• Extended Experimental Procedures
• Supplemental References

Figures

• Figure S1     FMRP-MediatedgH2A.X Induction and BRCA1 Phosphorylation, Related toFigures 1,3, and7. (A and B) Induction ofgH2A.X foci formation is impaired in FMRP KO (B) as compared to wild-type MEFs (A) in response to aphidicolin (APH) treatment.(C) Left graph, wild-type but not FMRP KO andDot1Lmutant MEFs exhibited a 3-fold (cells with < 10 foci) and 4-fold (cells with > 10 foci) increase ingH2A.X foci inresponse to APH. Asterisks, p < 0.05, Student t test. Data are represented as average of three independent experiments with SD. Western blot shows H3K79methylation levels in WT andDot1Lmutant MEFs. Right: induction ofgH2A.X foci formation in response to APH is impaired inDot1Lmutant MEFs as compared towild-type MEFs (A).(D) Western blot showing the levels of expression of exogenous Flag-HA-FMRP (lane 2) and FMRP Agenet domain mutants (lanes 3 and 4) in FMRP KO MEFs.(E) HeLa cells were subjected to scrambled or two independent FMRP shRNAs (FMRP shRNA1 and FMRP shRNA2). Cells were then treated with DMSO as acontrol or APH for 24 hr, followed by western blotting for BRCA1 phospho-Ser1423. In contrast to control RNAi cells (3.5-fold induction in Ser1423 phos-phorylation), FMRP RNAi cells showed no phosphorylation of BRCA1 at Ser1423 in response to APH treatment (compare lane 2 to 4 and 6). Asterisk indicatesanonspecific band.(F) Unlike in HeLa cells, in 293 cells FMRP RNAi-mediated dampening of phospho-Ser1423 could be partially attributed to the increased levels of phospho-Ser1423 even in the absence of APH treatment, suggesting that the FMRP effect on BRCA1 phosphorylation might be context dependent. Blots are repre-sentative of three experiments.
• Figure S2     FMRP and DOT1L-Dependent Cell Survival Assays, Related toFigures 1and7. (A) HeLa cells stably expressing an FMRP RNAi hairpin and reconstituted with an RNAi resistant wild-type FMRP-Flag-HA construct or empty Flag-HA vector.FMRP RNAi+Flag-HA cells exhibited more cell death (asterisks p = 0.05, n = 3, 2-way ANOVA) in response to the increased concentration of APH as comparedtoFMRP RNAi+FMRP-Flag-HA cells.(B) FMRP levels in cells used for the colony survival assays are shown in the box. Scrambled RNAi and FMRP RNAi are shown for comparison. Arrowheaddemarcates exogenous Flag-HA-FMRP.(C) Colony survival assays in the presence of APH using wild-type MEFs, FMRP KO MEFs andDot1Lmutant MEFs. FMRP KO MEFs andDot1Lmutant MEFswere more sensitive to the increased concentrations of APH as compared to wild-type MEFs (asterisks p = 0.05, n = 3, 2-way ANOVA)
• Figure S3     FMRP andgH2A.X Colocalization, Related toFigure 2. (A) FMRP immunostaining with anti-FMRP antibody (1) (Abcam, red) and a second anti-FMRP antibody (2) (Calbiochem, green) without leptomycin B (LPB)treatment, indicating low levels of nuclear FMRP. Scale bar, 10mm.(B) Overlap of FMRP andgH2A.X foci in the presence of leptomycin B (LPB) in DMSO or APH treated MEFs. a,b,c show various degrees ofgH2A.X/FMRP overlapindicated by solid arrowheads. d shows an example of a cell where FMRP andgH2A.X do not overlap (arrowhead). Scale bar, 10mm.(C) Graph shows that about 30% of cells displayed at least partial (at least 1 overlap event per cell)gH2A.X/FMRP overlap, which was increased upon APHtreatment, although it was not a statistically significant increase (p = 0.076, Student t test). Data are represented as average of three independent experiments with SD.
• Figure S4     Agenet Domain Binding to Methylated MLA Histone Substrates, Related toFigure 3. (A) In vitro pull-downs using FMRP GST-Agenet domain and MLA histones carrying methyl lysine analogs at various sites.(B–D) Equilibrium binding analysis of AgenetFMRPinteraction with MLA histone substrates using microscale themophoresis (MST). Independent binding reactionswere performed a minimum of three times. Data points were plotted with the x axis representing nM concentration of modified histone proteins and y axisrepresentingDFNorm[&], which is the reduction of fluorescence due to directed movement of molecules in a microscopic temperature gradient.
• Figure S5     Genetic Complementation Experiments Using Wild-Type FMRP and Its Agenet Domain Mutants in the AMPAR InternalizationAssay, Related toFigures 3 and 4. (A) In vitro H3K79me2 MLA nucleosome binding reactions using GST-AgenetKHKH domain. AgenetKHKHFMRPcontains 2 additional nucleic acid binding do-mains adjacent to the double-tudor domain allowing nucleosomal binding. Wild-type FMRP bound H3Kc79me2, whereas Agenet domain mutants T102A, Y103L,and R138Q (patient mutant, seeFigure S6A) did not (compare lanes 3 to 4,5, and 6).(B) Hippocampal primary neurons from a wild-type and anFmr1KO mouse, as well as neurons fromFmr1KO mouse reconstituted with lentiviral constructsexpressing wild-type FMRP or FMRP mutants T102A and Y103L were immunostained for surface and internalized AMPARs. Internalized AMPAR signal wasincreased and AMPAR signal remaining on the surface was reduced in FMRP KO neurons as compared to wild-type neurons (panels 1 and 2). FMRP KO neuronsinfected with lentivirus expressing either wild-type or mutant forms of FMRP displayed reversal of the excessive AMPAR internalization observed inFMRP KOneurons (panels 3–5), indicating that Agenet mutations T102A and Y103L do not affect FMRP function at neuronal dendrites.(C) Distribution of AMPARs in distal dendrites. Box-and-whisker plot showing that enhanced constitutive endocytosis of AMPARs in FMRP KO neurons (lane 2) iscorrected by lentivirus expressing either WT (lane 3), T102A mutant (lane 4), or Y103L mutant (lane 5) FMRP. Median: WT, 47.18; KO, 52.01; KO+FMRP(WT),47.44; KO+T102A, 49.86; KO+Y103L, 48.30; n = 30 each. p values were determined using one-way ANOVA (a= 0.05) with Bonferroni’s post hoc test.
• Figure S6     Analysis of the R138Q FMRP Patient Mutant in the Chromatin Interaction, DDR, and AMPAR Internalization Assays, Related to Figure 4. (A) Diagram of the FMRP Agenet double-tudor domain showing the location of the patient mutation R138Q.(B) Coomassie gel showing GST-AgenetKHKH and GST-R138QKHKH proteins used in MST experiments for Kd determination.(C) Western blot showing the levels of expression of exogenous Flag-HA-FMRP (lane 1) and FMRP (R138Q) Agenet domain patient mutant (lane 2) in FMRP KOMEFs, using anti-Flag antibody.(D) Western blot showing the levels of expression of exogenous wild-type FMRP (lane 2) and R138Q FMRP patient mutant (lane 3) rescue constructs in FMRPKOMEFs using anti-FMRP antibody. Lane 1 contains FMRP KO MEFs without the rescue for comparison.(E) Colony survival assay in the presence of hydroxyurea (HU) using FMRP KO MEFs reconstituted with wild-type FMRP or R138Q FMRP patient mutant. R138Q-reconstituted MEFs were more sensitive to the increased doses of HU as compared to wild-type FMRP reconstituted MEFs (p = 0.0045, 2-way ANOVA).(F) Images of hippocampal neurons immunostained for surface (S) and internalized (I) AMPARs. Panel 1: wild-type neurons, panel 2: FMRP KO neurons, panel 3:FMRP KO neurons rescued with wild-type FMRP, panel 4: FMRP KO neurons rescued with FMRP R138Q mutant.(G) Quantification of results in (F). Both rescue proteins functioned similarly in AMPARs internalization experiments, reversing AMPAR constitutive endocytosis inFMRP KO neurons. Compare lanes 3 and 4 (y axis is the ratio of internalized to total AMPARs). Median: WT, 52.60; KO, 74.20; KO+FMRP (WT), 51.57; KO+R138Q,51.44; n = 30 each. p values were determined using one-way ANOVA (a= 0.05) with Bonferroni's ost hoc test.
• Figure S7     Control Staining of FMRP in Mouse Spermatocytes, SPO11 Staining, and Analysis ofDot1LcKO Testes, Related toFigures 5 and 7. (A) Immunostaining of wild-type (WT) andFmr1KO meiotic spreads with anti-FMRP antibody (rabbit polyclonal anti-FMRP, Abcam) to confirm the specificity ofFMRP staining. SYCP3 marks the chromosomes. Scale bar, 5mm.(B) SPO11 recruitment and double-strand-break formation is not perturbed inFmr1KO cells. SPO11 catalyzes DSBs during the leptotene stage of meioticprophase. Staining for SPO11 in representative WT and representative KO leptotene cells reveals no difference between WT and KO. Equivalent staining with noprimary antibody is shown as a negative control. SPO11 staining alone is shown in the top row for clarity. Large green patches are sperm heads. Scale bar,10mm.(C–E)Dot1Lexpression is reduced inDot1LcKO testis.(C) Genotyping for theDot1LDallele using genomic DNA fromDot1Lfl/+;Mvh-Cre-(WT, fl/+) orDot1Lfl/D;Mvh-Cre+(cKO,D/D) adult testis tissue. The delta allele ispredicted to be700 bp.(D) qRT-PCR using two different primer sets specific forDot1L, using cDNA fromDot1Lfl/+;Mvh-Cre(WT) andDot1Lfl/D;Mvh-Cre+(cKO) adult testis. For eachqPCR primer set, one primer spans an exon-exon junction in theDot1LcDNA and the other primer is in a third, adjacent exon. Signal is normalized toActb. Errorbars signify SD for two biological replicates. Differences between means do not meet statistical significance. Note that cDNA was isolated from wholetestis,which contains both somatic and germ cells, but theMvh-Cretransgene is expressed only in germ cells. Therefore, someDot1Lexpression is expected in themixed somatic/germ cell population even in cKO testis.(E) Western blot for H3K79me1 and H3K79me2 on mouse testis lysates showing reduction of H3K79 methylation inDot1LcKO testes (lanes 3 and 4). Twosamples of WT (lanes 1 and 2) andDot1LcKO (lanes 3 and 4) testes are shown.