bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

1

2

3

4

5 Comprehensive Substrate Specificity Profiling of the

6 Human Nek Kinome Reveals Unexpected Signaling Outputs

7

8

9 Bert van de Kooij1, Pau Creixell1, Anne van Vlimmeren1, Brian A. Joughin1,

10 Chad J. Miller2,*, Nasir Haider3, Rune Linding4, Vuk Stambolic3,

11 Benjamin E. Turk2, Michael B. Yaffe1,5,6

12

13 14 15 16 17 18 19 20 21 22 23 24 1. Koch Institute for Integrative Cancer Research, MIT Center for Precision Cancer Medicine, 25 Departments of Biology and Bioengineering, Massachusetts Institute of Technology, 26 Cambridge, MA, USA. 27 2. Department of Pharmacology, Yale School of Medicine, New Haven, CT, USA 28 3. Department of Medical Biophysics, University of Toronto, Toronto, Canada 29 4. Biotech Research and Innovation Center, Faculty of Health and Medical Sciences, University 30 of Copenhagen, Copenhagen, Denmark 31 5. Department of Surgery, Beth Israel Deaconess Medical Center, Divisions of Acute Care 32 Surgery, Trauma, and Critical Care and Surgical Oncology, Harvard Medical School, Boston, 33 USA. 34 6. To whom correspondence should be addressed. E-mail [email protected] 35 * Current address: Department of Biochemistry, University of Washington, Seattle, WA, USA bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

36 Abstract

37 Human NimA-related kinases (Neks) have multiple mitotic and non-mitotic functions, but few

38 substrates are known. We systematically determined the phosphorylation-site motifs for the entire

39 Nek kinase family, except for Nek11. While all Nek kinases strongly select for hydrophobic

40 residues in the -3 position, the family separates into four distinct groups based on specificity for a

41 serine versus threonine phospho-acceptor, and preference for basic or acidic residues in other

42 positions. Unlike Nek1-Nek9, Nek10 is a dual-specificity kinase that efficiently phosphorylates

43 itself and peptide substrates on serine and tyrosine, and its activity is enhanced by tyrosine auto-

44 phosphorylation. Nek10 dual-specificity depends on residues in the HRD+2 and APE-4 positions

45 that are uncommon in either serine/threonine or tyrosine kinases. Finally, we show that the

46 phosphorylation-site motifs for the mitotic kinases Nek6, Nek7 and Nek9 are essentially identical

47 to that of their upstream activator Plk1, suggesting that Nek6/7/9 function as phospho-motif

48 amplifiers of Plk1 signaling.

2 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

49 Introduction

50 Protein phosphorylation by kinases plays an essential role in nearly all signaling events

51 and physiological processes within the cell. Protein serine/threonine kinases, together with

52 ubiquitin ligases, play a particularly important role in mitosis, during which thousands of sites are

53 phosphorylated (Dephoure et al., 2008). These mitotic phosphorylation events are predominantly

54 mediated by the kinases Cdk1, Plk1, Aurora A and Aurora B, together with several members of

55 the nimA-related kinase (Nek) family (Fry et al., 2017; Joukov et al., 2018). For Cdk1, Plk1 and

56 the Aurora kinases, multiple substrates and detailed downstream signaling functions have been

57 identified (Joukov et al., 2018), though much remains to be learned. In contrast, for the Nek family

58 kinases only a few substrates have been identified, and many of their functions within the complex

59 network of mitotic signaling are not well understood.

60 The Nek kinase family consists of eleven serine/threonine kinases that together form an

61 independent evolutionary branch of the human kinome (Fig. 1A). The first human mitotic Nek

62 kinase to be identified was Nek2, which is required for centrosomal disjunction following their

63 duplication during S-phase, in order to assemble a bipolar mitotic spindle (Fig. 1B; Fry et al., 2017)

64 A similar function was assigned to Nek5, which, in addition, also regulates centrosome

65 composition in interphase (Prosser et al., 2015). Furthermore, Nek6, Nek7 and Nek9 have been

66 shown to cooperate as a signaling module during various stages of mitosis (Belham et al., 2003).

67 In this process, Nek9 is first phosphorylated in early mitosis by Cdk1, creating a binding site for

68 the Plk1 polo-box domain. Plk1-mediated phosphorylation and activation of Nek9 then leads to

69 Nek-9-dependent phosphorylation and interaction-dependent activation of Nek6 and Nek7,

70 allowing them to perform their downstream functions (Roig et al., 2002; Belham et al., 2003; Yin

71 et al., 2003; Richards et al., 2009; Bertran et al., 2011). As a signaling module, Nek6/7/9 have

72 been reported to play a role in centrosome separation and maturation, nuclear envelope

73 breakdown, metaphase and anaphase progression, mitotic spindle formation and cytokinesis

3 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

74 (Fig. 1B; Roig et al., 2002; Yin et al., 2003; Yissachar et al., 2006; O’Regan et al., 2009; Salem

75 et al., 2010; Bertran et al., 2011; Kim et al., 2011; Laurell et al., 2011; Sdelci et al., 2012; Cullati

76 et al., 2017). Notably, the Nek kinase family is not exclusively involved in mitosis, but functionally

77 diverse. Nek kinases have also been reported to play a role in meiosis, ciliary biology, and the

78 response of cells to replication stress and DNA-damage (Fig. 1B; Quarmby et al., 2005; Melixetian

79 et al., 2009; Larissa S Moniz et al., 2011; Choi et al., 2013; Spies et al., 2016; Brieño-Enríquez et

80 al., 2017). The functions of Nek3 and Nek4 are not clear but the current literature suggests that

81 Nek3 may play a role in cell migration (Harrington et al., 2016), and Nek4 might be involved in

82 regulating microtubule stability (Fig. 1B; Doles et al., 2010).

83 We have previously determined the optimal substrate phosphorylation motifs, i.e. the

84 substrate amino acid sequence preferentially phosphorylated by a given kinase, for the mitotic

85 kinases Cdk1, Plk1, Aurora A, Aurora B and Nek2 (Alexander et al., 2011). This motif-information

86 strongly facilitates identification of kinase substrates, and can help to understand the organization

87 of complex phosphorylation networks like those in mitotic cells (Linding et al., 2007; Alexander et

88 al., 2011; Kang et al., 2013). Therefore, to further catalog the optimal substrate motif-atlas for

89 mitotic kinases, and to characterize substrate-specificity of a complete and functionally diverse

90 branch of the human kinome, we sought to determine the optimal phosphorylation-site motif for

91 the Nek-kinases using Oriented Peptide Library Screening (OPLS) (Hutti et al., 2004), followed

92 by direct experimental validation together with molecular modeling. This revealed a surprisingly

93 large amount of diversity in substrate-specificity among the Nek-kinases, and identified Nek10 as

94 a dual-specificity kinase, that can phosphorylate substrates on tyrosine in addition to serine.

95 Furthermore, these studies demonstrated that the mitotic kinases Nek6, Nek7 and Nek9, have a

96 phosphorylation-site motif that is almost identical to the motif of Plk1, suggesting that the Nek6/7/9

97 module functions as a Plk1-motif amplifier.

98

4 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

99 Results

100 Determination of the phosphorylation-site motif for all members of the Nek kinase family

101 To obtain the consensus phosphorylation-site motif for each individual Nek kinase, we

102 performed Oriented Peptide Library Screening (OPLS) as described previously (Hutti et al., 2004;

103 Alexander et al., 2011). In these experiments, each kinase is tested in 198 individual in vitro kinase

104 reactions, each with a different peptide library as substrate (Fig. S1). Each peptide library is

105 composed of a pool of 10-mer peptides containing an equimolar mix of serine and threonine at

106 the phospho-acceptor site, and one other fixed amino acid on a position ranging from five residues

107 upstream (-5 position) to four residues downstream (+4 position) of the phospho-acceptor residue.

108 The fixed amino acid is one of the 20 naturally occurring amino acids, or either phospho-threonine

109 (pT) or phospho-tyrosine (pY). The remaining eight unfixed positions in the library contain an

110 equimolar mix of all natural amino acids except cysteine, serine or threonine. The preference of

111 the protein kinase for each residue at each position is then determined by directly comparing its

112 activity against each individual peptide library. Lastly, to study phospho-acceptor site specificity

113 of the kinase, we included three completely degenerate peptide libraries that contain either a fixed

114 serine, threonine or tyrosine phospho-acceptor site, but no cysteine, serine, threonine or tyrosine

115 in the degenerate positions.

116 Each of the Nek kinases were expressed as 3xFLAG-tagged variants in HEK 293T cells,

117 isolated by anti-FLAG immunoprecipitation (IP), and assayed by OPLS. For Nek1, Nek3, Nek6,

118 Nek7, Nek8, and Nek9 the full-length protein was used. For Nek4 and Nek5, the kinase activity

119 of the full-length molecule was low, and OPLS experiments were therefore performed using the

120 isolated kinase domains, which displayed more activity. For Nek11, neither the full-length

121 construct nor the isolated kinase domain displayed sufficient activity for motif determination, but

122 clear motif data was obtained for all of the other Nek-kinases (Figs. 1C, 4A-B). To ensure that the

123 phosphorylating activity assayed by OPLS directly corresponded to that of the FLAG-tagged Nek

5 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

124 kinase of interest, inactivating point mutants of each Nek kinase were generated, expressed in

125 HEK 293T cells, purified and assayed as above by OPLS. As shown in Fig S2A, none of the

126 kinase-dead (KD) point mutants showed any detectable phosphorylation of the peptide libraries,

127 demonstrating that the signal obtained with the wild-type NEK kinases was generated by the

128 purified Neks and not by a contaminating kinase (Fig. S2A). Each OPLS-experiment was

129 performed at least twice, and the data from all replicates were pooled to obtain phosphorylation-

130 site motifs, which are presented as logos showing the positively selected residues in Fig. 1C, and

131 as complete logos including the negatively selected residues in Fig. S3.

132 These OPLS experiments revealed that all of the tested Nek kinases display a clear

133 preference for leucine, methionine, phenylalanine or tryptophan at the -3 position (Fig 1C). This

134 is consistent with what has been reported for Nek6, and similar to what we found previously for

135 Nek2 (Lizcano et al., 2002; Alexander et al., 2011). In addition, all of the Nek kinases strongly

136 selected against a proline in the +1 position (Fig. 1C, S3). However, the OPLS-experiments also

137 demonstrated clear specificity differences between the different family-members. Most notably,

138 our results revealed a dichotomy in phospho-acceptor site preference within the family, with

139 Nek1/3/4/5/8 preferentially phosphorylating threonine residues, while Nek6/7/9 primarily targeted

140 peptides with serine residues as the phospho-acceptor (Fig. 1C). Since the phosphorylation-site

141 motif determined previously for Nek2 did not contain information on acceptor site specificity, an

142 in vitro kinase assay of isolated Nek2 with degenerate peptide libraries containing either a serine

143 or threonine phospho-acceptor site was performed. Nek2 phosphorylated the serine phospho-

144 acceptor library more efficiently than the threonine library, indicating that Nek2 shares a

145 preference for a phospho-acceptor serine residue with Nek6/7/9 (Fig. S2B). Together, these data

146 reveal a Nek kinase family core consensus motif of [LMFW]-X-X-S/T-[no P], but with additional

147 specificity-differences. In general, Nek1/3/4/5/8 displayed a preference for hydrophobic and basic

148 residues on multiple positions within the motif, while Nek6/7/9 displayed a preference for acidic

6 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

149 residues and selected hydrophobic residues at various positions. The separation between these

150 groups of Nek kinases based on substrate-specificity is discussed in more detail below.

151

152 Validation of the phosphorylation-site motif for all members of the Nek family

153 To validate our OPLS-results, we designed peptides predicted by our OPLS-results to be

154 good substrates for either the Nek1/3/4/5/8 (13458-tide) group of kinases, or the Nek6/7/9 (679-

155 tide) group of kinases. In addition, we synthesized variants of these peptides that contained

156 residues predicted to be disfavored at specified positions (Fig. 2A). We tested the phosphorylation

157 of these peptides using in vitro kinase assays with Nek3 and Nek7 as representative members of

158 these Nek kinase groups (Fig. 2B-D). As shown in figure 2B and C, Nek3 could readily

159 phosphorylate the optimal Nek13458-tide, but was substantially impaired in its ability to

160 phosphorylate peptides in which either the -3 or -2 hydrophobic residues were mutated to aspartic

161 acid (Fig. 2B, 2C). As shown in figure 2D, Nek7 phosphorylated the optimal 679-tide much more

162 rapidly than a 679-tide variant containing a disfavored hydrophobic residue (I) in place of the

163 favored hydrophobic residue (L) in the -3 position (Fig. 2D). Similarly, substitution of the favored

164 aspartic acid residue in the -2 position for a basic residue (R), substantially reduced peptide

165 phosphorylation by Nek7 (Fig. 2D).

166 Interestingly a subset of Nek kinases, including Nek7, Nek8 and particularly Nek6, displayed

167 unexpected selection for phospho-tyrosine in the -1 position, suggesting that these kinases might

168 target substrates that have been primed by tyrosine kinase mediated phosphorylation. To validate

169 this specificity, we substituted the phospho-tyrosine in the -1 position of 679-tide to an OPLS-

170 disfavored isoleucine residue, which resulted in a reduced rate at which Nek6 phosphorylated

171 679-tide (Fig. 2E). Finally, all of the Nek kinases displayed a strong preference for a tryptophan

172 in the -4 position. Selection for tryptophan within kinase substrate motifs, especially at the -4

173 position, has rarely been observed (Miller et al., 2018). Corroborating our OPLS-data, the rate at

174 which Nek8 phosphorylated its optimal substrate Nek8-tide was reduced upon substitution of the

7 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

175 tryptophan in the -4 position to a valine (Fig. 2F). Furthermore, introduction of a tryptophan in the

176 -4 position in a peptide that is otherwise a very poor substrate for Nek8 (Nek8-poor) significantly

177 enhanced the rate at which Nek8 could phosphorylate Nek8-poor (Fig. 2F). Taken together, these

178 results with optimal and suboptimal peptide variants validate our OPLS-data, and demonstrate

179 that we can use the OPLS-data to predict good and poor substrates for the Nek kinases.

180

181 Nek10 is a dual-specificity kinase

182 Curiously, in these OPLS experiments, Nek10 displayed very little sequence preference for

183 any residue, except for a tyrosine at every position (Fig. 3A). We hypothesized that this result

184 could indicate that Nek10 is a tyrosine kinase. This could explain the effective phosphorylation of

185 all peptide libraries containing a fixed tyrosine, because the tyrosine itself could serve as a

186 phospho-acceptor site. In addition, it would explain the absence of additional preferences,

187 because each of the 198 peptide libraries will contain tyrosine residues on non-fixed positions

188 and therefore have many different target sites (Fig. S1). In agreement with this hypothesis, mass

189 spectrometric analysis of in vitro auto-phosphorylated Nek10 revealed the presence of several

190 phosphorylated tyrosine residues (Nasir Haider and Vuk Stambolic, unpublished data). To directly

191 determine if Nek10 phosphorylates tyrosine residues, we designed a Nek10 substrate peptide

192 based on the auto-phosphorylation site data, and introduced either a tyrosine, serine, or threonine

193 at the phospho-acceptor site (Fig. 3B, C). Surprisingly, Nek10 effectively phosphorylated peptides

194 with a serine phospho-acceptor site, but even more rapidly phosphorylated peptides with a

195 tyrosine phospho-acceptor site (Fig. 3B, C). In contrast, substitution of threonine as the phospho-

196 acceptor significantly impaired peptide phosphorylation by Nek10 (Fig. 3B, C). We therefore

197 conclude that Nek10 is a tyrosine/serine dual-specificity kinase.

198 None of the other Nek kinases phosphorylated peptides libraries containing tyrosine as the

199 phospho-acceptor (0 position in Fig. 1C, also Fig. S4A). Therefore, we reasoned that sequence

200 comparison between the Nek kinases might reveal residues that permit Nek10 to phosphorylate

8 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

201 both serine and tyrosine residues, since structural analysis has implicated specific regions of the

202 kinase domain that contribute to serine/threonine or tyrosine specificity (Taylor et al., 1995). One

203 of these regions is the P+1 loop, which lies directly N-terminal of the conserved APE-motif (Fig.

204 3D). We therefore aligned the P+1 loop sequences of the Nek kinase family members, and noted

205 that Nek10 has an isoleucine whereas all the other Nek kinases have a proline at the position four

206 residues N-terminal of the APE-motif (APE-4; Fig. 3E). Interestingly, an isoleucine in the APE-4

207 is typically found in tyrosine kinases, but in less than 5% of serine/threonine kinase, which more

208 commonly have a proline on this position (Fig. 3F). We therefore mutated the APE-4 isoleucine

209 in Nek10 to a proline (I693P), and found that this increased its capacity to phosphorylate a

210 substrate peptide on a serine residue but rendered Nek10 incapable of phosphorylating the same

211 peptide on a tyrosine (Fig. 3G).

212 Another region contributing to phospho-acceptor site specificity is the catalytic loop, which

213 contains the aspartic acid which serves as the catalytic base within the conserved HRD motif (Fig.

214 3D; Taylor et al., 1995). Intriguingly, Nek10 has a threonine in the HRD+2 position, which is

215 uncommon within the serine/threonine kinome. Instead, 86% of serine/threonine kinases contain

216 a lysine residue in the HRD+2 position, including all the other Nek kinases (Fig. 3H, S4B).

217 Mutation of the HRD+2 threonine to a lysine (T657K) generated a Nek10 variant that, like Nek10

218 I693P, could effectively phosphorylate serine residues but was incapable of phosphorylating

219 tyrosine residues (Fig. 3I). Finally, we sought to make a variant version of Nek10 that would

220 exclusively phosphorylate substrates on tyrosine residues. To this end, we mutated the threonine

221 in the APE-5 position to a proline, which is highly conserved in tyrosine kinases, but present in

222 only three serine/threonine kinases (Fig. 3F). In contrast to the I693P and T657K mutations, this

223 APE-5 mutation (T692P) severely reduced overall Nek10 kinase activity (Fig. 3J). However, the

224 resulting Nek10 T692P variant could still detectably phosphorylate tyrosine residues while it was

225 unable to phosphorylate serine residues (Fig. 3K). Taken together, these results identify specific

226 residues in the kinase domain of Nek10 that are essential for its dual-specificity, and show that

9 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

227 point-mutations in the catalytic or P+1 loop can toggle Nek10 between being a dual-specificity,

228 serine-specific, or tyrosine-specific kinase.

229

230 Nek10 phosphorylates serines and tyrosines in different sequence contexts

231 The positioning of the substrate in the catalytic site of the kinase is different for

232 serine/threonine kinases compared to tyrosine kinases (Taylor et al., 1995), raising the question

233 of how a dual-specificity kinase is able to correctly position substrates to catalyze phosphorylation

234 on serine as well as tyrosine residues. We reasoned that the phosphorylation-site sequence

235 context might play a major role, and therefore set out to determine whether Nek10 recognizes

236 distinct phosphorylation-site motifs for serine and tyrosine substrates. In these experiments we

237 used a serine-specific variant of Nek10 (I693P) to eliminate issues arising from tyrosine-

238 phosphorylation of peptide libraries by WT Nek10 that made the previous OPLS-results

239 uninterpretable (Fig. 3A). As shown in figure 4A, for Nek10 I693P the strong preference for

240 tyrosine in every position was absent, but instead it displayed a clear preference for hydrophobic

241 and basic residues, similar to the Nek1/3/4/5/8 group (Fig. 4A, 1C). However, in contrast to other

242 members of this group that preferentially select for threonine as the phospho-acceptor, Nek10

243 I693P preferentially phosphorylates serine (Fig. 4A). To eliminate the possibility that the I693P

244 mutation affected the intrinsic serine-motif specificity of Nek10, we validated these OPLS-results

245 with WT Nek10, rather than with the Nek10 I693P mutant. An optimal serine peptide substrate

246 ((S)-tide) was synthesized on basis of the OPLS-results, along with variants in which the favored

247 hydrophobic residue in the -3 position and the favored basic residue in the -1 positions were

248 replaced by a disfavored aspartic acid residue (-3D, -1D), and these peptides were used as

249 substrates in an in vitro kinase assays. As shown in figure 4B, WT Nek10 could effectively

250 phosphorylate (S)-tide but not the -3D and -1D peptides, therefore corroborating our OPLS-results

251 (Fig. 4B).

10 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

252 To investigate the phosphorylation-site motif for Nek10 on tyrosine substrates, we designed

253 a peptide library identical to the serine/threonine peptide library, but with a tyrosine instead of a

254 serine/threonine mixture in the phospho-acceptor site. Importantly, to prevent phosphorylation of

255 the degenerate positions, tyrosine, together with serine, threonine and cysteine residues, was

256 excluded from the amino acid mixture in the degenerate positions. These OPLS experiments

257 revealed a very different Nek10 phosphorylation-site motif for tyrosine substrates compared to

258 serine substrates. On tyrosine substrates, Nek10 does not select for the canonical Nek-family

259 motif signature [LMFW]-X-X-S/T-[no P], but instead shows a near-absolute requirement for

260 aromatic or selected hydrophobic residues in the +1 position (Fig. 4C). Thus, the serine

261 phosphorylation-site motif of the dual-specificity kinase Nek10 is similar to that of the other Nek

262 kinases, but the tyrosine phosphorylation-site motif of Nek10 is substantially more restricted.

263

264 Nek10 auto-phosphorylates on serine and tyrosine residues in trans, both in vitro and in

265 cells

266 Next, we investigated whether Nek10 can phosphorylate protein substrates, in addition to

267 peptide substrates, on both serine and tyrosine residues. Unfortunately, no Nek10 substrates

268 have been reported to date. However, preliminary mass spectrometry data indicated that Nek10

269 is a substrate for auto-phosphorylation (Nasir Haider and Vuk Stambolic, unpublished data).

270 Indeed, incubation of Nek10 with radiolabeled ATP revealed efficient auto-phosphorylation of WT

271 and serine-specific Nek10 (I693P(Ser)), but not of kinase-dead (KD) Nek10 in which the catalytic

272 aspartic acid was mutated to an asparagine residue (Fig. 5A). Furthermore, despite its reduced

273 activity, we could also clearly detect auto-phosphorylation of the tyrosine-specific (T692P(Tyr))

274 Nek10 variant, demonstrating that Nek10 can undergo auto-phosphorylation on both serine and

275 tyrosine residues in vitro (Fig. 5A).

276 Of the 518 protein kinases only 18 are recognized as dual-specificity. Members of the GSK3

277 and DYRK serine/threonine kinase families can also auto-phosphorylate on tyrosine residues, but

11 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

278 do this exclusively (GSK3) or primarily (DYRK) during folding in cis (Cole et al., 2004;

279 Soundararajan et al., 2013). To test whether Nek10 can auto-phosphorylate on tyrosine and

280 serine in trans, we performed in vitro kinase assays using FLAG-tagged WT, serine-specific or

281 tyrosine-specific Nek10 to phosphorylate an HA-tagged KD Nek10 as substrate. Following

282 incubation, Nek10 kinase and substrate were separated by anti-FLAG and anti-HA IP

283 respectively, and analyzed by immunoblotting and phosphorimaging. As shown in figure 5B,

284 phosphorylation of KD Nek10 was observed following incubation with either a serine-specific or

285 tyrosine-specific kinase-proficient Nek10, demonstrating that Nek10 can auto-phosphorylate in

286 trans on both serine and tyrosine residues (Fig. 5B).

287 Next, to investigate if serine and tyrosine auto-phosphorylation of Nek10 could also occur in

288 cells, we generated Nek10 knock-out U-2 OS cell lines using CRISPR technology (Fig. S5A).

289 These were then reconstituted with doxycycline-inducible, 3xFLAG-tagged WT, KD, serine-

290 specific (I693P(Ser)) and tyrosine-specific (T692P(Tyr)) variants of Nek10 (Fig. 5C). Tyrosine

291 phosphorylation was assessed by anti-FLAG IP followed by immunoblotting with an antibody

292 recognizing phospho-tyrosine. Clear tyrosine phosphorylation was observed of WT Nek10, but

293 not of a KD or serine-specific variant, consistent with Nek10 auto-phosphorylation on tyrosine

294 residues in cells (Fig. 5D). In contrast to what we found in vitro however, we could not detect any

295 tyrosine phosphorylation on tyrosine-specific Nek10 (Fig. 5D), likely due to a combination of the

296 very low activity of this Nek10 variant, and the higher threshold required for antibody detection

297 compared to 32[P]-radiolabeling.

298 To map the tyrosine(s) targeted for auto-phosphorylation by Nek10, we used the OPLS

299 phosphorylation-site motif to select seven candidate Nek10 target sites, and mutated each of

300 these sites to a phenylalanine. WT and mutant kinases containing 3xFLAG-tags were expressed

301 in U-2 OS cells, followed by anti-FLAG IP and immunoblotting for phospho-tyrosine. As shown in

302 figure 5E, tyrosine-phosphorylation was detected in each of the mutants, but the signal was

303 substantially reduced upon mutation of Y644, suggesting that Y644 is a major auto-

12 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

304 phosphorylation target site of Nek10 (Fig. 5E). An in vitro kinase assay using tyrosine-specific

305 Nek10 and a KD Nek10 substrate with or without the Y644F mutation, confirmed that Nek10

306 directly auto-phosphorylates Y644 in trans (Fig. 5F). Interestingly, the Y644F mutation

307 dramatically reduced the activity of Nek10 in an in vitro kinase assay with a serine peptide

308 substrate (Fig. 5G), and a similar result was obtained with a tyrosine peptide substrate (Fig. S5B).

309 These results strongly indicate that Nek10 auto-phosphorylation on Y644 contributes to its

310 activity.

311 Finally, to study serine auto-phosphorylation in cells, we isolated WT, KD, serine-specific and

312 tyrosine-specific Nek10 from the doxycycline-induced Nek10 knock-out cells (Fig. 5C), and

313 analyzed its phosphorylation status by mass spectrometry. Multiple phosphopeptides were

314 detected, but the region corresponding to N346-R365 of Nek10, which is outside of the kinase

315 domain, was the most abundant phosphopeptide in the WT and serine-specific samples, and was

316 completely absent in the KD and tyrosine-specific samples (Fig. S5C, S5D). This peptide contains

317 six serine residues, of which four were identified as phosphorylated by MASCOT protein

318 identification software (Fig. S5C). Mutation of these four serine sites in KD Nek10 reduced its

319 phosphorylation by a serine-specific Nek10 variant, demonstrating that one or more of these sites

320 can be auto-phosphorylated in trans by Nek10 (Fig. 5H). To determine which of these four sites

321 was the most likely target site, we manually annotated the MS/MS spectra, and identified S358 of

322 Nek10 as the phosphosite most consistent with the mass spectrometry data (Fig. S5E). Therefore,

323 we conclude from these results that Nek10 can auto-phosphorylate in trans, both in vitro and in

324 cells, and identified Y644 in the kinase domain, and S358 outside of the kinase domain, as auto-

325 phosphorylation target sites.

326

327 The Nek kinase family diverges into four specificity-groups

328 The OPLS results indicated that there are substantial differences in specificity among the

329 Nek kinase family members (Fig 1C). To examine this in more detail, we performed hierarchical

13 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

330 clustering of the Nek kinome based on quantitative OPLS motif information. To also include Nek10

331 in this analysis, we used the serine phosphorylation-site OPLS motif (Nek10(S)), because the

332 tyrosine phosphorylation-site motif is very distinct from the motifs of the other Nek kinases. This

333 clustering revealed that Nek1/3/4/5/8 were clearly separated from Nek2/6/7/9/10(S) (Fig. 6A).

334 Importantly, this division of the Nek kinases based on preferred substrate sequence cannot be

335 explained by amino acid sequence similarity between the Nek kinase domains themselves (Fig.

336 6B). Instead, an important contributor to this hierarchy is the preference for either serine or

337 threonine as the phospho-acceptor site (Fig. 6C). To assess non-phosphosite contributions, we

338 re-clustered the Nek kinase family based on motif specificity, but excluded the phospho-acceptor

339 site preference (Fig. 6D). This revealed that within the group of threonine-directed Nek kinases,

340 Nek 1/3/4 cluster separately from Nek 5/8 (Fig. 6D). Overall then, the motif data indicate that the

341 Nek kinase family can be separated into four specificity groups based on phospho-acceptor

342 preference and additional amino acids specificities (Fig. 6D). Interpretation of the sequence logos

343 shown in Fig. 1C indicates that Nek 1, 3 and 4 cluster in Group 1 based on their shared preference

344 for arginine in the -1 position. Within Group 1, Nek 1 and 4 also show strong selectivity for lysine

345 and arginine in the +2 position, and they are exceptional in their lack of preference for a tryptophan

346 in the -3 position. Group 2 consists of Nek 5 and 8, and is closely related to Group 1, but has a

347 less prominent selection for arginine in the -1 position. Group 3, which consists of Nek 2 and 10,

348 are distinguished by their serine phospho-acceptor specificity (Fig. 6C, S4A), although the

349 remaining motif specificity of Nek 2 is similar to that of kinases in Group 2 (Fig. S3). Finally, Group

350 4 consists of Nek6, 7, and 9, and clusters distantly from the rest (Fig. 6D) based on a preference

351 for acidic residues in the -5, -4, and particularly in the -2 position that is not shared with any of the

352 other Nek kinases (Fig. 6E).

353 To rationalize the unusual selectivity for acidic residues displayed by Group 4 Nek kinases,

354 we compared the X-ray crystal structures of the Nek1 and Nek7 kinase domains (Haq et al., 2015;

355 Melo-Hanchuk et al., 2017). This revealed the presence of two basic regions in Nek7 that are

14 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

356 absent from Nek1, and might accommodate acidic residues N-terminal of the phospho-acceptor

357 site (Fig. 6F). Additional crystal structures of these kinases with substrate peptides will be

358 necessary to experimentally validate these observations. To further investigate the relevance of

359 these motif differences between Group 4 and the other Nek kinases in substrate phosphorylation,

360 we then performed in vitro kinase assays using Nek3 and Nek7 with either a good peptide-

361 substrate for Groups 1 and 2 (13458-tide), or a good peptide-substrate for Group 4 (679-tide). As

362 predicted, 13458-tide was a considerably better substrate for Nek3 than 679-tide, and this was

363 reversed for Nek7 (Fig. 6G).

364

365 Nek6/7/9 share a common substrate motif with Polo-like Kinase 1 and can phosphorylate

366 the same substrates

367 We noticed a striking resemblance between the Nek6/7/9 motif, and the phosphorylation-

368 site motif previously obtained for Plk1 (Fig. 7A; Alexander et al., 2011). Hierarchical clustering of

369 the Nek-family and Plk1 based on substrate motif data from OPLS revealed that, with regards to

370 substrate specificity, Nek6/7/9 are more related to Plk1 than they are to the other Nek kinases

371 (Fig. 7B). Importantly Plk1, together with Cdk1, has previously been reported as the upstream

372 activator of Nek6/7/9 (Bertran et al., 2011). Hence, Plk1 and Nek6/7/9 not only operate together

373 in a linear signaling cascade during mitosis, but also share the same phosphorylation-site motif,

374 and hence could potentially phosphorylate a common set of substrates.

375 To assess potential substrate overlap, we used the quantitative specificity information derived

376 from the OPLS experiments to predict substrates for Plk1 and Nek6/7/9 using the Scansite scoring

377 algorithm (Obenauer et al., 2003). We restricted the scoring to sites that are present in the

378 Phophositeplus database of previously reported phosphorylation sites in the human proteome, to

379 exclude sites that are potentially not surface-accessible (Hornbeck et al., 2015). We considered

380 the top 10% of best scoring sites as potential targets. This analysis revealed a substantial overlap

381 in candidate substrates among the different kinases: over 50% of Plk1 target sites were also

15 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

382 predicted to be a good target site for at least one of the Nek kinases, and a total of 2801 sites

383 were predicted to be good target sites for all four kinases (Fig. 7C).

384 Next, we considered the possibility that substrates previously assigned as direct Plk1 targets

385 on the basis of inhibitor or knockdown studies might actually be either direct substrates of

386 Nek6/7/9, or are shared direct substrates of both Plk1 and Nek6/7/9. To address this possibility,

387 we analyzed the candidate Plk1 substrates at the mitotic spindle that were obtained in a

388 phosphoproteomic study, during which either a small molecule inhibitor or an shRNA was used

389 to block Plk1 function (Santamaria et al., 2011). Scansite scoring of these substrates showed a

390 strong correlation between the Nek7-scores and Plk1-scores (Fig. 7D). Using a cut-off for the top

391 25% of best-scoring sites, we identified 109 distinct sites on 77 different proteins as potential

392 targets of both Plk1 and Nek7. This included S453 on Cdc27, and S170 on RacGAP1, both of

393 which have been established as direct Plk1 substrates (Kraft et al., 2003; Burkard et al., 2009).

394 To test whether Nek7 could also directly phosphorylate these substrates, we generated WT

395 Cdc27 and S434A/S435A double mutant proteins using in vitro transcription/translation (IVTT),

396 and incubated them with isolated Plk1 or Nek7 in an in vitro kinase assay. As previously reported,

397 Cdc27 was readily phosphorylated by WT but not KD Plk1, and this phosphorylation was reduced

398 by mutation of the S434/S435 target site (Fig. 7E, upper panels; Kraft et al., 2003). Interestingly,

399 the same assay performed with Nek7 instead of Plk1 demonstrated that Nek7 could also

400 phosphorylate Cdc27 on the same site as Plk1 (Fig. 7E, lower panels). A second in vitro kinase

401 assay using RacGAP1 confirmed that this substrate is also directly phosphorylated by both Plk1

402 and Nek7, although both kinases appear to target sites other than or in addition to S170 (Fig. 7G).

403 Together, these experiments indicate that Plk1 and Nek6/7/9 share a strikingly similar

404 phosphorylation-site motif, have substantial overlap in potential substrates in the

405 phosphoproteome, and are capable of phosphorylating the same protein substrates on the same

406 target sites in vitro.

16 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

407 Discussion

408 While it is widely appreciated that the Nek kinases function in various stages of mitosis, and

409 also play roles in essential non-mitotic processes including cilia formation and the DNA damage

410 response (Moniz, Dutt, et al., 2011; Fry et al., 2017), only a limited number of direct substrates of

411 these kinases have been identified to date. To facilitate future substrate identification, we

412 systematically determined the phosphorylation-site motif for each kinase in the family, with the

413 exception of Nek11, using OPLS. This analysis revealed that all Nek kinases share a common

414 preference for selected hydrophobic residues in the -3 position, but differ substantially in their

415 preference for specific amino acids at other positions within the motif, including their preference

416 for either a threonine (Nek1/3/4/5/8) or serine (Nek2/6/7/9) as the phospho-acceptor residue. A

417 strong determinant of phospho-acceptor site specificity is the residue in the activation loop of the

418 kinase at the DFG+1 position (Chen et al., 2014), and the observed dichotomy of serine versus

419 threonine phospho-acceptor preference within the Nek family can be explained by the presence

420 of a leucine versus isoleucine/serine in this position (Fig. 6C).

421 Overall, the phosphorylation-site motifs indicate that the Nek family can be divided in four

422 specificity-groups (Fig. 6D, 7A). To date, all kinases in specificity Groups 3 (Nek2/10) and 4

423 (Nek6/7/9) have been implicated in one or more mitotic processes, with the exception of Nek10.

424 The exact function of Nek10 is not known, but it has previously been reported to play a role in the

425 response to UV-stress (Moniz and Stambolic, 2011). In the accompanying manuscript, we show

426 that Nek10 regulates the transcriptional activity of p53 by phosphorylating it on a specific tyrosine,

427 and that freely cycling A549 cells in which Nek10 has been knocked-out show a modest increase

428 in the percentage of mitotic cells (Haider et al., accompanying manuscript). It is therefore tempting

429 to speculate that the Nek kinase specificity clustering might reflect functional divergence between

430 mitotic and non-mitotic Nek kinases.

17 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

431 In addition to the four specificity-groups, we identified amino acid selection preferences that

432 are unique to individual Nek kinases. For example, Nek6 and Nek7 showed a clear preference

433 for substrate peptides that are already phosphorylated on tyrosine on the -1 position, suggesting

434 cross-talk between tyrosine and serine/threonine kinases during mitosis. It is becoming

435 increasingly clear that tyrosine phosphorylation events do, in fact, play a role in mitotic signaling

436 (Caron et al., 2016), and multiple tyrosine kinases have been shown to prefer an acidic residue

437 in the -1 position (Miller et al., 2018), which would generate good target sites for Nek6/7 (i.e. a

438 DE-pY-S motif).

439 In general, good substrates for a kinase usually contain phosphorylation site sequences that

440 closely match the optimal sequence motif for the kinase, though additional factors including

441 secondary docking sites, surface accessibility, and substrate protein conformation also play

442 important roles (Linding et al., 2007; Miller et al., 2018). It is therefore expected that the

443 phosphorylation sites on Nek kinase substrates should be enriched for the amino acids preferred

444 by the upstream Nek kinase. However, to date, too few substrates have been definitively and

445 reproducibly characterized for most of the individual Nek kinases to allow quantitative comparison

446 of those sites with the optimal motifs. The maximal number of substrates reported for any Nek

447 kinase is eleven (for Nek6), with less than 5 substrates known for all other Nek kinases except for

448 Nek 2. There are no published substrates for Nek 4, 5, 8, and 10. Nonetheless, a global analysis

449 of all previously reported Nek kinase target sites (Table S1) reveals an overrepresentation of

450 leucine and phenylalanine in the -3 position, consistent with the motif reported here. Of the eleven

451 published Nek6 substrates, three contain an asparagine on the -2 position, which we identified as

452 a specificity unique to Group 4 Nek kinases, and four others have a glycine, which is also

453 positively selected for by Nek6. It is important to emphasize that substrate phosphorylation by a

454 kinase does not require a perfect match between the target site and the optimal motif. Rather, the

455 better the site matches the optimal motif, the more efficiently it is phosphorylated by the kinase

18 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

456 (Obenauer et al., 2003), as we also demonstrate here in our peptide phosphorylation studies (Fig.

457 2B-F).

458 Importantly, our assays revealed that Nek10 is a dual-specificity kinase, and thus joins a

459 small group of human serine/threonine kinases with reported dual-specificity. These are the GSK-

460 3 kinases, the DYRK-kinases, the kinases of the MAP2K-family, TTK, CK2, Myt1 and the TESK-

461 kinases (Gómez et al., 1991; Lindberg et al., 1993; Wang et al., 1994; Kentrup et al., 1996; Liu et

462 al., 1997; Wilson et al., 1997; Toshima et al., 1999). GSK-3 and the DYRK kinases only

463 phosphorylate tyrosine residues when auto-phosphorylating during folding in cis, although it was

464 recently reported that the folded, mature DYRK1A retains a weak tyrosine auto-phosphorylation

465 capability (Cole et al., 2004; Lochhead et al., 2005; Soundararajan et al., 2013). TTK and CK2

466 have weak tyrosine phosphorylation activity compared to their serine phosphorylation activity, and

467 Myt1 and the kinases of the MAP2K family only phosphorylate on tyrosine and threonine in the

468 very distinct context of a TY-site on Cdk1 and a TxY-site on MAP kinases, respectively (Gómez

469 et al., 1991; Lindberg et al., 1993; Liu et al., 1997; Marin et al., 1999). The TESK-kinases are

470 therefore an exception within the group of dual-specificity kinases, because they can efficiently

471 phosphorylate multiple protein substrates in trans on both tyrosine and serine (Toshima et al.,

472 1999).

473 Here, we show that Nek10’s dual-specificity is most like that of the TESK-kinases, because

474 Nek10 phosphorylates tyrosine residues as efficiently as serine residue on substrates in trans.

475 Serine/threonine kinases are structurally different than tyrosine kinases, primarily in the catalytic

476 loop and the P+1 loop of the kinase domain. These regions make extensive contact with the

477 substrate, including with the phospho-acceptor site, and structural differences within these loops

478 dictate whether kinases exclusively phosphorylate either serine/threonine or tyrosine (Taylor et

479 al., 1995). We show here that an isoleucine in the APE-4 position, and a threonine in the HRD+2

480 position, makes the P+1 loop and catalytic loop of Nek10 unlike those typically found in either

481 tyrosine kinases or serine/threonine kinases, and these unique residues give Nek10 the capacity

19 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

482 to phosphorylate substrates on tyrosine in addition to serine. Serine/threonine kinases have a

483 conserved lysine on the HRD+2 position in the catalytic loop, which makes contact with the P+1

484 loop and the substrate backbone, and therefore helps positioning the substrate for

485 phosphorylation. The HRD+2 threonine, rather than lysine, in Nek10 is essential for its ability to

486 phosphorylate substrates on tyrosine, and within the dual-specificity kinome this residue is only

487 shared with the TESK kinases. This could potentially explain why of all dual-specificity kinases,

488 Nek10 and the TESK kinases appear to have the broadest tyrosine phosphorylation capacity. The

489 serine/threonine kinases BubR1, Plk4, PRPK and MISR2 also have a serine or threonine residue

490 in the HRD+2 position, suggesting that they might be dual-specificity. Notably, murine Nek1 has

491 been reported to be a dual-specificity kinase (Letwin et al., 1992), but we could not detect any

492 tyrosine-phosphorylation by human Nek1 (Fig. S4A), and Nek1 has neither the HRD+2 threonine,

493 nor the APE-4 isoleucine that distinguishes Nek10 from the other Nek family members (Figs. 3E,

494 H).

495 Additionally, the sequence surrounding the phosphorylation site also affects substrate

496 positioning, and can therefore affect phospho-acceptor site specificity. As an example of this

497 interdependence, we have previously noted that the preference of the kinases ATM/ATR for

498 phosphorylating serine-glutamine versus threonine-glutamine motifs appears to depend on the

499 residues found at other positions in the motif (Joughin et al., 2012). A similar observation of

500 interdependence between phospho-acceptor residue identity and specific amino acids in other

501 positions within the motif has been shown for the dual-specificity kinase CK2 (Marin et al., 1999).

502 Marin et al observed that tyrosine phosphorylation of a yeast substrate of CK2 depended on

503 specific residues in the -1 and +1 positions. However, phosphorylation of the same substrate with

504 the tyrosine replaced by a serine residue was considerably less influenced by substitutions of the

505 amino acids in the -1 and +1 positions and instead depended on the residue identity in the +3

506 position (Marin et al., 1999). Here, we identified distinct phosphorylation-site motifs for Nek10 on

507 tyrosine and on serine/threonine substrates (Fig. 4). This indicated that the ability of Nek10 to

20 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

508 phosphorylate peptides on tyrosine was strongly dependent on the presence of an aromatic

509 residue or a leucine in the +1 position of the substrate (Fig. 4C), but Nek10 was much more

510 tolerant of other amino acids in this position when phosphorylating substrates on serine (Fig. 4A).

511 Potentially, a bulky +1 residue orients the substrate within the catalytic cleft in a manner that

512 generates space for the large tyrosine phospho-acceptor site.

513 Finally, our study showed that Nek6/7/9 share a common phosphorylation motif with Plk1.

514 Plk1 interacts with Nek9 and functions as the upstream activating kinase in a linear mitotic

515 signaling pathway with Nek6, 7, and 9, raising the possibility that the common phosphorylation

516 motif we observed arises from cross-contamination in the isolated kinase preparations. However,

517 the fact that our Plk1 motif was determined using the recombinant kinase domain of Plk1

518 expressed in bacteria, and that catalytically dead versions of each of these Nek kinases displayed

519 no phosphorylation activity (Fig. S2) argues against this interpretation. Instead, it may be that

520 these mitotic kinases function together to form a coherent feed-forward loop analogous in some

521 ways to the coherent type 1-feed forward loop with OR gate function described by Uri Alon for

522 transcription factor signaling (Alon, 2007). Such a signaling network of kinases sharing a highly

523 related phosphorylation-site motif (Fig. 7G) might allow persistence of this phospho-motif for a

524 period of time after Plk1 inactivation, or could help spread this specific phospho-motif beyond

525 direct Plk1 substrates by acting as a “phospho-motif amplifier”. Clearly, many future experiments

526 investigating the spatial and temporal phosphorylation of in vivo substrates by kinases within this

527 network will be required to distinguish between these and other possibilities.

528

21 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

529 Materials and methods

530

531 Cloning

532 Coding DNA sequences of Nek1 and Nek3-Nek9 were obtained from the human ORFeome

533 collection. Using gateway cloning technology, Nek3-Nek9 were cloned into V1900, a gateway

534 destination vector derived from pCMV-Sp6 with a C-terminal 3xFLAG epitope tag

535 (DYKDHDGDYKDHDIDYKDDDDK). All other cloning was done by restriction ligation, Quick-

536 Fusion cloning (Biotool.com) or Gibson assembly using the NEBuilder HiFi DNA assembly

537 Cloning Kit (New England Biolabs). Nek1 was PCR-amplified from the ORFeome entry vector,

538 and cloned into pcDNA3 that was re-engineered to contain a C-terminal 3xFLAG tag (pcDNA3-

539 MCS-3xFLAG). The Nek4 kinase domain (aa: 1-300) was PCR amplified from the V1900-Nek4

540 plasmid, and cloned into pcDNA3-MCS-3xFLAG. The Nek5 kinase domain (aa: 2-300) was PCR

541 amplified from the V1900-Nek5 plasmid, and cloned into pcDNA3, which was reengineered with

542 an N-terminal MBP (maltose-binding-protein)-3xFLAG epitope tag. The Nek2 kinase domain (aa:

543 1-271) was PCR amplified from copy-DNA generated from U-2 OS mRNA, and cloned into

544 pcDNA3, which was reengineered with a C-terminal 3xFLAG-TEV-2xStreptagII-6xHIS epitope

545 tag. (TEV: Tobacco Etch Virus cleavage site (ENLYFQG), StreptagII: WSHPQFEK). The

546 construct pCMV-3xFLAG-Nek10 has been described previously (Moniz and Stambolic, 2011). A

547 small missing region in the Nek10 cDNA (bp: 2869-3012) was repaired by PCR, and the cDNA

548 including N-terminal 3xFLAG-tag was subcloned into pcDNA3, or the cDNA without 3xFLAG-tag

549 was subcloned into pcDNA3-2xHA-Strep, which is a pcDNA3-variant that was re-engineered for

550 epitope-tagging with an N-terminal 2xHA-StreptagII epitope tag (HA: YPYDVPDYA). The Plk1

551 kinase domain (aa: 38-346) was PCR-amplified from plasmids previously described and cloned

552 into pET-RSF (Elia et al., 2003). Cdc27 and RacGAP1 cDNA was PCR amplified from copy-DNA,

553 and cloned into pcDNA3-2xHA-Strep. All mutants of the Neks, and Cdc27 and RacGAP1 mutants,

22 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

554 were generated by PCR-mediated mutagenesis. All kinase-dead variants were made by mutation

555 of the aspartic acid in the conserved HRD motif into an asparagine, and in case of Nek4, an

556 additional aspartic acid to asparagine mutation was made in the conserved DFG-motif (DLG in

557 case of Nek4). In order to make the Nek10 KO cell-line, a Nek10-targeting sgRNA (sequence:

558 GTCTGAGCCCGCCATCAGGG) was cloned into lentiGuide-Puro (addgene #52963), and Cas9

559 was subcloned from pCW-Cas9 (addgene #50661) into pLVX-Cas9-ZsGreen. In addition, a Cas9-

560 targeting sgRNA (sequence: CTTGTACTCGTCGGTGATCA) was cloned into lentiguide-puro.

561 The inducible Nek10 construct was generated by replacing Cas9 in pCW-Cas9 with 3xFLAG-

562 Nek10-T2A-eGFP. Silent mutations were introduced in the PAM-sequence of the Nek10 sgRNA

563 target site, and the puromycin cassette in pCW was replaced with a blastidicin resistance

564 cassette.

565

566 Protein production and purification

567 All Nek kinases, unless indicated otherwise below, were purified from transiently transfected HEK

568 293T cells. In short, 7x106 HEK 293T cells were plated in 15 cm plates, and the next day

569 transfected with 20-25 µg kinase-DNA using polyethyleneimine (PEI). After O/N incubation,

570 transfection medium was replaced with fresh medium, and another 24h later cells were washed

571 with PBS and harvested by scraping in lysis buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM

572 EDTA, 1 mM EGTA, 1% Triton-X100, 1 mM DTT) supplemented with protease inhibitors

573 (cOmplete EDTA-free protease inhibitor cocktail, Roche) and phosphatase inhibitors (PhosSTOP,

574 Roche). Lysates were incubated on ice for 20 minutes, and cleared by centrifugation. Next, anti-

575 FLAG M2 affinity gel (Sigma) was added to the cleared lysate, followed by incubation for 2h at

576 4°C while rotating. After incubation, the beads were pelleted by centrifugation, washed twice with

577 lysis buffer and washed twice again with wash buffer (50 mM HEPES pH 7.4, 100 mM NaCl, 1

578 mM DTT, 0.01% NP-40, 10% glycerol). Protein was eluted from the beads by incubation for 1h at

23 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

579 RT (while rotating) with 150-250 µl wash buffer supplemented with phosphatase inhibitors

580 (PhosSTOP, Roche) and 0.5 mg/ml 3xFLAG peptide (APExBIO). To determine concentration and

581 purity, a BSA standard curve and a sample of the eluate were analyzed by SDS-PAGE and

582 Coomassie-staining (SimplyBlue Safe stain, Life Technologies). Eluted kinase was either used

583 directly, or snap-frozen and stored at -80°C. The 2xHA-Streptag-Nek10 KD (HA-tagged Nek10

584 KD, fig. 5) was isolated essentially exactly as described above. However, Strep-Tactin XT

585 superflow (IBA lifesciences) was used instead of anti-FLAG affinity gel, the composition of the

586 wash buffer was different (100 mM Tris pH 8, 150 mM NaCL, 1 mM EDTA) and the kinase was

587 eluted in 1x Biotin elution buffer BXT (IBA lifesciences). Plk1 was isolated from bacteria. A BL21

588 Rosetta 2 strain was transformed and induced to express the 6xHIS-MBP-TEV-tagged Plk1

589 kinase domain. The bacteria were lysed, followed by incubation with an Amylose resin (New

590 England Biolabs)) and elution with maltose. In case of constitutively active T210D Plk1, a second

591 purification step on a nickel-column (Ni Sepharose 6 Fast Flow, GE Healthcare) was performed,

592 followed by imidazole-elution.

593

594 Oriented Peptide Library Screen

595 Oriented Peptide Library Screens were performed as described in Hutti et al, but with a slightly

596 adapted protocol so that the assay could be done in 1536-well plates (Hutti et al., 2004). The

597 serine/threonine peptide library (Kinase Substrates Library, Group I and Group II, Anaspec)

598 consists of 198 peptide pools with the sequence Y-A-Z-X-X-X-X-S/T-X-X-X-X-A-G-K-K-(LC-

599 Biotin)-NH2, were X= mixture of all natural amino acids except Cys, Ser and Thr, S/T=equal

600 mixture of Ser and Thr, and LC-Biotin= long chain biotin. In each peptide pool, the Z is fixed to be

601 a single amino acid, which can be any of the 20 natural amino acids or phospho-Thr or phospho-

602 Tyr. In the example, the Z is positioned on the P-5 relative to the S/T, but in the complete library

603 the Z would be fixed in a position ranging from -5 to +4 relative to the S/T. Hence, there are nine

24 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

604 different positions and 22 different amino acids, generating 198 different peptide pools. In

605 addition, the library contains three peptide pools that are completely degenerate with the

606 exception of a fixed Ser, Thr or Tyr as phospho-acceptor site for use in determining the 0 position

607 (Fig. S1). The tyrosine-peptide library (custom order, JPT) consisted of peptide pools with the

608 sequence G-A-Z-X-X-X-X-Y-X-X-X-X-A-G-K-K-(LC-Biotin)-NH2, and in this library Cys, Ser, Thr

609 and Tyr were absent from the amino acid mixtures on the X positions. All peptides were dissolved

610 in DMSO to 7.5 mM for the Ser/Thr library, and 5mM for the Tyr-library. The peptides were arrayed

611 in a 1536-well plate by multi-channel pipetting, and diluted to 500 µM with water.

612 For the kinase assay, 2.2 µl kinase buffer per well was dispensed in 1536-well plates using

613 a Mantis nanodispenser robot (Formulatrix). The core kinase buffer consisted of 50 mM Tris-HCl

614 pH 7.5, 0.1 mM EGTA, 1 mM DTT, 50 µM ATP, 1 mM PKA-inhibitor (Sigma) and 0.1% Tween-

615 20, and was supplemented with 10 mM MgCl2 and 2mM MnCl2 for Nek1/2/3/4/6/7/9, 10 mM

616 MnCl2 for Nek5/8, and 10 mM MgCl1 and 10 mM MnCl2 for Nek10. Next, 250 nl peptide was

617 transferred from the peptide-stock plate to the kinase assay plate using a pintool with slot tips

618 (V&P Scientific, Pin type FP3S200). Subsequently, 200 nl of a mix of kinase and [g-33P)-ATP was

619 added using the Mantis, with a final concentration in each well ranging from 5-50 nM of the kinase

620 and 0.027 µCi/µl for the [g-33P]-ATP. Between every step, the 1536-well plate was centrifuged

621 briefly to collect all liquid, and cooled on ice as needed to prevent evaporation. After addition of

622 the kinase, the plate was incubated at 30°C for 2h. Subsequently, 250 nl of each reaction was

623 spotted on streptavidin-membrane (Promega) using the pintool, the membrane was washed and

624 dried, exposed to a phosphorimaging screen, and imaged on a Typhoon 9400 imager (GE

625 Healthcare).

626 Spot intensities were quantified using ImageStudio (Li-COR BioSciences), and normalized

627 by dividing the measured intensity for an individual spot by the average spot intensity for all amino

628 acids in that position. For each kinase, OPLS-experiments were done at least twice, and

25 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

629 normalized values were averaged for all replicates. In case of a fixed serine or threonine on the

630 Z-position, the fixed residue presents a second acceptor site in addition to the 0-position S/T, and

631 spot intensities are therefore high at those positions, which does not necessarily reflect real

632 sequence preferences. To correct for this, normalized intensities for fixed S or T were changed to

633 1 in case of intensity>1. For Nek10 on the Tyr-library, the same was done but for intensities

634 measured for fixed-Tyr peptides. The exception was the +1 Tyr, because we reasoned the

635 normalized intensity>1 most likely reflected a true preference, considering its strong preference

636 for aromatic amino acids. Next, normalized values were log2-transformed. These values were

637 used to generate heatmaps using Morpheus (https://software.broadinstitute.org/ morpheus/).

638 These values were transformed to sequence logos with Seq2Logo (Thomsen et al., 2012).

639

640 In vitro kinase assays

641 For in vitro kinase assays with a peptide-substrate, 5-50 nM of kinase was pre-incubated at 30°C

642 for 30 minutes in the same kinase buffer as used for OPLS, but without the 0.1% Tween-20.

643 Subsequently, [g-33P]-ATP (or in some cases [g-32P]-ATP) was added to a final concentration of

644 0.075-1 µCi/µl, and peptide added to a final concentration of 50 µM. Peptides were synthesized

645 in-house at the 5 µmol starting resin scale per 96 well-plate well, using an MultiPep small scale

646 peptide synthesizer (Intavis) and standard FMOC chemistry. In addition to the peptide-specific

647 10-mer sequences indicated in the figure panels, each peptide had an N-terminal Y-A (or W-A for

648 peptides of Nek10) sequence to allow concentration determination of peptide-solutions by UV-

649 spectrophotometry. Also, each peptide had a C-terminal G-K-K-K sequence to increase solubility

650 and interaction with phosphocellulose. After adding substrate and [g-33P]-ATP, the mixture was

651 incubated at 30°C for 20 minutes, and every 5 minutes a sample was spotted on p81

652 phosphocellulose papers (Reaction Biology Corp.), with the exception of some of the repeats with

653 the Nek10 T693P variant, that were incubated for 1h or 2h, with spotting every 20 min or 30 min

26 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

654 respectively. After spotting, the phosphocellulose papers were immediately soaked in 0.5%

655 phosphoric acid. At the end of the assay, all papers were washed three times in 0.5% phosphoric

656 acid, air-dried, transferred to vials with scintillation counter fluid and counted by a LS 6500

657 scintillation counter (Beckman Coulter). The resulting data points were analyzed by linear

658 regression, and the phosphorylation rate was defined as the slope of the linear regression curve.

659 To study Nek10 auto-phosphorylation, 250 ng kinase was incubated in OPLS-kinase buffer

660 with 1 µCi/µl [g-32P]-ATP for 30 minutes at 30°C. The reaction was stopped by adding 6x sample

661 buffer (SB) for SDS-PAGE (208 mM Tris-HCl pH 6.8, 42% glycerol, 3 mM beta-mercapto-ethanol,

662 10% SDS, 5 mg/ml bromophenol blue) and boiling for 5 minutes at 95°C. The sample was split,

663 and analyzed by SDS-PAGE followed by Coomassie staining to asses total Nek10 levels, or

664 analyzed by SDS-PAGE followed by phosphorimaging to analyze protein phosphorylation.

665 To study Nek10 auto-phosphorylation in trans, 250 ng of FLAG-tagged kinase-proficient

666 Nek10 was mixed with 250 ng of HA-tagged kinase-deficient Nek10 and 0.06 µCi/µl [g-32P]-ATP

667 in kinase buffer. The mixture was incubated at 30°C for 15 min to 2h, depending on the assay,

668 and the reaction was stopped by adding lysis buffer supplemented with EDTA to a final

669 concentration of 20 mM. The sample was split in half and incubated for 1h at RT (while rotating)

670 with either anti-FLAG M2 affinity gel (Sigma) or anti-HA affinity matrix (mAb 3F10; Roche). The

671 beads were washed once with lysis buffer, dried, and resuspended in 1xSB. Nek10 was eluted of

672 the beads by boiling in 1xSB, the sample was split, and 1/4th was analyzed by SDS-PAGE

673 followed by immunoblotting to asses total Nek10 levels, while 3/4th was analyzed by SDS-PAGE

674 followed by phosphorimaging to assess protein phosphorylation.

675

676 U-2 OS KO cell-line generation

677 All cell-lines were grown at 37°C in a humidified incubator supplied with 5% CO2, in DMEM

678 supplemented with 10% FBS. Lentivirus was produced by transient transfection of HEK 293T

27 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

679 cells using the CalPhos mammalian transfection kit (Clontech laboratories) to introduce packaging

680 vectors (VSVg and GAG/POL/D8.2) and either pLVX-Cas9-Zsgreen, or lentiguide-PURO-Nek10

681 sgRNA, or pCW-Nek10-T2A-GFP-Blasticidin. The viral supernatant was filtered through a 0.45

682 µm filter and used to transduce U-2 OS target cells. First, U-2 OS cells were transduced with

683 Cas9, followed by sorting for Zsgreen positive cells. These cells were then transduced with Nek10

684 sgRNA, followed by selection with puromycin (2 µg/ml; Invivogen). Doubly-transduced cells were

685 subsequently cultured for two weeks to allow editing, and single cell sorted into 96-well plates.

686 After expansion of the single-cell clones, gene editing was analyzed by PCR-amplifying the edited

687 locus from gDNA, followed by Sanger sequencing of the PCR product.

688 After selecting a Nek10-KO clone, Cas9 was removed by transfecting with a Cas9-sgRNA

689 construct using lipofectamine 2000 (Thermo Fisher Scientific). Cells were cultured for ten days

690 and sorted for ZsGreen-negative cells. Immunoblot analysis indicated that although all cells were

691 negative for Zsgreen, about 30% of Cas9-expression still remained. Next, the cells were

692 transduced with the pCW-Nek10 constructs, followed by selection with Blasticidin S (10 µg/ml;

693 Invitrogen). To select for cells without leaky expression of Nek10, but good induction upon

694 treatment with doxycycline, the following was done: first, GFP-negative cells were sorted, followed

695 by induction of Nek10 expression with 2 µg/ml doxycycline (Sigma) for several days, and a second

696 round of sorting, this time on GFP-positive cells.

697

698 Antibodies and western blotting

699 Cell lysates was analyzed by SDS-PAGE, and transferred to nitrocellulose membranes by tank

700 electrotransfer. Membranes were blocked using either 5% skim milk in Tris-Buffered-Saline

701 (TBS), or with Odyssey blocking buffer (LI-COR BioSciences) for 1h at RT. Primary antibody

702 probing was done either in 1% skim milk in TBS-T (TBS with 0.1% Tween-20), or in Odyssey

703 blocking buffer, for 1h-2h at RT, or O/N at 4°C. Primary antibodies used are Mouse-anti-FLAG

28 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

704 M2 (1:1000; Sigma), Rat-anti-HA 3F10 (1:500; Roche), Mouse-anti-b-actin (1:10,000; Sigma),

705 Rabbit-anti-b-Actin (1:1000, Cell Signaling Technology), Rabbit-anti-phospho-tyrosine P-Y-1000

706 (1:2000; Cell Signaling Technology) and Rabbit-anti-Plk1 (1:100; Santa Cruz Biotechnology, Inc.).

707 After primary antibody incubation, the membranes were washed 3x with TBS-T, followed by

708 staining with secondary antibody in Odyssey blocking buffer for 1h at RT. Secondary antibodies

709 used were Goat-anti-Mouse, Goat-anti-Rabbit, or Goat-anti-Rat, all from LI-COR, all 1:10,000,

710 and all labeled with either IRDye 680 or IRDye 800. After secondary antibody staining, the

711 membranes were washed again 3x with TBS-T, and imaged with an Odyssey CLx scanner (LI-

712 COR BioSciences). Image analysis was done using ImageStudio (LI-COR BioSciences).

713

714 Nek10 auto-phosphorylation in cells

715 To study phosphorylation in cells, U-2 OS cells were transfected with the 3xFLAG-Nek10 vector

716 using lipofectamine 2000 (Thermo Fisher Scientific) for transient expression, or 3xFLAG-Nek10

717 expression was induced in the stable U-2 OS Nek10 KO cell lines with inducible Nek10 by

718 treatment with docxycycline for 48h. The cells were lysed and Nek10 was immobilized on anti-

719 FLAG beads as described in the “protein purification’” section above. Next, beads were washed

720 three times with lysis buffer, and all bound proteins eluted by boiling in 1xSB. To study tyrosine

721 phosphorylation, eluates and total cell lysates were analyzed by SDS-PAGE followed by

722 immunoblotting. To analyze serine phosphorylation by mass spectrometry, eluates were

723 separated by SDS-PAGE, followed by Coomassie staining of the gel. The band containing

724 3xFLAG-Nek10 was excised, reduced and alkylated, followed by in-gel trypsin digestion. Tryptic

725 peptides were extracted, desalted and lyophilized. Next, phosphopeptides were enriched using

726 the High-Select Fe-NTA phosphopeptide enrichment kit (ThermoFisher Scientific) per

727 manufacturer’s instructions. The elution tubes were coated with trypsinized BSA to prevent loss

728 of phospho-peptides from binding to the tube wall. Eluted phosphopeptides were re-suspended

29 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

729 in 0.1% acetic acid and separated by reverse phase HPLC using an EASY- nLC1000 (Thermo)

730 with a self-pack 5 µm tip analytical column (12 cm of 5 µm C18, New Objective). Next, peptides

731 were ionized by nanoelectrospray and analyzed using a QExactive HF-X mass spectrometer

732 (ThermoFisher Scientific). The full MS-scan was followed by MS/MS for the top 15 precursor ions

733 in each cycle. Raw mass spectral data files (.raw) were searched using Proteome Discoverer 2.2

734 (Thermo) and Mascot version 2.4.1 (Matrix Science). In addition, for the phosphopeptide

735 corresponding to N346-R365 of Nek10, the spectrum was manually annotated by comparison to

736 predicted products using Protein Prospector (http://prospector.ucsf.edu/prospector/mshome.htm)

737 to ensure correct phosphorylation site identification.

738

739 Motif analysis and substrate prediction

740 The hierarchical clustering based on the OPLS-datasets was done using the averaged,

741 normalized intensities, with the values for fixed Ser and Thr (except in the 0 position) set to 1 if

742 >1 (also see OPLS section). Hierarchical clustering was done using Morpheus

743 (https://software.broadinstitute.org/ morpheus/) according to a One minus Pearson correlation

744 metric with an average linkage method. For the clustering based on sequence, Nek kinase domain

745 amino acid sequences were uploaded into JalVIEW, aligned with ClustalWS, and the % sequence

746 identity used to derive a dendrogram and similarity matrix.

747 To predict substrates of Plk1 and Nek6/7/9, we took the averaged, column-normalized

748 intensities of the OPLS-experiments to generate a Position-Specific Scoring Matrix (PSSM), and

749 scored every phosphorylated site in the Phosphositeplus database for their match to the PSSM

750 using Scansite (Obenauer et al., 2003; Hornbeck et al., 2015). A Venn diagram was generated

751 for the 10% best (i.e. lowest) scoring sites using the Venn diagram maker online

752 (https://www.meta-chart.com/venn#/display). To predict mitotic substrates of Plk1 and Nek7, table

753 S2 from Santamaria et al. was used (Santamaria et al., 2011). The phosphosites that are not

754 Class I phosphosites were removed, and duplicate sites were taken out, generating a hitlist of

30 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

755 694 sites consistent with what was reported in Santamaria et al. As above, this list of sites was

756 scored using Scansite version 4 for their match to the Plk1 and Nek6/7/9 PSSM.

757

758 Statistical analysis

759 All statistical analysis was done using Graphpad Prism. If the experiment contained two datasets,

760 significance was calculated using a Students t-test, either paired or unpaired, and either regular

761 or ratiometric, depending on what was most appropriate given the experimental conditions. If the

762 experiment contained more than two datasets, an ANOVA with post hoc multiple comparison was

763 performed.

764

31 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

765 Acknowledgements

766 We thank Dan Lim and other members of the Yaffe laboratory for helpful discussion and advice,

767 Forest White (Massachusetts Institute of Technology, Cambridge, MA, USA) for advice on Mass

768 Spectrometry experiments and Prasad Jallepalli (Memorial Sloan Kettering, New York, NY, USA)

769 for kindly sharing reagents. We also thank the Biopolymers and Proteomics core facility, and

770 Genomics: High throughput Screening core facility of the Swanson Biotechnology Center for their

771 assistance. This research was supported by National Institute of Health grants R01-GM104047

772 (M.B.Y. and B.E.T.), R01-ES015339 (M.B.Y.), R35-ES028374 (M.B.Y.), the Charles and Marjorie

773 Holloway foundation (M.B.Y.), the MIT Center for Precision Cancer Medicine, and by fellowships

774 for B.v.d.K. from the Dutch Cancer Society (BUIT 2015-7546) and Ludwig Center Fund. Support

775 was also provided by the Cancer Center Support Grant P30-CA14051 from the National Cancer

776 Institute and the Center for Environmental Health Sciences Support Grant P30-ES002109 from

777 the National Institute of Environmental Health Sciences.

778 779

32 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

780 References

781 Alexander, J., Lim, D., Joughin, B. A., Hegemann, B., Hutchins, J. R. A., Ehrenberger, 782 T., Ivins, F., Sessa, F., Hudecz, O., Nigg, E. A., Fry, A. M., Musacchio, A., 783 Stukenberg, P. T., Mechtler, K., Peters, J.-M., Smerdon, S. J. and Yaffe, M. B. 784 (2011) ‘Spatial Exclusivity Combined with Positive and Negative Selection of 785 Phosphorylation Motifs Is the Basis for Context-Dependent Mitotic Signaling’, Sci. 786 Signal., 4(179), pp. ra42-ra42. 787 788 Alon, U. (2007) ‘Network motifs: theory and experimental approaches’, Nat. Rev. Genet., 789 8, pp. 450–461. 790 791 Belham, C., Roig, J., Caldwell, J. A., Aoyama, Y., Kemp, B. E., Comb, M. and Avruch, J. 792 (2003) ‘A mitotic cascade of NIMA family kinases: Nercc1/Nek9 activates the Nek6 793 and Nek7 kinases’, J. Biol. Chem., 278(37), pp. 34897–34909. 794 795 Bertran, M. T., Sdelci, S., Regué, L., Avruch, J., Caelles, C. and Roig, J. (2011) ‘Nek9 is 796 a Plk1-activated kinase that controls early centrosome separation through Nek6/7 797 and Eg5’, EMBO J., 30(13), pp. 2634–2647. 798 799 Brieño-Enríquez, M. A., Moak, S. L., Holloway, J. K. and Cohen, P. E. (2017) ‘NIMA- 800 related kinase 1 (NEK1) regulates meiosis I spindle assembly by altering the 801 balance between α-Adducin and Myosin X’, PLoS One, 12(10), pp. 1–18. 802 803 Burkard, M. E., Maciejowski, J., Rodriguez-Bravo, V., Repka, M., Lowery, D. M., 804 Clauser, K. R., Zhang, C., Shokat, K. M., Carr, S. A., Yaffe, M. B. and Jallepalli, P. 805 V. (2009) ‘Plk1 self-organization and priming phosphorylation of HsCYK-4 at the 806 spindle midzone regulate the onset of division in human cells’, PLoS Biol., 7(5), p. 807 e1000111. 808 809 Caron, D., Byrne, D. P., Thebault, P., Soulet, D., Landry, C. R., Eyers, P. A. and Elowe, 810 S. (2016) ‘Mitotic phosphotyrosine network analysis reveals that tyrosine 811 phosphorylation regulates Polo-like kinase 1 (PLK1)’, Sci. Signal., 9(458), pp. rs14- 812 rs14. 813 814 Chen, C., Ha, B. H., Thévenin, A. F., Lou, H. J., Zhang, R., Yip, K. Y., Peterson, J. R., 815 Gerstein, M., Kim, P. M., Filippakopoulos, P., Knapp, S., Boggon, T. J. and Turk, B. 816 E. (2014) ‘Identification of a major determinant for serine-threonine kinase 817 phosphoacceptor specificity.’, Mol. Cell, 53(1), pp. 140–7. 818 819 Choi, H. J. C., Lin, J. R., Vannier, J. B., Slaats, G. G., Kile, A. C., Paulsen, R. D., 820 Manning, D. K., Beier, D. R., Giles, R. H., Boulton, S. J. and Cimprich, K. A. (2013) 821 ‘NEK8 links the ATR-regulated replication stress response and S phase CDK 822 activity to renal ciliopathies’, Mol. Cell, 51(4), pp. 423–439. 823 824 Cole, A., Frame, S. and Cohen, P. (2004) ‘Further evidence that the tyrosine 825 phosphorylation of glycogen synthase kinase-3 (GSK3) in mammalian cells is an 826 autophosphorylation event’, Biochem. J., 377(1), pp. 249–255. 827 828 Cullati, S. N., Kabeche, L., Kettenbach, A. N. and Gerber, S. A. (2017) ‘A bifurcated

33 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

829 signaling cascade of NIMA-related kinases controls distinct kinesins in anaphase’, J. 830 Cell Biol., 216(8), pp. 2339–2354. 831 832 Dephoure, N., Zhou, C., Villen, J., Beausoleil, S. A., Bakalarski, C. E., Elledge, S. J. and 833 Gygi, S. P. (2008) ‘A quantitative atlas of mitotic phosphorylation’, Proc. Natl. Acad. 834 Sci., 105(31), pp. 10762–10767. 835 836 Doles, J. and Hemann, M. T. (2010) ‘Nek4 status differentially alters sensitivity to distinct 837 microtubule poisons’, Cancer Res., 70(3), pp. 1033–1041. 838 839 Elia, A. E. H., Cantley, L. C. and Yaffe, M. B. (2003) ‘Proteomic screen finds pSer/pThr- 840 binding domain localizing Plk1 to mitotic substrates’, Science (80-. )., 299(5610), pp. 841 1228–1231. 842 843 Fry, A. M., Bayliss, R. and Roig, J. (2017) ‘Mitotic Regulation by NEK Kinase Networks’, 844 Front. Cell Dev. Biol., 5(102). 845 846 Gómez, N. and Cohen, P. (1991) ‘Dissection of the protein kinase cascade by which 847 nerve growth factor activates MAP kinases’, Nature, 353, pp. 170–173. 848 849 Haq, T., Richards, M. W., Burgess, S. G., Gallego, P., Yeoh, S., O’Regan, L., Reverter, 850 D., Roig, J., Fry, A. M. and Bayliss, R. (2015) ‘Mechanistic basis of Nek7 activation 851 through Nek9 binding and induced dimerization’, Nat. Commun., 6(8771), pp. 1–12. 852 853 Harrington, K. M. and Clevenger, C. V (2016) ‘Identification of NEK3 Kinase Threonine 854 165 as a Novel Regulatory Phosphorylation Site That Modulates Focal Adhesion 855 Remodeling Necessary for Breast Cancer Cell Migration’, J. Biol. Chem. , 291(41), 856 pp. 21388–21406. 857 858 Hornbeck, P. V., Zhang, B., Murray, B., Kornhauser, J. M., Latham, V. and Skrzypek, E. 859 (2015) ‘PhosphoSitePlus, 2014: Mutations, PTMs and recalibrations’, Nucleic Acids 860 Res., 43(D1), pp. D513–D520. 861 862 Hutti, J. E., Jarrell, E. T., Chang, J. D., Abbott, D. W., Storz, P., Toker, A., Cantley, L. C. 863 and Turk, B. E. (2004) ‘A rapid method for determining protein kinase 864 phosphorylation specificity’, Nat. Methods, 1(1), pp. 27–29. 865 866 Joughin, B. A., Chengcheng, L., Lauffenburger, D. A., Hogue, C. W. V. and Yaffe, M. B. 867 (2012) ‘Protein kinases display minimal interpositional dependence on substrate 868 sequence: potential implications for the evolution of signalling networks’, Philos. 869 Trans. R. Soc. B Biol. Sci. Royal Society, 367(1602), pp. 2574–2583. 870 871 Joukov, V. and De Nicolo, A. (2018) ‘Aurora-PLK1 cascades as key signaling modules in 872 the regulation of mitosis’, Sci. Signal., 11(45), p. eaar4195. 873 874 Kang, S. A., Pacold, M. E., Cervantes, C. L., Lim, D., Lou, H. J., Ottina, K., Gray, N. S., 875 Turk, B. E., Yaffe, M. B. and Sabatini, D. M. (2013) ‘mTORC1 Phosphorylation Sites 876 Encode Their Sensitivity to Starvation and Rapamycin’, Science (80-. )., 341(6144), 877 p. 1236566. 878 879 Kentrup, H., Becker, W., Heukelbach, J., Wilmes, A., Schürmann, A., Huppertz, C.,

34 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

880 Kainulainen, H. and Joost, H.-G. (1996) ‘Dyrk, a Dual Specificity Protein Kinase with 881 Unique Structural Features Whose Activity Is Dependent on Tyrosine Residues 882 between Subdomains VII and VIII’, J. Biol. Chem. , 271(7), pp. 3488–3495. 883 884 Kim, S., Kim, S. and Rhee, K. (2011) ‘NEK7 is essential for centriole duplication and 885 centrosomal accumulation of pericentriolar material proteins in interphase cells’, J. 886 Cell Sci., 124(22), pp. 3760–3770. 887 888 Kraft, C., Herzog, F., Gieffers, C., Mechtler, K., Hagting, A., Pines, J. and Peters, J. 889 (2003) ‘Mitotic regulation of the human anaphase-promoting complex by 890 phosphorylation’, EMBO J., 22(24), pp. 6598–6609. 891 892 Laurell, E., Beck, K., Krupina, K., Theerthagiri, G., Bodenmiller, B., Horvath, P., 893 Aebersold, R., Antonin, W. and Kutay, U. (2011) ‘Phosphorylation of Nup98 by 894 multiple kinases is crucial for NPC disassembly during mitotic entry’, Cell, 144(4), 895 pp. 539–550. 896 897 Letwin, K., Mizzen, L., Motro, B., Ben-david, Y., Bernstein, A., Pawson, T., Letwinl, K., 898 Mizzen, L., Motro, B., Ben-david, Y., Bernstein, A., Pawsonl, T., Letwin, K., Mizzen, 899 L., Motro, B., Ben-david, Y., Bernstein, A. and Pawson, T. (1992) ‘A mammalian 900 dual specificity protein kinase, Nek1, is related to the NIMA cell cycle regulator and 901 highly expressed in meiotic germ cells’, EMBO J., 11(10), pp. 3521–3531. 902 903 Lindberg, R. A., Fischer, W. H. and Hunter, T. (1993) ‘Characterization of a human 904 protein threonine kinase isolated by screening an expression library with antibodies 905 to phosphotyrosine’, Oncogene, 8(2), pp. 351–359. 906 907 Linding, R. et al. (2007) ‘Systematic Discovery of In Vivo Phosphorylation Networks’, 908 Cell, 129(7), pp. 1415–1426. 909 910 Liu, F., Stanton, J. J., Wu, Z. and Piwnica-Worms, H. (1997) ‘The human Myt1 kinase 911 preferentially phosphorylates Cdc2 on threonine 14 and localizes to the 912 endoplasmic reticulum and Golgi complex.’, Mol. Cell. Biol., 17(2), pp. 571–583. 913 914 Lizcano, J. M., Deak, M., Morrice, N., Kieloch, A., Hastie, C. J., Dong, L., Schutkowski, 915 M., Reimer, U. and Alessi, D. R. (2002) ‘Molecular Basis for the Substrate 916 Specificity of NIMA-related Kinase-6 (NEK6): evidence that Nek6 does not 917 phosphorylate the hydrophobic motif of ribosomal S6 protein kinase and serum- and 918 glucocorticoid-induced protein kinase in vivo’, J. Biol. Chem., 277(31), pp. 27839– 919 27849. 920 921 Lochhead, P. A., Sibbet, G., Morrice, N. and Cleghon, V. (2005) ‘Activation-Loop 922 Autophosphorylation Is Mediated by a Novel Transitional Intermediate Form of 923 DYRKs’, Cell, 121(6), pp. 925–936. 924 925 Marin, O., Meggio, F., Sarno, S., Cesaro, L., Pagano, M. A. and Pinna, L. A. (1999) 926 ‘Tyrosine versus serine/threonine phosphorylation by protein kinase casein kinase- 927 2. A study with peptide substrates derived from immunophilin Fpr3’, J. Biol. Chem., 928 274(41), pp. 29260–29265. 929 930 Melixetian, M., Klein, D. K., Sørensen, C. S. and Helin, K. (2009) ‘NEK11 regulates

35 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

931 CDC25A degradation and the IR-induced G2/M checkpoint.’, Nat. Cell Biol., 11(10), 932 pp. 1247–1253. 933 934 Melo-Hanchuk, T. D., Slepicka, P. F., Meirelles, G. V., Basei, F. L., Lovato, D. V., 935 Granato, D. C., Pauletti, B. A., Domingues, R. R., Leme, A. F. P., Pelegrini, A. L., 936 Lenz, G., Knapp, S., Elkins, J. M. and Kobarg, J. (2017) ‘NEK1 kinase domain 937 structure and its dynamic protein interactome after exposure to Cisplatin’, Sci. Rep., 938 7(1), p. Art. 5445. 939 940 Miller, C. J. and Turk, B. E. (2018) ‘Homing in: Mechanisms of Substrate Targeting by 941 Protein Kinases’, Trends Biochem. Sci., 43(5), pp. 380–394. 942 943 Moniz, L. S., Dutt, P., Haider, N. and Stambolic, V. (2011) ‘Nek family of kinases in cell 944 cycle, checkpoint control and cancer’, Cell Div., 6(18). 945 946 Moniz, L. S. and Stambolic, V. (2011) ‘Nek10 Mediates G2/M Cell Cycle Arrest and MEK 947 Autoactivation in Response to UV Irradiation’, Mol. Cell. Biol. 948 949 O’Regan, L. and Fry, A. M. (2009) ‘The Nek6 and Nek7 Protein Kinases Are Required 950 for Robust Mitotic Spindle Formation and Cytokinesis’, Mol. Cell. Biol., 29(14), pp. 951 3975–3990. 952 953 Obenauer, J. C., Cantley, L. C. and Yaffe, M. B. (2003) ‘Scansite 2.0: Proteome-wide 954 prediction of cell signaling interactions using short sequence motifs.’, Nucleic Acids 955 Res., 31(13), pp. 3635–3641. 956 957 Prosser, S. L., Sahota, N. K., Pelletier, L., Morrison, C. G. and Fry, A. M. (2015) ‘Nek5 958 promotes centrosome integrity in interphase and loss of centrosome cohesion in 959 mitosis’, J. Cell Biol., 209(3), pp. 339–348. 960 961 Quarmby, L. M. and Mahjoub, M. R. (2005) ‘Caught Nek-ing: cilia and centrioles’, J. Cell 962 Sci., 118(22), pp. 5161–5169. 963 964 Richards, M. W., O’Regan, L., Mas-Droux, C., Blot, J. M. Y., Cheung, J., Hoelder, S., 965 Fry, A. M. and Bayliss, R. (2009) ‘An Autoinhibitory Tyrosine Motif in the Cell-Cycle- 966 Regulated Nek7 Kinase Is Released through Binding of Nek9’, Mol. Cell, 36(4), pp. 967 560–570. 968 969 Roig, J., Mikhailov, A., Belham, C. and Avruch, J. (2002) ‘Nercc1, a mammalian NIMA- 970 family kinase, binds the Ran GTPase and regulates mitotic progression’, Genes 971 Dev., 16(13), pp. 1640–1658. 972 973 Salem, H., Rachmin, I., Yissachar, N., Cohen, S., Amiel, a, Haffner, R., Lavi, L. and 974 Motro, B. (2010) ‘Nek7 kinase targeting leads to early mortality, cytokinesis 975 disturbance and polyploidy.’, Oncogene, 29(28), pp. 4046–57. 976 977 Santamaria, A., Wang, B., Elowe, S., Malik, R., Zhang, F., Bauer, M., Schmidt, A., Silljé, 978 H. H. W., Körner, R. and Nigg, E. A. (2011) ‘The Plk1-dependent Phosphoproteome 979 of the Early Mitotic Spindle’, Mol. Cell. Proteomics, 10(1), p. M110.00457. 980 981 Sdelci, S., Schütz, M., Pinyol, R., Bertran, M. T., Regué, L., Caelles, C., Vernos, I. and

36 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

982 Roig, J. (2012) ‘Nek9 phosphorylation of NEDD1/GCP-WD contributes to Plk1 983 control of γ-tubulin recruitment to the mitotic centrosome’, Curr. Biol., 22(16), pp. 984 1516–1523. 985 986 Soundararajan, M., Roos, A. K., Savitsky, P., Filippakopoulos, P., Kettenbach, A. N., 987 Olsen, J. V., Gerber, S. A., Eswaran, J., Knapp, S. and Elkins, J. M. (2013) 988 ‘Structures of down syndrome kinases, DYRKs, reveal mechanisms of kinase 989 activation and substrate recognition’, Structure, 21(6), pp. 986–996. 990 991 Spies, J., Waizenegger, A., Barton, O., Sürder, M., Wright, W. D., Heyer, W. D. and 992 Löbrich, M. (2016) ‘Nek1 Regulates Rad54 to Orchestrate Homologous 993 Recombination and Replication Fork Stability’, Mol. Cell, 62(6), pp. 903–917. 994 995 Taylor, S. S., Radzio-Andzelm, E. and Hunter, T. (1995) ‘How do protein kinases 996 discriminate between serine/threonine and tyrosine? Structural insights from the 997 insulin receptor protein-tyrosine kinase.’, FASEB J., 9(13), pp. 1255–1266. 998 999 Thomsen, M. C. F. and Nielsen, M. (2012) ‘Seq2Logo: A method for construction and 1000 visualization of amino acid binding motifs and sequence profiles including sequence 1001 weighting, pseudo counts and two-sided representation of amino acid enrichment 1002 and depletion’, Nucleic Acids Res., 40(W1), pp. W281–W287. 1003 1004 Toshima, J., Tanaka, T. and Mizuno, K. (1999) ‘Dual Specificity Protein Kinase Activity 1005 of Testis-specific Protein Kinase 1 and Its Regulation by Autophosphorylation of 1006 Serine-215 within the Activation Loop’, J. Biol. Chem. , 274(17), pp. 12171–12176. 1007 1008 Wang, Q. M., Fiol, C. J., DePaoli-Roach, A. A. and Roach, P. J. (1994) ‘Glycogen 1009 synthase kinase-3 beta is a dual specificity kinase differentially regulated by tyrosine 1010 and serine/threonine phosphorylation.’, J. Biol. Chem. , 269(20), pp. 14566–14574. 1011 1012 Wilson, L. K., Dhillon, N., Thorner, J. and Martin, G. S. (1997) ‘Casein Kinase II 1013 Catalyzes Tyrosine Phosphorylation of the Yeast Nucleolar Immunophilin Fpr3’, J. 1014 Biol. Chem. , 272(20), pp. 12961–12967. 1015 1016 Yin, M. J., Shao, L., Voehringer, D., Smeal, T. and Jallal, B. (2003) ‘The 1017 Serine/Threonine Kinase Nek6 Is Required for Cell Cycle Progression through 1018 Mitosis’, J. Biol. Chem., 278(52), pp. 52454–52460. 1019 1020 Yissachar, N., Salem, H., Tennenbaum, T. and Motro, B. (2006) ‘Nek7 kinase is 1021 enriched at the centrosome, and is required for proper spindle assembly and mitotic 1022 progression’, FEBS Lett., 580, pp. 6489–6495. 1023 1024

37 bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. Figure 1 A C Nek1 Nek3 Nek4 Nek5 -5 -4 -3 -2 -1 0 +1+2+3+4 -5 -4 -3 -2 -1 0 +1+2+3+4 -5 -4 -3 -2 -1 0 +1+2+3+4 -5 -4 -3 -2 -1 0 +1+2+3+4 P P P P TK G G G G A A A A TKL C C C C S S S S T T T T V V V V I I I I CMGC L L L L STE M M M M F F F F Y Y Y Y W W W W H H H H K K K K CK R R R R Q Q Q Q N N N N AGC D D D D E E E E pT pT pT pT CAMK pY pY pY pY Y W M R I R Y R W M M F F M L R E MRI Nek9 Nek6 M Y L L I K FMM QK Nek8 Nek7 MF LRL M F F RW R pT F WL MF KWF V V MK F F Y RI KRFM WK HR F Y norm. intensity) N F W IY K W Nek2 RK W V LF L MD II P 2 V YpT L Nek11 Nek10 L GL WHKpTF CWLMY WHMGW KMR W G LK F K N K C QpTL pT Y H V pTL pT F V Y I L L Nek3 Y VRR LNF N R YY M F C V Favored residues Q R pT E L Y L Nek4 N pT V R W V N H L N QpT HW W H QN W YY W K N H FMCN Y GGpT QIM MM W WMRpT Q Y RL Q L H Q GA Q pY EW Y N TQ K F VHM E GH Y (Log T T G pYH R M M W K G pT M Y R R S Y C A ST L T Q FG N R CT I GH T SL L H T Nek5 T QT T G GY S G AT H A F S T T L TMT D L I N K G TE G W C M T I R T T R E T Y S pT G S TM T I M T S S S N C Q L Y I M H L pT S Q F N T T H T pT S Q D Y G S H S WS V pYS FTW I A T K T L T L S S M Nek1 T

B Nek6 Nek7 Nek8 Nek9 -5 -4 -3 -2 -1 0 +1+2+3+4 -5 -4 -3 -2 -1 0 +1+2+3+4 -5 -4 -3 -2 -1 0 +1+2+3+4 -5 -4 -3 -2 -1 0 +1+2+3+4 P P P P G G G G A A A A C C C C S S S S Nek1/8 T T T T Nek2/5 V V V V Nek4 I I I I L L L L M M M M Nek1 F F F F Nek6/7/9 Y Y Y Y Nek10/11 W W W W H H H H Nek8 K K K K R R R R Q Q Q Q N N N N Nek3 D D D D E E E E pT pT pT pT pY pY pY pY

Nek6/7 LDpY Y pY Y pY M N RR E LDE MW R DM Y D WNQ WN R Q F WPLM H KK Nek6/7/9 CD pY G F LY YY Q N WWF W F W FED W W R C FW DWGN M HW norm. intensity) N K NF FC K ER C H Y D H F H WK A H pT 2 N M DDMN M R D Y EC E H P WW H W D CD QF R N MpY pY P H C G HR W G pY pY MYW N N GW GR P C K G H GN AR EWpY F pY M MP H LF QN I K MY K Favored residues H CG CF W Y PYD RW pT G Y R PpTMHR QC YLG CWY W I C G H F D W W EPP G pY L V E KA Y Q E M N T R P T MG CT RF E V L pT G K A N T T pTW QS Q AF H G R HH I F K GA S C TYA pT N F QC Y G (Log pT T A N pY Q T G A QS E GGL S L T T N S T Y S RYTY H S E pY S T pY C Q T T S pT S L D S S S G T F M G S Q T S T pYpT HQ S I C pY MCS G S CCAYS HS M pTNK HTY C SPA C C M S HS F C S S C H TH P W SS

Fig. 1. Phosphorylation-site motifs for the human NimA-related kinases. (A) Dendrogram showing the phylogenetic relationship between all human kinases. The branch containing the Nek-family is magnified in the inset. Adapted from Man- ning et al., 2002. (B) Cartoon depicting the identified and suggested functions for the Nek kinases in interphase cells (top left panel) and cells in different stages of mitosis (other three panels). See introduction for details. (C) Phosphorylation-site motifs of the Nek-family members were determined by OPLS. Shown are representative OPLS dot blots, and sequence logos of the favored residues quantified over multiple assays (n≥2).. bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. Figure 2 A B C D Nek3 phosphorylation Nek3 phosphorylation Nek7 phosphorylation 13458-tide RFFLRTMRFW of peptides of peptides of peptides 13458-tide -3D RFDLRTMRFW 20000 1.2 1.2 ) 13458-tide ) e T e 13458-tide -2D RFFDR MRFW M t t 1 13458-tide -3D ) P

a 1.0 a 1.0 1 r r

C

679-tide DDLDpYSYNWW ( 15000 13458-tide -2D t o on 0.8 on 0.8 e t o i S i

679-tide -3I I on t DD DpY YNWW t i e t a a l l

679-tide -2R DDLRpYSYNWW a l y 10000 0.6 y 0.6 ** de s y i t de s 697-tide -1I DDLD I SYNWW i 0.4 t 0.4 pho r hpo r Nek8-tide WWMWQTMRFW pho r 5000 ** 0.2 679 - 0.2 ( Nek8-tide -4V WVMWQTMRFW 13458 - *** ( Pho s Pho s *** Nek8-poor HYGGKTHVII Pho s 0 0.0 0.0 Nek8-poor -4W HWGGKTHVII 0 5 10 15 20 Time (min) 679-tide 13458-tide 679-tide679-tide -3I -2R E F 13458-tide13458-tide -3D -2D Nek6 phosphorylation Nek8 phosphorylation of peptides of peptides 1.2 1.2 e e ) * t t ) 1 a 1.0 a 1.0

1 r r

t o on 0.8 on 0.8 t o i i e t t e a a l * l y 0.6 y 0.6 de s i de s i t t 0.4 - 0.4 8 hpo r hpo r * k e

679 - 0.2 0.2 ( N ( Pho s Pho s 0.0 0.0

-tide 8 679-tide Nek Nek8-poor 679-tide -1I Nek8-tide -4V Nek8-poor -4W

Fig.2 Validation of the phosphorylation-site motifs for the Nek kinases. (A) Peptides used in the studies depicted in panels B-F. Small font S or T indicates the phospho-acceptor site, pY indicates a phosphorylated tyrosine residue. (B) Purified Nek3 was incubated with the indicated peptide substrates and radiolabeled ATP. Peptide phosphorylation was quantified by scintillation counting (Mean ± SEM, n=3; CPM=counts per minute). (C) The data depicted in panel (B) were analyzed by linear regression and the slopes of the regression curves were calculated to determine the phosphorylation rate (Mean ± SEM, n=3, **P<0.005, ***P<0.0005). (D) As in panel (C), but for Nek7 (Mean ± SEM, n=3, **P<0.005, ***P<0.0005). (E) As in panel (C), but for Nek6 (Mean ± SEM, n=3, *P<0.05). (F) As in panel (C), but for Nek8 (Mean ± SEM, n=3, *P<0.05). bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. Figure 3 A B C D Nek10 Nek10 phosphorylation Nek10 phosphorylation -5 -4 -3 -2 -1 0 +1+2+3+4 P of peptides of peptides G

A ) 6000 Y KVMMKyIFNL 1.2 C M e

S t P S KVMMKSIFNL

a 1.0 C T r

( T

V T KVMMK IFNL )

I 1 4000 on 0.8

i on i L t t a

M l a l t o

F y 0.6 y

e ** Y W 2000 s 0.4 Y pho r

H ( K pho r R 0.2 ** Q Pho s N Pho s D 0 0.0 E 0 5 10 15 20 Y S T pT pY Time (min) HRD+2 APE-4

E F 100 80 E LF Y P+1 loop APE 60 V M Y AW -8 -7 -6 -5 -4 -3 -2 -1 V P R I G Nek1 (163-173) CIGTPYYLSPE 40 S RD PC F L G Nek10 phosphorylation RIA P ET LN A -kinome (%) A Nek2 (176-186) FVGTPYYMSPE ALV IL QFA KR of peptides 20 Y RQ V M S VQ TPYCE Frequency in SW K TDQ KFN ECSD (162-172) D YH P

S T Nek3 YVGTPYYVSPE D K ST IT NRE V GHKAF TKS W IYN -Q V ILD 3.5 N E R MLRT C P LHM Q H N M HS GM -I G I F D -G C K C Y -CS -CF Y

Nek4 (166-176) LIGTPYYMSPE G e t Y ) 3.0 S 100 a T 1 (164-174) Nek5 CIGTPYYLSPE r **

Nek6 (207-217) LVGTPYYMSPE 80 2.5 on t o i t Nek7 (196-206) LVGTPYYMSPE e

a 2.0 60 l s Nek8 (163-173) VVGTPCYISPE y Y

L 1.5 Nek9 (211-212) LVGTPYYMSPE 40 I M o *

K t LR I pho r V 1.0 Nek10 (689-699) VVGTILYSCPE Q T -kinome (%) G Y 20 MFV ATW K I ST Y S

Frequency in A NM W RF ST L E P ( DS L KQPTW 0.5 Nek10 I693P VVGTPLYSCPE PIR Y EED FFA C DACLRAL V 0 LVR Pho s Nek10 T692P VVGPILYSCPE -8 -7 -6 -5 -4 -3 -2 -1 0.0 Position relative to APE WT I693P H Catalytic I Nek10 T657K PJ Nek10 phosphorylation K Nek10 T692P loop W phosphorylation of peptides phosphorylation +1+2+3+4 Nek1 (123-132) KILHRDIKSQ of peptides of peptides Nek2 (136-145) TVLHRDLKPA 1.2 8000 2.5 ) WT on Y )

(122-131) e Nek3 RVLHRDIKSK e M * t t WT on S 1

P a 1.0 (126-135) a r Nek4 HILHRDLKTQ r 2.0

C T692P on Y (

) 6000

(123-132) t o Nek5 KILHRDIKAQ 1 T692P on S on

0.8 on

e i i t t on

Nek6 (167-176) RVMHRDIKPA i 1.5 a a t l l t o a

y 0.6 y (156-165) RVMHRDIKPA l 4000 Nek7 de s e ns i y 1.0 Nek8 (123-132) LILHRDLKTQ s 0.4 S pho r pho r

( **

(171-180) pep t Nek9 GILHRDIKTL pho r 2000 0.5 Nek10 (650-659) RIVHRDLTPN 0.2 o ** N Pho s ( Pho s

Nek10 T657K RIVHRDLKPN 0.0 Pho s 0 0.0 Y S T 0 5 10 15 20 No Y S Time (min) peptide Fig.3 Nek10 is a dual-specificity kinase. (A) Representative blot showing the OPLS results for wild-type Nek10 on the serine-threonine peptide library. (B) Purified Nek10 was incubated with the indicated peptide substrates and radio-labeled ATP. Peptide phosphorylation was quantified by scintillation counting. (Mean ± SEM, n=3). (C) The data depicted in panel (B) were analyzed by linear regression and the slopes of the regression curves were calculated to determine the phosphory- lation rate (Mean ± SEM, n=3, **P<0.005). (D) Cartoon representation of the Akt/GSK3-peptide structure (PDB: 1O6L). The kinase domain, activation loop, and the P+1 loop are shown in gray, green and magenta respectively. The GSK3 peptide is shown in cyan, side chain shown is of the phospho-acceptor serine. (E) Amino acid sequence alignment of the P+1 loop and APE motif of Nek1 to Nek10, and including the Nek10 I693P and T692P variants. (F) Logo depicting the frequency of amino acids in the P+1 loop of either all human serine/threonine (ST) kinases, but excluding the tyrosine kinase like group (upper panel), or all human tyrosine (Y) kinases (lower panel). (G) As in panel (C), but for Nek10 WT and the I693P variant (Mean ± SEM, n=3, *P<0.05, **P<0.005). (H) Amino acid sequence alignment of the catalytic loop of Nek1 to Nek10, including the Nek10 T657K variant. (I) As in panel (C), but for the Nek10 T657K variant (Mean ± SEM, n=3, **P<0.005). (J) As in panel (B), but for Nek10 WT and the T692P variant. (K) As in panel (C) but for the Nek10 T692P variant. In order to obtain accurate rate calculations, the experiment shown in panel J was repeated x3 with longer incubation times (Mean ± SEM, n=4, *P<0.05). The signal observed in the “no peptide” sample results from binding of auto-phosphorylated kinase to the phos- phocellulose paper. bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. Figure 4 A B C Nek10 I693P Nek10 WT on S-peptide on Y-peptide -5 -4 -3 -2 -1 0 +1+2+3+4 -5 -4 -3 -2 -1 0 +1+2+3+4 -5 -4 -3 -2 -1 0 +1+2+3+4 -5 -4 -3 -2 -1 0 +1+2+3+4 P P G Nek10 phosphorylation G A R of peptides A 2 W C 2 Y 1.2 C ) S Y S N 1 e Y T T R t Q M R 3x Log Y Favored N F H V M a 1.0 V W V RW W H r W R 3x Log Q HM Favored S HWR t o H W F (S)-tide NWMNR MRGF I W HM I L EDM SRL G H LW L S R T Q G RM P F D F (norm. intensity) SL e FI KKQQ FLFpYR N K Q H S E pY IQ pY Y E M N IY L Y Y Y E Y Y G F V MR H V L A E KI Y pY K M I N E N I D E S L S F G V V L G C C Q N L P V M K N K E D G L pT G K H TP SA I T VN S E on T E Y pT C V D 0.8 T P W S C T F H L D M CA A T i -3D NWDNR MRGF A pT

(norm. intensity) F G W I F pYPC pY F AN pY M pT SC M G G T K t G V D A P DY G A F N R P W FF CpT CAV PDV pT H pT FK A HT W pT WT a F S R F C I WN Y T C G R

Q GL l GTTS GSMH S P IP HM V T L N

S S de s N S L -1D AS D R V L NWMN MRGF HYS WI H AW QP E RW A A N PpYC TIAK i NpY C MM pYD Y C Y C Y pY D y 0.6 pTRC E VC QA Q DE QK E N pY LCVI pTVQpY S T W W K I pYY K )- t YpTEpTG K A A H LQH L PQ H K pY T S D I E D 0.4 R pY 2 K I pTPI hpo r K T K pT pYpTV R 2 V pT R 10 ( K

D Log Q P E k 0.2 Q P e N Log D D N

Pho s **** **** Disfavored G D pT N D (

Disfavored D E 0.0 E (norm. intensity) pY pT (norm. intensity) E pT pY P P pY pT

Nek10(S)-tide Nek10(S)-tideNek10(S)-tide -3D -1D

Fig.4. Phosphorylation-site motifs for Nek10 on tyrosine and serine target sites. (A) The phosphorylation-site motif of Nek10 I693P was determined by OPLS. Shown are a representative blot, and a sequence logo of data quantified over two independent OPLS-experiments, showing both the favored and disfavored amino acids. (B) Purified Nek10 was incubated with the indicated peptide substrates and radiolabeled ATP. Peptide phosphorylation was quantified by scintillation counting. The phosphorylation rate was determined by linear regression (Mean ± SEM, n=3, ****P<0.00005). (C) As in panel (A), but now for WT Nek10, and using a peptide library with a fixed tyrosine as the phosphoacceptor site (n=3). Note that in this case the 0 position column contains no peptide libraries. bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. Figure 5 A B C

FLAG-Nek10: KD I693P(Ser)T692P(Tyr) HA-Nek10 KD: + + + I693P T692P (P)FLAG-Nek10 32P dark Nek10: WT KD I693P(Ser)T692P(Tyr) WT KD (Ser) (Tyr) α-FLAG (P)FLAG-Nek10 32 Doxy.: - + - + - + - + (P)Nek10 32P dark P light IP α-FLAG FLAG-Nek10 α-FLAG FLAG-Nek10 (P)Nek10 32 P light Actin α-Actin Nek10 Coomassie (P)HA-Nek10 KD 32P dark α-HA (P)HA-Nek10 KD 32 P light IP HA-Nek10 KD α-HA

D WT KD I693P(Ser)T692P(Tyr) E Doxy.: - + + + + EV WT KD Y172FY583FY644FY679FY815FY1120F1158F (pY)FLAG-Nek10 α-pY (pY)FLAG-Nek10 α-pY α-FLAG α-FLAG IP FLAG-Nek10 α-FLAG IP FLAG-Nek10 α-FLAG

(pY)FLAG-Nek10 α-pY FLAG-Nek10 α-FLAG TCL FLAG-Nek10 α-FLAG TCL Actin α-Actin Actin α-Actin F G H Nek10 peptide FLAG-Nek10 KD: + - - phosphorylation FLAG-Nek10 KD: + - - FLAG-Nek10 T692P(Tyr): - + + 1.2 FLAG-Nek10 I693P(Ser): - + +

HA-Nek10 KD: + + - e HA-Nek10 KD: + + - t

a 1.0

HA-Nek10 KD Y644F: - - + r WMNRSMRGF HA-Nek10 KD SSSS/AAAA: - - + ) 1

32 32 (P)FLAG-Nek10 on 0.8 (P)FLAG-Nek10 P α-FLAG i P α-FLAG t a t o IP l IP e FLAG-Nek10 α-FLAG y 0.6 FLAG-Nek10 α-FLAG s

T 0.4 hpo r

32 W 32 (P)HA-Nek10 KD P α-HA ( (P)HA-Nek10 KD P α-HA 0.2 IP IP HA-Nek10 KD α-HA Pho s *** HA-Nek10 KD α-HA 0.0 T

10 W Nek Nek10 Y644F Fig.5 Nek10 autophosphorylates on both tyrosine and serine residues (A) Purified wild-type (WT), kinase-dead (KD), serine-specific (I693P(Ser)), or tyrosine-specific (T692P(Tyr)) Nek10 was incubated in vitro with radiolabeled ATP. Total Nek10 levels were analyzed by coomassie-staining, while phosphorylated (P) Nek10 was detected by phosphorimaging. 32P dark and 32P light respectively indicate a longer and shorter exposure of the same gel. (B) In vitro kinase assay with indicat- ed FLAG-tagged Nek10 variants as the kinase, and HA-tagged Nek10 KD as the substrate. Kinase-proficient and kinase-dead Nek10 were separated by anti-FLAG and anti-HA IP respectively, and analyzed by immunoblotting and phos- phorimaging. (C) Nek10 KO U-2 OS cells were reconstituted with doxycycline (Doxy) inducible FLAG-Nek10 variants and treated with 2 µg/ml doxycycline for 7 days, or left untreated. FLAG-Nek10 levels were analyzed by immunoblotting. (D) Cells described in panel (C) were treated with doxycycline for 2 days, and FLAG-Nek10 was isolated by α-FLAG IP, followed by immunoblotting to detect total Nek10 (α-FLAG) and tyrosine-phosphorylated Nek10 (α-pY) (TCL=total cell lysate). (E) U-2 OS cells were transfected with an empty vector control (EV) or the indicated FLAG-Nek10 variants. For WT and KD Nek10, less plasmid was transfected than for the Nek10 Y-to-F mutants. FLAG-Nek10 was isolated by α-FLAG IP and analyzed by immunoblotting. (F) As in panel (B), now using FLAG-tagged Nek10, either KD or tyrosine-specific (T692P(Tyr)), as the kinase, and HA-tagged Nek10 KD, either lacking additional mutations or with a Y644F mutation, as the substrate. (G) Purified Nek10 WT or the Y644F mutant was incubated with a peptide substrate and radiolabeled ATP. Peptide phosphorylation was quantified by scintillation counting. The phosphorylation rate was determined by linear regres- sion (Mean ± SEM, n=3, ***P<0.0005). Note that the autophosphorylation-defective Y644F mutant has dramatically reduced activity. (H) As in panel (B), now using FLAG-tagged Nek10, either KD or serine-specific (I693P(Ser)), as the kinase, and HA-tagged Nek10 KD, either lacking additional mutations or with a S353A/S356A/S358A/S359A quadruple mutation (SSSS/AAAA), as the substrate.. bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. Figure 6 A B C Clustering based Pearson correlation Clustering based % sequence identity on specificity 0 0.5 1 on sequence 0 100 (S) Nek1 Nek4 Nek3 Nek5 Nek8 Nek2 Nek10 Nek6 Nek7 Nek9 Nek1 Nek5 Nek3 Nek4 Nek2 Nek9 Nek8 Nek6 Nek7 Nek10 Nek1 Nek4 Nek3 Nek5 Nek8 Nek2 Nek10(S) Nek6 Nek7 Nek9 Nek1 Nek1 0S Nek4 Nek5 0T Nek3 Nek3 0Y Nek5 Nek4 DFG+1: I I S I I L L L L L Nek8 Nek2 Nek2 Nek9 Nek10(S) Nek8 Log (Normalized intensity) Nek6 Nek6 2 Nek7 Nek7 Nek9 Nek10 -1.5 0 1.5

D E (S) (S) (S)

Clustering based Pearson correlation Nek1 Nek4 Nek3 Nek5 Nek8 Nek2 Nek10 Nek6 Nek7 Nek9 Nek1 Nek4 Nek3 Nek5 Nek8 Nek2 Nek10 Nek6 Nek7 Nek9 Nek1 Nek4 Nek3 Nek5 Nek8 Nek2 Nek10 Nek6 Nek7 Nek9 -1.5 -5P -4P -2P on specificity: 0 0.5 1 -5G -4G -2G No S vs T -5A -4A -2A -5C -4C -2C 0 -5S -4S -2S -5T -4T -2T Nek1 Nek4 Nek3 Nek5 Nek8 Nek2 Nek10(S) Nek6 Nek7 Nek9 -5V -4V -2V Nek1

(Normalized intensity) -5I -4I -2I Nek4 Group 1 2 -5L -4L -2L Nek3 1.5 Log -5M -4M -2M Nek5 Group 2 -5F -4F -2F Nek8 -5Y -4Y -2Y Nek2 Group 3 -5W -4W -2W Nek10(S) -5H -4H -2H Nek6 -5K -4K -2K Nek7 Group 4 -5R -4R -2R Nek9 -5Q -4Q -2Q -5N -4N -2N -5D -4D -2D -5E -4E -2E F -5pT -4pT -2pT kcal/(mol*e) -5pY -4pY -2pY

-10 0 10 G Nek3 phosphorylation Nek7 phosphorylation of peptides of peptides Nek1 Nek7 1.2 1.2 ) e e t t ) 1

a a 1.0 1.0 1 r r

t o on on 0.8 0.8 t o e i i t t e a a l l y y 0.6 0.6 de s i de s t i 0.4 t 0.4 pho r pho r

0.2 ** 679 - 0.2 ( 13458 - ( Pho s Pho s ** 0.0 0.0 13458- 679- 13458- 679- tide tide tide tide Fig.6 The Nek family diverges into four specificity-groups. (A) Similarity matrix and dendrogram depicting the Pearson correlation between the phosphorylation-site motifs of the Nek kinase family members. (B) Similarity matrix and dendrogram showing the percentage identity between the amino acid sequences of the Nek kinase domains. (C) Heatmap depicting the OPLS data for the phospho-acceptor site position. Measured intensities of individual dots on OPLS dot blots were normal- ized to the average of all values in the position (e.g. -1 position), averaged over multiple experiments, and log2-transformed (n≥2). Nek family members are ordered according to the clustering shown in panel (A). For each Nek the amino acid C-ter- minal to the DFG-motif (DFG+1) is depicted underneath the heatmap. (D) As in panel (A), but the phospho-acceptor speci- ficity data was excluded from the clustering analysis. (E) As in panel (C), but for the -5 (left panel), -4 (middle panel) and -2 (right panel) position to emphasize the distinct specificities of the Nek6/7/9 group (boxed sections). (F) Surface representa- tion of the substrate binding sites of Nek1 (PDB: 4APC) and Nek7 (PDB:5DE2). Surface is shaded according to electrostatic potential. The dotted yellow lines indicate specific basic patches in the Nek7 structure that potentially interact with acidic residues in the substrate. Peptide and ANP-Mn was modelled into the Nek1 and Nek7 structures by alignment with the Akt/GSK3-peptide structure (PDB: 1O6L; Fig 3D). (G) Purified Nek3 (left panel) or Nek7 (right panel) was incubated with peptide substrates and radio-labeled ATP. Peptide phosphorylation was quantified by scintillation counting. The phosphory- lation rate was determined by linear regression (Mean ± SEM, n=3, **P<0.005).). bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. Figure 7 A B Consensus optimal phosphorylation-site motif Nek1 Nek4 Nek3 Nek5 Nek8 Nek2 Nek10(S) Nek6 Nek7 Nek9 Plk1 Nek1 0 for Nek kinase groups and Plk1 Nek4 Nek3 P-5 P-4 P-3 P-2 P-1 P0 P+1 P+2 P+3 P+4 Nek5 Group 1 Nek1/3/4 [X] [W] [LMFW] [X] [R] [T] [Ø] [KR] [Ø] [X] Nek8 [X] [W] [LMFW] [X] [X] [T] [MF] [KR] [Ø] [X] Nek2 0.5 Group 2 Nek5/8 Nek10(S) Group 3 Nek2/10(S) [X] [W] [LMF] [X] [R] [S] [Ø] [R] [X] [X] Nek6 Group 4 Nek6/7/9 [D] [D] [LMF] [DEN] [pY] [S] [Ø] [X] [X] [X] Nek7 Nek9 Plk1 [X] [X] [LYFH] [DENY] [pY] [S] [Ø] [X] [X] [X] Plk1 Pearson correlation 1

C Predicted substrate overlap D Plk1 and Nek7 Scansite score comparison 2.5 # overlapping Poor targets

Plk1 Worst

Kinases target sites e r Plk1, Nek6 6439 2.0 Good Plk1 targets

Plk1, Nek7 7156 sc o

e Good Nek7 targets t

Plk1, Nek9 4274 i 1.5 Nek6 Plk1, Nek6, Nek7 4998 Good Plk1 and Nek7 Plk1, Nek6, Nek9 3315 1.0 Nek7 targets

Plk1, Nek7, Nek9 3171 Sc an s 7

Nek6, Nek7 11281 k

e 0.5

Nek6, Nek9 9848 N RacGAP1(S170) Nek6, Nek7, Nek9 6851 CDC27(S435) Nek9 Nek7, Nek9 8287 0.0 All 2801 Best 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Best Worst Plk1 Scansite score E F G Plk1 or Nek7 WT: + - + + Plk1 or Nek7 WT: + - + + Plk1 or Nek7 KD: - + - - Plk1 or Nek7 KD: - + - - CDC27 WT: - + + - RacGAP1 WT: - + + - CDC27 SS/AA: - - - + RacGAP1 S170A: - - - + Plk1 P (P)CDC27 32P (P)RacGAP1 32P P CDC27 α-HA Plk1 RacGAP1 α-HA Plk1 LDXSFXX P Mitotic Plk1 α-Plk1 Plk1 α-Plk1 Nek9 substrates

(P)CDC27 32P (P)RacGAP1 32P P P P CDC27 α-HA Nek7 RacGAP1 α-HA Nek7 Nek6 Nek7 P Nek7 α-FLAG Nek7 α-FLAG

Fig. 7 Plk1 and Nek6, 7 and 9 share a common phosphorylation-site motif. (A) Table depicting the consensus phos- phorylation-site motifs for the different Nek specificity groups, and for Plk1, with Ø indicating hydrophobic amino acids. The Plk1 motif was generated from data published in Alexander et al., 2011. (B) Similarity matrix and dendrogram depicting the Pearson correlation between the phosphorylation-site motifs of the Nek family members, and Plk1. (C) Sites in the Phospho- siteplus database were scored using the Scansite algorithm for their match to the phosphorylation-site motifs of Plk1, Nek6, Nek7 and Nek9. The best scoring 10% were plotted in the Venn diagram to indicate the fraction of good potential target sites shared between the kinases. (D) Sites with reduced phosphorylation upon Plk1 inhibition or knockdown, as reported by Santamaria et al. 2011, were scored for their match to the Plk1 or Nek7 phosphorylation-site motif using the Scansite algo- rithm. Note the lower the score, the better the phosphorylation site sequence matches the optimal motif, with a score of 0 indicating a perfect match. Dotted lines indicate the 25% cut-off for best scoring sites. (E) Purified wild-type (WT) or kinase-dead (KD) Plk1 or Nek7 was incubated with HA-tagged Cdc27 WT or S434A/ S435A double mutant (SS/AA), followed by phosphorimaging (32P) and immunoblotting for kinase and Cdc27 substrate levels. (F) As in panel (E), now using RacGAP1 WT or S170A mutant as substrate. (G) A model of the signaling interactions and common phosphorylation site motif shared between Plk1, Nek9, Nek6, Nek7 that result in potential phospho-motif amplification during mitosis. bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. Supplementary Figure 1

1a. Design 198 peptide libraries... 1b. ... and design three 3. Add kinase and [γ-33]-ATP phospho-acceptor peptide libraries Analyze phoshorylation of peptide -5 -4 -3 -2 -1 0 +1 +2 +3 +4 ~Z-X-X-X-X-S/T-X-X-X-X~(biotin) -5 -4 -3 -2 -1 0 +1+2+3+4 -5 -4 -3 -2 -1 0 +1+2+3+4 ~X-Z-X-X-X-S/T-X-X-X-X~(biotin) ~X-X-X-X-X-S-X-X-X-X~(biotin) P G ~X-X-X-X-X-T-X-X-X-X~(biotin) A C ~X-X-X-X-X-Y-X-X-X-X~(biotin) S ~X-X-X-X-X- -X-X-Z-X~(biotin) T S/T X = Mix of all amino acids V (no C, S, T or Y) I ~X-X-X-X-X-S/T-X-X-X-Z~(biotin) L M Z = fixed amino acid F X = Mix of all amino acids (no C, S, or T) Y W H K R Q 2. Array in multiwell plate N D E -5 -4 -3 -2 -1 0 +1+2+3+4 pT P ~P-X-X-X-X-S/T-X-X-X-X~(biotin) pY Dot blot G A C S ~X-X-X-X-X-S-X-X-X-X~(biotin) T V RK I MS C SQW R A CTMNH YL QF E DWD K QG K E L T A P H L M C CEF DS Q WC RMW I S I P P C HC F HS G T HA MV F F E T L R G PV Q D D G N FP P Q KL VG P A CI I Y TYN L ~X-M-X-X-X- -X-X-X-X~(biotin) DE SLPT S/T A N M KD K VW R AK E Q KSFE R N H F Y Y V W H Logo K ~X-X-X-X-X-S/T-X-X-K-X~(biotin) P R Q N D E pT pY

Supplementary figure 1. Experimental lay-out of an OPLS-experiment. Experimental lay-out of an OPLS-experiment. Biotinylated peptide libraries are incubated in vitro with an isolated kinase and radiolabeled ATP, in independent reactions in a multiwell plate. The peptide libraries are then transferred to a streptavidin membrane, and the level of phosphorylation quantified by measuring the intensity of radiolabel incorporation using a phosphorimager. The example shows a basophilic kinase with a preference for basic residues on the -2 and -3 positions, a preference for a serine as the phospho-acceptor site, and a strong selection against a proline in the +1 position.. bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. Supplementary Figure 2 A Nek1 Nek1 KD Nek3 Nek3 KD Nek4 KD Nek4 Position: -5 -4 -3 -2 -1 0 +1+2+3 +4 -5 -4 -3 -2 -1 0 +1+2+3 +4 Position: -5 -4 -3 -2 -1 0 +1+2+3 +4 -5 -4 -3 -2 -1 0 +1+2+3 +4 Position: -5 -4 -3 -2 -1 0 +1+2+3 +4 -5 -4 -3 -2 -1 0 +1+2+3 +4 P P P G G G A A A C C C S S S T T T V V V I I I L L L M M M F F F Y Y Y W W W H H H K K K R R R Q Q Q N N N D D D E E E pT pT pT pY pY pY

Nek5 Nek5 KD Nek6 Nek6 KD Nek7 Nek7 KD Position: -5 -4 -3 -2 -1 0 +1+2+3 +4 -5 -4 -3 -2 -1 0 +1+2+3 +4 Position: -5 -4 -3 -2 -1 0 +1+2+3 +4 -5 -4 -3 -2 -1 0 +1+2+3 +4 Position: -5 -4 -3 -2 -1 0 +1+2+3 +4 -5 -4 -3 -2 -1 0 +1+2+3 +4 P P P G G G A A A C C C S S S T T T V V V I I I L L L M M M F F F Y Y Y W W W H H H K K K R R R Q Q Q N N N D D D E E E pT pT pT pY pY pY

Nek8 Nek8 KD Nek9 Nek9 KD B Position: -5 -4 -3 -2 -1 0 +1+2+3 +4 -5 -4 -3 -2 -1 0 +1+2+3 +4 Position: -5 -4 -3 -2 -1 0 +1+2+3 +4 -5 -4 -3 -2 -1 0 +1+2+3 +4 P P Nek2 phosphorylation G G A A of peptides C C

) 300 S S S T T M V V P T C I I ( L L 200 on

M M i t

F F a l

Y Y y W W H H 100

K K pho r R R Q Q

N N Pho s D D 0 E E 0 15 30 pT pT pY pY Time (min)

Supplementary figure 2. OPLS-results for kinase-dead Nek mutants. (A) Representative OPLS-blots of every Nek kinase together with their kinase-dead (KD) mutants on the same blot. (B) Purified Nek2 was incubated in vitro with radiola- beled ATP and a degenerate peptide substrate containing either a serine (S) or threonine (T) phospho-acceptor site, as described in supplementary figure 1. Peptide phosphorylation was quantified by scintillation counting (CPM= counts per minute). bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. Supplementary Figure 3

Nek1 Nek2 Nek3 Nek4

R Y W M Y R F L M F M I F Y FR F F W IYI M W V Y MW R M M CWLMH WH G Y LF F NVR N V R M pT R L L YY W K N MMQW GA Y N TQ K K GA T pT R C S L L T L T H A MH L A G S K Q T T pTT S H W S I F S S S S L M F V Y I V S T I F A L C A I F K T pY VA pT M E T M L P L V R GpT R R I R I V G H C W KWF H A V KpT pT NI R MK Q pY F N R pY C QI V R RK F F EE C RF K GL KVF P H Q K M L WHpT V E pY C E KWW QpTLpTLpT pTY Y QG Q E LF NY Q VRR M I pY CD R L YpT N H L H QNIW A DpYN D FKM K GGpT Q M A pT EW I T G Y M T W LQ pT M R S Y V T SLT R Y S G L pY Y T E E G Y R R I Q M H W W F T N M Y T GpT Favored D F D N A pY I S R pT A R A K C W A H T WAK L N QpT H pY Y K E H H Q P A N I W Y C K K Y SS F Y MpT K I W R Q F W Q pT N FY M C pT H K C FF R F V M S G pT E M R F G T pT T CT Q I V QC A C M RT N G S KR T TM C pTN Y C T L D T I N T T QT S N T VF S H WSF HLF C S P A N pY L R S G VpT G S L H C S N C DY H RL W T I Y G C MT Y W FG WT Q V G AT KT H P A AS PGS WV S C pY S V A A S K K L pY Q KYH H E N N Q H N Q K H IT Y C I C V A I K pY pY W L A M CE pY D E YG L M VQGG N H V T PV D S N N QpY S H P F PpT E H H pY CV W pT M I W H A K L L P E VP D DC pY IpY S Q TLY DpY P S S V Q E Q VC C pY NP H KpT Y F pY pY V ERN EM P C GCHA E Q KpT D QN K P A D D W pY G D K W D Q S N S P pYC Q P I E S IGR W GLpT PK IA AA pY P P Q EQEG G N G PVV GD Y VPpY E A S pY G pYW pY CpY I Y II D P D A IEE pYAC pYH I pYP VpY pTpY EEA NPDE D CD K N D I FN (norm. intensity) P VD 2 DD E E P D IPpT D P E EE log pT pT EpT P DPD EGP Disfavored D P D P D -5 -4 -3 -2P-1 0 P+1 +2 +3 +4 -5 -4 -3 -2 -1 0 +1 +2 +3 +4 -5 -4 -3 -2 -1 0 D+1 +2 +3 +4 -5 -4 -3 -2 -1 0 D+1 +2 +3 +4 Nek5 Nek6 Nek7 Nek8

W pY pY E R L pY Y M FM MKI D Y LDE MWQ R WLM Q W E FR LY FYY H MF D NQ WN FED W W R C FW R F Y D F FC WK K MD II P C FpY FG DDWMNF MHN R D CDH F R W GL K W W W HR W GW FpY pY QMYWP N K C L W W E K pY M M V K W MY R L L N C pYW I YK PYD W W pT C F CWY I C V Y D G pY L V E Y L H P YL TE R L RF L P G V E T N T M M N H W V H I F W C S HW N MG T QC E GL S LpY C G G T Y E R T N S C I F Y N H Y pT F pT K M N G MT H S L A S T N H C T T Q P S R H H L C Q H F L L Q A YS C H M N P H CD FQ S C S V C H pY E I ST D G Y C Q H Y G H pT E P pT N pYH R pY MG R G A R M Y G F N H KA N H T C pY Q GY T YT G T G P C I C D S T G AE pTT pY S F G T S H Q D T R SW W pY V E N H L A G C A pY C Y W N K W V V G C pT C pT F I K Y M pYL W C pY pT pY S V Q M H A H G QD C H C AA PpT QC Q Q pT R T A M L pY ML EF A KA E W W S T I R D N A NM A D R AQ AF YHA F pTN pT K T A D A pTG QI P N G R Y Y pY C pYG M S M A pY C S S S CpT S Q N HS T C PQ WT K H pT S S R Q G SpT T N pT V F R T M YT N L Y F P A F N LM A S S P F pY V V R K S A T SW I SG A V E pT DFK IDPD L Favored pT pY S Q V A F pY G pT SI S Y Y pY LL Q R pYY D CV R E N T I V H Y H L RVpT L M L EV A S A F G K PK A pTpY I K D V pTI NQNV M AL NPT pT E pT pT V R I Q PK K A pT pYI V I A R N A Q G AE T Q I F R DR IE pY E P K DM I V IK pT E QV pY S LK pT AF Y K ALT K G N N LVQK K VpY K IK pTP PPK GQ DKpT K Q V LM T D V G G IR VQE I K P I pTD IT V pTI HI P VP E E Q E I pY V pY P D VPP D R TI D V ED DEE VP EE KP EE E

(norm. intensity) Q 2 IRP R D log P Disfavored -5 -4 -3 -2 -1 0P+1 +2 +3 +4 -5 -4 -3 -2 -1 0 P+1 +2 +3 +4 -5 -4 -3 -2 -1 0 P+1 +2 +3 +4 -5 -4 -3 -2 -1 0 P+1 +2 +3 +4 Nek9 Nek10[S] Nek10[Y] Plk1

DpY R W pYpYH D FY W NV Y Y pT N W pT N Y AF I M pTW R WD EH F DIL E EpY Y R IMC N Y HC V E Q HYM Y R CpT W C EK L LD D L WT G Q WHD F A R N MT C N S F R H P F C S V FA G T T T R T MY S S A S K Y S H H L V M S DM F Q E W pY GT W P C L G M R I Q RW M Y W N C W R V C I K HR NF QN HWR H W QY Q M M V W H G Q N DM SM I E H LW L GP R Y H R S Q Q T F A P R F P M A D SLF pY K FI KQ Q H F R R L pY NSK pY Q D pY I K H N ERY Y E F Y Y G H L I F L Y I A A E M pY W M I pY M I N D E KN S H L G V I L E R K L F V Q SL GC C N L GQ V K G V M V Q GK K Q PED G M pT H V K T N A I PT L I N S Q N Y W V S E R Y pT C K N E T V D E D FW MH T P W G F W T A W F C L D E M H F G CAT A R G A pT A K I E WN FWFR PCAN pY R C H PDHV pY pY Q P pT F T W pT SC A H K H A G Y WT S D Q NR T I FT G L Y C C G G GLS MH pY G T T VS M V A H pY IT L D S AS S N S pY P S G N I G R L M W W L D G N HY P N W G H A N W QP P G G R ER A C V pY N CA pT H LF Q APpYC TIAK pT

Favored N R K C G Y R MM pT pT YA N H pY YS C G D F Q DC pT E KA Y pY pTP G T D pTWK K QS pTR E C GA C Q YT T G V E T Q pY S C S C pT S S CS T PF A pY C SW ST Y A H R P SV H W W F E K MYT Q E P MP E QK V K K CpY pT D V T V VAN N pY LC I AF SI pT pTV pY T NpY pT P A C VRQF H pYV KQ I pYY pT FF pTC WMM YpT K Q pT EpTG K A V YLDP LMLL LQH PQ QKTpYL PDP KL pY T A pT R E A I I D K M AQ NLE R E PIpY K D RVE IpT pT D LT D pYIV KV pYpTV T I pTI D pT K I II DpY P E P G (norm. intensity) VVI D D E 2 K D DpT G

log E IRP F pY P Disfavored P P E pT pT P -5 -4 -3 -2 -1 0 +1E+2 +3 +4 -5 -4 -3 -2 -1 0 P+1 +2 +3 +4 -5 -4 -3 -2 -1 0 +1 +2 +3 +4 -5 -4 -3 -2 -1 0 +1 +2 +3 +4 Supplementary figure 3. Phosphorylation site-motifs for the Nek kinases and Plk1. Complete sequence logos of the phosphorylation-site motifs for Nek1-Nek10, and for Plk1 emphasizing both positive and negative selection. Nek10[S] and Nek10[Y] indicate the phosphorylation-site motif of Nek10 on a serine-substrate library or a tyrosine-substrate library respectively. The phosphorylation-site motifs for Nek1 and Nek3-Nek10 were determined in this study, the phosphoryla- tion-site motifs for Nek2 and Plk1 were determined in Alexander et al., 2011. bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. Supplementary Figure 4

A Nek: 1 3 4 5 6 7 8 9 B S T 100 ST-kinases Back- 80 TKL-kinases 60 Y-kinases ground 40

Y 20 e C Nek10 phosphorylation no m ) of peptides i % k ( 15 n i (S)-tide NWMNRSMRGF t 80000 -3D e ) NWDNRSMRGF ub s M -1D S NWMND MRGF s P

C 10

( 60000 equen cy r F on i t a

l 40000 y 5

pho r 20000

Pho s 0 0 0 5 10 15 20 A C D E F G H I K L M N P Q R S T V W Y Time (min) Amino acid

Supplementary figure 4. The ability of Nek10 to phosphorylate on tyrosine is unique within the Nek family. (A) Reproduction of the column in the 0 position of the OPLS-dot blots shown in figure 1C, which contain the degenerate peptide libraries used to determine phospho-acceptor site specificity, are depicted again for ease of comparison. (B) Frequency of each amino acid in the HRD+2 position, which is the position equivalent to Nek10 T657, in all human serine/threonine (ST) kinases, all human tyrosine-kinase-like (TKL) kinases, or all human tyrosine (Y) kinases. Note that a lysine in this position is extremely common in serine/threonine kinases, while the threonine residue found in Nek10 is quite rare. (C) Purified Nek10 was incubated with the indicated peptide substrates and radiolabeled ATP. Peptide phosphorylation was quantified by scintillation counting (Mean ± SEM, n=3; CPM=counts per minute). bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. Supplementary Figure 5

A Cas9 cut site sgRNA gDNA template Sequence result 1

Sequence result 2

B Nek10 phosphorylation C of peptide

) 50000 Nek10 WT

M Peptide count

P Nek10 Y644F

C 40000 ( Peptide WT I693P(Ser)T692P(Tyr) KD on i 30000 t ASLsssssGAASLK 5 8 6 5 a l y 20000 LTsVVGTILYSCPEVLK 1 0 0 0 NFVSDHSsIGsLssANAAGR 50 27 0 0 pho r 10000 YHsYPWGTK 2 0 0 0

Pho s (R)SFsASGGER 17 11 13 7 0 0 5 10 15 20 Time (min) D E 14000 +3 a2 [R].NFVSDHSSIGSLpSSANAAGR.[I] 12000 S358 Y644 10000

8000 y -H PO -H O kinase 8 3 4 2 y8-H3PO4

1 1172 Intensity 6000 b11-H2O y y13-H3PO4-H2O y2 5 y7-H2O y -H PO -H O 4000 y b2 y4 10 3 4 2 1 y3 y13-H3PO4 y6 y y10-H3PO4 b 8 2000 3 y14-H3PO4-H2O y7 b7 b8 b10 0 0 200 400 600 800 1000 1200 1400 m/z

Supplementary figure 5. Nek10 autophosphorylates on Y644 and S358/359 (A) Sequencing results of the U-2 OS Nek10 knockout cells confirm out-of-frame editing of both alleles of Nek10 using CRISPR. Chromatograms and DNA sequence alignment of the sgRNA target site (exon 5) to the sequencing result of the U-2 OS Nek10 KO clone are present- ed. Shown are bases 23,286 till 23,592, with the exception of a 220 bp stretch as indicated by the gap between the two sequence panels. Sequencing of multiple PCR products indicated that all Nek10 alleles were edited. Sequence 1 contains a 19 bp deletion in combination with a 5 bp insertion further downstream (not shown). Sequence 2 contains a large deletion of 281 bp. (B) Purified Nek10 WT or a Y644F mutant were incubated with a tyrosine peptide substrate with the sequence EWMNWYWQRR and radiolabeled ATP. Peptide phosphorylation was quantified by scintillation counting (CPM=counts per minute). (C) Nek10 wild-type (WT), serine-specific (I693P(Ser)), tyrosine-specific (T692P(Tyr)) or kinase-dead (KD) was isolated from the cells described in figure 5C and analyzed by mass spectrometry. Table shows the peptide count of each identified Nek10 phosphopeptide for each of the Nek10 variants. Phosphorylated sites are indicated by red lower case letters. (D) Cartoon representation of the Nek10 protein domain structure. Sites of Nek10 autophosphorylation are indicated. (E) Representative MS/MS spectrogram of the Nek10 peptide containing phosphorylated S358. bioRxiv preprint doi: https://doi.org/10.1101/515221; this version posted January 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. Table S1 Table S1 List of sites phosphorylated by the Nek kinases. The Phosphositeplus database was mined for sites on human substrates that have been reported to be phosphorylated by any of the Nek kinases for which we determined the motif in this manuscript. The -3 column is shaded yellow. Leucine, methionine and phenylalanine are colored in purple in the -3 and +1 positions for all Nek kinases, asparigine is colored in red in the -2 position for Nek6, and basic residues are colored in blue in the -1 and +2 positions for Nek1 and Nek3.