1. Volume 2, Issue 4, Pages 427-436 (October 1998)
SH3GL3 Associates with the Huntingtin Exon 1 Protein and Promotes the Formation of Polygln-Containing Protein Aggregates 
Annie Sittler, Stephanie Wälter, Niels Wedemeyer, Renate Hasenbank, Eberhard Scherzinger, Holger Eickhoff, Gillian P Bates, Hans Lehrach, Erich E Wanker 
Molecular Cell 
Volume 2, Issue 4, Pages (October 1998)
DOI: /S (00)

2. Figure 1
Specific Interaction of SH3GL3 with N-Terminal Huntingtin Fragments
(A) Schematic representation of LexA fusion proteins used as baits. The C-terminal part of LexA is indicated as a gray box, the polygln stretch is shown as a black box. The peripheral amino acid positions of the proteins fused to LexA are given below the diagrams. Yeast strains expressing the proteins LexA, LexA-SIM1, LexA-HDex1Q51, or LexA-HD588Q23 were transformed with pGAD-359 encoding the protein GAL4ad-SH3GL3. Individual Leu+Trp+ transformants were spotted in duplicate on SD-leu-trp and SD-leu-trp-his plates and tested for β-GAL activity.
(B) Elongated polygln domains in HDex1p enhance its binding to the SH3GL3 protein. The plasmids encoding the LexA fusion proteins SIM1, HD588Q23, HDex1Q51, and HDex1Q20 were transformed in the yeast strain encoding the protein GAL4ad-SH3GL3. The transformants were assayed for lacZ reporter gene expression. Each column represents the mean value of three measurements.
(C) In vitro binding to GST-HDex1p of His-SH3GL3 produced in E. coli. Equal amounts of purified GST-HDex1 proteins with 20, 30, and 51 glutamines were incubated with E. coli crude extract containing the His-SH3GL3 protein. Bound proteins were analyzed by Western blotting and probed with monoclonal RGS.His antibody (ECL detection). Quantification of the GST-HDex1 bound-His-SH3GL3 protein was done by densitometry of the autoradiogram. The lowest measured value was arbitrarily set as 1.
Molecular Cell 1998 2, DOI: ( /S (00) )

3. Figure 2
Identification of Protein Domains Responsible for the Interaction between HDex1Q51 and SH3GL3
(A) Structure of fusion proteins used to determine the domains responsible for the interaction between SH3GL3 and HDex1Q51. The polygln stretch (Q) is indicated as a black box, the SH3 domain is indicated as a dark gray box, and the polyproline sequence (P) is shown as a light gray box. The peripheral amino acid positions of the proteins fused to LexA and GAL4ad are given below the diagrams.
(B) Determination of the protein domains essential for the SH3GL3-HDex1Q51 interaction using the yeast two-hybrid system. Two independent yeast clones each expressing the indicated LexA and GAL4ad fusion proteins were transferred to SD-leu-trp and SD-leu-trp-his plates. Yeast colonies grown on the SD-leu-trp-his plates were then assayed for β-GAL activity.
Molecular Cell 1998 2, DOI: ( /S (00) )

4. Figure 3 Coimmunoprecipitation of Huntingtin and SH3GL3
In A and B, COS-1 cells were transfected with the HD514Q68, HDex1Q51, and SH3GL3 constructs as indicated in the table above the gels. (A) The total extract was immunoprecipitated with anti-HD1, anti-Q51, anti-GL3, anti-GAPDH, or Nis-Q51 preimmune serum, and the immunoprecipitated material was analyzed by Western blotting. For immunodetection, the huntingtin-specific monoclonal anti-HD-4C8 was used. (B) HDex1Q51 or HD514Q68 was expressed along with SH3GL3, immunoprecipitated with anti-Q51 or Nis-Q51 preimmune serum, and revealed with anti-HA-12CA5, thus detecting the HA-tagged SH3GL3 protein. (C) Human brain extract from an HD-affected individual was subjected to immunoprecipitation with increasing amounts of anti-GL3 (2, 5, and 10 μl in lanes 1–3) or with increasing amounts of the corresponding preimmune serum (lanes 4–6). As controls, immunoprecipitations were carried out with anti-Q51 (positive) or Nis-Q51 (negative). Immunodetection was performed with anti-HD-4C8.
Molecular Cell 1998 2, DOI: ( /S (00) )

5. Figure 4
Expression of SH3GL3 in Human Tissues and Subcellular Fractionation
(A) Identification of SH3GL3 protein in various human tissues by Western blot analysis using the anti-GL3 antibody. (B) Biochemical subcellular fractionation of huntingtin and SH3GL3 from human cortex. Protein fractions were prepared as described (Wanker et al. 1997), resolved by SDS-PAGE, and immunoblotted using the monoclonal anti-HD-4C8 or the polyclonal anti-GL3 antibody. The fractions are marked as P1, P2, P3, S1, S2, and S3.
Molecular Cell 1998 2, DOI: ( /S (00) )

6. Figure 5
Aggregates Containing Huntingtin and SH3GL3 Are Formed in COS Cells
(A) Western blot analysis of the soluble fractions of transfected COS cells. COS-1 cells were transfected with different HDex1 constructs and/or the SH3GL3 construct as indicated in the table above the gel. Cleared cell extracts were prepared and 10 μg of each extract was separated by SDS-PAGE and analyzed by Western blotting using anti-HD1 or anti-GL3 antibody.
(B) Filter retardation assay performed on the insoluble fraction of the transfected cell extracts. The pellet fractions obtained after centrifugation of whole cell lysates were subjected to DNaseI treatment, boiled in 1% SDS and 50 mM DTT, and filtered through a cellulose acetate membrane. The aggregated proteins retained on the filter were then detected with the antibodies anti-HD1 and anti-GL3. In a and b, 3 and 5 μl of the pellet fraction were applied to the filter, respectively.
(C) Western blot analysis of the pellet fraction of the transfected cell extracts. Ten μl of the fractions analyzed in B was subjected to SDS-PAGE and immunoblotted with the anti-HD1 and anti-GL3 antibodies. The arrows point to the aggregated HD exon 1 protein (containing also SH3GL3) that failed to enter the stacking gel.
Molecular Cell 1998 2, DOI: ( /S (00) )

7. Figure 6
Subcellular Localization of HDex1p and SH3GL3 by Immunofluorescence Microscopy
COS cells were transfected with the indicated constructs and immunolabeled with anti-HD1 coupled to FITC-conjugated antibody (green) and with anti-HA-12CA5 coupled to CY3-conjugated antibody (red). Cells were counterstained with Hoechst allowing detection of the nuclei with a triple filter (panels 3, 5, and 6). Colocalization of HDex1Q93 and SH3GL3 is illustrated in panels 1–3, where the intranuclear and perinuclear aggregates show the expected overlap (panel 3) of the green (panel 2) and red signals (panel 1), indicating the presence of SH3GL3 in the polygln-containing aggregates. The cellular distribution of SH3GL3 transfected alone is shown in panel 4. Colocalization is also shown for soluble huntingtin with a polygln repeat in the normal range (HDex1Q20) and SH3GL3 (panel 5). No colocalization is detected between HDex1Q51 and the unrelated FMRP Iso7 protein used as control (panel 6). Shown are representative fields in all cases.
Molecular Cell 1998 2, DOI: ( /S (00) )