Naa 12 rescues embryonic lethality in Naa 10-Deficient Mice in the amino-1 terminal acetylation pathway 2 3

Hyae Yon Kweon1¶, Mi-Ni Lee1,#a¶*, Max Dörfel2, Seungwoon Seo1, Leah Gottlieb3,4, Thomas Papazyan2, Nina 4 McTiernan5, Rasmus Ree5, Andrew Garcia6, Michael Flory7, Jonathan Crain2, Alison Sebold2, Scott Lyons2, 5 Ahmed Ismail2, Elaine Marchi8, Seong-keun Sonn1, Se-Jin Jeong9, Sejin Jeon1, Shinyeong Ju10, Simon J. 6 Conway17, TaeSoo Kim1, Hyun-Seok Kim1, Cheolju Lee10,11, Tae-Young Roh12, Thomas Arnesen5,13,14, Ronen 7 Marmorstein3,4,15, Gholson J. Lyon2,6,16*, and Goo Taeg Oh1* 8


65
N-terminal (Nt-) acetylation (NTA) is one of the most common protein modifications, 66 occurring co-and post-translationally. Approximately 80% of cytosolic proteins are N-67 terminally acetylated in humans and ~50% in yeast (1), while NTA is less common in 68 prokaryotes and archaea (2). NTA is catalyzed by a set of enzymes, the N-terminal 69 acetyltransferases (NATs), which transfer an acetyl group from acetyl-coenzyme A (Ac-CoA) 70 to the free α-amino group of a protein's N-terminus. To date, eight distinct NATs (NatA -NatH) 71 have been identified in eukaryotes that are classified based on different subunit compositions 72 and substrate specificities (3)(4)(5). NTA has been implicated in steering protein folding, stability 73 or degradation, subcellular targeting, and complex formation (6-10). The vital role of NATs 74 and NTA in development has also emerged (11). 75 NatA, the major NAT complex, targets ~40% of the human proteome, acetylating 76 Ala-, Gly-, Thr-, Val-and Cys N-termini after removal of the initiator methionine (1, 5). Human 77 NatA consists of two main subunits, the catalytic subunit N-α-acetyltransferase 10 (NAA10) 78 (Ard1) and the auxiliary subunit NAA15 (Nat1), and a regulatory subunit HYPK (12)(13)(14). 79 NAA15 function has been linked to cell survival, tumor progression, and retinal development 80 (15,16). In addition, Naa10-catalyzed N-terminal acetylation has been reported to be essential 81 for development in many species (11,(17)(18)(19)(20)(21)(22), and although NatA is not essential in S. cerevisiae, 82 depletion of Naa10 or Naa15 has strong effects, including slow growth and decreased survival 83 when exposed to various stresses (23,24). 84 NAA10 mutations were found to be associated with several human diseases 85 characterized by severe phenotypes, including global developmental defects (11). Among 86 these, the X-linked Ogden syndrome (OS) (25,26) shows the most severe pathological 87 features such as infant lethality and has reduced NatA catalytic activity. In a Saccharomyces 88 cerevisiae model for the Naa10 Ser37Pro mutant, the mutation impairs NatA complex 89 formation and leads to a reduction in NatA catalytic activity and functionality (27,28). 90 Further, OS patient-derived cells have impaired NTA in vivo of some NatA substrates (25). 91 Over the years, many additional pathogenic NAA10 variants have been identified in NAA10 92 (29)(30)(31)(32)(33)(34)(35) and the collection of presenting symptoms is currently referred to as Ogden syndrome 93 or NAA10-related syndrome (36). Other families have also been identified with NAA15 94 mutations (29)(30)(31)(32)(33)(34)(35). 95 The autosomal NAA10 homolog, NAA11 (ARD2), has been reported to be present in 96 mice and humans, and is co-expressed with NAA10 in human cell lines (37). Therefore, NAA11 97 could conceivably compensate when NAA10 is reduced or lacking (11). However, NAA11 was 98 only found in testis and placenta in human and gonadal tissues in mouse, while NAA10 showed 99 widespread expression in various tissues in embryos and adults (38). Thus, any functional 100 redundancy or compensation might be limited to certain tissues only. 101 To elucidate the functional role of Naa10 during development in mice, we used two 102 different Naa10-deficient mouse lines: one, referred to as Naa10 knockout (KO), which was 103 previously reported specifically related to bone density in postnatal day 3 (P3) mice (39), and 104 another denoted as Naa10 tm1a(EUCOMM)Hmgu (Naa10 tm1a ), generated in this study. These Naa10-105 deficient mice exhibit pleiotropic developmental abnormalities at a range of different ages, 106 overlapping with some of the phenotypes seen in human disease involving NAA10 impairment. 107 Because we did not discover major changes in the overall Nt-acetylome in Naa10 KO mice, 108 we hypothesized that there might be a compensating gene in mice, which led us to the 109 identification of a new paralog of Naa10, which we name Naa12. Naa12 is expressed in several 110 organs (liver, kidney, heart and testis) and, like Naa10, binds to Naa15 to mediate NatA activity. 111 Furthermore, lethality was observed in Naa10; Naa12 double-KO mice, which supports the 112 compensatory role of Naa12 in vivo. Thus, we demonstrate that Naa10 is essential for proper 113 development and Naa12, a newly-identified paralog of Naa10, can play a compensatory role 114 in mice. 115

116
Naa10 knockout mice can be born, but display pleiotropic developmental defects 117 To explore the role of Naa10 in development, most analyses were carried out using 118 our Naa10 KO model mice that had been generated previously (39) using a targeting vector 119 deleting Exon1, including the start codon, and Exon2 to Exon4 containing the GNAT domain 120 including the Acetyl-CoA binding motif, which is crucial for Naa10 function. We also 121 generated another Naa10-deficient mouse which we called Naa10 tm1a , expressing β-122 galactosidase rather than the Naa10 gene (S1A Fig). Naa10 expression was deficient in 123 Naa10 tm1a mice (S1B and S1C Fig). Especially strong β-gal staining was observed during 124 embryonic stages in the brain, heart and spinal cord (S1D Fig). Male Naa10 KO (Naa10 -/Y ) 125 embryos displayed mild to severe developmental defects compared to wild-type (WT) 126 (Naa10 +/Y ) embryos. Some Naa10 -/Y mice had lower levels of somites and developmental 127 delay. Additionally, some Naa10 -/Y embryos had a normal number of somites but were 128 retarded in growth ( Fig 1A). Some of the embryos underwent lysis or remained arrested at an 129 earlier stage than embryonic day 10.5 (E10.5), with no turning, an abnormal trunk, and 130 underdeveloped facial features. These phenotypes also reproduced in Naa10 tm1a/Y embryos. 131 Next, we assessed whether Naa10 is essential for viability and counted the Mendelian ratios. 132 Both Naa10 -/Y and Naa10 tm1a/Y mice were under-represented after birth, while there was no 133 significant reduction in the embryonic stage in both mouse lines (S1 and S2 Table). We 134 monitored the pups daily at postnatal day 0 (P0) to postnatal day 3 (P3) and beyond, and the 135 survival rate of Naa10 -/Y mice dramatically decreased relative to either WT (Naa10 +/Y and 136 Naa10 +/+ ) or heterozygous female (Naa10 +/-) mice after the first few days of life (Fig 1B), 137 and a few Naa10 -/Y mice with postnatal lethality exhibited severe developmental defects such 138 as craniofacial anomaly, an undeveloped lower body, whole-body edema, and ocular 139 malformations ( Fig 1C). 140 Congenital heart defects are one of the main causes of infant lethality, and cardiac 141 diseases are a common developmental anomaly in OS patients (31), with some OS males 142 dying in infancy with cardiac arrhythmias (26). Therefore, we investigated whether Naa10 143 KO affects cardiac development. Development of a four-chambered septated heart is 144 normally complete at E14.5, therefore we examined the cardiovascular system at E14.5. We 145 identified ventricular septal defects (VSD) in several Naa10 -/Y embryos, as well as 146 concomitant double outlet right ventricle (DORV) at E14.5 (Fig 1D, upper). Ventricular septal 147 defects (VSD) and atrial septal defects (ASD) were also observed at E18.5 (Fig 1D,  inversus, but we did not observe this in any adult (>4 weeks) Naa10 -/Y mice examined (n=19). 159 Combined, these data suggest that Naa10 mutant CHDs are mainly confined to aberrant 160 remodeling of the great vessels of the heart, leading to pulmonary overload at birth resulting 161 in lethality. 162 Some of the surviving homozygous mice (Naa10 -/Y and Naa10 -/-) had reduced body 163 weight (Fig 2A). This reduced body weight continued through weaning, although some mice 164 did lose weight as they developed progressive hydrocephaly. If the analysis was restricted to 165 only include litters in which there was at least one Naa10 +/Y and one Naa10 -/Y mouse in the 166 same litter alive to be weighed at P4 and beyond, it was consistently observed that the 167 smallest weight animal among the two genotypes was almost always the Naa10 -/Y mice. For 168 example, 13 litters met this criteria from the mating (Naa10 +/x Naa10 +/Y ), and 12/13 of the 169 litters had the Naa10 -/Y as the lower weight (Fisher's exact test, two-tailed, P value <0.0001). 170 5 litters met this criteria from the mating (Naa10 +/x Naa10 -/Y ), and of these, all of them had 171 the Naa10 -/Y as the lower weight (Fisher's exact test, two-tailed, P value = 0.0079). Therefore, 172 despite the known variability in weight data as a function of genetic background, 173 environment, and stochastic variation (41), it does appear at least for "within-litter" analysis 174 that Naa10 -/Y males are born at a smaller weight than Naa10 +/Y males and remain the smallest 175 male in the litter throughout their life. 176 Although piebaldism has never been reported in humans with OS, all (100%) of 177 Naa10 -/Y and Naa10 tm1a/Y mice exhibited hypopigmentation on their belly (Fig 2B,upper),178 with this piebaldism quite varied in its extent but not appearing to correlate in any way with 179 other phenotypes, such as hydrocephaly. Another phenotype with complete penetrance was 180 bilateral supernumerary ribs (14 pairs of rib instead of 13) in all Naa10 -/Y and Naa10 tm1a/Y 181 mice (Fig 2B, middle and bottom, Table 1). This extra pair of ribs linking to the sternum  182  transforms the T8 vertebrae into an anterior T7-like phenotype (S3A-S3D Fig, Table 1). 183

190
(C) Representative images of Naa10 -/Y pups during early postnatal days compared with Naa10 +/Y .

191
Severe developmental defects such as malformations of head and lower body (one leg; black 192 arrowheads), whole-body edema and anophthalmia (black arrows) are shown (N=1 each). (D)

216
A majority of the Naa10 -/Y and Naa10 -/mice also had four instead of the usual three 217 sternebrae, which were sometimes fused (Table 1). Cervical vertebrae fusion was also 218 demonstrated in Naa10 -/Y mice, particularly involving C1 and C2, suggesting possible 219 anteriorization of C2 into a C1-like phenotype (S3E and S3F Fig, and Table S3). The number 220 of lumbar vertebrae remained the same, thus suggesting an anterior transformation of the first 221 sacral vertebra to a lumbar-like phenotype. These combined observations suggest possible 222 anterior transformations in the Naa10 mutant skeletal phenotype, with an anteriorization of 223 C2, a T8 transformation to a T7-like phenotype with ribs connecting to the sternum, an extra 224 pair of ribs on L1 likely due to an L1 transformation to a T13-like phenotype, and an anterior 225 transformation of the first sacral vertebra to a lumbar-like phenotype with loss of fusion to 226 the sacral wings. 227 Some Naa10 KO (Naa10 -/Y and Naa10 -/-) mice survived more than 4 months, with 228 some of these then developing hydronephrosis (Fig 2C,middle). They had some hollowed 229 space in the kidney, which had been filled with fluid and their ureter was thickened already at 230 P3 stage of prenatal development in some Naa10 -/Y mice (S4A Fig). Commonly, 231 hydronephrosis is caused by a blockage or obstruction in the urinary tract. We speculate that 232 this swelling in Naa10 KO (Naa10 -/Y and Naa10 -/-) mice is likely caused by ureteral defects 233 rather than the kidney itself. Moreover, some Naa10 KO mice displayed genital defects, such 234 as seminal vesicle malformation and hydrometrocolpos, respectively (Fig 2C, bottom). Many 235 Naa10 -/female mice appeared to have decreased fecundity, although they were fertile upon 236 the first mating, and this decrease in fecundity is possibly due to the development of 237 hydrometrocolpos ( Fig 2C, bottom), which might result from structural issues, like vaginal 238 atresia or a retained vaginal septum, although this requires further investigation. Additionally, 239 hydrocephaly became clinically apparent with a round-shaped head (Fig 2C,upper) in some 240 Naa10 -/Y mice. CT scanning of some of these mice confirmed hydrocephaly as the primary 241 cause of their rapid deteriorating condition, usually within the first 3 months of life (S4B 242 Fig). CT scanning did not reveal any obstructive lesions (such as a tumor) in any of the 243 ventricles that could account for the hydrocephaly. Taken together, these results indicate that 244 Naa10 contributes to overall development and is particularly important for viability. 245 Litter sizes and offspring from other matings were also investigated, as shown in 246 Table S4. Matings were set up between Naa10 -/females and C57bl6J WT (Naa10 +/Y ) males, 247 involving eleven mating pairs with 7 unique females and 7 unique males. Of a total of 127 248 pups that were born, 37 died in the first day of life and were degraded and/or cannibalized 249 prior to any tail sample being retrieved, thus not being genotyped. This was a relatively high 250 death rate in the first 24 hours of life (29%), more so than with the other matings, except for 251 the one between Naa10 -/females and Naa10 -/Y males (Table S4). However, this is 252 substantially less than the death rate of 90% (46/51) reported for the same mating in the Lee 253 et al. paper (42), and we currently do not have an explanation for this discrepancy. Of the 254 remaining 90 pups that could be genotyped, 59 of these were Naa10 +/females and 31 were 255 Naa10 -/Y males. Seven of the 59 Naa10 +/females and two of the 31 Naa10 -/Y males died in 256 the first 3 days of life (for a total death rate in the first three days for all born pups of 46/127, 257 or 36%), and after this time, none of the remaining Naa10 +/females died in the first ten 258 weeks of life (88% overall survival), whereas ten of the 31 Naa10 -/Y males developed 259 hydrocephaly and died in the first ten weeks of life (61% overall survival). The death rate of 260 36% in the first three days of life is similar to the rate of 42.4% seen with the mating of 261 Naa10 -/females with Naa10 -/Y males (Table S4), whereas this rate is higher than that seen for 262 Naa10 +/females with Naa10 +/Y males (15.8%) and Naa10 +/females with Naa10 -/Y males 263 (13.6% between Naa10 -/Y and Naa10 +/Y mice ( Fig 3A). To further investigate this, we measured the in 282 vitro Nt-acetylation activity of NatA via immunoprecipitation of the large auxiliary subunit 283 Naa15 from mouse tissues. This analysis showed normal expression of Naa15 in Naa10 KO 284 liver tissue as in WT tissues (Fig 3B), and we isolated a physical complex composed of 285 Naa15 and undefined partners that retains NatA activity from Naa10 KO tissues (Fig 3C). 286 These data suggest that despite the loss of Naa10 in mice, the NatA complex remains active. 287

A Naa10 paralog exists in mice 304
Naa10 disruption is lethal in a variety of organisms, including D. melanogaster (17), 305 C. elegans (18) and T. brucei (45). Given the relatively mild phenotype and no reduction of 306 the Nt-acetylome in Naa10 KO mice, we hypothesized that there might be a yet unidentified 307 paralog of Naa10, which can compensate for loss of function in mice. A Blast search for 308 genomic sequences with homology to Naa10 exposed several Naa10 pseudogenes on 309 chromosome 2, 3, 7, 12, 15 and 18. Additionally, southern blot analysis from C57BL/6J DNA 310 with Naa10 cDNA probe detected bands of the expected sizes on the X chromosome (S5A 311 and S5B Fig), while other bands of unexpected sizes appeared on other chromosomes 2, 5, 15 312 and 18. The previously identified Naa10 paralog Naa11 is located on chromosome 5, 313 however, this paralog is only expressed in testes (38). We identified a homologous sequence 314 on chromosome 18, which we named Naa12. This predicted gene shows high similarity to 315 Naa10, and RiboSeq and mRNA traces of this region suggest possible transcription and 316 translation of this gene (S5C Fig). The protein sequence of Naa12 is >80% identical to 317 Naa10. In addition, Naa12 is almost 90% identical with Naa11 (37). 318 Quantitative PCR (q-PCR) analysis also confirmed the expression of this transcript in 319 all tested tissues (S6A Fig), with the expression of Naa12 unchanged in the corresponding 320 Naa10 KO tissues. We performed a sequence alignment of the two known mNaa10 isoforms, 321 mNaa11 and mNaa12 and selected a unique Naa12 peptide for immunization and antibody 322 generation ( To assure that the antibody is specific, we also probed the 326 membrane with Naa12 antibody that was blocked by the peptide it was raised against. The 327 band at 35 kDa disappears when probed with the blocked antibody, which appears to confirm 328 the specificity. 329 To test whether Naa12 has a similar enzymatic activity as Naa10, we performed a 330 radioactive-based acetyltransferase assay using synthetic peptides ( Fig 4A). Since 331 monomeric Naa10 preferentially acetylates N-termini with acidic side chains (46-48), we 332 used peptides representing the N-termini of γ -actin (starting DDDIA-) and γ-actin (starting 333 EEEIA-), the two known Naa10 in vitro substrates. Additionally, we used a peptide starting 334 with SESSSKS-, an in vitro NatA complex substrate High mobility group protein A1. As 335 expected for the monomeric proteins, we could not detect any activity towards the SESSSKS-336 substrate. Importantly, both Naa10 and Naa12 significantly Nt-acetylated the acidic N-337 terminal peptides demonstrating the intrinsic capacity of Naa12 to catalyze Nt-acetylation 338 ( Fig 4A). 339 Across species, Naa10 is bound to its auxiliary subunit, Naa15, which links the 340 catalytic subunit to the ribosome to facilitate co-translational Nt-acetylation of proteins as 341 they emerge from the exit tunnel (23, 49-53). Due to its high sequence similarity (see also 342 S4B Fig), we suspected that Naa12 may also interact with Naa15. To test this hypothesis, we 343 performed co-immunoprecipitation assays in HEK 293 cells. Apart from Naa10 (isoform 1, 344 Naa10 235 ) and Naa12, we also included the second isoform of mNaa10, mNaa10 225 that has 345 been described earlier (12, 50, 54) as well as Naa11. Both Naa10 isoforms as well as Naa11 346 and Naa12 co-precipitated with V5-Naa15 but not V5 alone, suggesting that all tested 347 proteins could form a stable complex with Naa15 in mouse ( Fig 4B). In a mass spectrometry 348 analysis of a similar setup to that shown in Fig 3B, immunoprecipitation of endogenous 349 Naa15 from liver tissue derived from WT and Naa10 KO mice, we identified one peptide in 350 the Naa10 KO sample that is identical between Naa12 and Naa11, but differing from Naa10 351 (Table 2). Further, we detected a number of peptides shared between Naa10, Naa11 and 352 Naa12 in the Naa15 immunoprecipitates from the Naa10 KO sample. In contrast, peptides 353 unique for Naa10 were only detected in the WT mouse sample. This together with data 354 demonstrating that Naa11 is only expressed in testes and that no Naa10 is present in the 355 Naa10 KO mice, strongly suggest that endogenous Naa15-Naa12 complexes are present in 356 Naa10 KO mice.

372
Naa12 may rescue loss of Naa10 in mice 373 To investigate whether Naa12 can rescue the loss of the function of Naa10 in vivo, 374 Naa12 KO mice were generated using CRISPR technology (55). One 95-base pair deletion 375 Δ131-225 in Naa12 was characterized in depth ( Fig 5A). This mutation introduces a 376 frameshift, leading to a termination codon at amino acid 67, which should either result in 377 complete knockout of the protein or, at best, the expression of a truncated mini-protein that 378 would be far shorter than the usual 220 amino acid Naa12. We confirmed the deletion by 379 PCR with genomic DNA (Fig 5B). QPCR further showed deletion of Naa12 in the tested 380 tissues of Naa12 KO mice ( Fig 5C), however, it seemed that Naa12 might be slightly 381 expressed in testis. Due to the high similarity between Naa11 and Naa12, the expression 382 shown in Naa12 KO testis could actually be Naa11 rather than Naa12, and this was 383 confirmed by RT-PCR showing definite deletion (S7A Fig). This was confirmed at the protein 384 level with the Naa12 antibody in heart tissue lysates (S7B Fig). However, this antibody was 385 unable to consistently detect protein even in WT liver, kidney, or brain tissue lysates, which 386 could be due to very low expression or post-translational modification of Naa12 in these 387 tissues. Furthermore, given that this antibody is raised against a peptide at the C-terminus of 388 Naa12

400
Relative expression level of WT (white bars) and Naa12 KO (black bars) after normalizing to that of 401 GAPDH.

403
Homologs that play a compensatory role usually show similar phenotypes to each 404 other when one of them is deficient (56). Therefore, we analyzed Naa12 KO mice to see if 405 they produced similar developmental defects to those in Naa10 KO mice. Unexpectedly, 406 knockout mice for this gene were viable (Table S5), albeit at a reduced rate of fertility (Table  407 S4). Necropsy and inspection of testes and seminal vesicles under a stereomicroscope, along 408 with weighing these tissues, did not reveal any macroscopic differences that could readily 409 explain the decreased fertility of the Naa12 KO males. Furthermore, the phenotypes 410 (piebaldism and bilateral supernumerary ribs; Fig.2b) observed in Naa10 KO mice with 411 complete penetrance were not present in Naa12 KO mice (S7C Fig). Overall, there were not 412 any obvious phenotypes in these mice, except for reduced fertility. 413 Matings between compound heterozygous (Naa10 +/-; Naa12 +/-) females and 414 Naa10 +/Y ; Naa12 +/males demonstrate that no live births occurred for Naa10; Naa12 double-415 knockout (DKO) males (Naa10 -/Y ; Naa12 -/-) ( Fig 6A). In addition, the average litter size was 416 very small when compared to the control (WT x WT) matings, suggesting embryonic 417 lethality (Table 3 and Table S4). In order to determine whether lethality occurs during the 418 embryonic stage, we genotyped E18.5 litters -just before birth. Consistent with our previous 419 observations, we could not obtain any Naa10 -/Y ; Naa12 -/embryos, and many embryos could 420 not be genotyped because they were already in the midst of resorption (n=23) (Fig 6). We 421 checked an even earlier stage at E10.5 and also found zero Naa10 -/Y ; Naa12 -/embryos, and 422 also with far fewer resorptions at this stage (N=3). Interestingly, we did observe Naa10 -/Y ; 423 Naa12 +/embryos where two of them displayed delayed developmental stage (appearing 424 younger than E10.5) and another two embryos were lysed and had already begun 425 degenerating (but despite this, we could at least genotype these embryos). This helps explain 426 why only one Naa10 -/Y ; Naa12 +/embryo was observed at E18.5. Furthermore, Naa10 +/-; 427 Naa12 -/female embryos were also lysed/degenerating at E10.5 and were not observed from 428 that day onward. Matings between compound heterozygous females and Naa10 +/Y ; Naa12 -/-429 males also did not yield Naa10 -/Y ; Naa12 -/male or Naa10 +/-; Naa12 -/female mice during 430 development (S8 Fig). Consistent with our previous findings, we noted many resorptions at 431 E12.5 and E18.5 that could not be genotyped. The number of living postnatal compound 432 heterozygous female mice was also considerably lower than the predicted Mendelian ratios 433 ( Fig. 6 and S8 Fig), and the surviving Naa10 +/-; Naa12 +/females were smaller in size than 434 littermate controls (Fig. 7D). Due to the severe embryonic lethality observed in the Naa10; 435 Naa12 DKO male mice and the Naa10 +/-; Naa12 -/female mice, which was not seen in each 436 single KO (Naa10 KO or Naa12 KO), it seems likely that, without compensation by Naa12, 437 NTA is disrupted in Naa10; Naa12 DKO mice. Together, these data support the 438 compensatory role of Naa12 in vivo. 439 440

441
Pedigree and genotypes of pups and embryos at E10.5 and E18.5 from Naa10 +/-; Naa12 +/female 442 mice crossed to the Naa10 +/Y ; Naa12 +/male mice.    Table S7 shows the results in which weight of Naa10 mice in grams is regressed upon 462 age, Naa10 knockout status, and their interaction. Unsurprisingly, age predicts weight for 463 males and females strongly, with growth slowing with age (first column). Though a strong 464 negative effect of the knockout is seen in for both males and females (second column), when 465 both age and knockout status are modeled together (third column) this effect all but 466 disappears in females. Moreover, in females there is no interaction of knockout status with 467 age (fourth column), suggesting that the Naa10 knockout status itself has no significant effect 468 on the growth rate in females. For males, however, the main effect of the knockout remains 469 when age is included in the model (third column) and the interaction is significant (fourth 470 column), indicating that the Naa10 knockout both reduces weight of males overall and lowers 471 the rate of growth. 472

444
Results of analyses of mixed-genetic background Naa10/Naa12 mice are shown in 473 does not appear to reduce the weight or growth rate of females, while a heterozygous Naa10 484 KO is sufficient to reduce both weight and growth rate. Interestingly, when modeled together, 485 both the Naa10 and the Naa12 KOs significantly reduced weight (above, fourth column) and 486 the interaction of the Naa10 and the Naa12 heterozygous KOs significantly reduced weight 487 (above, fifth column). As no female mice were both KO for Naa10 and homozygous KO for 488 Naa12, the effect of the interaction of those two factors could not be determined. The triple 489 interaction of heterozygous Naa10 KO, Naa12 KO and age was weakly significant, 490 suggesting that the presence of both knockouts affects growth rate above and beyond the 491 effects of each knockout independently (below, fourth column). No males with knockouts of 492 both Naa10 and Naa12 were born, so no test of their interaction was possible. An effect was 493 seen for the Naa10 knockout on weight when modeled with age and age 2 (second column), 494 and the significant interaction of the Naa10 knockout with age and age 2 (third column) shows 495 that the Naa10 KO in males reduces the growth rate. As with females, no significant effect of 496 a Naa12 KO, whether heterozygous or homozygous, was seen in males, nor is there a 497 significant interaction with age (fourth column). When the interactions of age with both 498 Naa10 and Naa12 knockout status are entered in one model, Naa10 alone is seen to reduce 499 growth rates (fifth column). 500 501

502
We have shown that Naa10 deficiency results in pleotropic developmental defects in 503 two different Naa10-deficient mouse models. Similar to infant mortality in some OS males, 504 the lethality of Naa10 KO mice increased dramatically in pups in the first 3 days of life (Fig  505  1B). Defects in kidney, brain, pigmentation (piebaldism), and ribs were observed during 506 embryonic or early postnatal stages in some mice (Fig 2B and 2C). These observed 507 phenotypes overlap with some of the phenotypes found in surviving humans with OS, 508 including supernumerary vertebrae and hydrocephaly, although piebaldism has not been 509 reported to date in any humans. However, the puzzling lack of embryonic lethality in the 510 Naa10 KO mice prompted us to discover Naa12 as a possible compensatory NAT, with 511 Naa10-like NTA activity (Fig 4A), with an interaction between Naa15 and Naa12 (Fig 4B). In 512 addition, co-immunoprecipitation of endogenous Naa15 from Naa10 KO mouse tissues 513 followed by mass spectrometry analysis (Table 2) and Nt-acetylation assays (Fig 3C) fully 514 support that the endogenous Naa12-Naa15 complexes produces NatA activity. Finally, we 515 found genetic proof of the compensatory activity of Naa12 in mice when we observed 516 embryonic lethality in in Naa10; Naa12 DKO male and Naa10 +/-; Naa12 -/female mice ( Fig  517  6 and S7 Fig). This compensation by Naa12 explains the mouse proteomics data indicating 518 normal Nt-acetylation in Naa10 KO mice (Fig 3A). We have confirmed the expression of 519 Naa12 in various tissues using qPCR and Western blot analyses (S5A- S5E Fig). The band for 520 Naa12 runs a little higher in the Western blot than the calculated molecular weight would 521 suggest, which is consistent with previous observations of NAA10 gel migration. Future 522 characterization of Naa12 may define possible post-translational modifications specific for 523 Naa12 that might account for the variability of detection by the antibody. 524 Gene duplication has long been believed to be a major driving force in evolution that 525 provides genetic novelty in organisms. Paralogous genes, originating by small-scale or 526 whole-genome duplication, overlap functional roles for each other and can completely or 527 partially compensate for the loss of the duplicate gene (56, 57). There is not yet any human 528 reported with complete knockout for NAA10. There is one published truncating variant in the 529 C-terminal portion of NAA10 in a male patient with microphthalmia (35), but unfortunately 530 there are no cell lines available from this family to confirm whether any truncated NAA10 531 protein is expressed, as was shown with a splice-site mutation in a Lenz microphthalmia 532 family (29). NAA10 was also identified in screens for essential genes in human cell lines (58, 533 59), so it seems unlikely that an unknown NAA10-like paralogous gene exists in humans, 534 other than the already known NAA11. 535 The pleiotropic phenotypes shown in Naa10 KO mice, including hypopigmentation 536 and supernumerary ribs with a penetrance of 100%, were not observed in the Naa12 KO 537 mice. Naa10 itself has been described to have N-ɛ-acetyl-activity towards internal lysine 538 residues of proteins involved in various disease-and development-related signaling pathways 539 (11), although its acetylation of some substrates is controversial (60, 61). Since the Nt-540 acetylome appears to be globally intact in MEFs from Naa10 KO mice, it is possible that the 541 presented phenotypes could be due to the loss of Naa10-specific N-ɛ-acetyl-activity or non-542 catalytic roles of Naa10 (4). Alternatively, the quantitative expression of Naa10 and Naa12 543 might be different within or between tissues, which might then explain why there is clearly a 544 phenotype for Naa10 +/-; Naa12 + /female mice (not born at Mendelian ratios and the few that 545 are born are usually much smaller) but no apparent phenotype in Naa10 +/+ ; Naa12 -/female 546 mice. It seems likely that the mechanism cannot be simply additive between two equally 547 expressed proteins, because if the expression of each protein is theoretically set at an arbitrary 548 unit of 10, then Naa10 +/-; Naa12 + /female mice might possibly have half as much of each 549 protein, so that the total dose of both proteins together would be 10, instead of 20. Likewise, 550 the total dose of both proteins together would also be predicted to be 10 in a Naa10 +/+ ; 551 Naa12 -/female. Yet, the Naa10 +/-; Naa12 + /female mice have a phenotype, whereas the 552 Naa10 +/+ ; Naa12 -/female mice do not (Fig 6). Therefore, other explanations could include 553 different tissue-specific dosages of each protein, different expression between different 554 tissues, possible X-chromosome skewing for the X-linked Naa10 in different tissues, or 555 different functions of the two enzymes, including Naa10-specific N-ɛ-acetyl-activity or non-556 catalytic roles of Naa10 (4). These questions remain unanswered and are worth exploring in 557 future studies. 558 There are several clinical features that were presented in the original description of 559 OS (26) which can now be better understood in light of the phenotypes found in the knockout 560 mouse model. For example, all of the affected children in the first families with OS were 561 noted to have large and, in some cases, persistently open fontanels (25, 26). For one child 562 (Family 1, Individual II-1), CT scanning revealed cerebral atrophy with enlarged ventricles, 563 and in another child (Family 1, Individual III-4), there was evidence on magnetic resonance 564 imaging (MRI) of "moderate lateral and third ventricular dilatation without identified cause". 565 Lastly, all of the children had respiratory depression and apneic episodes, along with varying 566 course of hypotonia and/or hypertonia (including documented hyperreflexia in at least one 567 case (Family 2, Individual III-2)). In retrospect, it seems that these clinical features could be 568 consistent with mild hydrocephaly in these probands with Ogden syndrome, which resolved 569 over time. This is also consistent with the ventriculomegaly reported in several female OS 570 probands with missense mutations in NAA10, along with ventriculomegaly in one other male 571 proband who died in the first week of life, with generalized hypotonia and lack of 572 spontaneous respirations (62). One of the female patients with an Arg83Cys mutation in 573 Naa10 (#9 in table 1 of that paper) was reported as having intraventricular hemorrhage in the 574 occipital horn, hypoxic-ischemic encephalopathy, and a ventriculo-peritoneal shunt. It is 575 possible that this sequence of events is compatible with hydrocephaly with clinical signs and 576 symptoms that required the placement of the shunt. 577 There are additional cardiac and skeletal features that are also worth re-examining in 578 light of these new findings. In some of the original cases of OS, there were varying levels of 579 pulmonary valve stenosis detected on echocardiography, along with some documentation of 580 pulmonary hypoplasia (26). For example, individual III-7 in Family 1 was found on 581 echocardiography to have small persistent ductus arteriosus, a mildly decreased left 582 ventricular systolic function, an abnormal appearing aortic valve, an enlargement of the right 583 ventricle, decreased right ventricular systolic function, and persistence of the foramen ovale. 584 Individual III-6 from this same extended family was found on echocardiography to have a 585 thickened bicuspid aortic valve and mild pulmonary hypertension. One of the OS female 586 patients with an Arg83Cys mutation in NAA10 was reported to have "supernumerary 587 vertebrae" (62). Prompted by our findings of supernumerary ribs in the mice, we obtained an 588 MRI report for this patient, in which the radiologist concluded that there appeared to be 25 589 distinct vertebrae, as opposed to the usual 24, with a suggestion of a 13th rib, at least on the 590 right. The report went on to state that "the vertebrae represent 7 cervical vertebrae, 13 rib-591 bearing thoracic vertebrae, and 5 lumbar vertebrae, and the L1 vertebra is mildly dysmorphic, 592 with a suggestion of anterior beaking". In addition, chest and abdominal X-rays from two of 593  (73) phenotype is also close to the Rpl38 -/-phenotype (82), except for the sacral fusion described 628 in Rpl38 -/-mice. Interestingly, it was shown that Hox genes were dysregulated in this 629 genotype. The skeletal findings and comparison to other mutant mice suggest a pattern 630 consistent with a homeotic anterior transformation hypothesis. 631 The developmental role of Naa10 in mice has been previously described (42). Lee et 632 al. reported embryonic lethality at E12.5-14.5 and beyond (due to placental defects), 633 hydrocephaly, postnatal growth retardation, and maternal effect lethality in Naa10 KO mice 634 and suggested that genomic imprinting dysregulation is associated with those developmental 635 phenotypes. In the present study, hydrocephaly and postnatal growth retardation were also 636 apparent, but embryonic lethality was not observed, which prompted the search for and 637 discovery of Naa12. The previous paper (42) did not report the piebaldism, homeotic anterior 638 transformation, hydronephrosis, and genital defects (such as seminal vesicle malformation 639 and hydrometrocolpos), nor did they explain the cause of death in the first day of life, which 640 is at least partly due to congenital heart defects, as reported herein. A more recent paper from 641 the same group reported that conventional and adipose-specific Naa10p deletions in mice 642 resulted in increased energy expenditure, thermogenesis, and beige adipocyte differentiation 643 in the surviving mice (83). Although the Lee et al. paper reported a very high maternal effect 644 lethality rate of 90% (46/51) (otherwise stated as a survival rate of 10% (5/51)) for newborns 645 in matings following Naa10 -/female and C57BL/6J wild type male intercrossing, this rate 646 was only 29% (37/127) in this same mating herein in the first 24 hours of life and with a total 647 death rate in the first three days for all newborns of 46/127, or 36% (Table S4), with this 648 result deriving from a larger number of mating pairs, litters and pups. Although this rate of 649 36% is higher than that seen with matings involving Naa10 +/females (15.8% and 13.6%) 650 (Table S4), the explanation for this ~20% difference in survival in the first 3 days of life 651 could involve differences in maternal care provided by the Naa10 +/and Naa10 -/females, but 652 this would have to be investigated in future studies, involving detailed behavioral and 653 cognitive assessment of the dams. 654 The reasons for the differences between the studies in regards to maternal effect 655 lethality and in utero lethality are unknown at present. Whilst Lee et al. deleted Naa10 exons 656 2-6 (42), the current study deleted Naa10 exons 1-4 or used an allele Naa10 tm1a expressing β-657 galactosidase instead of the Naa10 gene (S1 Fig.), and there was not any significant 658 embryonic lethality in either line (S1 and S2 Table). All three of these mouse models were 659 made using 129Sv/Ev ES cells, and all three are nulls lacking Naa10 protein. It is the case 660 that the previous study used the Cre/loxP system to generate the Naa10 KO mice, where a 661 floxed Naa10 female mouse was crossed with the Ella-Cre transgenic male mouse expressing 662 Cre recombinase for germ line deletion of loxP-flanked Naa10, whereas our mice were made 663 using standard gene-targeting methods without the use of Cre recombinase, but it is not clear 664 how this would have resulted in embryonic lethality, particularly as these mice were only 665 used after "at least six generations of backcross with C57BL/6 mice". The explanation for 666 differences in embryonic lethality might be more likely due to different combinations of 667 modifying alleles that are present in the different C57BL/6J genetic backgrounds, rather than 668 differences in our model systems, and future plans will address this after back-crossing more 669 than 20 generations to C57BL/6J to achieve an entirely inbred line. The impact of genetic 670 background is supported by the observation that additional null alleles on mixed genetic 671 backgrounds, made during the process of generating missense mouse models for OS, have far 672 less penetrance for a range of the various phenotypes, including much less perinatal lethality 673 (unpublished observations). 674 In conclusion, our study provides strong evidence that Naa10, the catalytic subunit of 675 N-acetyltransferase A (NatA), is critical for normal development in mice. Furthermore, this 676 study explains the puzzle regarding the lack of complete embryonic lethality in the Naa10 677 knockout mice due to the discovery of a second mouse Naa10 paralog, which, unlike Naa11, 678 is expressed in the heart as well as other tissues. Taken together, our findings suggest that the 679 newly identified Naa12 can functionally rescue Naa10 loss and act as a catalytic subunit in 680 mouse NatA complexes. 681 682 Generation of Naa12 knockout mice. The mice were made using standard methods by 700 microinjection of CRISPR reagent mix into zygotes obtained from the mating of B6D2F1 701 females (i.e. 50% C57BL/6J, 50% DBA/2J (D2)) females to inbred C57BL/6J males. The 702