed qRT-PCR to assess mRNA expression levels of the TDP-43 transgene in all suppressor mutant backgrounds. We found that none of the suppressor strains have decreased expression of the transgene. These data have been included as Supplemental Figure  1A-B.

Finally, does hse-5 affect lifespan? Also good to know in terms of modifiers of age-dependent phenotypes.
We have tested the lifespan of OH1487 hse-5(-) and found it is long-lived relative to N2 (non-Tg); however, it does not rescue the shortened lifespan of TDP-43 tg animals. We have added these data as Supplemental Figure 3E and Supplemental Table 2. Table 2 is missing. We apologize for this and have now included Table 2 with the resubmission materials.

Minor issues
p. 5 "We found that 5 of the mutants tested had decreased levels of TDP-43 protein accompanied by reduced phosphorylation (Fig. 2C-G). Interestingly, these results indicate the remaining 4 suppressor genes that improve TDP-43 motor dysfunction do so without a direct impact on TDP-43 phosphorylation or protein levels." It looks like there is more of a mix of reduced total TDP-43 versus pTDP-43. It is clear that something is happening, but perhaps a simple relative quantification would aid the figure. Also, what happened to TDP-43 in the gly-8 mutants?
We have added quantitation of immunoblots to Figure 2, and have added additional immunoblots and quantitation as Supplemental Figure 1C-F). We have not explored the mechanism of TDP-43 turnover in the gly-8 mutant as this paper focuses primarily on hse-5. We agree that these are interesting points to add to the discussion. We have included details of the few genes previously identified in either ALS patient tissue or TDP-43 model organisms to the discussion. We have also expanded discussion of the C. elegans model, including some of the unique features that make it a useful model for studying TDP-43 phosphorylation in particular. It is unlikely that the endogenous C. elegans tdp-1 gene is affecting the phenotype of the TDP-43 tg animals. We have previously published characterization of the TDP-43 tg strain in a tdp-1 deletion mutant background, and found no effect of loss of tdp-1 on the TDP-43 tg phenotypes [2]. We have also added more discussion of potential relationships between TDP-43 and ECM proteins, and included more details about previously published work studying hse-5 in neuronal migration and axon regeneration. The description of motility changes in the non-Tg background is not applicable to this screen and so we have removed it from the flow chart (see revised Figure 1A). Fig 2A and 2B.

Along these lines, it would be useful to see the motility of WT in
We have added motility of N2 in Supplemental Fig. 1A, and included details of N2 motility into the figure legend of Fig. 2.
2. Figure 2 C-G, the authors show that 5 loss of function mutants, including hse-5, exhibit reduced phosphorylated tdp-43(M337V) protein levels, as well as total tdp-43(M337V) levels (although paqr mutants look to have similar total tdp but reduced phospho but the authors do not discuss this). The reduction in phospho-tdp could be due to several less interesting effects that have nothing to do with tdp-43, including transgene suppression (ie gly-8 mutants appear to have NO tdp expression) or reduced snb-1 promoter activity and subsequent reductions in tdp mRNA. Were any of these possibilities examined? This is important not only for the authors to interpret their data but also to prevent others from chasing nonexistent tdp-43 toxicity mechanisms. All three reviewers raised this important point about whether suppressors were affecting mRNA expression of the TDP-43 transgene. To address it, we have performed qRT-PCR to assess mRNA expression levels of the TDP-43 transgene in all suppressor mutant backgrounds. We found that none of the suppressor strains have decreased expression of the transgene. These data have been included as Supplemental Figure 1A-B. Fig 2C-G "indicate the remaining 4 suppressor genes improve TDP-43 motor dysfunction without a direct impact on TDP-43 phosphorylation or protein levels." Given that no data were presented for these 4 mutants, there are no data in the manuscript to support this conclusion. We now include the immunoblot and quantitation data for the suppressor genes that do not change TDP-43 protein levels as Supplemental Figure 1.