Sumoylation- and GAR1-Dependent Regulation of Dyskerin Nuclear And

bioRxiv preprint doi: https://doi.org/10.1101/2020.09.02.280198; this version posted September 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 SUMOylation- and GAR1-dependent regulation of dyskerin nuclear and subnuclear 2 localization 3 4 MacNeil, D.E.1,2, Lambert-Lanteigne, P.1, Qin, J.1,2, McManus, F.3, Bonneil, E.3, Thibault, P.3, & 5 Autexier, C.1,2* 6

7 1 Lady Davis Institute for Medical Research, Jewish General Hospital, 2McGill University, 8 Montréal, QC, Canada, 3Université de Montréal, Montréal, QC, Canada

9 *Correspondence: [email protected]

10 Summary

11 Dyskerin, a telomerase-associated protein and H/ACA ribonucleoprotein complex

12 component plays an essential role in human telomerase assembly and activity. The nuclear and

13 subnuclear compartmentalization of dyskerin and the H/ACA complex is an important though

14 incompletely understood aspect of H/ACA ribonucleoprotein function. The posttranslational

15 modification, SUMOylation, targets a wide variety of proteins, including numerous RNA-

16 binding proteins, and most identified targets reported to date localize to the nucleus. Four

17 SUMOylation sites were previously identified in the C-terminal Nuclear/Nucleolar Localization

18 Signal (N/NoLS) of dyskerin, each located within one of two lysine-rich clusters. We found that

19 a cytoplasmic localized C-terminal truncation variant of dyskerin lacking most of the C-terminal

20 N/NoLS and both lysine-rich clusters represents an under-SUMOylated variant of dyskerin

21 compared to wildtype dyskerin. We demonstrate that mimicking constitutive SUMOylation of

22 dyskerin using a SUMO3-fusion construct can drive nuclear accumulation of this variant, and

23 that the SUMO site K467 in this N/NoLS is particularly important for the subnuclear localization

24 of dyskerin to the nucleolus in a mature H/ACA complex assembly- and SUMO-dependent

25 manner. We also characterize a novel SUMO-interacting motif in the mature H/ACA complex

26 component GAR1 that mediates the interaction between dyskerin and GAR1. Mislocalization of

27 dyskerin, either in the cytoplasm or excluded from the nucleolus, disrupts dyskerin function and

28 leads to reduced interaction of dyskerin with the telomerase RNA. These data indicate a role for

29 dyskerin C-terminal N/NoLS SUMOylation in regulating the nuclear and subnuclear localization

30 of dyskerin, which is essential for dyskerin function as both a telomerase-associated protein and

31 as an H/ACA ribonucleoprotein involved in rRNA and snRNA biogenesis.

46 Introduction

47 The H/ACA ribonucleoprotein (RNP) complex is responsible for pseudouridine synthesis

48 at specific bases in ribosomal (r)RNA and small nuclear (sn)RNA in subnuclear compartments,

49 specifically the nucleolus and Cajal bodies, respectively (1-5). The protein components of this

50 complex at maturity are dyskerin (6-8), NOP10, NHP2 (9), and GAR1 (1, 10). The mature

51 H/ACA complex assembles with noncoding (nc)RNA members of the H/ACA family, such as

52 small nucleolar (sno)RNAs and small Cajal body specific (sca)RNAs that provide target

53 pseudouridine synthesis specificity to dyskerin, the pseudouridine synthase of the H/ACA

54 complex. The H/ACA motif is also a conserved biogenesis domain in telomerase RNAs of

55 metazoans (11), including the human telomerase RNA (hTR) which relies on the H/ACA

56 complex proteins for stability, processing, and function (12-16).

57 While hTR has no known target for guiding pseudouridine synthesis by dyskerin, the

58 importance of the H/ACA complex in hTR biogenesis is demonstrated by mutations causing the

59 premature aging disease and telomere syndrome dyskeratosis congenita (DC), with reported

60 patient mutations in the genes encoding each protein component of the mature complex,

61 excluding GAR1, as well as in the H/ACA biogenesis domain of hTR itself (17). Patients with

62 DC have characteristic accelerated telomere shortening which leads to pathology in proliferative

63 tissues, and results in bone marrow failure as the leading cause of mortality in this disease (18-

64 21). In particular, DC patients with mutations disrupting the H/ACA complex components or

65 H/ACA domain of hTR have reduced hTR accumulation which drives telomerase activity defects

66 and accelerated telomere shortening (12). There have also been several reports of DC mutations

67 in the H/ACA complex components affecting H/ACA RNA biogenesis beyond hTR (22, 23), and

68 the essentiality of dyskerin and the H/ACA complex is likely due to its importance in rRNA and

69 snRNA posttranscriptional modification. The dkc1 gene encoding dyskerin is a core essential

70 gene that is highly conserved, with phylogenetic roots in bacteria and archaea. Knockout of this

71 gene is lethal in fungi (6), flies (24, 25), mice (26), and human cells (27, 28). Though X-linked

72 dyskeratosis congenita (X-DC) is a commonly inherited form of the disease caused by mutations

73 in dkc1, a complete deletion or loss of the gene has never been reported in X-DC, further

74 demonstrating the essentiality of dyskerin.

75 The compartmentalization of dyskerin and the H/ACA complex is an important though

76 incompletely understood aspect of H/ACA RNP function. Dyskerin has been reported to rely on

77 two nuclear/nucleolar localization sequences (N/NoLSs) for complete nuclear import and

78 retention, as well as for nucleolar accumulation (29). With the exception of GAR1, the H/ACA

79 RNP components are present at sites of transcription of H/ACA RNAs in the nucleoplasm, along

80 with the assembly factor NAF1 which is replaced by GAR1 upon complex maturation (30-32).

81 Mature H/ACA complexes localize in the dense fibrillar component (DFC) of the nucleolus and

82 in the Cajal bodies (7) where they guide posttranscriptional modification of rRNA and snRNA,

83 respectively, dependent upon the H/ACA RNA with which the complex is assembled. The

84 stepwise assembly of H/ACA RNPs has been proposed to play a role in localization of the

85 complex to its sites of function (32). Although the mechanism governing subnuclear

86 compartmentalization of the mature H/ACA complex remains incompletely characterized, it is

87 likely to rely on regulation of miscibility with these discrete membrane-free regions of the

88 nucleus. This has been recently demonstrated for other nucleolar proteins resident in the DFC

89 such as fibrillarin, which relies on an intrinsically disordered glycine and arginine rich (GAR)

90 domain and RNA interactions for miscibility with the DFC (33, 34).

91 The posttranslational modification SUMOylation has been demonstrated to affect nuclear

92 and subnuclear localization of a number of protein targets, including resident proteins of the

93 nucleolus (35-37). This modification involves conjugation of small ubiquitin-like modifier

94 (SUMO) protein to lysine residues of target proteins in an E1 activating (SAE1/SAE2) and E2

95 conjugating (Ubc9) enzyme-dependent manner, often with the help of one of many E3 SUMO

96 ligases, and promoted by a SUMOylation consensus motif in target proteins (ψKXE/D – where ψ

97 is a hydrophobic residue and X is any residue) (38). While SUMOylation has been reported to

98 regulate various functions of target proteins, a key aspect of SUMOylation is mediating protein-

99 protein interactions between SUMO targets and proteins containing SUMO-interacting motifs

100 (SIMs) which non-covalently bind SUMO (39-41). SUMOylation is a reversible modification,

101 with several identified SUMO-specific proteases cleaving immediately after the C-terminal

102 diglycine repeat in SUMO moieties, and therefore being responsible both for maturation of free

103 SUMO and for removal of SUMO from target lysines (42-45). Typically, at steady state, only a

104 small proportion of a SUMO target is conjugated to SUMO moieties. We previously

105 demonstrated that dyskerin is a SUMOylation target of SUMO1 and SUMO2/3 isoforms, and

106 that substituting either of two N-terminal X-DC-implicated lysine residues to arginine reduces

107 the proportion of SUMOylated dyskerin in cells, leading to reductions in hTR, reduced

108 telomerase activity, and accelerated telomere shortening (46). We have since shown that these

109 two X-DC residues impact the dyskerin-hTR interaction, though the SUMO dependence of this

110 interaction was not investigated (16).

111 Here we further investigate a regulatory role for SUMOylation of dyskerin. Using

112 mutational analyses and SUMO-fusion constructs, we demonstrate that the C-terminal N/NoLS

113 of dyskerin is a SUMO3 target, and that mimicking constitutive SUMOylation of a cytoplasmic

114 truncation variant of dyskerin is sufficient to drive nuclear accumulation but not proper

115 subnuclear localization of dyskerin. We also demonstrate that the nucleolar localization of

116 dyskerin is mediated by the SUMO3 site K467 in this C-terminal N/NoLS, and that K467 is

117 required for the interaction between dyskerin and GAR1 in a SUMO3-dependent manner, and

118 novelly identify a SIM in GAR1 that is important for this interaction.

119 Results

120 The C-terminal nuclear/nucleolar localization sequence of dyskerin is a SUMOylation target

121 Many proteome-wide studies performed in human cell lines have identified dyskerin as a

122 target of SUMOylation, both by SUMO1 and SUMO2/3 (47-55). Compiling the results of these

123 studies, it is evident that dyskerin is a highly decorated target for SUMOylation, with 24 sites

124 identified by mass spectrometry (MS) analyses (Figure 1A). For the purpose of this study, we

125 focused on four SUMO2/3 sites in particular due to the placement of these lysines in the C-

126 terminal N/NoLS (K467, K468, K498, and K507), which was previously reported to mediate

127 efficient localization of dyskerin to the nucleus alone and in combination with an N-terminal

128 N/NoLS (29). Importantly, truncation of the C-terminal N/NoLS by replacing K446 with a stop

129 codon (X), and thus removal of all four SUMO3 sites and the lysine-rich (K-rich) clusters in

130 which they are situated, substantially reduces the amount of SUMOylated dyskerin detectable by

131 immunoblotting following Ni-NTA purification from HEK293 cells expressing FLAG-tagged

132 dyskerin and 6xHis-SUMO3 (Figure 1B, wildtype vs. K446X). Indeed, while FLAG-tagged

133 wildtype dyskerin co-localizes with the nucleolar marker fibrillarin in HEK293 cells assessed by

134 immunofluorescence (IF), the FLAG-tagged K446X accumulates predominantly in the

135 cytoplasm (Figure 1C, top and middle panels). Interestingly, mimicking constitutive

136 SUMOylation of K446X by fusing a SUMO3 moiety to the N-terminus of this dyskerin variant

137 allows for detection of high molecular weight products by Ni-NTA from HEK293 cells co-

138 expressing FLAG-tagged SUMO3-K446X and 6xHis-SUMO3, indicating that this fusion protein

139 is highly SUMOylated (Figure 1B). This SUMO3-fusion is also sufficient to drive the K446X

140 truncation variant into the nucleus (Figure 1C, bottom two panels). However, the SUMO3-

141 fusion variant remains excluded from the nucleolar compartment, suggesting that mimicking

142 permanent SUMOylation of dyskerin disrupts proper subnuclear localization. This hypothesis is

143 supported by our observation that fusion of SUMO3 to either the N-terminus or the C-terminus

144 of FLAG-tagged wildtype dyskerin also leads to disrupted subnuclear localization (Figure 1D).

145 These data suggest that the C-terminal N/NoLS of dyskerin regulates nuclear localization in a

146 SUMO3-dependent manner, though the reversibility of SUMOylation after nuclear import is

147 likely important for mediating proper subnuclear trafficking of dyskerin. This would be

148 consistent with a previous proposal that “balanced SUMOylation levels” may be required for

149 nucleolar regulation (56).

150 Dyskerin nuclear and subnuclear localization is mediated by SUMOylation

151 The C-terminal N/NoLS of dyskerin contains two K-rich clusters, each of which contains

152 two MS-identified SUMO3 target sites (Figure 1A). To elucidate which of these four SUMO3-

153 sites, if any, may be mediating localization of dyskerin, and to reduce potential compensation for

154 a loss of a single SUMOylation site by SUMO conjugation to neighbouring lysine residues, a

155 stop codon was introduced at A481 in the FLAG-dyskerin construct, thus removing the entire

156 second K-rich cluster while leaving the first intact. In contrast to the K446X variant, FLAG-

157 tagged A481X efficiently localizes in the nucleus and the nucleolus, as observed by co-

158 localization with fibrillarin assessed by IF (Figure 2A). This suggests that the second K-rich

159 cluster, thus the SUMO3 sites within it are not critical regulators of dyskerin nuclear localization

160 per se, and these data are consistent with previous localization analysis of a truncation variant at

161 D493 (29). However, full length FLAG-tagged dyskerin in which the SUMO3 site K467 in the

162 first K-rich cluster is substituted to an arginine (K467R) displays an apparent nucleolar

163 exclusion/nucleoplasmic accumulation phenotype when assessed by IF (Figure 2A). This is in

164 contrast to the K468R variant that localizes comparably to wildtype dyskerin (Figure 2A).

165 Substituting both of these lysines to arginine (K467/468R) leads to a localization phenotype

166 similar to the single K467R substitution variant (Figure 2A). FLAG-positive cells were scored

167 based on localization phenotype as a percentage of FLAG-positive cells counted, and

168 localization of each FLAG-tagged dyskerin (wildtype or variant) was assessed from three

169 independent experimental replicates (Figure 2B, C). Importantly, while the major localization

170 phenotype of the K446X truncation variant is cytoplasmic, FLAG-signal in cells expressing this

171 variant was also observed in both the cytoplasm and nucleolar fraction concomitantly (Figure

172 2B, blue bar). This is consistent with previous time course experiments demonstrating through

173 microinjection of EGFP-tagged K446X into cells that this truncation impairs but does not

174 entirely prevent nuclear and subnuclear localization of dyskerin (29). While still able to localize

175 within the nucleus and to the nucleolus, the truncation variant of dyskerin at A481, A481X, does

176 display an increase in concomitant nucleoplasmic and nucleolar localization of dyskerin

177 compared to wildtype, suggesting that this truncation modestly affects localization of dyskerin,

178 albeit to a lesser extent than K446X or K467R (Figure 2B, brown bar). Indeed, K467R has a

179 substantial reduction in nucleolar and corresponding increase in nucleoplasmic localization

180 compared to wildtype dyskerin, as assessed by exclusion from co-localization with fibrillarin

181 signal, but co-localization with DAPI signal (Figure 2B, red and black bars, respectively). In

182 contrast, no substantial differences in localization were observed between K468R and wildtype

183 (Figure 2B). The double substitution variant K467/468R does not differ in localization

184 compared to the single K467R variant, and thus has a reduction in nucleolar and increase in

185 nucleoplasmic localization compared to wildtype dyskerin (Figure 2B). As previously shown

186 (Figure 1C), fusing K446X to SUMO3 is sufficient to drive this truncation into the nucleus, but

187 this fusion variant has reduced nucleolar localization compared to wildtype dyskerin (Figure

188 2B). However, SUMO3-fusion of dyskerin differs from K467R in nucleolar exclusion, as

189 SUMO3-K446X and SUMO3-wildtype dyskerin form distinct puncta in the nucleoplasm while

190 K467R localization in the nucleoplasm is diffuse (Figure 1C and D, Figure 2A). Some of these

191 puncta may represent CB’s given their occasional overlap with fibrillarin puncta outside of

192 nucleolar clusters (Figure 1C arrowheads), but are most likely nucleoplasmic aggregates driven

193 by the permanent nature of the SUMO3-fusion (Figure 1C and D arrows). Importantly, neither

194 the FLAG-tag nor another tag (eGFP) disrupt localization of wildtype dyskerin, as eGFP-tagged

195 wildtype dyskerin (Figure 2D) and endogenous dyskerin examined by IF (Figure 2E) display

196 comparable localization patterns to exogenously expressed FLAG-tagged wildtype dyskerin.

197 Taken together, these data tell us that in addition to SUMO3 mediating nuclear localization of

198 the K446X truncation of dyskerin, the SUMO3 site K467 plays an important regulatory role for

199 the nucleolar localization of dyskerin.

200 Nuclear and subnuclear localization of dyskerin affects mature H/ACA RNP assembly

201 As a functional readout for H/ACA complex assembly and localization, co-

202 immunoprecipitation (co-IP) of FLAG-tagged dyskerin and interacting components was

203 performed from HEK293 cell lysate. Following FLAG-IP, interactions of FLAG-tagged dyskerin

204 wildtype and N/NoLS variants were assessed by immunoblotting for endogenous H/ACA RNP

205 assembly factors and components. Comparable to wildtype dyskerin, all FLAG-tagged N/NoLS

206 variants were able to interact with the pre-H/ACA RNP component NAF1 and the pre- and

207 mature H/ACA RNP components NOP10 and NHP2 (Figure 3A). Strikingly, the N/NoLS

208 variants with nucleolar exclusion phenotypes (K467R and K467/468R) were unable to interact

209 with the mature H/ACA RNP component GAR1 (Figure 3B). Importantly, disruption of either

210 nuclear or subnuclear localization of dyskerin leads to impaired hTR-dyskerin interaction as

211 measured by qPCR following RNA extraction and reverse transcription from IP fractions; neither

212 FLAG-tagged K467R nor K446X interact with hTR relative to wildtype dyskerin (Figure 3C).

213 This is in contrast to the N/NoLS variants with little to no localization defects, K468R and

214 A481X which do not display defective interactions with hTR relative to wildtype dyskerin

215 (Figure 3C). These data indicate that proper localization of dyskerin is tied to H/ACA RNP

216 complex assembly, connect GAR1-dyskerin interaction defects to the nucleolar exclusion of

217 dyskerin, and demonstrate that improper dyskerin localization disrupts the ability of dyskerin to

218 interact with H/ACA RNAs like hTR.

219 Defects in dyskerin localization affect telomerase activity

220 We asked whether H/ACA complex assembly defects disrupted dyskerin function in the

221 context of telomerase activity and H/ACA RNA biogenesis. In order to assess telomerase

222 activity, endogenous dyskerin was depleted via siRNA targeting the 3’ UTR of dyskerin in

223 HEK293 cells with or without stable expression of FLAG-tagged dyskerin wildtype or N/NoLS

224 variants. After 72h of depletion, telomerase activity was measured in the cell lysate using Q-

225 TRAP. HEK293 cells depleted of endogenous dyskerin have significantly reduced telomerase

226 activity compared to untreated (mock) cells, cells treated with a scramble siRNA, and cells with

227 stable exogenous expression of FLAG-tagged wildtype dyskerin that are depleted of endogenous

228 dyskerin (Figure 3D). Additionally, HEK293 cells with stable expression of FLAG-tagged

229 dyskerin localization variants that have defective localization (K467R, K467/8R, and K446X)

230 display significantly reduced telomerase activity following depletion of endogenous dyskerin. In

231 contrast, cells with stable expression of FLAG-tagged dyskerin localization variants that are

232 competent for nuclear and nucleolar localization (K468R and A481X) display telomerase

233 activity similar to cells with stable expression of FLAG-tagged wildtype dyskerin following

234 depletion of endogenous dyskerin.

235 GAR1 interaction with dyskerin mediates nucleolar localization in a SUMO-dependent manner

236 While the dyskerin variants that are excluded from the nucleolus do not interact with

237 GAR1, the mainly cytoplasmic K446X truncation which is competent for nucleolar localization

238 is capable of interacting with endogenous GAR1 (Figure 2B, Figure 3B). Due to this

239 observation that nucleolar localization of dyskerin is connected to the dyskerin-GAR1 interaction

240 and the SUMO3 site K467, we asked whether the interaction between GAR1 and dyskerin may

241 be SUMO3-mediated, and whether this interaction is responsible for mediating dyskerin

242 nucleolar localization. To test this, we performed FLAG co-IPs from HEK293 cells expressing

243 FLAG-tagged dyskerin fused to SUMO3 at the C-terminus. Interestingly, fusing SUMO3 to the

244 C-terminus of the K467R variant (Figure 4A) is able to rescue the robust GAR1 interaction

245 defect of K467R (Figure 3B), and wildtype dyskerin with C-terminal SUMO3 fusion is also able

246 to interact with GAR1 comparably to wildtype dyskerin alone (Figure 4A). However, the

247 subnuclear localization of both of these SUMO3-fusions does differ from that of wildtype

248 dyskerin, as assessed by IF, consistent with a predicted requirement for SUMOylation

249 reversibility for proper regulation of dyskerin subnuclear localization. Importantly, compared to

250 fusion of SUMO3 to the N-terminus of K446X or the non-fusion K467R variant, these C-

251 terminal SUMO3 fusions display more co-localization with fibrillarin in the nucleolar

252 compartment, though less frequent exclusive nucleolar localization of these fusions is observed

253 compared to wildtype dyskerin (Figure 2B, Figure 4B). This suggests that the nucleolar

254 localization of K467R, as well as interaction between GAR1 and dyskerin may indeed be

255 SUMO3-dependent. To further elucidate a potential SUMO3-mediated GAR1-dyskerin

256 interaction, and using the prediction software GPS-SUMO 4.0, we identified a single predicted

257 SIM within GAR1 at residues 70-74 (70-VVLLG-74) proximal to the previously predicted

258 dyskerin-GAR1 interface (Figure 4C). In order to assess whether this predicted SIM could

259 mediate the interaction between GAR1 and dyskerin, we substituted each residue in the predicted

260 SIM to alanine in a 3xFLAG-tagged GAR1 construct (annotated as 5A), and assessed the

261 interaction of 3xFLAG-tagged GAR1 with endogenous dyskerin. Wildtype 3xFLAG-tagged

262 GAR1 is able to interact with endogenous dyskerin, as assessed by FLAG co-IP from HEK293

263 cells expressing 3xFLAG-GAR1, however GAR1 5A displays a reduced interaction with

264 endogenous dyskerin (Figure 4D). These data demonstrate that GAR1 contains a SIM which

265 mediates the efficient interaction between dyskerin and GAR1 in a SUMO3-dependent manner,

266 relying on the SUMO3 site K467 in the C-terminal N/NoLS of dyskerin, and that this interaction

267 with GAR1 governs the localization of dyskerin in the nucleolus.

268 Discussion

269 Dyskerin and the H/ACA RNP complex play essential roles in H/ACA RNA biogenesis,

270 posttranscriptional modification of rRNA and snRNA, and in human telomerase assembly and

271 activity. The ability of dyskerin to carry out its various functions relies heavily on its nuclear and

272 subnuclear compartmentalization, where it assembles with H/ACA RNA and localizes to sites of

273 function, including the nucleolus where pseudouridine synthesis occurs on rRNA. In this study

274 we further demonstrate the interconnectedness of dyskerin localization, H/ACA RNP assembly,

275 and complex function. More specifically, we demonstrate that efficient nuclear localization of

276 dyskerin, driven by the K-rich C-terminal N/NoLS, as well as mature H/ACA complex assembly

277 and nucleolar localization mediated by K467 in this N/NoLS are crucial for dyskerin assembly

278 with H/ACA RNA like hTR. Furthermore, we demonstrate that the localization of dyskerin can

279 be mediated by SUMOylation sites in the C-terminal N/NoLS.

280 A previous study of dyskerin nuclear localization characterized two N/NoLS regions, one

281 in the N-terminus (amino acids 11-20) and one in the C-terminus (amino acids 446-514) (29).

282 This foundational study reported that removing or mutating the N-terminal region alone did not

283 disrupt localization of dyskerin, whereas removal of the C-terminal region alone drastically

284 impeded nuclear localization, and combinatorial removal of both regions abolished nuclear

285 localization altogether. As such, we focused on this C-terminal N/NoLS region as the primary

286 driver of dyskerin nuclear localization. Strikingly, we found that efficient localization of

287 dyskerin to the nucleus, while impaired by truncation of the C-terminal N/NoLS at K446, can be

288 driven by mimicking SUMOylation through fusing dyskerin to a SUMO3 moiety. This suggests

289 that the loss of SUMOylation sites from this truncation variant may be responsible for inefficient

290 dyskerin nuclear import and/or retention. Given the multitude of dyskerin SUMOylation sites

291 identified by MS which fall outside of the C-terminal N/NoLS of dyskerin, it is also likely that

292 dyskerin SUMOylation takes place within the nucleus for some sites following nuclear import.

293 Establishing which SUMOylation sites in particular govern nuclear localization requires further

294 investigation, as pinpointing the MS-identified SUMOylation sites in this region responsible for

295 nuclear localization was not evident by removal or substitution of K467, K468, K498, or K507.

296 However, this mutational analysis instead revealed that K467 plays an important regulatory role

297 in subnuclear localization of dyskerin to the nucleoli. Importantly, in our study and in the

298 previous work by Heiss et al., full truncation of the C-terminal region does not prevent nucleolar

299 localization of dyskerin per se (29). As such, we postulate that the C-terminal N/NoLS may

300 govern several aspects of stepwise dyskerin localization (nuclear import, nucleoplasmic

301 assembly with H/ACA RNA, and nucleolar miscibility) through regulated conformational

302 changes. More specifically, we speculate that a conformational change of this C-terminal region

303 governed by SUMOylation at K467 may be responsible for licensing dyskerin nucleolar

304 localization. However, the absence of this region as a whole allows for dyskerin nucleolar

305 localization in the absence of K467 SUMOylation, albeit in the context of inefficient nuclear

306 localization, because no conformational change is required for the K446X truncation variant.

307 This would be consistent with reports that full length dyskerin and dyskerin homologues are

308 difficult to purify in vitro due to insolubility issues which can be resolved by removal of this C-

309 terminal region (32, 57), and also in agreement with a lack of reported structure of this

310 functionally required tail due to its apparent intrinsic low complexity (6, 58, 59). It also seems

311 likely that a conformational change in dyskerin may be responsible for regulating the exchange

312 of GAR1 for NAF1 upon H/ACA complex maturation, though this needs further investigation.

313 Meanwhile, subnuclear localization of dyskerin-SUMO3 fusion proteins, variant or

314 wildtype, was observed to differ from wildtype dyskerin alone. We postulate that the constitutive

315 nature of this SUMOylation mimic disrupts nucleolar localization due to the inability of

316 deSUMOylating proteases to reverse this imitated posttranslational modification. This proposal

317 is based on not only the abundance of nucleolar SUMO-targets and SUMOylation machinery

318 involvement in nucleolar integrity (60-64), but also on the nucleolar localization of SUMO-

319 specific proteases (SENP3 and SENP5) involved in deconjugation of SUMO2/3 from target

320 proteins (65). SENP3 in particular has been demonstrated to interact with the nucleolar resident

321 protein nucleophosmin (NPM1), the 60S maturation factors PELP1, TEX10, WDR18, and

322 Las1L, and is capable of deSUMOylating NPM1, PELP1, and Las1L (66-68). Consistent with

323 the hypothesis that SUMO removal may regulate nucleolar localization of SUMO-target

324 proteins, depletion of SENP3, and thus reduction of nucleolar deSUMOylation, has been

325 reported to lead to nucleolar release of the PELP1-TEX10-WDR18 complex (68). Furthermore,

326 in yeast the nucleolar SUMO-specific protease Ulp2 has been demonstrated to reverse

327 SUMOylation of rDNA-bound SUMO-targets, and engineered increased SUMOylation by

328 depletion of Ulp2 leads to a reduction of several nucleolar proteins bound to rDNA (69, 70).

329 Intriguingly, NPM1 is responsible for localization of SENP3 to the nucleolus (71). NPM1 is a

330 resident protein of the outer-most nucleolar component where ribosomal subunit maturation

331 takes place, the granular component (GC), and as such would make a good candidate for

332 gatekeeping localization of nucleolar proteins in a SUMO removal-dependent manner. This

333 remains uninvestigated but would also fit into models of phase-mediated nucleolar

334 compartmentalization (33, 72), as discussed in greater detail below.

335 Here we also report that the efficient interaction between dyskerin and GAR1 is mediated

336 through a newly characterized SIM in GAR1 (amino acids 70-VVLLG-74). SIMs are typically

337 short hydrophobic stretches of residues that can form an extended β-strand backbone, which then

338 non-covalently interacts with SUMO moieties to foster stronger or more frequent SUMO-

339 mediated protein-protein interactions (38). It is important to note that this predicted motif is not

340 well conserved in lower eukaryotes or archaea (58). We found that substituting all five of these

341 GAR1 residues to alanine impairs the interaction of GAR1 with endogenous dyskerin, indicating

342 that an efficient interaction between GAR1 and dyskerin relies on this SIM, which is proximal to

343 but does not overlap with any of the residues structurally identified previously to mediate the

344 interaction between these two proteins in yeast and archaea (58, 73). Anecdotally, this SUMO-

345 mediated interaction between GAR1 and dyskerin may also offer some explanation for the

346 reported difficulty of in vitro reconstitution of H/ACA complexes using full length proteins, and

347 indeed the GAR1-dyskerin interface that has been identified structurally using homologues from

348 other organisms does not account for the C-terminal N/NoLS of human dyskerin as this region

349 was absent from the dyskerin homologues used for crystallization (58, 59, 74, 75). These

350 structural data also indicate that the GAR1-dyskerin interaction does take place in the absence of

351 SUMOylation and without the dyskerin C-terminal N/NoLS in vitro, indicating that while this

352 GAR1 SIM contributes to the efficient interaction between dyskerin and GAR1 in a cellular

353 context, this SIM is not required per se. We also observed that substituting the dyskerin SUMO3

354 site K467 to arginine abolishes the interaction between GAR1 and dyskerin, and that this GAR1

355 interaction defect of the K467R variant can be rescued by fusing K467R to SUMO3. It is not

356 known if K467 directly interfaces with GAR1, due to the absence of data on this C-terminal

357 region of dyskerin from structural studies. However, the observation that fusion of K467R to a

358 SUMO3 moiety can recover the ability of this variant to interact with GAR1 strongly implies

359 that SUMOylation of K467 mediates the efficient interaction between GAR1 and dyskerin.

360 Finally, we postulate that the SUMO-mediated interaction between GAR1 and dyskerin is

361 required for dyskerin localization to the nucleolus. Along with the data we present here, this

362 hypothesis is rooted in recent analyses of the nucleolar resident protein fibrillarin. Fibrillarin is a

363 small nucleolar RNP counterpart to dyskerin responsible for the 2ʹO-methylation

364 posttranscriptional modification of rRNA in the DFC, guided by C/D box snoRNA rather than

365 H/ACA box snoRNA (76-80). Several studies have demonstrated that localization of fibrillarin

366 to the DFC is mediated by an intrinsically disordered GAR domain, as well as by interactions

367 with nascent pre-rRNA as the RNA is sorted radially from its site of transcription through the

368 three nucleolar components, of which the DFC is the centre (33, 34). These studies and others

369 have shown that the nucleolus represents a complex membrane-free compartment with three

370 distinctly liquid-liquid phase separated components, which as a whole are phase separated from

371 the surrounding nucleoplasm (33, 72, 81-83). This context is important to bear in mind when

372 considering dynamic localization of resident nucleolar proteins in and out of these separated

373 phases. The regulated miscibility of fibrillarin with the DFC relies on its GAR domain and

374 protein-RNA interactions. As such, we propose that dyskerin miscibility with the DFC relies on

375 its interaction with GAR1, not only through acting as a GAR domain for dyskerin and the entire

376 H/ACA complex in trans, but also by providing high H/ACA complex-to-guide RNA affinity

377 which facilitates accurate H/ACA complex placement on target RNA, like rRNA in the nucleolus

378 (84, 85). This hypothesis is supported by our observations that 1) the K467R dyskerin variant is

379 unable to interact with GAR1 and the H/ACA box RNA hTR; 2) this K467R variant is unable to

380 co-localize with fibrillarin in the nucleolus; and 3) improving the interaction between the K467R

381 variant and GAR1 by fusing K467R to SUMO3 also allows for partial co-localization of the

382 K467R variant with fibrillarin in the nucleolus. Furthermore, the ability of the nucleolar-miscible

383 K446X truncation to fully assemble with H/ACA pre- and mature RNP components, including

384 interacting with GAR1 also lends support to this hypothesis. We also speculate that the lack of

385 GAR domains in the archaeal homologues of GAR1 and fibrillarin provides evolutionary support

386 for the notion that GAR domains mediate membrane-free compartmentalization of these

387 complexes in eukaryotes, as archaea lack nuclear compartmentalization altogether and would

388 have no need for GAR domain-mediated nucleolar miscibility of the otherwise evolutionarily

389 conserved H/ACA or C/D RNP complexes (86). Further confirmation of the phase dynamics of

390 human dyskerin with or without GAR1 is needed to elucidate this hypothesis.

391 Methods

392 Plasmids, Cell Culture, and Transfections

393 The plasmid pcDNA3.1-FLAG-dyskerinWT from the lab of Dr. François Dragon was used

394 to generate point mutations or truncations via site directed mutagenesis, as previously described

395 (16, 46). Specifically, primers (Table S1) were designed to generate K467R, K468R,

396 K467/468R, A481X, and K446X. For expression of 3xFLAG-GAR1 in human cells, the

397 pcDNA3.1 3xF-GAR1 plasmid was purchased from Addgene (#126873), and the predicted SIM

398 70-VVLLG-74 was substituted to 70-AAAAA-74 by site directed mutagenesis. The construct

399 pcDNA3.1-6xHis-SUMO3 was obtained from Dr. Frédérick Antoine Mallette (Université de

400 Montréal). The plasmid encoding eGFP-tagged dyskerin (pmEGFP-C1-DKC1) was a gift from

401 the lab of Dr. Ling-Ling Chen (Shanghai Institute of Biochemistry and Cell Biology) (34). All

402 constructs underwent Sanger DNA sequencing at Génome Québec CES.

403 Human embryonic kidney (HEK293) cells were maintained in Dulbecco’s Modification

404 Eagle’s Medium DMEM (Wisent) supplemented with 10% fetal bovine serum FBS (Wisent),

405 and Antibiotic-Antimycotic (Gibco), at 37˚C 5% CO2. Polyclonal FLAG-dyskerin stable cells

406 were maintained under selective pressure in G418 (750µg/ml). Transfection of pmEGFP-C1-

407 DKC1, pcDNA3.1-FLAG-dyskerin constructs, pcDNA3.1-6xHis-SUMO3, and/or pcDNA3.1

408 3xFLAG-GAR1 was performed using Lipofectamine 2000 Transfection Reagent (Invitrogen)

409 according to the reagent protocol. Prior to transfection, media was changed to DMEM with 10%

410 FBS and lacking Antibiotic-Antimycotic, and 5 hours after transfection the media was replaced

411 with DMEM containing both FBS and Antibiotic-Antimycotic.

412 Transfection of siRNA was performed with Lipofectamine RNAiMAX Transfection

413 Reagent (Invitrogen) according to the manufacturer’s user protocol. siRNA targeting the 3’ UTR

414 (24 nM, 72h treatment, sidkc1) was used for depletion of endogenous dyskerin (Table S1). The

415 siRNA sequence targeting the 3’ UTR was previously described (87). A mock transfection (no

416 siRNA) and transfection of a scramble siRNA were used as negative controls in each

417 experiment. siRNAs were ordered through ThermoFisher Scientific.

418 SUMO-interaction Motif Prediction

419 The GPS-SUMO 4.0 prediction tool was used to predict possible SUMO-interacting

420 motifs in GAR1. The coding amino acid sequence for isoform 1 of GAR1 was obtained in

421 FASTA format through Uniprot (identifier Q9NY12-1). The “SUMO Interaction Threshold” was

422 set to “Medium”. The SUMO Interaction prediction score obtained for residues 70-VVLLG-74

423 was 31.605, with a cutoff of 29.92 and P-value 0.112.

424 Nickel Affinity Purification of SUMOylated FLAG-dyskerin

425 For analysis of SUMOylated FLAG-dyskerin by immunoblotting, HEK293 cells

426 expressing 6xHis-SUMO3 and/or FLAG-dyskerin (wildtype, K446X, or SUMO3-K446X) were

427 lysed under denaturing conditions. Briefly, cells were washed with 1XPBS and collected by

428 scraping. One fifth of cells per condition were kept for input and lysed in 2xLaemmli followed

429 by boiling. The remainder of the cell pellet was lysed in 6M GuHCl buffer (10mM Tris-HCl

430 pH8, 6M GuHCl, 10mM imidazole, 0.1M NaH2PO4, adjusted to pH8 with NaOH) at room

431 temperature by passage through a 21G1¼ syringe (5x) followed by passage through an insulin

432 syringe (3x). Cell lysate was cleared by centrifugation at 13000rpm for 20min at 4˚C. The

433 supernatant was incubated with NiNTA resin (pre-washed 2x with 1XPBS and 1x with GuHCl

434 buffer) on a rotator at room temperature overnight. Resin was then washed 1x with GuHCl

435 buffer, 1x with wash buffer 1 (10mM Tris-HCl pH8, 8M urea, 10mM imidazole, 0.1M Na

436 H2PO4, adjusted to pH8 with NaOH), and 2x with wash buffer 2 (10mM Tris-HCl pH8, 8M urea,

437 10mM imidazole, 0.1M NaH2PO4, 0.1% v/v Triton X-100, adjusted to pH6.3 with NaOH). For

438 elution, resin was incubated in elution buffer (50mM NaH2PO4, 300mM NaCl, 500mM

439 imidazole, adjusted to pH8) for 3h on a rotator at 4˚C. The eluate was collected by centrifugation

440 and resin discarded.

441 Immunofluorescence

442 To assess localization of FLAG-dyskerin to the nucleolus, HEK293 cells expressing

443 FLAG-dyskerin constructs were fixed with 4% formaldehyde-PBS for 10 minutes at room

444 temperature. The fixing solution was removed and coverslips were briefly rinsed with PBS,

445 followed by permeabilization of cells with 0.1% Triton X-100-PBS for 5 minutes at 4˚C.

446 Permeabilized cells were then washed with PBS before blocking in 5% BSA-PBS for 1 hour at

447 room temperature. Cells were probed for FLAG-dyskerin with rabbit anti-FLAG (Sigma-Aldrich

448 F7425, 1:500) or mouse anti-dyskerin (Santa Cruz H-3, 1:25) in PBG (1% cold fish water

449 gelatin, 0.5% bovine serum albumin (BSA), in PBS) overnight at 4˚C in a humidity chamber. In

450 the case of assessing localization of only exogenous FLAG-tagged dyskerin, this was followed

451 by probing with mouse anti-fibrillarin (monoclonal antibody 72B9 obtained from Dr. Kenneth

452 Michael Pollard, 1:30) as a nucleolar marker, in PBG at 37˚C for 1 hour. Coverslips were

453 washed with PBS and immunostained in PBG with secondary antibodies conjugated to

454 fluorescein isothiocyanate (FITC) (donkey anti-mouse IgG; Jackson ImmunoResearch Lab, Inc.,

455 1:125) or Cy3 (donkey anti-rabbit; Jackson ImmunoResearch Lab, Inc., 1:125). Coverslips were

456 washed with PBS and mounted in Vectashield with DAPI (Vector Laboratories). Cells with

457 FLAG-dyskerin signal were manually scored based on localization phenotype as a percentage of

458 the number of cells with FLAG signal detected, ≥50 cells were counted in experimental triplicate

459 for scoring of localization of each FLAG-tagged dyskerin construct. Images were captured using

460 an Axio Imager 2 microscope (63X; Carl Zeiss, Jena, Germany). Nucleolar localization was

461 determined by co-localization with fibrillarin clusters, nucleoplasmic localization was

462 determined by co-localization with DAPI, and cytoplasmic localization was determined by

463 concentrated signal outside of and surrounding DAPI.

464 Immunoprecipitation

465 Protein-protein interactions were assessed by immunoprecipitating FLAG-dyskerin

466 wildtype or N/NoLS variants from HEK293 cells and immunoblotting for endogenous dyskerin-

467 interacting proteins; by immunoprecipitating 3xFLAG-GAR1 wildtype or 5A and

468 immunoblotting for endogenous dyskerin. Monoclonal M2 mouse anti-FLAG antibody (Sigma-

469 Aldrich F3165) and Protein G Sepharose (GE Healthcare) pre-blocked in 1% BSA-PBS were

470 used to immunoprecipitate (IP) FLAG-tagged and 3xFLAG-tagged proteins. The protocol used

471 to assess protein-protein interactions was the same used to analyze the interaction between

472 FLAG-dyskerin and hTR, and was modified based on a protocol that has been previously

473 described for another hTR-interacting protein (88), as well as used for FLAG-tagged dyskerin

474 (16). Briefly, cells were first lysed in low salt buffer (25mM HEPES-KCl pH7.9, 5mM KCl,

475 0.5mM MgCl2, 0.5% NP-40, 1X protease inhibitor cocktail from Roche, 20mM N-

476 ethylmaleimide, and 40U/ml RNAseOut) for 10min on ice. Lysates were cleared by

477 centrifugation at 5000rpm for 5min at 4˚C, supernatants were kept on ice, and pellets underwent

478 a second lysis in high salt buffer (25mM HEPES-KCl pH7.9, 350mM NaCl, 10% w/v sucrose,

479 0.01% NP-40, 1X protease inhibitor cocktail from Roche, 20mM N-ethylmaleimide, and 40U/ml

480 RNAseOut) with 30sec vortex followed by 30min on a rotator at 4˚C. Both low salt and high salt

481 lysates were then cleared by centrifugation at 13000rpm for 30min at 4˚C, supernatants were

482 pooled, and total lysate was pre-cleared at 4˚C on a rotator for 30min using Protein G Sepharose

483 that was pre-washed with 1XPBS. Bradford analysis was used to calculate total protein

484 concentration prior to IP. Lysates were incubated with anti-FLAG antibody for 2h at 4˚C on a

485 rotator before pre-blocked Protein G Sepharose was added, followed by an additional 1h

486 incubation at 4˚C on a rotator. IPs were washed 4x with 1ml of modified RIPA buffer (50mM

487 Tris-HCl pH8, 150mM NaCl, 10mM MgCl2, 1% NP-40, 0.5% sodium deoxycholate, 1mM

488 PMSF, 0.1X protease inhibitor cocktail from Rocher, and 20mM N-ethylmaleimide). For

489 protein-protein interactions, elution from Protein G Sepharose was performed with Laemmli

490 buffer and boiling. For protein-RNA interactions, elution was performed with TRIzol reagent

491 (Invitrogen), followed by chloroform extraction and reverse transcription. Inputs (10% of lysate

492 volume used for IP) were collected after pre-clearing with Protein G Sepharose and prior to IP,

493 and treated with either Laemmli buffer and boiled, or with TRIzol reagent.

494 Immunoblotting and Antibodies

495 Analysis of protein expression and IP experiments was performed by resolving proteins

496 by SDS-PAGE, transfer to PVDF and immunoblotting. Primary antibodies used for

497 immunoblotting were: anti-FLAG (Proteintech, 20543-1-AP, 1:4000) or anti-FLAG (Proteintech,

498 66008-3-Ig, 1:4000), anti-NAF1 (Abcam, ab157106, 1:1000), anti-NHP2 (Proteintech, 15128-1-

499 AP, 1:5000), anti-NOP10 (Abcam, ab134902, 1:500), anti-dyskerin (Santa Cruz, sc-373956, H-3,

500 1:1500), anti-GAR1 (Proteintech, 11711-1-AP, 1:1000), anti-His (Santa Cruz, sc-8036, H-3,

501 1:500), and anti-alpha tubulin (Sigma, T5168, 1:5000).

502

503 RNA Extraction and RT-qPCR

504 RNA was extracted using TRIzol reagent (Invitrogen), according to the reagent protocol.

505 Reverse transcription was performed with SuperScript II Reverse Transcriptase (Invitrogen)

506 according to the user protocol, with hexameric random primers. PerfeCTa SYBR Green FastMix

507 with Low ROX (Quanta) was used for all qPCR analyses, in a 7500FAST real-time PCR system

508 (ABI) as previously described (46). The comparative ΔΔCT method was used to compare RNA

509 enrichment between samples. For analysis of protein-RNA interactions, 5 µl of RNA from input

510 and 5 µl of RNA from IP fractions were reverse transcribed into cDNA and subjected to qPCR

511 using specific primers for target RNAs (Table S1). The ΔΔCT was calculated between the mean

512 CT of the IP and the mean CT of the input for each sample.

513 Q-TRAP

514 Quantitative analysis of telomerase activity was done using the Q-TRAP protocol

515 previously described (89). Briefly, HEK293 cells with or without expression of FLAG-dyskerin

516 constructs were treated with scramble siRNA or siRNA to deplete endogenous dyskerin for 72h

517 prior to harvesting by scraping and lysis in NP-40 lysis buffer. A standard curve was generated

518 with a serial dilution of mock lysate (HEK293 cells untreated with siRNA and not expressing

519 FLAG-dyskerin) for each experimental replicate (n=3), with 1 µg, 0.2 µg, 0.04 µg, 0.008 µg, and

520 0.0016 µg of total protein. For comparison of telomerase activity between conditions, 0.2 µg of

521 total protein was used for each sample.

522 Statistical Analyses

523 All statistical analyses were performed using GraphPad Prism 7. One-way ANOVA tests

524 (p < 0.01) were used to compare RNA enrichment when assessing interaction between FLAG-

525 dyskerin and hTR, and for comparison of relative telomerase activity (RTA) in Q-TRAP

526 experiments. For analysis of RNA interaction, the enrichment of hTR in each N/NoLS variant IP

527 fraction was compared to the enrichment of hTR in the FLAG-dyskerin wildtype IP fraction. For

528 analysis of telomerase activity, the RTA percentage of each condition was separately compared

529 to the mock (untreated HEK293 cells) RTA percentage. Each experiment was performed in

530 triplicate, and error bars represent the standard error of the mean between experimental

531 replicates. Dunnet’s test was used to correct for multiple comparisons.

532 Acknowledgements

533 We thank Dr. Frédérick Antoine Mallette, Dr. François Dragon, and Dr. Ling-Ling Chen

534 for providing us with pcDNA3.1-6xHis-SUMO3, pcDNA3.1-FLAG-dyskerinWT, and pmEGFP-

535 C1-DKC1 plasmids, respectively. We thank Dr. Kenneth Michael Pollard for providing us with

536 mouse anti-fibrillarin antibody for IF. We thank Dr. Stéphane Richard for use of the Axio Imager

537 2 microscope.

538 Author Contributions

539 Author contributions: D.E. M., and C. A. designed research; D.E. M, P. L.-L., J. Q., F.

540 M., and E. B. performed experiments; D.E. M. and C. A. wrote the manuscript; all authors

541 contributed to reviewing and editing the manuscript.

542 Declaration of Interests

543 References

544 1. Balakin AG, Smith L, Fournier MJ. 1996. The RNA world of the nucleolus: two major

545 families of small RNAs defined by different box elements with related functions. Cell

546 86:823-34.

547 2. Ganot P, Bortolin ML, Kiss T. 1997. Site-specific pseudouridine formation in

548 preribosomal RNA is guided by small nucleolar RNAs. Cell 89:799-809.

549 3. Ganot P, Caizergues-Ferrer M, Kiss T. 1997. The family of box ACA small nucleolar

550 RNAs is defined by an evolutionarily conserved secondary structure and ubiquitous

551 sequence elements essential for RNA accumulation. Genes Dev 11:941-56.

552 4. Ni J, Tien AL, Fournier MJ. 1997. Small nucleolar RNAs direct site-specific synthesis of

553 pseudouridine in ribosomal RNA. Cell 89:565-73.

554 5. Darzacq X, Jady BE, Verheggen C, Kiss AM, Bertrand E, Kiss T. 2002. Cajal body-

555 specific small nuclear RNAs: a novel class of 2'-O-methylation and pseudouridylation

556 guide RNAs. EMBO J 21:2746-56.

557 6. Jiang W, Middleton K, Yoon HJ, Fouquet C, Carbon J. 1993. An essential yeast protein,

558 CBF5p, binds in vitro to centromeres and microtubules. Mol Cell Biol 13:4884-93.

559 7. Meier UT, Blobel G. 1994. NAP57, a mammalian nucleolar protein with a putative

560 homolog in yeast and bacteria. J Cell Biol 127:1505-14.

561 8. Heiss NS, Knight SW, Vulliamy TJ, Klauck SM, Wiemann S, Mason PJ, Poustka A,

562 Dokal I. 1998. X-linked dyskeratosis congenita is caused by mutations in a highly

563 conserved gene with putative nucleolar functions. Nat Genet 19:32-8.

564 9. Henras A, Henry Y, Bousquet-Antonelli C, Noaillac-Depeyre J, Gelugne JP, Caizergues-

565 Ferrer M. 1998. Nhp2p and Nop10p are essential for the function of H/ACA snoRNPs.

566 EMBO J 17:7078-90.

567 10. Girard JP, Lehtonen H, Caizergues-Ferrer M, Amalric F, Tollervey D, Lapeyre B. 1992.

568 GAR1 is an essential small nucleolar RNP protein required for pre-rRNA processing in

569 yeast. EMBO J 11:673-82.

570 11. Podlevsky JD, Chen JJ. 2016. Evolutionary perspectives of telomerase RNA structure

571 and function. RNA Biol 13:720-32.

572 12. Mitchell JR, Cheng J, Collins K. 1999. A box H/ACA small nucleolar RNA-like domain

573 at the human telomerase RNA 3' end. Mol Cell Biol 19:567-76.

574 13. Jady BE, Bertrand E, Kiss T. 2004. Human telomerase RNA and box H/ACA scaRNAs

575 share a common Cajal body-specific localization signal. J Cell Biol 164:647-52.

576 14. Tseng CK, Wang HF, Schroeder MR, Baumann P. 2018. The H/ACA complex disrupts

577 triplex in hTR precursor to permit processing by RRP6 and PARN. Nat Commun 9:5430.

578 15. Roake CM, Chen L, Chakravarthy AL, Ferrell JE, Jr., Raffa GD, Artandi SE. 2019.

579 Disruption of Telomerase RNA Maturation Kinetics Precipitates Disease. Mol Cell

580 74:688-700 e3.

581 16. MacNeil DE, Lambert-Lanteigne P, Autexier C. 2019. N-terminal residues of human

582 dyskerin are required for interactions with telomerase RNA that prevent RNA

583 degradation. Nucleic Acids Res 47:5368-5380.

584 17. Podlevsky JD, Bley CJ, Omana RV, Qi X, Chen JJ. 2008. The telomerase database.

585 Nucleic Acids Res 36:D339-43.

586 18. Connor JM, Gatherer D, Gray FC, Pirrit LA, Affara NA. 1986. Assignment of the gene

587 for dyskeratosis congenita to Xq28. Hum Genet 72:348-51.

588 19. Drachtman RA, Alter BP. 1992. Dyskeratosis congenita: clinical and genetic

589 heterogeneity. Report of a new case and review of the literature. Am J Pediatr Hematol

590 Oncol 14:297-304.

591 20. Arngrimsson R, Dokal I, Luzzatto L, Connor JM. 1993. Dyskeratosis congenita: three

592 additional families show linkage to a locus in Xq28. J Med Genet 30:618-9.

593 21. Dokal I. 1996. Dyskeratosis congenita: an inherited bone marrow failure syndrome. Br J

594 Haematol 92:775-9.

595 22. Bellodi C, McMahon M, Contreras A, Juliano D, Kopmar N, Nakamura T, Maltby D,

596 Burlingame A, Savage SA, Shimamura A, Ruggero D. 2013. H/ACA small RNA

597 dysfunctions in disease reveal key roles for noncoding RNA modifications in

598 hematopoietic stem cell differentiation. Cell Rep 3:1493-502.

599 23. Benyelles M, O'Donohue MF, Kermasson L, Lainey E, Borie R, Lagresle-Peyrou C,

600 Nunes H, Cazelles C, Fourrage C, Ollivier E, Marcais A, Gamez AS, Morice-Picard F,

601 Caillaud D, Pottier N, Menard C, Ba I, Fernandes A, Crestani B, de Villartay JP, Gleizes

602 PE, Callebaut I, Kannengiesser C, Revy P. 2020. NHP2 deficiency impairs rRNA

603 biogenesis and causes pulmonary fibrosis and Hoyeraal-Hreidarsson syndrome. Hum Mol

604 Genet doi:10.1093/hmg/ddaa011.

605 24. Phillips B, Billin AN, Cadwell C, Buchholz R, Erickson C, Merriam JR, Carbon J, Poole

606 SJ. 1998. The Nop60B gene of Drosophila encodes an essential nucleolar protein that

607 functions in yeast. Mol Gen Genet 260:20-9.

608 25. Giordano E, Peluso I, Senger S, Furia M. 1999. minifly, a Drosophila gene required for

609 ribosome biogenesis. J Cell Biol 144:1123-33.

610 26. He J, Navarrete S, Jasinski M, Vulliamy T, Dokal I, Bessler M, Mason PJ. 2002.

611 Targeted disruption of Dkc1, the gene mutated in X-linked dyskeratosis congenita, causes

612 embryonic lethality in mice. Oncogene 21:7740-4.

613 27. Hart T, Chandrashekhar M, Aregger M, Steinhart Z, Brown KR, MacLeod G, Mis M,

614 Zimmermann M, Fradet-Turcotte A, Sun S, Mero P, Dirks P, Sidhu S, Roth FP, Rissland

615 OS, Durocher D, Angers S, Moffat J. 2015. High-Resolution CRISPR Screens Reveal

616 Fitness Genes and Genotype-Specific Cancer Liabilities. Cell 163:1515-26.

617 28. Bertomeu T, Coulombe-Huntington J, Chatr-Aryamontri A, Bourdages KG, Coyaud E,

618 Raught B, Xia Y, Tyers M. 2018. A High-Resolution Genome-Wide CRISPR/Cas9

619 Viability Screen Reveals Structural Features and Contextual Diversity of the Human

620 Cell-Essential Proteome. Mol Cell Biol 38.

621 29. Heiss NS, Girod A, Salowsky R, Wiemann S, Pepperkok R, Poustka A. 1999. Dyskerin

622 localizes to the nucleolus and its mislocalization is unlikely to play a role in the

623 pathogenesis of dyskeratosis congenita. Hum Mol Genet 8:2515-24.

624 30. Ballarino M, Morlando M, Pagano F, Fatica A, Bozzoni I. 2005. The cotranscriptional

625 assembly of snoRNPs controls the biosynthesis of H/ACA snoRNAs in Saccharomyces

626 cerevisiae. Mol Cell Biol 25:5396-403.

627 31. Yang PK, Hoareau C, Froment C, Monsarrat B, Henry Y, Chanfreau G. 2005.

628 Cotranscriptional recruitment of the pseudouridylsynthetase Cbf5p and of the RNA

629 binding protein Naf1p during H/ACA snoRNP assembly. Mol Cell Biol 25:3295-304.

630 32. Darzacq X, Kittur N, Roy S, Shav-Tal Y, Singer RH, Meier UT. 2006. Stepwise RNP

631 assembly at the site of H/ACA RNA transcription in human cells. J Cell Biol 173:207-18.

632 33. Feric M, Vaidya N, Harmon TS, Mitrea DM, Zhu L, Richardson TM, Kriwacki RW,

633 Pappu RV, Brangwynne CP. 2016. Coexisting Liquid Phases Underlie Nucleolar

634 Subcompartments. Cell 165:1686-1697.

635 34. Yao RW, Xu G, Wang Y, Shan L, Luan PF, Wang Y, Wu M, Yang LZ, Xing YH, Yang

636 L, Chen LL. 2019. Nascent Pre-rRNA Sorting via Phase Separation Drives the Assembly

637 of Dense Fibrillar Components in the Human Nucleolus. Mol Cell 76:767-783 e11.

638 35. Pichler A, Melchior F. 2002. Ubiquitin-related modifier SUMO1 and nucleocytoplasmic

639 transport. Traffic 3:381-7.

640 36. Melchior F, Schergaut M, Pichler A. 2003. SUMO: ligases, isopeptidases and nuclear

641 pores. Trends Biochem Sci 28:612-8.

642 37. Heun P. 2007. SUMOrganization of the nucleus. Curr Opin Cell Biol 19:350-5.

643 38. Varejao N, Lascorz J, Li Y, Reverter D. 2019. Molecular mechanisms in SUMO

644 conjugation. Biochem Soc Trans doi:10.1042/BST20190357.

645 39. Song J, Durrin LK, Wilkinson TA, Krontiris TG, Chen Y. 2004. Identification of a

646 SUMO-binding motif that recognizes SUMO-modified proteins. Proc Natl Acad Sci U S

647 A 101:14373-8.

648 40. Song J, Zhang Z, Hu W, Chen Y. 2005. Small ubiquitin-like modifier (SUMO)

649 recognition of a SUMO binding motif: a reversal of the bound orientation. J Biol Chem

650 280:40122-9.

651 41. Hecker CM, Rabiller M, Haglund K, Bayer P, Dikic I. 2006. Specification of SUMO1-

652 and SUMO2-interacting motifs. J Biol Chem 281:16117-27.

653 42. Li SJ, Hochstrasser M. 1999. A new protease required for cell-cycle progression in yeast.

654 Nature 398:246-51.

655 43. Li SJ, Hochstrasser M. 2000. The yeast ULP2 (SMT4) gene encodes a novel protease

656 specific for the ubiquitin-like Smt3 protein. Mol Cell Biol 20:2367-77.

657 44. Takahashi Y, Mizoi J, Toh EA, Kikuchi Y. 2000. Yeast Ulp1, an Smt3-specific protease,

658 associates with nucleoporins. J Biochem 128:723-5.

659 45. Gong L, Millas S, Maul GG, Yeh ET. 2000. Differential regulation of sentrinized

660 proteins by a novel sentrin-specific protease. J Biol Chem 275:3355-9.

661 46. Brault ME, Lauzon C, Autexier C. 2013. Dyskeratosis congenita mutations in dyskerin

662 SUMOylation consensus sites lead to impaired telomerase RNA accumulation and

663 telomere defects. Hum Mol Genet 22:3498-507.

664 47. Becker J, Barysch SV, Karaca S, Dittner C, Hsiao HH, Berriel Diaz M, Herzig S, Urlaub

665 H, Melchior F. 2013. Detecting endogenous SUMO targets in mammalian cells and

666 tissues. Nat Struct Mol Biol 20:525-31.

667 48. Impens F, Radoshevich L, Cossart P, Ribet D. 2014. Mapping of SUMO sites and

668 analysis of SUMOylation changes induced by external stimuli. Proc Natl Acad Sci U S A

669 111:12432-7.

670 49. Hendriks IA, D'Souza RC, Yang B, Verlaan-de Vries M, Mann M, Vertegaal AC. 2014.

671 Uncovering global SUMOylation signaling networks in a site-specific manner. Nat Struct

672 Mol Biol 21:927-36.

673 50. Xiao Z, Chang JG, Hendriks IA, Sigurethsson JO, Olsen JV, Vertegaal AC. 2015.

674 System-wide Analysis of SUMOylation Dynamics in Response to Replication Stress

675 Reveals Novel Small Ubiquitin-like Modified Target Proteins and Acceptor Lysines

676 Relevant for Genome Stability. Mol Cell Proteomics 14:1419-34.

677 51. Hendriks IA, Treffers LW, Verlaan-de Vries M, Olsen JV, Vertegaal ACO. 2015.

678 SUMO-2 Orchestrates Chromatin Modifiers in Response to DNA Damage. Cell Rep

679 10:1778-1791.

680 52. Lamoliatte F, McManus FP, Maarifi G, Chelbi-Alix MK, Thibault P. 2017. Uncovering

681 the SUMOylation and ubiquitylation crosstalk in human cells using sequential peptide

682 immunopurification. Nat Commun 8:14109.

683 53. Hendriks IA, Lyon D, Young C, Jensen LJ, Vertegaal AC, Nielsen ML. 2017. Site-

684 specific mapping of the human SUMO proteome reveals co-modification with

685 phosphorylation. Nat Struct Mol Biol 24:325-336.

686 54. Hendriks IA, Lyon D, Su D, Skotte NH, Daniel JA, Jensen LJ, Nielsen ML. 2018. Site-

687 specific characterization of endogenous SUMOylation across species and organs. Nat

688 Commun 9:2456.

689 55. El-Asmi F, McManus FP, Brantis-de-Carvalho CE, Valle-Casuso JC, Thibault P, Chelbi-

690 Alix MK. 2020. Cross-talk between SUMOylation and ISGylation in response to

691 interferon. Cytokine 129:155025.

692 56. Zhao X. 2018. SUMO-Mediated Regulation of Nuclear Functions and Signaling

693 Processes. Mol Cell 71:409-418.

694 57. Normand C, Capeyrou R, Quevillon-Cheruel S, Mougin A, Henry Y, Caizergues-Ferrer

695 M. 2006. Analysis of the binding of the N-terminal conserved domain of yeast Cbf5p to a

696 box H/ACA snoRNA. RNA 12:1868-82.

697 58. Li S, Duan J, Li D, Yang B, Dong M, Ye K. 2011. Reconstitution and structural analysis

698 of the yeast box H/ACA RNA-guided pseudouridine synthase. Genes Dev 25:2409-21.

699 59. Li S, Duan J, Li D, Ma S, Ye K. 2011. Structure of the Shq1-Cbf5-Nop10-Gar1 complex

700 and implications for H/ACA RNP biogenesis and dyskeratosis congenita. EMBO J

701 30:5010-20.

702 60. Ayaydin F, Dasso M. 2004. Distinct in vivo dynamics of vertebrate SUMO paralogues.

703 Mol Biol Cell 15:5208-18.

704 61. Matafora V, D'Amato A, Mori S, Blasi F, Bachi A. 2009. Proteomics analysis of

705 nucleolar SUMO-1 target proteins upon proteasome inhibition. Mol Cell Proteomics

706 8:2243-55.

707 62. Srikumar T, Lewicki MC, Raught B. 2013. A global S. cerevisiae small ubiquitin-related

708 modifier (SUMO) system interactome. Mol Syst Biol 9:668.

709 63. Takahashi Y, Dulev S, Liu X, Hiller NJ, Zhao X, Strunnikov A. 2008. Cooperation of

710 sumoylated chromosomal proteins in rDNA maintenance. PLoS Genet 4:e1000215.

711 64. Zhao X, Blobel G. 2005. A SUMO ligase is part of a nuclear multiprotein complex that

712 affects DNA repair and chromosomal organization. Proc Natl Acad Sci U S A 102:4777-

713 82.

714 65. Yun C, Wang Y, Mukhopadhyay D, Backlund P, Kolli N, Yergey A, Wilkinson KD,

715 Dasso M. 2008. Nucleolar protein B23/nucleophosmin regulates the vertebrate SUMO

716 pathway through SENP3 and SENP5 proteases. J Cell Biol 183:589-95.

717 66. Haindl M, Harasim T, Eick D, Muller S. 2008. The nucleolar SUMO-specific protease

718 SENP3 reverses SUMO modification of nucleophosmin and is required for rRNA

719 processing. EMBO Rep 9:273-9.

720 67. Castle CD, Cassimere EK, Denicourt C. 2012. LAS1L interacts with the mammalian

721 Rix1 complex to regulate ribosome biogenesis. Mol Biol Cell 23:716-28.

722 68. Finkbeiner E, Haindl M, Muller S. 2011. The SUMO system controls nucleolar

723 partitioning of a novel mammalian ribosome biogenesis complex. EMBO J 30:1067-78.

724 69. Gillies J, Hickey CM, Su D, Wu Z, Peng J, Hochstrasser M. 2016. SUMO Pathway

725 Modulation of Regulatory Protein Binding at the Ribosomal DNA Locus in

726 Saccharomyces cerevisiae. Genetics 202:1377-94.

727 70. Liang J, Singh N, Carlson CR, Albuquerque CP, Corbett KD, Zhou H. 2017. Recruitment

728 of a SUMO isopeptidase to rDNA stabilizes silencing complexes by opposing SUMO

729 targeted ubiquitin ligase activity. Genes Dev 31:802-815.

730 71. Raman N, Nayak A, Muller S. 2014. mTOR signaling regulates nucleolar targeting of the

731 SUMO-specific isopeptidase SENP3. Mol Cell Biol 34:4474-84.

732 72. Brangwynne CP, Mitchison TJ, Hyman AA. 2011. Active liquid-like behavior of nucleoli

733 determines their size and shape in Xenopus laevis oocytes. Proc Natl Acad Sci U S A

734 108:4334-9.

735 73. Rashid R, Liang B, Baker DL, Youssef OA, He Y, Phipps K, Terns RM, Terns MP, Li H.

736 2006. Crystal structure of a Cbf5-Nop10-Gar1 complex and implications in RNA-guided

737 pseudouridylation and dyskeratosis congenita. Mol Cell 21:249-60.

738 74. Walbott H, Machado-Pinilla R, Liger D, Blaud M, Rety S, Grozdanov PN, Godin K, van

739 Tilbeurgh H, Varani G, Meier UT, Leulliot N. 2011. The H/ACA RNP assembly factor

740 SHQ1 functions as an RNA mimic. Genes Dev 25:2398-408.

741 75. Singh M, Wang Z, Cascio D, Feigon J. 2015. Structure and interactions of the CS domain

742 of human H/ACA RNP assembly protein Shq1. J Mol Biol 427:807-23.

743 76. Kiss-Laszlo Z, Henry Y, Bachellerie JP, Caizergues-Ferrer M, Kiss T. 1996. Site-specific

744 ribose methylation of preribosomal RNA: a novel function for small nucleolar RNAs.

745 Cell 85:1077-88.

746 77. Cavaille J, Nicoloso M, Bachellerie JP. 1996. Targeted ribose methylation of RNA in

747 vivo directed by tailored antisense RNA guides. Nature 383:732-5.

748 78. Tycowski KT, You ZH, Graham PJ, Steitz JA. 1998. Modification of U6 spliceosomal

749 RNA is guided by other small RNAs. Mol Cell 2:629-38.

750 79. Jady BE, Kiss T. 2001. A small nucleolar guide RNA functions both in 2'-O-ribose

751 methylation and pseudouridylation of the U5 spliceosomal RNA. EMBO J 20:541-51.

752 80. Ganot P, Jady BE, Bortolin ML, Darzacq X, Kiss T. 1999. Nucleolar factors direct the 2'-

753 O-ribose methylation and pseudouridylation of U6 spliceosomal RNA. Mol Cell Biol

754 19:6906-17.

755 81. Weber SC, Brangwynne CP. 2015. Inverse size scaling of the nucleolus by a

756 concentration-dependent phase transition. Curr Biol 25:641-6.

757 82. Scheer U, Weisenberger D. 1994. The nucleolus. Curr Opin Cell Biol 6:354-9.

758 83. Boisvert FM, van Koningsbruggen S, Navascues J, Lamond AI. 2007. The

759 multifunctional nucleolus. Nat Rev Mol Cell Biol 8:574-85.

760 84. Caton EA, Kelly EK, Kamalampeta R, Kothe U. 2018. Efficient RNA pseudouridylation

761 by eukaryotic H/ACA ribonucleoproteins requires high affinity binding and correct

762 positioning of guide RNA. Nucleic Acids Res 46:905-916.

763 85. Wang P, Yang L, Gao YQ, Zhao XS. 2015. Accurate placement of substrate RNA by

764 Gar1 in H/ACA RNA-guided pseudouridylation. Nucleic Acids Res 43:7207-16.

765 86. Lafontaine DL, Tollervey D. 1998. Birth of the snoRNPs: the evolution of the

766 modification-guide snoRNAs. Trends Biochem Sci 23:383-8.

767 87. Lin P, Mobasher ME, Alawi F. 2014. Acute dyskerin depletion triggers cellular

768 senescence and renders osteosarcoma cells resistant to genotoxic stress-induced

769 apoptosis. Biochem Biophys Res Commun 446:1268-75.

770 88. Booy EP, Meier M, Okun N, Novakowski SK, Xiong S, Stetefeld J, McKenna SA. 2012.

771 The RNA helicase RHAU (DHX36) unwinds a G4-quadruplex in human telomerase

772 RNA and promotes the formation of the P1 helix template boundary. Nucleic Acids Res

773 40:4110-24.

774 89. Herbert BS, Hochreiter AE, Wright WE, Shay JW. 2006. Nonradioactive detection of

775 telomerase activity using the telomeric repeat amplification protocol. Nat Protoc 1:1583-

776 90.

777 Figure Titles and Legends

778 Figure 1: Residues in the C-terminal nuclear/nucleolar localization sequence of dyskerin

779 are SUMO3 targets that govern nuclear accumulation. a. A linear schematic of human

780 dyskerin domains. The amino acid range corresponding to the predicted C-terminal

781 nuclear/nucleolar localization sequence (N/NoLS) (438-514) is denoted below the schematic,

782 indicating the MS-identified SUMO3 sites in this region (K467, K468, K498, and K507) with

783 solid black circles, and the two lysine (K)-rich clusters (K467-K480, and K498-K507) are

784 underlined. MS-identified SUMO3 sites reported in proteome-wide studies cited in the text are

785 indicated by solid black circles above the schematic. Residues A481 and K446 are indicated by

786 arrowheads. b. FLAG-dyskerin (wildtype WT and dyskerin truncation variant K446X without or

787 with N-terminal SUMO3 fusion) and 6xHis-SUMO3 were expressed in HEK293 cells. His-

788 SUMO3 conjugates were purified using Ni-NTA agarose beads following lysis under denaturing

789 conditions, and SUMOylated FLAG-dyskerin was assessed in the elution by immunoblotting

790 with an anti-FLAG antibody. A fraction of each HEK293 cell pellet used for purification was

791 kept prior to lysis (Input) for checking expression of FLAG-dyskerin and His-SUMO3 by

792 immunoblotting. The K446X truncation runs at the expected lower molecular weight than WT

793 dyskerin, while SUMO3-K446X runs at the expected higher molecular weight than WT due to

794 the SUMO3 fusion. FLAG-dyskerin is indicated by arrows, while asterisks indicate non-specific

795 antibody signal. The higher molecular weight smear observed in the SUMO3-K446X elution

796 (indicated by the bracket to the right of the blot) indicates SUMOylated species of dyskerin. c.

797 Representative images of the co-localization of FLAG-dyskerin (wildtype WT and dyskerin

798 truncation variant K446X without or with N-terminal SUMO3 fusion – Cy3 shown in red) with

799 nucleolar marker fibrillarin (FITC shown in green), as observed by indirect immunofluorescence.

800 The nucleus is indicated by DAPI staining of nuclear DNA (in blue). d. Localization of FLAG-

801 tagged WT, N-terminal SUMO3 fusion WT dyskerin (SUMO3-WT), and C-terminal SUMO3

802 fusion WT dyskerin (WT-SUMO3) was assessed by indirect immunofluorescence (Cy3 shown in

803 red). Fibrillarin was used as a nucleolar marker (FITC shown in green), and the nucleus is

804 indicated by DAPI staining of nuclear DNA (in blue). In c. and d. examples of nucleoplasmic

805 FLAG-dyskerin foci that co-localize with nucleoplasmic fibrillarin foci are indicated by

806 arrowheads, while arrows indicate nucleoplasmic FLAG-dyskerin foci that do not co-localize

807 with nucleoplasmic fibrillarin foci. All scale bars indicate 10µm.

808 Figure 2: Nuclear and subnuclear localization of dyskerin is mediated by SUMO3 sites in

809 the C-terminal nuclear/nucleolar localization sequence. a. FLAG-dyskerin was transiently

810 expressed in HEK293 cells, and localization was assessed in fixed cells by indirect

811 immunofluorescence. Representative images of the most prevalent localization phenotype of

812 FLAG-dyskerin (wildtype WT, dyskerin truncation variant A481X, and substitution variants

813 K467R, K468R, and double K467/468R – Cy3 in red) and the nucleolar marker fibrillarin (FITC

814 in green) are shown. Nucleolar localization is represented for WT and K468R, concomitant

815 nucleoplasmic & nucleolar is represented by A481X, and nucleoplasmic localization is

816 represented by K467R and K467/468R. The nucleus is indicated by DAPI staining of nuclear

817 DNA (in blue) b. Quantification of localization phenotype scoring for FLAG-dyskerin WT and

818 localization variants as a percentage of FLAG-positive HEK293 cells is indicated (≥50 cells per

819 condition were counted, in experiment replicate n=3). c. Examples of how localization

820 phenotype was scored are shown. Nucleolar localization was determined by co-localization with

821 clustered fibrillarin signal, nucleoplasmic localization was determined by co-localization with

822 DAPI outside of the clustered fibrillarin signal, and cytoplasmic localization was determined by

823 concentrated signal outside of and surrounding DAPI. d. Localization of eGFP and eGFP-tagged

824 WT dyskerin (in green) was assessed in fixed HEK293 cells following transient transfection,

825 using a FITC filter. e. Localization of endogenous dyskerin (Cy3, in red) was assessed by IF in

826 fixed HEK293 cells. As a negative control, fixed HEK293 cells were assessed by IF using only

827 secondary Cy3-conjugated antibody. The nucleus is indicated by DAPI staining of nuclear DNA

828 (in blue). These are representative images.All scale bars indicate 10µm.

829 Figure 3: Dyskerin nuclear and nucleolar localization is linked to mature H/ACA complex

830 assembly and function. Interactions of FLAG-dyskerin WT and localization variants with

831 endogenous pre- and mature H/ACA ribonucleoprotein complex components were assessed by

832 co-immunoprecipitation (IP) from HEK293 cell lysates. Assembly of the a. H/ACA pre-RNP

833 complex involving NAF1, NHP2, and NOP10 was investigated by immunoblotting for the

834 endogenous H/ACA pre-RNP components and FLAG-dyskerin proteins following IP. b.

835 Interaction of dyskerin with the mature H/ACA complex component GAR1 was examined

836 following IP by immunoblotting for endogenous GAR1 and FLAG-dyskerin. Localization

837 variants that are excluded from the nucleolus (K467R and K467/468R) do not interact with

838 GAR1. Immunoblotting targets are indicated to the right of each panel as “WB: α target”, and a

839 list of antibodies can be found in the materials and methods section. A non-specific band

840 revealed by the anti-GAR1 antibody at 25kDa is indicated by an asterisk. Each co-IP and

841 immunoblotting was performed in experimental replicate a minimum of n=2, representative blots

842 are shown. c. Dyskerin-hTR interactions were assessed by IP of FLAG-tagged dyskerin followed

843 by RNA extraction and qPCR. Relative to wildtype IP fractions, dyskerin variants with

844 substantial localization defects (K467R and K446X) display significantly reduced enrichment of

845 hTR following IP. HEK293 cells lacking FLAG-tagged dyskerin (indicated as mock) were used

846 as a negative control for RNA binding to the FLAG antibody and/or Protein G Sepharose. Mock

847 cells were subject to the same IP protocol detailed for fractions containing FLAG-tagged

848 dyskerin. These data represent experimental replicates of n=3. Statistically significant reductions

849 in enrichment relative to wildtype are indicated by * (P value < 0.01). Error bars represent SEM.

850 d. HEK293 cells without or with stable expression of FLAG-dyskerin (WT or N/NoLS variants)

851 were depleted of endogenous dyskerin using siRNA, and confirmed by immunoblotting against

852 endogenous dyskerin. The same membrane was first probed using an anti-dyskerin antibody,

853 followed by an anti-tubulin antibody as a loading control. Arrows indicate FLAG-tagged

854 dyskerin, the asterisk indicates endogenous dyskerin, and the arrowhead indicated tubulin. Q-

855 TRAP was performed using cell lysate. Statistically significant reductions in relative telomerase

856 activity (RTA) compared to mock (untreated) HEK293 cells are indicated by * (P value <

857 0.0001), and this was repeated in experimental triplicate. Error bars represent SEM.

858 Figure 4: Efficient interaction between dyskerin and GAR1 is mediated by SUMO3. a.

859 Following transient transfection of FLAG-tagged dyskerin in HEK293 cells and co-

860 immunoprecipitation (IP) from cell lysates using FLAG antibody, the interaction between

861 endogenous GAR1 and wildtype (WT)-SUMO3 fusion, K467R-SUMO3 fusion, or WT dyskerin

862 was assessed by immunoblotting. SUMO3 fusion variants run at the expected higher molecular

863 weight than WT dyskerin alone due to the SUMO3 moiety. A non-specific band revealed by the

864 anti-GAR1 antibody at 25kDa is indicated by an asterisk. b. C-terminal SUMO3 fusions of

865 FLAG-dyskerin wildtype and K467R variant were transiently expressed in HEK293 cells, and

866 localization was assessed in fixed cells by indirect immunofluorescence. Representative images

867 of the most prevalent localization phenotype of FLAG-dyskerin (Cy3 in red) and the nucleolar

868 marker fibrillarin (FITC in green) are shown. The nucleus is indicated by DAPI staining of

869 nuclear DNA (in blue). Quantification of localization phenotype scoring as a percentage of

870 FLAG-positive HEK293 cells is indicated (≥50 cells per condition were counted, in experiment

871 replicate n=3), and scale bars indicate 10µm. c. A linear schematic of human GAR1, with

872 glycine and arginine rich domains indicated (RGG-1 and RGG-2). The predicted SIM (70-

873 VVLLG-74) is indicated, and residues expected to physically interface with dyskerin based on

874 previous homologue structural studies in yeast are bolded and italicized. d. Substitution of all

875 five predicted SIM residues to alanine (GAR1 5A) impairs the interaction between GAR1 and

876 dyskerin, as demonstrated by co-IP of 3xFLAG-GAR1 and endogenous dyskerin from HEK293

877 cells transiently expressing 3xFLAG-GAR1. Interaction between endogenous dyskerin and WT

878 or 5A GAR1 was assessed by immunoblotting following FLAG IP from cell lysates.

879 Immunoblotting targets are indicated to the right of each panel as “WB: α target”, and a list of

880 antibodies can be found in the materials and methods section. Each co-IP and immunoblotting

881 was performed in experimental replicate a minimum of n=2, representative blots are shown.

882

883

A K-rich 1 K-rich 2 NLS DKCLD TruB PUA NLS

438 514 VAEVVKAPQVVAEAAKTAKRKRESESESDETPPAAPQLIKKEKKKSKKDKKAKAGLESGAEPGDGDSDTTKKKKKKKKAKEVELVSE K-rich 1 K-rich 2

B - - : FLAG -dyskerin C Merged Cy3 FITC + - + - + + + kDa : 6xHis -SUMO3 (FLAG) (fibrillarin) DAPI

180 140 WT 100

Elution 70 65 *

45 WB: αFLAG K446X 180 140 100

70 bioRxiv preprint doi: https://doi.org/10.1101/2020.09.02.280198* ; this version posted September 3, 2020. The copyright holder for this preprint 65 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Input

45 WB: αFLAG SUMO3 20 -K446X

15 WB: αHis

D Merged Cy3 FITC + (FLAG) (fibrillarin) DAPI

SUMO3- WT

WT- SUMO3

Figure 1: Residues in the C-terminal nuclear/nucleolar localization sequence of dyskerin are SUMO3 targets that govern nuclear accumulation. a. A linear schematic of human dyskerin domains. The amino acid range corresponding to the predicted C-terminal nuclear/nucleolar localization sequence (N/NoLS) (438-514) is denoted below the schematic, indicating the MS-identified SUMO3 sites in this region (K467, K468, K498, and K507) with solid black circles, and the two lysine (K)-rich clusters (K467-K480, and K498-K507) are underlined. MS-identified SUMO3 sites reported in proteome-wide studies cited in the text are indicated by solid black circles above the schematic. Residues A481 and K446 are indicated by arrowheads. b. FLAG-dyskerin (wildtype WT and dyskerin truncation variant K446X without or with N-terminal SUMO3 fusion) and 6xHis-SUMO3 were expressed in HEK293 cells. His- SUMO3 conjugates were purified using Ni-NTA agarose beads following lysis under denaturing conditions, and SUMOylated FLAG-dyskerin was assessed in the elution by immunoblotting with an anti-FLAG antibody. A fraction of each HEK293 cell pellet used for purification was kept prior to lysis (Input) for checking expression of FLAG-dyskerin and His-SUMO3 by immunoblotting. The K446X truncation runs at the expected lower molecular weight than WT dyskerin, while SUMO3-K446X runs at the expected higher molecular weight than WT due to the SUMO3 fusion. FLAG-dyskerin is indicated by arrows, while asterisks indicate non-specific antibody signal. The higher molecular weight smear observed in the SUMO3-K446X elution (indicated by the bracket to the right of the blot) indicates SUMOylated species of dyskerin. c. Representative images of the co-localization of FLAG-dyskerin (wildtype WT and dyskerin truncation variant K446X without or with N-terminal SUMO3 fusion – Cy3 shown in red) with nucleolar marker fibrillarin (FITC shown in green), as observed by indirect immunofluorescence. The nucleus is indicated by DAPI staining of nuclear DNA (in blue). d. Localization of FLAG- tagged WT, N-terminal SUMO3 fusion WT dyskerin (SUMO3-WT), and C-terminal SUMO3 fusion WT dyskerin (WT-SUMO3) was assessed by indirect immunofluorescence (Cy3 shown in red). Fibrillarin was used as a nucleolar marker (FITC shown in green), and the nucleus is indicated by DAPI staining of nuclear DNA (in blue). In c. and d. examples of nucleoplasmic FLAG-dyskerin foci that co-localize with nucleoplasmic fibrillarin foci are indicated by arrowheads, while arrows indicate nucleoplasmic FLAG-dyskerin foci that do not co-localize with nucleoplasmic fibrillarin foci. All scale bars indicate 10µm. A Merged B Cy3 FITC + Dyskerin Localization (FLAG) (fibrillarin) DAPI 140

WT 120 100 80 60 Cells (%) A481X 40 20 0 Percentage FLAG-positiveof T R R X X X 1 K467R W 46 467 468 48 446 4 K K A K K 67/468R - 4 O3 K M U S

K468R FLAG-dyskerin

Everywhere Nucleolar Cytoplasmic & Nucleoplasmic K467/ Nucleoplasmic & Nucleolar 468R Cytoplasmic Nucleoplasmic Cytoplasmic & Nucleolar

Cytoplasmic Cytoplasmic Nucleoplasmic C and and and bioRxiv preprintEverywhere doi: https://doi.org/10.1101/2020.09.02.280198Cytoplasmic ; this version posted SeptemberNucleolar 3, 2020. The copyright holder forNucleoplasmic this preprint (which was not certifiedNucleoplasmic by peer review) is the author/funder.Nucleolar All rights reserved. No reuse allowed Nucleolarwithout permission.

Merged + DAPI

FITC (fibrillarin)

Cy3 (FLAG)

D FITC E Cy3 DAPI (eGFP) Merged DAPI (dyskerin) Merge

mock 2˚ only

eGFP Endogenous dyskerin

eGFP-WT dyskerin

Figure 2: Nuclear and subnuclear localization of dyskerin is mediated by SUMO3 sites in the C-terminal nuclear/nucleolar localization sequence. a. FLAG-dyskerin was transiently expressed in HEK293 cells, and localization was assessed in fixed cells by indirect immunofluorescence. Representative images of the most prevalent localization phenotype of FLAG-dyskerin (wildtype WT, dyskerin truncation variant A481X, and substitution variants K467R, K468R, and double K467/468R – Cy3 in red) and the nucleolar marker fibrillarin (FITC in green) are shown. Nucleolar localization is represented for WT and K468R, concomitant nucleoplasmic & nucleolar is represented by A481X, and nucleoplasmic localization is represented by K467R and K467/468R. The nucleus is indicated by DAPI staining of nuclear DNA (in blue) b. Quantification of localization phenotype scoring for FLAG-dyskerin WT and localization variants as a percentage of FLAG-positive HEK293 cells is indicated (≥50 cells per condition were counted, in experiment replicate n=3). c. Examples of how localization phenotype was scored are shown. Nucleolar localization was determined by co-localization with clustered fibrillarin signal, nucleoplasmic localization was determined by co-localization with DAPI outside of the clustered fibrillarin signal, and cytoplasmic localization was determined by concentrated signal outside of and surrounding DAPI. d. Localization of eGFP and eGFP-tagged WT dyskerin (in green) was assessed in fixed HEK293 cells following transient transfection, using a FITC filter. e. Localization of endogenous dyskerin (Cy3, in red) was assessed by IF in fixed HEK293 cells. As a negative control, fixed HEK293 cells were assessed by IF using only secondary Cy3-conjugated antibody. The nucleus is indicated by DAPI staining of nuclear DNA (in blue). These are representative images.All scale bars indicate 10µm. A B kDa - : FLAG-dyskerin kDa - : FLAG-dyskerin 100 70 WB: α FLAG 75 WB: α NAF1 60 FLAG

α 35 75 WB: α GAR1 WB: α FLAG 25 *

FLAG 60 70 α WB: α FLAG IP: IP: 60 15 WB: α NHP2

Input 35 IP: WB: α GAR1 WB: α NOP10 25 * 10 100 C 75 WB: α NAF1 3 Dyskerin-hTR Interaction 75

WB: α FLAG 60 2 Input

15 WB: α NHP2 1 RNA Enrichment bioRxiv preprint doi: https://doi.org/10.1101/2020.09.02.280198; this version posted September 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder.WB: α NOP10 All rights reserved. No* reuse allowed without* permission. 10 * Relative Relative to Wildtype Dyskerin 0 WT mock K446X K467R D K468R A481X 200 Q-TRAP sidkc1

150 kDa --- K446X K467R K468R K467/ K468R A481X WT :FLAG-dyskerin 70 100 60 * WB: α dyskerin RTA (%) RTA 50 60 * * * * WB: α tubulin 0 45 1 ck A c1 o N kc1 kc1 kc kc1 k m idkc1 dkc1 d siR s si sid sid sid sid le + + + + si + R R X + X mb 8R a WT / cr s K467 K468 A481 K446 K467 Figure 3: Dyskerin nuclear and nucleolar localization is linked to mature H/ACA complex assembly and function. Interactions of FLAG-dyskerin WT and localization variants with endogenous pre- and mature H/ACA ribonucleoprotein complex components were assessed by co-immunoprecipitation (IP) from HEK293 cell lysates. Assembly of the a. H/ACA pre-RNP complex involving NAF1, NHP2, and NOP10 was investigated by immunoblotting for the endogenous H/ACA pre-RNP components and FLAG-dyskerin proteins following IP. b. Interaction of dyskerin with the mature H/ACA complex component GAR1 was examined following IP by immunoblotting for endogenous GAR1 and FLAG-dyskerin. Localization variants that are excluded from the nucleolus (K467R and K467/468R) do not interact with GAR1. Immunoblotting targets are indicated to the right of each panel as “WB: α target”, and a list of antibodies can be found in the materials and methods section. A non-specific band revealed by the anti-GAR1 antibody at 25kDa is indicated by an asterisk. Each co-IP and immunoblotting was performed in experimental replicate a minimum of n=2, representative blots are shown. c. Dyskerin-hTR interactions were assessed by IP of FLAG-tagged dyskerin followed by RNA extraction and qPCR. Relative to wildtype IP fractions, dyskerin variants with substantial localization defects (K467R and K446X) display significantly reduced enrichment of hTR following IP. HEK293 cells lacking FLAG-tagged dyskerin (indicated as mock) were used as a negative control for RNA binding to the FLAG antibody and/or Protein G Sepharose. Mock cells were subject to the same IP protocol detailed for fractions containing FLAG-tagged dyskerin. These data represent experimental replicates of n=3. Statistically significant reductions in enrichment relative to wildtype are indicated by * (P value < 0.01). Error bars represent SEM. d. HEK293 cells without or with stable expression of FLAG-dyskerin (WT or N/NoLS variants) were depleted of endogenous dyskerin using siRNA, and confirmed by immunoblotting against endogenous dyskerin. The same membrane was first probed using an anti-dyskerin antibody, followed by an anti-tubulin antibody as a loading control. Arrows indicate FLAG-tagged dyskerin, the asterisk indicates endogenous dyskerin, and the arrowhead indicated tubulin. Q- TRAP was performed using cell lysate. Statistically significant reductions in relative telomerase activity (RTA) compared to mock (untreated) HEK293 cells are indicated by * (P value < 0.0001), and this was repeated in experimental triplicate. Error bars represent SEM. A C

RGG-1 RGG-2 kDa - : FLAG-dyskerin

180 70 127 VVLLGEFLHPCEDDIVCKCTTDENKVPYFNAPVYLENKEQIGKVDEIFGQLRDFYFSV 140 SIM

100 WB: α FLAG FLAG α 70 D IP: 65 kDa - : 3xFLAG-GAR1 35 35 WB: α GAR1 WB: α FLAG FLAG

25 * α 25 180 IP: 75 140 60 WB: α dyskerin

35 100 WB: α FLAG WB: α FLAG 25 Input Input Input 70 75 65 60 WB: α dyskerin bioRxiv 35preprint doi: https://doi.org/10.1101/2020.09.02.280198; this version posted September 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. WB: α GAR1 25 *

Merged B Cy3 FITC Dyskerin Localization + (FLAG) (fibrillarin) DAPI 140 120 WT- 100 SUMO3 80 60 Cells (%) 40 20 K467R- SUMO3 0 Percentage FLAG-positiveof WT-SUMO3 K467R-SUMO3 FLAG-dyskerin Everywhere Cytoplasmic & Nucleoplasmic Nucleolar Nucleoplasmic & Nucleolar Nucleoplasmic

Figure 4: Efficient interaction between dyskerin and GAR1 is mediated by SUMO3. a. Following transient transfection of FLAG-tagged dyskerin in HEK293 cells and co- immunoprecipitation (IP) from cell lysates using FLAG antibody, the interaction between endogenous GAR1 and wildtype (WT)-SUMO3 fusion, K467R-SUMO3 fusion, or WT dyskerin was assessed by immunoblotting. SUMO3 fusion variants run at the expected higher molecular weight than WT dyskerin alone due to the SUMO3 moiety. A non-specific band revealed by the anti-GAR1 antibody at 25kDa is indicated by an asterisk. b. C-terminal SUMO3 fusions of FLAG-dyskerin wildtype and K467R variant were transiently expressed in HEK293 cells, and localization was assessed in fixed cells by indirect immunofluorescence. Representative images of the most prevalent localization phenotype of FLAG-dyskerin (Cy3 in red) and the nucleolar marker fibrillarin (FITC in green) are shown. The nucleus is indicated by DAPI staining of nuclear DNA (in blue). Quantification of localization phenotype scoring as a percentage of FLAG-positive HEK293 cells is indicated (≥50 cells per condition were counted, in experiment replicate n=3), and scale bars indicate 10µm. c. A linear schematic of human GAR1, with glycine and arginine rich domains indicated (RGG-1 and RGG-2). The predicted SIM (70- VVLLG-74) is indicated, and residues expected to physically interface with dyskerin based on previous homologue structural studies in yeast are bolded and italicized. d. Substitution of all five predicted SIM residues to alanine (GAR1 5A) impairs the interaction between GAR1 and dyskerin, as demonstrated by co-IP of 3xFLAG-GAR1 and endogenous dyskerin from HEK293 cells transiently expressing 3xFLAG-GAR1. Interaction between endogenous dyskerin and WT or 5A GAR1 was assessed by immunoblotting following FLAG IP from cell lysates. Immunoblotting targets are indicated to the right of each panel as “WB: α target”, and a list of antibodies can be found in the materials and methods section. Each co-IP and immunoblotting was performed in experimental replicate a minimum of n=2, representative blots are shown.