Modifiers of mutant huntingtin aggregation

Protein aggregation is a common hallmark of a number of age-related neurodegenerative diseases, including Alzheimer’s, Parkinson’s, and polyglutamine-expansion disorders such as Huntington’s disease, but how aggregation-prone proteins lead to pathology is not known. Using a genome-wide RNAi screen in a C. elegansmodel for polyglutamine aggregation, we previously identified 186 genes that suppress aggregation. Using an RNAi screen for human orthologs of these genes, we here present 26 human genes that suppress aggregation of mutant huntingtin in a human cell line. Among these are genes that have not been previously linked to mutant huntingtin aggregation. They include those encoding eukaryotic translation initiation, elongation and translation factors, and genes that have been previously associated with other neurodegenerative diseases, like the ATP-ase family gene 3-like 2 (AFG3L2) and ubiquitin-like modifier activating enzyme 1 (UBA1). Unravelling the role of these genes will broaden our understanding of the pathogenesis of Huntington’s disease.


Introduction
A number of age-related neurodegenerative disorders, like Huntington's, Parkinson's and Alzheimer's disease, are characterized by the appearance of insoluble protein aggregates in specific brain areas.In Huntington's disease and other polyglutamine disorders, a genetically-encoded expansion of a polyglutamine stretch in the disease-causing protein renders the mutant protein prone to aggregation [1].There is a direct inverse correlation between the polyglutamine tract length and the aggregation kinetics, as well as the age of disease onset.In addition, environmental factors and other genetic factors are known to contribute to the disease phenotype [1] [2].

Introduction
A number of age-related neurodegenerative disorders, like Huntington's, Parkinson's and Alzheimer's disease, are characterized by the appearance of insoluble protein aggregates in specific brain areas.In Huntington's disease and other polyglutamine disorders, a genetically-encoded expansion of a polyglutamine stretch in the disease-causing protein renders the mutant protein prone to aggregation [1].There is a direct inverse correlation between the polyglutamine tract length and the aggregation kinetics, as well as the age of disease onset.In addition, environmental factors and other genetic factors are known to contribute to the disease phenotype [1] [2].

Introduction
A number of age-related neurodegenerative disorders, like Huntington's, Parkinson's and Alzheimer's disease, are characterized by the appearance of insoluble protein aggregates in specific brain areas.In Huntington's disease and other polyglutamine disorders, a genetically-encoded expansion of a polyglutamine stretch in the disease-causing protein renders the mutant protein prone to aggregation [1].There is a direct inverse correlation between the polyglutamine tract length and the aggregation kinetics, as well as the age of disease onset.In addition, environmental factors and other genetic factors are known to contribute to the disease phenotype [1] [2].

PLOS Currents Huntington Disease
The microscopically visible structures, consisting of disease-specific misfolded proteins, are formed through putative toxic intermediates.While ample evidence suggests that the process of aggregation is coupled to pathology, the toxicity of the different aggregating species remains a matter of debate [3] [4] [5].Genes that modify protein aggregation might be disease modifiers in Huntington's disease, in other polyglutamine disorders, or in other related neurodegenerative diseases.Finding these genes will broaden our understanding of the pathogenesis of these neurodegenerative diseases, and may provide novel therapeutic insights.
Small model organisms are especially useful for genetic screening purposes as they are suitable for highthroughput genetic screens.Because their genetic material is highly conserved to humans, such studies can yield results that are relevant for human neurodegenerative diseases as well [6].A small nematode, Caenorhabditis elegans, which transgenically expresses polyglutamine-proteins fused to green fluorescent protein (GFP), shows an age-and polyglutamine-length-dependent aggregation similar to the human disease situation [7].This model has previously been used for a genome-wide screen in which expression of about 85% of the 17,000 genes of the worm was knocked down by RNAi through feeding [8].This study revealed 186 genes that, when knocked down, resulted in an increase in aggregation; these genes are therefore potential suppressors of protein aggregation.This set of genes includes a number of genes that function in processes that were already implicated in protein folding and misfolding in different model systems, like a number of molecular chaperones and proteasomal subunits [9] [10] [11] [12].In addition, genes functioning in processes that were not previously linked to protein aggregation disorders were found, as well as some genes with an unknown function.
In the current study we screened all the human homologs of the modifier genes identified in the C. elegansscreen [8].Several studies have already confirmed the validity of C. elegans-genetic screens to study human protein aggregation disorders, by assessing the effect of knockdown of the homolog genes on protein aggregation in other human cell models [13] [14] [15].Here, we investigated the role of orthologs of all the modifiers of polyglutamine aggregation found in C. elegans in mutant huntingtin aggregation.
In human cultured cells, huntingtin aggregation can be modelled by using expression constructs containing the first 67 amino acids of the huntingtin protein with an internal stretch with a variable number of glutamines fused to GFP [16].In this study, we expressed Htt-Q74-GFP and specifically knocked down the human orthologs of the C. elegans-genes by using shRNA constructs.We show that 26 human orthologs of the 186 C. elegansmodifier genes from our previous study have a robust effect on the aggregation of mutant huntingtin.Among these are genes which already have confirmed roles in mutant huntingtin aggregation, validating our experimental approach.In addition, we identified a number of genes with no previously known involvement in the aggregation of mutant huntingtin, such as the eukaryotic translation initiation, elongation and termination factors.

Orthology of C. elegans genes
To determine the human homologs of the 186 C. elegans genes identified in the genome-wide RNAi-screen, the KEGG-database was accessed through the KEGG API by a program script (source code available at http://www.molgenis.org/browser/molgenis_apps/trunk/apps/xgap/miscellaneous/kegg/Run.java?rev=8265).It attempts to map any gene identifier to the corresponding KEGG entry for the gene, followed by finding its closest orthologous gene in a specified target organism.The closest (or 'best match') ortholog was found by using a combination of the 'KEGG Orthology' and 'KEGG Sequence Similarity Database' sources.Homology was confirmed by reverse orthology (e.g.searching for the C. elegans ortholog of the human gene); some genes were discarded if they: (1) did not have a human ortholog, (2) showed reverse orthology to another C. elegans gene, (3) had homology with more than one C. elegans gene, or (4) the homology-score was lower than 0.20.In The microscopically visible structures, consisting of disease-specific misfolded proteins, are formed through putative toxic intermediates.While ample evidence suggests that the process of aggregation is coupled to pathology, the toxicity of the different aggregating species remains a matter of debate [3] [4] [5].Genes that modify protein aggregation might be disease modifiers in Huntington's disease, in other polyglutamine disorders, or in other related neurodegenerative diseases.Finding these genes will broaden our understanding of the pathogenesis of these neurodegenerative diseases, and may provide novel therapeutic insights.
Small model organisms are especially useful for genetic screening purposes as they are suitable for highthroughput genetic screens.Because their genetic material is highly conserved to humans, such studies can yield results that are relevant for human neurodegenerative diseases as well [6].A small nematode, Caenorhabditis elegans, which transgenically expresses polyglutamine-proteins fused to green fluorescent protein (GFP), shows an age-and polyglutamine-length-dependent aggregation similar to the human disease situation [7].This model has previously been used for a genome-wide screen in which expression of about 85% of the 17,000 genes of the worm was knocked down by RNAi through feeding [8].This study revealed 186 genes that, when knocked down, resulted in an increase in aggregation; these genes are therefore potential suppressors of protein aggregation.This set of genes includes a number of genes that function in processes that were already implicated in protein folding and misfolding in different model systems, like a number of molecular chaperones and proteasomal subunits [9] [10] [11] [12].In addition, genes functioning in processes that were not previously linked to protein aggregation disorders were found, as well as some genes with an unknown function.
In the current study we screened all the human homologs of the modifier genes identified in the C. elegansscreen [8].Several studies have already confirmed the validity of C. elegans-genetic screens to study human protein aggregation disorders, by assessing the effect of knockdown of the homolog genes on protein aggregation in other human cell models [13] [14] [15].Here, we investigated the role of orthologs of all the modifiers of polyglutamine aggregation found in C. elegans in mutant huntingtin aggregation.
In human cultured cells, huntingtin aggregation can be modelled by using expression constructs containing the first 67 amino acids of the huntingtin protein with an internal stretch with a variable number of glutamines fused to GFP [16].In this study, we expressed Htt-Q74-GFP and specifically knocked down the human orthologs of the C. elegans-genes by using shRNA constructs.We show that 26 human orthologs of the 186 C. elegansmodifier genes from our previous study have a robust effect on the aggregation of mutant huntingtin.Among these are genes which already have confirmed roles in mutant huntingtin aggregation, validating our experimental approach.In addition, we identified a number of genes with no previously known involvement in the aggregation of mutant huntingtin, such as the eukaryotic translation initiation, elongation and termination factors.

Orthology of C. elegans genes
To determine the human homologs of the 186 C. elegans genes identified in the genome-wide RNAi-screen, the KEGG-database was accessed through the KEGG API by a program script (source code available at http://www.molgenis.org/browser/molgenis_apps/trunk/apps/xgap/miscellaneous/kegg/Run.java?rev=8265).It attempts to map any gene identifier to the corresponding KEGG entry for the gene, followed by finding its closest orthologous gene in a specified target organism.The closest (or 'best match') ortholog was found by using a combination of the 'KEGG Orthology' and 'KEGG Sequence Similarity Database' sources.Homology was confirmed by reverse orthology (e.g.searching for the C. elegans ortholog of the human gene); some genes were discarded if they: (1) did not have a human ortholog, (2) showed reverse orthology to another C. elegans gene, (3) had homology with more than one C. elegans gene, or (4) the homology-score was lower than 0.20.In The microscopically visible structures, consisting of disease-specific misfolded proteins, are formed through putative toxic intermediates.While ample evidence suggests that the process of aggregation is coupled to pathology, the toxicity of the different aggregating species remains a matter of debate [3] [4] [5].Genes that modify protein aggregation might be disease modifiers in Huntington's disease, in other polyglutamine disorders, or in other related neurodegenerative diseases.Finding these genes will broaden our understanding of the pathogenesis of these neurodegenerative diseases, and may provide novel therapeutic insights.
Small model organisms are especially useful for genetic screening purposes as they are suitable for highthroughput genetic screens.Because their genetic material is highly conserved to humans, such studies can yield results that are relevant for human neurodegenerative diseases as well [6].A small nematode, Caenorhabditis elegans, which transgenically expresses polyglutamine-proteins fused to green fluorescent protein (GFP), shows an age-and polyglutamine-length-dependent aggregation similar to the human disease situation [7].This model has previously been used for a genome-wide screen in which expression of about 85% of the 17,000 genes of the worm was knocked down by RNAi through feeding [8].This study revealed 186 genes that, when knocked down, resulted in an increase in aggregation; these genes are therefore potential suppressors of protein aggregation.This set of genes includes a number of genes that function in processes that were already implicated in protein folding and misfolding in different model systems, like a number of molecular chaperones and proteasomal subunits [9] [10] [11] [12].In addition, genes functioning in processes that were not previously linked to protein aggregation disorders were found, as well as some genes with an unknown function.
In the current study we screened all the human homologs of the modifier genes identified in the C. elegansscreen [8].Several studies have already confirmed the validity of C. elegans-genetic screens to study human protein aggregation disorders, by assessing the effect of knockdown of the homolog genes on protein aggregation in other human cell models [13] [14] [15].Here, we investigated the role of orthologs of all the modifiers of polyglutamine aggregation found in C. elegans in mutant huntingtin aggregation.
In human cultured cells, huntingtin aggregation can be modelled by using expression constructs containing the first 67 amino acids of the huntingtin protein with an internal stretch with a variable number of glutamines fused to GFP [16].In this study, we expressed Htt-Q74-GFP and specifically knocked down the human orthologs of the C. elegans-genes by using shRNA constructs.We show that 26 human orthologs of the 186 C. elegansmodifier genes from our previous study have a robust effect on the aggregation of mutant huntingtin.Among these are genes which already have confirmed roles in mutant huntingtin aggregation, validating our experimental approach.In addition, we identified a number of genes with no previously known involvement in the aggregation of mutant huntingtin, such as the eukaryotic translation initiation, elongation and termination factors.

Orthology of C. elegans genes
To determine the human homologs of the 186 C. elegans genes identified in the genome-wide RNAi-screen, the KEGG-database was accessed through the KEGG API by a program script (source code available at http://www.molgenis.org/browser/molgenis_apps/trunk/apps/xgap/miscellaneous/kegg/Run.java?rev=8265).It attempts to map any gene identifier to the corresponding KEGG entry for the gene, followed by finding its closest orthologous gene in a specified target organism.The closest (or 'best match') ortholog was found by using a combination of the 'KEGG Orthology' and 'KEGG Sequence Similarity Database' sources.Homology was confirmed by reverse orthology (e.g.searching for the C. elegans ortholog of the human gene); some genes were discarded if they: (1) did not have a human ortholog, (2) showed reverse orthology to another C. elegans gene, (3) had homology with more than one C. elegans gene, or (4) the homology-score was lower than 0.20.In

PLOS Currents Huntington Disease
total, 177 human homologs were identified, the identity between human and C. elegans genes was calculated on the amino-acid level.
Design siRNA-targets and shRNA-oligonucleotides and cloning strategy Three siRNA-targets per gene were designed through the OligoEngine website using the Gene Silencing RNAi Design Tool optimized for pSUPER vectors.Target sequences were designed on the cDNA-sequences as found in the NCBI database.Three 19-nucleotide target sequences were selected based on the 'general rules' of RNAidesign [17] [18], the 'recommended targets' selected by the program, by the number of Blast hits, and by the potency and distribution of the target regions over the mRNA-sequence.For some genes, especially small ones, only one or two siRNA-sequences could be designed.The 19-nucleotide-sequences were used in a standard pSUPER-cloning strategy [19].
The pSUPER-vector was digested with BglII and HindIII overnight at 37°C.60 base-long oligonucleotide sequences were ordered in 96-well plate format at a concentration of 50µg/µl.For annealing, forward and reverse oligos were combined with milliQ and NEB buffer 2, heated for 15 min to 95°C and cooled overnight.
The annealed sequences were diluted 1:50 in milliQ, and 1 µl was used in an overnight ligation at 16°C with 50 ng digested vector.5 µl of the ligation mix was transformed to 25 µl competent DH5α by heat-shock and plated on 6 cm LB-Amp-plates.Three colonies per ligation were used for colony-PCR with primers pRS for (CCCTTGAACCTCCTCGTTCGACC) and pRSrev (GAGACGTGCTACTTCCATTTGTC). Positive clones showed a 650 bpinsert, compared to negative clones showing a 600 bp-insert, this difference was observed on a 1.5% agarose gel.Positive clones were grown in liquid cultures to obtain plasmid DNA and sequences were verified by sequencing with pRSseq (GCTGACGTCATCAACCCGCT).

Cell culture and transfection
HEK293-cells were cultured using standard conditions (DMEM+10%FBS+1% penstrep).For transfection, 15×10 5 cells were plated into 12-well plates on glass coverslips coated with poly-D-lysine.After 16 hours, cells were transfected with equal amounts of shRNA-vectors and Htt-exon1-Q74-GFP (16) using PEI as the transfection agent (a total of 1 µg DNA per well).After 48 hours, cells were fixed with 4% PFA for 20 minutes at room temperature, followed by washing with milliQ and dehydration in 70% and 100% alcohol, and mounted on glass slides using Vectashield containing DAPI.In every experiment, shRNA-constructs with a scrambled (SCR) 19ntsequence, which did not target any part of the human genome, shRNA targeting green fluorescent protein (GFP), and shRNA targeting chaperonin-containing TCP1, Subunit 2 (CCT2) were used as positive controls.

Aggregate counting and selecting positive candidates
Aggregates were counted using two different techniques.One technique was making multiple images of the GFP-signal and the DAPI-staining per coverslip by regular fluorescence microscopy.Using a batch macro made in ImageJ, pictures were analyzed to count the number of cells and the number of aggregates.The number of aggregates was calculated as percentage aggregates per total number of cells, averaged over 5 images.
Coverslips were also photographed using TissueFAXS (TissueGnostics, Vienna, Austria) and DAPI-staining, and aggregates were monitored using TissueQuest software.Comparing ImageJ/TissueQuest results of the same experiment showed similar results.
To be able to compare the results of different experiments, the level of aggregation with the SCR shRNA was set to 100%; the results of the knockdown experiments in the same experiment were normalized to this number.
When shRNA-constructs targeting GFP or CCT2 did not lead to a reduction or induction of aggregation, the experiment was repeated.After the initial screen, positive shRNA-constructs were identified if they showed an induction of the percentage of aggregates of at least 150% compared to the SCR-shRNA.The experiments with these shRNA-constructs were repeated in triplicate.
Figure 1 presents a graphic overview of the experiments.total, 177 human homologs were identified, the identity between human and C. elegans genes was calculated on the amino-acid level.

Design siRNA-targets and shRNA-oligonucleotides and cloning strategy
Three siRNA-targets per gene were designed through the OligoEngine website using the Gene Silencing RNAi Design Tool optimized for pSUPER vectors.Target sequences were designed on the cDNA-sequences as found in the NCBI database.Three 19-nucleotide target sequences were selected based on the 'general rules' of RNAidesign [17] [18], the 'recommended targets' selected by the program, by the number of Blast hits, and by the potency and distribution of the target regions over the mRNA-sequence.For some genes, especially small ones, only one or two siRNA-sequences could be designed.The 19-nucleotide-sequences were used in a standard pSUPER-cloning strategy [19].
The pSUPER-vector was digested with BglII and HindIII overnight at 37°C.60 base-long oligonucleotide sequences were ordered in 96-well plate format at a concentration of 50µg/µl.For annealing, forward and reverse oligos were combined with milliQ and NEB buffer 2, heated for 15 min to 95°C and cooled overnight.
The annealed sequences were diluted 1:50 in milliQ, and 1 µl was used in an overnight ligation at 16°C with 50 ng digested vector.5 µl of the ligation mix was transformed to 25 µl competent DH5α by heat-shock and plated on 6 cm LB-Amp-plates.Three colonies per ligation were used for colony-PCR with primers pRS for (CCCTTGAACCTCCTCGTTCGACC) and pRSrev (GAGACGTGCTACTTCCATTTGTC). Positive clones showed a 650 bpinsert, compared to negative clones showing a 600 bp-insert, this difference was observed on a 1.5% agarose gel.Positive clones were grown in liquid cultures to obtain plasmid DNA and sequences were verified by sequencing with pRSseq (GCTGACGTCATCAACCCGCT).

Cell culture and transfection
HEK293-cells were cultured using standard conditions (DMEM+10%FBS+1% penstrep).For transfection, 15×10 5 cells were plated into 12-well plates on glass coverslips coated with poly-D-lysine.After 16 hours, cells were transfected with equal amounts of shRNA-vectors and Htt-exon1-Q74-GFP (16) using PEI as the transfection agent (a total of 1 µg DNA per well).After 48 hours, cells were fixed with 4% PFA for 20 minutes at room temperature, followed by washing with milliQ and dehydration in 70% and 100% alcohol, and mounted on glass slides using Vectashield containing DAPI.In every experiment, shRNA-constructs with a scrambled (SCR) 19ntsequence, which did not target any part of the human genome, shRNA targeting green fluorescent protein (GFP), and shRNA targeting chaperonin-containing TCP1, Subunit 2 (CCT2) were used as positive controls.

Aggregate counting and selecting positive candidates
Aggregates were counted using two different techniques.One technique was making multiple images of the GFP-signal and the DAPI-staining per coverslip by regular fluorescence microscopy.Using a batch macro made in ImageJ, pictures were analyzed to count the number of cells and the number of aggregates.The number of aggregates was calculated as percentage aggregates per total number of cells, averaged over 5 images.
Coverslips were also photographed using TissueFAXS (TissueGnostics, Vienna, Austria) and DAPI-staining, and aggregates were monitored using TissueQuest software.Comparing ImageJ/TissueQuest results of the same experiment showed similar results.
To be able to compare the results of different experiments, the level of aggregation with the SCR shRNA was set to 100%; the results of the knockdown experiments in the same experiment were normalized to this number.
When shRNA-constructs targeting GFP or CCT2 did not lead to a reduction or induction of aggregation, the experiment was repeated.After the initial screen, positive shRNA-constructs were identified if they showed an induction of the percentage of aggregates of at least 150% compared to the SCR-shRNA.The experiments with these shRNA-constructs were repeated in triplicate.
Figure 1 presents a graphic overview of the experiments.total, 177 human homologs were identified, the identity between human and C. elegans genes was calculated on the amino-acid level.

Design siRNA-targets and shRNA-oligonucleotides and cloning strategy
Three siRNA-targets per gene were designed through the OligoEngine website using the Gene Silencing RNAi Design Tool optimized for pSUPER vectors.Target sequences were designed on the cDNA-sequences as found in the NCBI database.Three 19-nucleotide target sequences were selected based on the 'general rules' of RNAidesign [17] [18], the 'recommended targets' selected by the program, by the number of Blast hits, and by the potency and distribution of the target regions over the mRNA-sequence.For some genes, especially small ones, only one or two siRNA-sequences could be designed.The 19-nucleotide-sequences were used in a standard pSUPER-cloning strategy [19].
The pSUPER-vector was digested with BglII and HindIII overnight at 37°C.60 base-long oligonucleotide sequences were ordered in 96-well plate format at a concentration of 50µg/µl.For annealing, forward and reverse oligos were combined with milliQ and NEB buffer 2, heated for 15 min to 95°C and cooled overnight.
The annealed sequences were diluted 1:50 in milliQ, and 1 µl was used in an overnight ligation at 16°C with 50 ng digested vector.5 µl of the ligation mix was transformed to 25 µl competent DH5α by heat-shock and plated on 6 cm LB-Amp-plates.Three colonies per ligation were used for colony-PCR with primers pRS for (CCCTTGAACCTCCTCGTTCGACC) and pRSrev (GAGACGTGCTACTTCCATTTGTC). Positive clones showed a 650 bpinsert, compared to negative clones showing a 600 bp-insert, this difference was observed on a 1.5% agarose gel.Positive clones were grown in liquid cultures to obtain plasmid DNA and sequences were verified by sequencing with pRSseq (GCTGACGTCATCAACCCGCT).

Cell culture and transfection
HEK293-cells were cultured using standard conditions (DMEM+10%FBS+1% penstrep).For transfection, 15×10 5 cells were plated into 12-well plates on glass coverslips coated with poly-D-lysine.After 16 hours, cells were transfected with equal amounts of shRNA-vectors and Htt-exon1-Q74-GFP (16) using PEI as the transfection agent (a total of 1 µg DNA per well).After 48 hours, cells were fixed with 4% PFA for 20 minutes at room temperature, followed by washing with milliQ and dehydration in 70% and 100% alcohol, and mounted on glass slides using Vectashield containing DAPI.In every experiment, shRNA-constructs with a scrambled (SCR) 19ntsequence, which did not target any part of the human genome, shRNA targeting green fluorescent protein (GFP), and shRNA targeting chaperonin-containing TCP1, Subunit 2 (CCT2) were used as positive controls.

Aggregate counting and selecting positive candidates
Aggregates were counted using two different techniques.One technique was making multiple images of the GFP-signal and the DAPI-staining per coverslip by regular fluorescence microscopy.Using a batch macro made in ImageJ, pictures were analyzed to count the number of cells and the number of aggregates.The number of aggregates was calculated as percentage aggregates per total number of cells, averaged over 5 images.
Coverslips were also photographed using TissueFAXS (TissueGnostics, Vienna, Austria) and DAPI-staining, and aggregates were monitored using TissueQuest software.Comparing ImageJ/TissueQuest results of the same experiment showed similar results.
To be able to compare the results of different experiments, the level of aggregation with the SCR shRNA was set to 100%; the results of the knockdown experiments in the same experiment were normalized to this number.
When shRNA-constructs targeting GFP or CCT2 did not lead to a reduction or induction of aggregation, the experiment was repeated.After the initial screen, positive shRNA-constructs were identified if they showed an induction of the percentage of aggregates of at least 150% compared to the SCR-shRNA.The experiments with these shRNA-constructs were repeated in triplicate.
Figure 1 presents a graphic overview of the experiments.

C. elegans modifiers of protein aggregation are highly conserved to humans
In a C. elegans screen for modifiers of polyglutamine aggregation, we previously identified 186 C. elegans genes that modify polyglutamine aggregation [8].To establish functional conservation of the human counterparts of these genes, we identified the closest human orthologs of these genes by automated analyses.
The KEGG database retrieval script provided us with information about BLAST identity, gene name, definition, NCBI-GI, GeneID, GenbankID, and amino acid and nucleotide sequences.A reverse test was performed by searching the C. elegans homolog of the human gene to avoid paralogy.
In total, we found 177 human genes that had at least 20% amino acid sequence homology to a human gene (listed in table 1).70% (124/177) of these human genes had more than 40% identity on the amino acid level with their C. elegans homolog (table 1).High identity was found within the groups of protein degradation, folding, synthesis and transport (mean identity 60%, of which 85% of the genes had an identity >40%).These are essential cellular processes that are highly conserved throughout evolution.For nine of the C. elegans genes, no human homolog could be assigned for various reasons: for five genes, there was another gene in the list that led to the same human homolog (these were doubles); three genes were not conserved to humans at all, and one gene no longer existed in the databases searched.

C. elegans modifiers of protein aggregation are highly conserved to humans
In a C. elegans screen for modifiers of polyglutamine aggregation, we previously identified 186 C. elegans genes that modify polyglutamine aggregation [8].To establish functional conservation of the human counterparts of these genes, we identified the closest human orthologs of these genes by automated analyses.
The KEGG database retrieval script provided us with information about BLAST identity, gene name, definition, NCBI-GI, GeneID, GenbankID, and amino acid and nucleotide sequences.A reverse test was performed by searching the C. elegans homolog of the human gene to avoid paralogy.
In total, we found 177 human genes that had at least 20% amino acid sequence homology to a human gene (listed in table 1).70% (124/177) of these human genes had more than 40% identity on the amino acid level with their C. elegans homolog (table 1).High identity was found within the groups of protein degradation, folding, synthesis and transport (mean identity 60%, of which 85% of the genes had an identity >40%).These are essential cellular processes that are highly conserved throughout evolution.For nine of the C. elegans genes, no human homolog could be assigned for various reasons: for five genes, there was another gene in the list that led to the same human homolog (these were doubles); three genes were not conserved to humans at all, and one gene no longer existed in the databases searched.

C. elegans modifiers of protein aggregation are highly conserved to humans
In a C. elegans screen for modifiers of polyglutamine aggregation, we previously identified 186 C. elegans genes that modify polyglutamine aggregation [8].To establish functional conservation of the human counterparts of these genes, we identified the closest human orthologs of these genes by automated analyses.
The KEGG database retrieval script provided us with information about BLAST identity, gene name, definition, NCBI-GI, GeneID, GenbankID, and amino acid and nucleotide sequences.A reverse test was performed by searching the C. elegans homolog of the human gene to avoid paralogy.
In total, we found 177 human genes that had at least 20% amino acid sequence homology to a human gene (listed in table 1).70% (124/177) of these human genes had more than 40% identity on the amino acid level with their C. elegans homolog (table 1).High identity was found within the groups of protein degradation, folding, synthesis and transport (mean identity 60%, of which 85% of the genes had an identity >40%).These are essential cellular processes that are highly conserved throughout evolution.For nine of the C. elegans genes, no human homolog could be assigned for various reasons: for five genes, there was another gene in the list that led to the same human homolog (these were doubles); three genes were not conserved to humans at all, and one gene no longer existed in the databases searched.

Huntingtin aggregation
To establish the role of these genes in the aggregation of mutant huntingtin (Htt) in human cultured cells, we generated three shRNA-constructs per gene and transfected these in combination with a huntingtin-exon1 construct with 74 glutamines, fused to GFP (Htt-Q74-GFP) to HEK293-cells.We measured the number of aggregates per total number of cells.As a negative control, we used an SCR sequence that did not target any human genes as verified by BLAST.In addition, shRNA-constructs targeting the DNA sequence of GFP were used as positive controls; these shRNA-constructs resulted in a reduction of protein aggregation (figure 2).
In the original screen in C. elegans, six members of the chaperonin TRiC were found to be modifiers of polyglutamine aggregation [8].Earlier studies have confirmed the role of these genes in the aggregation of polyglutamine and mutant huntingtin in yeast [20], neuron-like cells [14], and cultured human cell-lines [13], including HEK-293 cells.To test the validity of our screen, we first designed shRNA-constructs against the human homologs of these genes and observed that knockdown of the majority of these subunits led to an increase in mutant huntingtin aggregation (table 1, figure 2).

Huntingtin aggregation
To establish the role of these genes in the aggregation of mutant huntingtin (Htt) in human cultured cells, we generated three shRNA-constructs per gene and transfected these in combination with a huntingtin-exon1 construct with 74 glutamines, fused to GFP (Htt-Q74-GFP) to HEK293-cells.We measured the number of aggregates per total number of cells.As a negative control, we used an SCR sequence that did not target any human genes as verified by BLAST.In addition, shRNA-constructs targeting the DNA sequence of GFP were used as positive controls; these shRNA-constructs resulted in a reduction of protein aggregation (figure 2).
In the original screen in C. elegans, six members of the chaperonin TRiC were found to be modifiers of polyglutamine aggregation [8].Earlier studies have confirmed the role of these genes in the aggregation of polyglutamine and mutant huntingtin in yeast [20], neuron-like cells [14], and cultured human cell-lines [13], including HEK-293 cells.To test the validity of our screen, we first designed shRNA-constructs against the human homologs of these genes and observed that knockdown of the majority of these subunits led to an increase in mutant huntingtin aggregation (table 1, figure 2).

Huntingtin aggregation
To establish the role of these genes in the aggregation of mutant huntingtin (Htt) in human cultured cells, we generated three shRNA-constructs per gene and transfected these in combination with a huntingtin-exon1 construct with 74 glutamines, fused to GFP (Htt-Q74-GFP) to HEK293-cells.We measured the number of aggregates per total number of cells.As a negative control, we used an SCR sequence that did not target any human genes as verified by BLAST.In addition, shRNA-constructs targeting the DNA sequence of GFP were used as positive controls; these shRNA-constructs resulted in a reduction of protein aggregation (figure 2).
In the original screen in C. elegans, six members of the chaperonin TRiC were found to be modifiers of polyglutamine aggregation [8].Earlier studies have confirmed the role of these genes in the aggregation of polyglutamine and mutant huntingtin in yeast [20], neuron-like cells [14], and cultured human cell-lines [13], including HEK-293 cells.To test the validity of our screen, we first designed shRNA-constructs against the human homologs of these genes and observed that knockdown of the majority of these subunits led to an increase in mutant huntingtin aggregation (table 1, figure 2).We evaluated alterations of aggregation of mutant huntingtin in cells transfected with a total of 524 shRNAvectors (table 1).The cut-off for induction of aggregation was set at 150% compared to the SCR shRNA within the same experiment.Of the original 524 shRNA-constructs corresponding to the 177 genes, 147 shRNAconstructs, representing 90 different genes, induced aggregation of more than 150% (table 1).Knockdown of a number of genes showed an opposite effect compared to their effect on polyglutamine aggregation in C. elegans .For example, knockdown of many ribosomal subunits led to lower aggregation levels.This was also the case for some ATPases and ATP-synthases.Further, knockdown of the genes previously predicted to be involved in protein transport did not have an obvious effect on the aggregation of mutant huntingtin (table 1).

PLOS Currents Huntington Disease
In sum, not all the genes identified as suppressors of polyglutamine aggregation in C. elegans also function as suppressors of mutant huntingtin aggregation in cultured mammalian cells.

human genes were validated as suppressors of mutant Huntingtin aggregation
In the first set of experiments, for 30 genes two shRNA-constructs led to an induction of mutant huntingtin aggregation of >150%, and for 13 genes this was the case with all three shRNA-constructs (see also table 1).
Thereafter, experiments with shRNA-constructs for these 43 genes were repeated in triplicate, for all three shRNA-constructs.The results were averaged and after three repeats, 58 shRNA-constructs, corresponding to 26 human genes, showed an induction of mutant huntingtin aggregation of >150%.The percentage of aggregation of the three repeats and the averages can be found in figure 3 and table 2. We evaluated alterations of aggregation of mutant huntingtin in cells transfected with a total of 524 shRNAvectors (table 1).The cut-off for induction of aggregation was set at 150% compared to the SCR shRNA within the same experiment.Of the original 524 shRNA-constructs corresponding to the 177 genes, 147 shRNAconstructs, representing 90 different genes, induced aggregation of more than 150% (table 1).Knockdown of a number of genes showed an opposite effect compared to their effect on polyglutamine aggregation in C. elegans .For example, knockdown of many ribosomal subunits led to lower aggregation levels.This was also the case for some ATPases and ATP-synthases.Further, knockdown of the genes previously predicted to be involved in protein transport did not have an obvious effect on the aggregation of mutant huntingtin (table 1).
In sum, not all the genes identified as suppressors of polyglutamine aggregation in C. elegans also function as suppressors of mutant huntingtin aggregation in cultured mammalian cells.

human genes were validated as suppressors of mutant Huntingtin aggregation
In the first set of experiments, for 30 genes two shRNA-constructs led to an induction of mutant huntingtin aggregation of >150%, and for 13 genes this was the case with all three shRNA-constructs (see also table 1).
Thereafter, experiments with shRNA-constructs for these 43 genes were repeated in triplicate, for all three shRNA-constructs.The results were averaged and after three repeats, 58 shRNA-constructs, corresponding to 26 human genes, showed an induction of mutant huntingtin aggregation of >150%.The percentage of aggregation of the three repeats and the averages can be found in figure 3 and table 2. We evaluated alterations of aggregation of mutant huntingtin in cells transfected with a total of 524 shRNAvectors (table 1).The cut-off for induction of aggregation was set at 150% compared to the SCR shRNA within the same experiment.Of the original 524 shRNA-constructs corresponding to the 177 genes, 147 shRNAconstructs, representing 90 different genes, induced aggregation of more than 150% (table 1).Knockdown of a number of genes showed an opposite effect compared to their effect on polyglutamine aggregation in C. elegans .For example, knockdown of many ribosomal subunits led to lower aggregation levels.This was also the case for some ATPases and ATP-synthases.Further, knockdown of the genes previously predicted to be involved in protein transport did not have an obvious effect on the aggregation of mutant huntingtin (table 1).
In sum, not all the genes identified as suppressors of polyglutamine aggregation in C. elegans also function as suppressors of mutant huntingtin aggregation in cultured mammalian cells.

human genes were validated as suppressors of mutant Huntingtin aggregation
In the first set of experiments, for 30 genes two shRNA-constructs led to an induction of mutant huntingtin aggregation of >150%, and for 13 genes this was the case with all three shRNA-constructs (see also table 1).
Thereafter, experiments with shRNA-constructs for these 43 genes were repeated in triplicate, for all three shRNA-constructs.The results were averaged and after three repeats, 58 shRNA-constructs, corresponding to 26 human genes, showed an induction of mutant huntingtin aggregation of >150%.The percentage of aggregation of the three repeats and the averages can be found in figure 3 and table 2. It can be seen that several functional classes are overrepresented among the final positives.Some of these cellular pathways, like protein folding and degradation, play a role in protein quality control and are therefore obvious regulators of protein aggregation.For example, two or three shRNA-constructs targeting the genes for four of the eleven proteasomal subunits in the C. elegans screen ( PSMA7, PSMB3, PSMD11 and PSMD14 ), led to an induction of mutant huntingtin aggregation of >150%.And of the six CCT subunits in the C. elegansscreen, five ( CCT2, CCT4, CCT5, CCT7 and TCP-1 ) met the criteria for being a suppressor of mutant huntingtin aggregation in this study.In our C. elegans screen, two hsp70-chapersones (homologs of HSPA8 and HSPA9 ) were found to be modifiers of aggregation, in the human cells only HSPA8 was a modifier of huntingtin aggregation.However, one shRNA for HSPA9 also induces aggregation to 140%.shRNA-constructs targeting four eukaryotic translation initiation factors and two eukaryotic translation elongation factors also led to an induction of mutant huntingtin aggregation.Their role in protein aggregation, or in mutant huntingtin aggregation more specifically, is novel.Furthermore, shRNA-constructs targeting two nuclear ribonucleoproteins, SNRNP200 and SNRPF, led to an induction of mutant huntingtin aggregation.The proteins encoded by these genes function in a complex with the SMN protein; mutations in the SMN gene are the causative factor in spinal muscular atrophy, implicating a general role for the protein in neurodegeneration [21] .It can be seen that several functional classes are overrepresented among the final positives.Some of these cellular pathways, like protein folding and degradation, play a role in protein quality control and are therefore obvious regulators of protein aggregation.For example, two or three shRNA-constructs targeting the genes for four of the eleven proteasomal subunits in the C. elegans screen ( PSMA7, PSMB3, PSMD11 and PSMD14 ), led to an induction of mutant huntingtin aggregation of >150%.And of the six CCT subunits in the C. elegansscreen, five ( CCT2, CCT4, CCT5, CCT7 and TCP-1 ) met the criteria for being a suppressor of mutant huntingtin aggregation in this study.In our C. elegans screen, two hsp70-chapersones (homologs of HSPA8 and HSPA9 ) were found to be modifiers of aggregation, in the human cells only HSPA8 was a modifier of huntingtin aggregation.However, one shRNA for HSPA9 also induces aggregation to 140%.shRNA-constructs targeting four eukaryotic translation initiation factors and two eukaryotic translation elongation factors also led to an induction of mutant huntingtin aggregation.Their role in protein aggregation, or in mutant huntingtin aggregation more specifically, is novel.Furthermore, shRNA-constructs targeting two nuclear ribonucleoproteins, SNRNP200 and SNRPF, led to an induction of mutant huntingtin aggregation.The proteins encoded by these genes function in a complex with the SMN protein; mutations in the SMN gene are the causative factor in spinal muscular atrophy, implicating a general role for the protein in neurodegeneration [21] .It can be seen that several functional classes are overrepresented among the final positives.Some of these cellular pathways, like protein folding and degradation, play a role in protein quality control and are therefore obvious regulators of protein aggregation.For example, two or three shRNA-constructs targeting the genes for four of the eleven proteasomal subunits in the C. elegans screen ( PSMA7, PSMB3, PSMD11 and PSMD14 ), led to an induction of mutant huntingtin aggregation of >150%.And of the six CCT subunits in the C. elegansscreen, five ( CCT2, CCT4, CCT5, CCT7 and TCP-1 ) met the criteria for being a suppressor of mutant huntingtin aggregation in this study.In our C. elegans screen, two hsp70-chapersones (homologs of HSPA8 and HSPA9 ) were found to be modifiers of aggregation, in the human cells only HSPA8 was a modifier of huntingtin aggregation.However, one shRNA for HSPA9 also induces aggregation to 140%.shRNA-constructs targeting four eukaryotic translation initiation factors and two eukaryotic translation elongation factors also led to an induction of mutant huntingtin aggregation.Their role in protein aggregation, or in mutant huntingtin aggregation more specifically, is novel.Furthermore, shRNA-constructs targeting two nuclear ribonucleoproteins, SNRNP200 and SNRPF, led to an induction of mutant huntingtin aggregation.The proteins encoded by these genes function in a complex with the SMN protein; mutations in the SMN gene are the causative factor in spinal muscular atrophy, implicating a general role for the protein in neurodegeneration [21] .

Table 2. List of genes in which knocking down by two or three shRNA-constructs leads to an induction in mutant huntingtin aggregation
Functional classes as previously assigned (8), C. elegans cosmid and gene, human gene name, Gene ID, full name and identity are shown, as in table 1. #shRNAs represent the number of shRNA-constructs that lead to an increase in aggregation of >150% after three experiments.The last column shows the highest average induction observed for that gene In sum, in this study we have shown that 26 modifiers of polyglutamine aggregation in C. elegans have a conserved function towards modifying mutant huntingtin aggregation in cultured human cells and we identified some genes not previously implicated in this pathway.

Previously identified modifiers of (huntingtin) aggregation
In the current study, we have monitored the effect on protein aggregation of C. elegans genes shown to modify polyglutamine aggregation, in a human cell line [8].The C. elegans screen was focused on finding genes that, upon knockdown, increase aggregation and are therefore suppressors of protein aggregation.The original screen in C. elegans revealed the aggregation-suppressing function of some genes that were previously linked to protein aggregation, including a number of chaperonins [8] known to assist in protein folding.Earlier studies had already confirmed their functional conservation towards modifying mutant huntingtin aggregation in human cultured cells [13] [14] [15] and in yeast [20].In the study by Kitamura et al [13], knockdown of CCTζ (CCT6A) and CCTα (TCP-1) led to stimulating aggregation of expanded polyglutamine and mutant huntingtin in HEK293and HeLa cells, respectively.Tam et al [14] reported knockdown of subunits of the Chaperonin-Containing TCP1 Complex (CCT), which is believed to reduce total CCT-levels, leading to higher levels of aggregation in N2Acells.This is in line with our results, which show that five out of the six chaperonin-subunits found in the C. elegans screen are suppressors of mutant Huntingtin aggregation (table 2).
Of the eleven proteasomal subunits in the C. elegans screen, four were also confirmed as aggregationmodifying genes in our study.It is well known that the proteasome is involved in the targeting and degradation of misfolded proteins, but the exact role of the proteasome in the degradation of expanded polyglutamine proteins remains debated [22] [23].Inclusions in neurons from Huntington disease models can be labelled with antibodies against ubiquitin [24] [25], but it is unclear whether the proteasome can degrade expanded polyglutamine proteins [26] [27].Similar conflicting results have been obtained when proteasomal activity in brain was measured -global activity does not seem to be altered but there may be changes on the cellular/subcellular level [28] [29] [30].Chemical reduction of proteasome activity leads to an increase in the amount in huntingtin fragment levels [22] and, in addition, proteasome activity has been show to decline with age, consistent with the late onset of Huntington's disease [23].This is in line with our study, in which we observed an increase in mutant huntingtin aggregation upon knockdown of four proteasomal subunits.
We identified ubiquitin-like modifier activating enzyme 1 (UBA1) and ubiquitin C (UBC) as modifiers of mutant Table 2. List of genes in which knocking down by two or three shRNA-constructs leads to an induction in mutant huntingtin aggregation Functional classes as previously assigned (8), C. elegans cosmid and gene, human gene name, Gene ID, full name and identity are shown, as in table 1. #shRNAs represent the number of shRNA-constructs that lead to an increase in aggregation of >150% after three experiments.The last column shows the highest average induction observed for that gene In sum, in this study we have shown that 26 modifiers of polyglutamine aggregation in C. elegans have a conserved function towards modifying mutant huntingtin aggregation in cultured human cells and we identified some genes not previously implicated in this pathway.

Previously identified modifiers of (huntingtin) aggregation
In the current study, we have monitored the effect on protein aggregation of C. elegans genes shown to modify polyglutamine aggregation, in a human cell line [8].The C. elegans screen was focused on finding genes that, upon knockdown, increase aggregation and are therefore suppressors of protein aggregation.The original screen in C. elegans revealed the aggregation-suppressing function of some genes that were previously linked to protein aggregation, including a number of chaperonins [8] known to assist in protein folding.Earlier studies had already confirmed their functional conservation towards modifying mutant huntingtin aggregation in human cultured cells [13] [14] [15] and in yeast [20].In the study by Kitamura et al [13], knockdown of CCTζ (CCT6A) and CCTα (TCP-1) led to stimulating aggregation of expanded polyglutamine and mutant huntingtin in HEK293and HeLa cells, respectively.Tam et al [14] reported knockdown of subunits of the Chaperonin-Containing TCP1 Complex (CCT), which is believed to reduce total CCT-levels, leading to higher levels of aggregation in N2Acells.This is in line with our results, which show that five out of the six chaperonin-subunits found in the C. elegans screen are suppressors of mutant Huntingtin aggregation (table 2).
Of the eleven proteasomal subunits in the C. elegans screen, four were also confirmed as aggregationmodifying genes in our study.It is well known that the proteasome is involved in the targeting and degradation of misfolded proteins, but the exact role of the proteasome in the degradation of expanded polyglutamine proteins remains debated [22] [23].Inclusions in neurons from Huntington disease models can be labelled with antibodies against ubiquitin [24] [25], but it is unclear whether the proteasome can degrade expanded polyglutamine proteins [26] [27].Similar conflicting results have been obtained when proteasomal activity in brain was measured -global activity does not seem to be altered but there may be changes on the cellular/subcellular level [28] [29] [30].Chemical reduction of proteasome activity leads to an increase in the amount in huntingtin fragment levels [22] and, in addition, proteasome activity has been show to decline with age, consistent with the late onset of Huntington's disease [23].This is in line with our study, in which we observed an increase in mutant huntingtin aggregation upon knockdown of four proteasomal subunits.
We identified ubiquitin-like modifier activating enzyme 1 (UBA1) and ubiquitin C (UBC) as modifiers of mutant

Previously identified modifiers of (huntingtin) aggregation
In the current study, we have monitored the effect on protein aggregation of C. elegans genes shown to modify polyglutamine aggregation, in a human cell line [8].The C. elegans screen was focused on finding genes that, upon knockdown, increase aggregation and are therefore suppressors of protein aggregation.The original screen in C. elegans revealed the aggregation-suppressing function of some genes that were previously linked to protein aggregation, including a number of chaperonins [8] known to assist in protein folding.Earlier studies had already confirmed their functional conservation towards modifying mutant huntingtin aggregation in human cultured cells [13] [14] [15] and in yeast [20].In the study by Kitamura et al [13], knockdown of CCTζ (CCT6A) and CCTα (TCP-1) led to stimulating aggregation of expanded polyglutamine and mutant huntingtin in HEK293and HeLa cells, respectively.Tam et al [14] reported knockdown of subunits of the Chaperonin-Containing TCP1 Complex (CCT), which is believed to reduce total CCT-levels, leading to higher levels of aggregation in N2Acells.This is in line with our results, which show that five out of the six chaperonin-subunits found in the C. elegans screen are suppressors of mutant Huntingtin aggregation (table 2).
Of the eleven proteasomal subunits in the C. elegans screen, four were also confirmed as aggregationmodifying genes in our study.It is well known that the proteasome is involved in the targeting and degradation of misfolded proteins, but the exact role of the proteasome in the degradation of expanded polyglutamine proteins remains debated [22] [23].Inclusions in neurons from Huntington disease models can be labelled with antibodies against ubiquitin [24] [25], but it is unclear whether the proteasome can degrade expanded polyglutamine proteins [26] [27].Similar conflicting results have been obtained when proteasomal activity in brain was measured -global activity does not seem to be altered but there may be changes on the cellular/subcellular level [28] [29] [30].Chemical reduction of proteasome activity leads to an increase in the amount in huntingtin fragment levels [22] and, in addition, proteasome activity has been show to decline with age, consistent with the late onset of Huntington's disease [23].This is in line with our study, in which we observed an increase in mutant huntingtin aggregation upon knockdown of four proteasomal subunits.
We identified ubiquitin-like modifier activating enzyme 1 (UBA1) and ubiquitin C (UBC) as modifiers of mutant huntingtin aggregation.Variations in the ubiquitin system have been associated with Huntington's disease and other neurodegenerative diseases [31].More specifically, mutations in the UBA1-gene (or UBE1) have been associated with X-linked infantile spinal muscular atrophy [32], a neurodegenerative disease of the motor neurons.Another link between ubiquitin-genes and Huntington's disease comes from heterozygous Ubc knockout mice: when these animals were crossed with a Huntington's disease mouse model, there was an improvement of the disease phenotype [33].This would suggest that knockdown of UBC leading to an increase in mutant huntingtin aggregation may be beneficial to the cells, but our study demonstrates the opposite.However, genes involved in protein folding, degradation and ubiquitination, which are linked to protein aggregation in general, are also modifiers of mutant Huntingtin aggregation in cultured mammalian cells.

Novel modifiers of huntingtin aggregation
Our study also showed the involvement of a number of genes in mutant Huntingtin aggregation that were not previously linked to protein aggregation.One of these genes is the ATPase family gene 3-like 2 protein (AFG3L2).No role for this gene has been reported so far in mutant huntingtin aggregation, but mutations in AFG3L2 cause the dominant spinocerebellar ataxia, SCA28 [34], a neurodegenerative disorder characterized by Purkinje cell degeneration.This gene is closely related to paraplegin, which, when mutated, causes an autosomal recessive form of hereditary spastic paraplegia (HSP) [35].AFG3L2 and paraplegin together form the mitochondrial AAA-protease, its yeast homologs function in mitochondrial protein quality control [36] [37].
Altered assembly of the protease-components is the pathological feature in HSP.Lower expression levels of subunits of this system are mimicked in our experiments by shRNA-constructs.This might lead to an altered assembly of the protease, causing aberrations in the protein quality control system in general, which leads to higher aggregation levels.
A major group of modifiers of mutant huntingtin aggregation are the eukaryotic translation initiation and elongation factors.Knockdown of EEF1A2, EFTUD2, EIF2S2, EIF2S3, EIF3A and EIF3E led to induction in mutant huntingtin aggregation in our screen.More subunits of this complex were found in the C. elegans screen as modifier genes, but just did not reach >150% aggregation induction, indicating that they might induce aggregation on lower levels.Subunits of this complex, like EIF2 -alpha, have been implicated in autophagy and protein aggregation.For example, heat shock protein HspB8 induces the phosphorylation of the alpha-subunit of eiF2, which is involved in the suppression of aggregation through autophagy [38] [39].In addition, altered levels of eiF-proteins are found in degenerating brain regions of Alzheimer's patients [40], further implicating a function for these proteins in stress response.So, similar to lower expression of heat shock proteins, shRNAconstructs targeting eiF-subunits could also reduce the autophagic response of cells to different stressors and thereby cause higher levels of protein aggregation.
In addition, knockdown of SNRP200 and SNRPF , encoding two small nuclear ribonucleoproteins (snRNPs), led to an induction in mutant huntingtin aggregation in the present study.snRNPs have been implicated in neurodegenerative disorders before: for example, snRNPs are in a complex with the SMN protein [21].
Furthermore, a mutation in SNRP200 causes autosomal dominant retinitis pigmentosa [41], a degenerative disorder of the retina.Our results imply that there might be a relationship between SMA and autosomal dominant retinitis pigmentosa and protein aggregation.
Another interesting new candidate gene for modifying protein aggregation is FARSB (phenylalanine-t-RNA synthetase, beta-subunit).All three shRNA-constructs targeting this gene led to a induction of mutant huntingtin aggregation of >150%.It was previously shown that mutations in another tRNA-synthetase in mice (alanyl-tRNA synthetase, Aars) led to protein misfolding and neurodegeneration [42].In this study it was also demonstrated that low levels of mischarged tRNAs already lead to an accumulation of misfolded proteins in neurons, indicating that general alterations in this pathway might be deleterious to these cells.The human homolog of this gene, AARS, was present in our screen, but only one shRNA-construct led to an induction of protein aggregation of >150%, so it was excluded after the first screening round.huntingtin aggregation.Variations in the ubiquitin system have been associated with Huntington's disease and other neurodegenerative diseases [31].More specifically, mutations in the UBA1-gene (or UBE1) have been associated with X-linked infantile spinal muscular atrophy [32], a neurodegenerative disease of the motor neurons.Another link between ubiquitin-genes and Huntington's disease comes from heterozygous Ubc knockout mice: when these animals were crossed with a Huntington's disease mouse model, there was an improvement of the disease phenotype [33].This would suggest that knockdown of UBC leading to an increase in mutant huntingtin aggregation may be beneficial to the cells, but our study demonstrates the opposite.However, genes involved in protein folding, degradation and ubiquitination, which are linked to protein aggregation in general, are also modifiers of mutant Huntingtin aggregation in cultured mammalian cells.

Novel modifiers of huntingtin aggregation
Our study also showed the involvement of a number of genes in mutant Huntingtin aggregation that were not previously linked to protein aggregation.One of these genes is the ATPase family gene 3-like 2 protein (AFG3L2).No role for this gene has been reported so far in mutant huntingtin aggregation, but mutations in AFG3L2 cause the dominant spinocerebellar ataxia, SCA28 [34], a neurodegenerative disorder characterized by Purkinje cell degeneration.This gene is closely related to paraplegin, which, when mutated, causes an autosomal recessive form of hereditary spastic paraplegia (HSP) [35].AFG3L2 and paraplegin together form the mitochondrial AAA-protease, its yeast homologs function in mitochondrial protein quality control [36] [37].
Altered assembly of the protease-components is the pathological feature in HSP.Lower expression levels of subunits of this system are mimicked in our experiments by shRNA-constructs.This might lead to an altered assembly of the protease, causing aberrations in the protein quality control system in general, which leads to higher aggregation levels.
A major group of modifiers of mutant huntingtin aggregation are the eukaryotic translation initiation and elongation factors.Knockdown of EEF1A2, EFTUD2, EIF2S2, EIF2S3, EIF3A and EIF3E led to induction in mutant huntingtin aggregation in our screen.More subunits of this complex were found in the C. elegans screen as modifier genes, but just did not reach >150% aggregation induction, indicating that they might induce aggregation on lower levels.Subunits of this complex, like EIF2 -alpha, have been implicated in autophagy and protein aggregation.For example, heat shock protein HspB8 induces the phosphorylation of the alpha-subunit of eiF2, which is involved in the suppression of aggregation through autophagy [38] [39].In addition, altered levels of eiF-proteins are found in degenerating brain regions of Alzheimer's patients [40], further implicating a function for these proteins in stress response.So, similar to lower expression of heat shock proteins, shRNAconstructs targeting eiF-subunits could also reduce the autophagic response of cells to different stressors and thereby cause higher levels of protein aggregation.
In addition, knockdown of SNRP200 and SNRPF , encoding two small nuclear ribonucleoproteins (snRNPs), led to an induction in mutant huntingtin aggregation in the present study.snRNPs have been implicated in neurodegenerative disorders before: for example, snRNPs are in a complex with the SMN protein [21].
Furthermore, a mutation in SNRP200 causes autosomal dominant retinitis pigmentosa [41], a degenerative disorder of the retina.Our results imply that there might be a relationship between SMA and autosomal dominant retinitis pigmentosa and protein aggregation.
Another interesting new candidate gene for modifying protein aggregation is FARSB (phenylalanine-t-RNA synthetase, beta-subunit).All three shRNA-constructs targeting this gene led to a induction of mutant huntingtin aggregation of >150%.It was previously shown that mutations in another tRNA-synthetase in mice (alanyl-tRNA synthetase, Aars) led to protein misfolding and neurodegeneration [42].In this study it was also demonstrated that low levels of mischarged tRNAs already lead to an accumulation of misfolded proteins in neurons, indicating that general alterations in this pathway might be deleterious to these cells.The human homolog of this gene, AARS, was present in our screen, but only one shRNA-construct led to an induction of protein aggregation of >150%, so it was excluded after the first screening round.huntingtin aggregation.Variations in the ubiquitin system have been associated with Huntington's disease and other neurodegenerative diseases [31].More specifically, mutations in the UBA1-gene (or UBE1) have been associated with X-linked infantile spinal muscular atrophy [32], a neurodegenerative disease of the motor neurons.Another link between ubiquitin-genes and Huntington's disease comes from heterozygous Ubc knockout mice: when these animals were crossed with a Huntington's disease mouse model, there was an improvement of the disease phenotype [33].This would suggest that knockdown of UBC leading to an increase in mutant huntingtin aggregation may be beneficial to the cells, but our study demonstrates the opposite.However, genes involved in protein folding, degradation and ubiquitination, which are linked to protein aggregation in general, are also modifiers of mutant Huntingtin aggregation in cultured mammalian cells.

Novel modifiers of huntingtin aggregation
Our study also showed the involvement of a number of genes in mutant Huntingtin aggregation that were not previously linked to protein aggregation.One of these genes is the ATPase family gene 3-like 2 protein (AFG3L2).No role for this gene has been reported so far in mutant huntingtin aggregation, but mutations in AFG3L2 cause the dominant spinocerebellar ataxia, SCA28 [34], a neurodegenerative disorder characterized by Purkinje cell degeneration.This gene is closely related to paraplegin, which, when mutated, causes an autosomal recessive form of hereditary spastic paraplegia (HSP) [35].AFG3L2 and paraplegin together form the mitochondrial AAA-protease, its yeast homologs function in mitochondrial protein quality control [36] [37].
Altered assembly of the protease-components is the pathological feature in HSP.Lower expression levels of subunits of this system are mimicked in our experiments by shRNA-constructs.This might lead to an altered assembly of the protease, causing aberrations in the protein quality control system in general, which leads to higher aggregation levels.
A major group of modifiers of mutant huntingtin aggregation are the eukaryotic translation initiation and elongation factors.Knockdown of EEF1A2, EFTUD2, EIF2S2, EIF2S3, EIF3A and EIF3E led to induction in mutant huntingtin aggregation in our screen.More subunits of this complex were found in the C. elegans screen as modifier genes, but just did not reach >150% aggregation induction, indicating that they might induce aggregation on lower levels.Subunits of this complex, like EIF2 -alpha, have been implicated in autophagy and protein aggregation.For example, heat shock protein HspB8 induces the phosphorylation of the alpha-subunit of eiF2, which is involved in the suppression of aggregation through autophagy [38] [39].In addition, altered levels of eiF-proteins are found in degenerating brain regions of Alzheimer's patients [40], further implicating a function for these proteins in stress response.So, similar to lower expression of heat shock proteins, shRNAconstructs targeting eiF-subunits could also reduce the autophagic response of cells to different stressors and thereby cause higher levels of protein aggregation.
In addition, knockdown of SNRP200 and SNRPF , encoding two small nuclear ribonucleoproteins (snRNPs), led to an induction in mutant huntingtin aggregation in the present study.snRNPs have been implicated in neurodegenerative disorders before: for example, snRNPs are in a complex with the SMN protein [21].
Furthermore, a mutation in SNRP200 causes autosomal dominant retinitis pigmentosa [41], a degenerative disorder of the retina.Our results imply that there might be a relationship between SMA and autosomal dominant retinitis pigmentosa and protein aggregation.
Another interesting new candidate gene for modifying protein aggregation is FARSB (phenylalanine-t-RNA synthetase, beta-subunit).All three shRNA-constructs targeting this gene led to a induction of mutant huntingtin aggregation of >150%.It was previously shown that mutations in another tRNA-synthetase in mice (alanyl-tRNA synthetase, Aars) led to protein misfolding and neurodegeneration [42].In this study it was also demonstrated that low levels of mischarged tRNAs already lead to an accumulation of misfolded proteins in neurons, indicating that general alterations in this pathway might be deleterious to these cells.The human homolog of this gene, AARS, was present in our screen, but only one shRNA-construct led to an induction of protein aggregation of >150%, so it was excluded after the first screening round.

Fig. 1 :
Fig. 1: presents a graphic overview of the experiments.Flowchart of experimental procedures.Experiments performed to test 186 C. elegans suppressors of polyglutamine aggregation for their conserved function on mutant huntingtin aggregation in mammalian cells.See text for further explanations and selection criteria.

Fig. 1 :
Fig. 1: presents a graphic overview of the experiments.Flowchart of experimental procedures.Experiments performed to test 186 C. elegans suppressors of polyglutamine aggregation for their conserved function on mutant huntingtin aggregation in mammalian cells.See text for further explanations and selection criteria.

Fig. 1 :
Fig. 1: presents a graphic overview of the experiments.Flowchart of experimental procedures.Experiments performed to test 186 C. elegans suppressors of polyglutamine aggregation for their conserved function on mutant huntingtin aggregation in mammalian cells.See text for further explanations and selection criteria.

Table 1 .
Human orthologs of C. elegans modifiers of polyglutamine aggregation and their effect on mutant Huntingin aggregation in cultured mammalian cells Genes are grouped in functional classes as previously assigned (8).C. elegans cosmid and gene, human gene name, Gene ID, and full name are shown, as well as their identity on the amino acid level.Columns a, b and c represent different individual shRNA-constructs and the percentage of mutant huntingtin aggregation in that experiment compared to the scrambled control.The number of shRNA-constructs with an aggregation induction >150% for that gene is given in the last column.

Table 1 .
Human orthologs of C. elegans modifiers of polyglutamine aggregation and their effect on mutant Huntingin aggregation in cultured mammalian cells Genes are grouped in functional classes as previously assigned (8).C. elegans cosmid and gene, human gene name, Gene ID, and full name are shown, as well as their identity on the amino acid level.Columns a, b and c represent different individual shRNA-constructs and the percentage of mutant huntingtin aggregation in that experiment compared to the scrambled control.The number of shRNA-constructs with an aggregation induction >150% for that gene is given in the last column.

Table 1 .
Human orthologs of C. elegans modifiers of polyglutamine aggregation and their effect on mutant Huntingin aggregation in cultured mammalian cells Genes are grouped in functional classes as previously assigned (8).C. elegans cosmid and gene, human gene name, Gene ID, and full name are shown, as well as their identity on the amino acid level.Columns a, b and c represent different individual shRNA-constructs and the percentage of mutant huntingtin aggregation in that experiment compared to the scrambled control.The number of shRNA-constructs with an aggregation induction >150% for that gene is given in the last column.

Fig. 2 :
Fig. 2: Validation of experimental procedures (A) Percentage of cells with Htt-Q74-GFP aggregates using control shRNA-constructs and vectors targeting GFP and CCT2.The error bar represents the standard error of the mean (SEM) of 10 individual experiments.

Fig. 2 :
Fig. 2: Validation of experimental procedures (A) Percentage of cells with Htt-Q74-GFP aggregates using control shRNA-constructs and vectors targeting GFP and CCT2.The error bar represents the standard error of the mean (SEM) of 10 individual experiments.

Fig. 2 :
Fig. 2: Validation of experimental procedures (A) Percentage of cells with Htt-Q74-GFP aggregates using control shRNA-constructs and vectors targeting GFP and CCT2.The error bar represents the standard error of the mean (SEM) of 10 individual experiments.

Fig. 3 :
Fig. 3: Mutant huntingtin-aggregation induction by 58 different shRNA-constructs representing 26 individual genes.Average induction of aggregation versus control of final positives after three repeats.Only the genes where >2 shRNA-constructs showed induction of aggregation are shown.Bars represent the standard error of the mean (SEM) of the three experiments.* = p-value

Fig. 3 :
Fig. 3: Mutant huntingtin-aggregation induction by 58 different shRNA-constructs representing 26 individual genes.Average induction of aggregation versus control of final positives after three repeats.Only the genes where >2 shRNA-constructs showed induction of aggregation are shown.Bars represent the standard error of the mean (SEM) of the three experiments.* = p-value

Fig. 3 :
Fig. 3: Mutant huntingtin-aggregation induction by 58 different shRNA-constructs representing 26 individual genes.Average induction of aggregation versus control of final positives after three repeats.Only the genes where >2 shRNA-constructs showed induction of aggregation are shown.Bars represent the standard error of the mean (SEM) of the three experiments.* = p-value

Table 2 .
List of genes in which knocking down by two or three shRNA-constructs leads to an induction in mutant huntingtin aggregation Functional classes as previously assigned(8), C. elegans cosmid and gene, human gene name, Gene ID, full name and identity are shown, as in table1.#shRNAs represent the number of shRNA-constructs that lead to an increase in aggregation of >150% after three experiments.The last column shows the highest average induction observed for that geneIn sum, in this study we have shown that 26 modifiers of polyglutamine aggregation in C. elegans have a conserved function towards modifying mutant huntingtin aggregation in cultured human cells and we identified some genes not previously implicated in this pathway.