I've left out the the majority of the paper as it's enormous, but anyone inclined should read it. The very tl;dr is that inserting a gene that expresses a humanised NOVA1 protein causes rats to emit vocalisations at a different frequency and with far greater complexity than normal rat squeaks. The understanding is that this protein is responsible for the human ability to speak.
Article follows.
Archive | live (nature)
NOVA1, a neuronal RNA-binding protein expressed in the central nervous system, is essential for survival in mice and normal development in humans. A single amino acid change (I197V) in NOVA1’s second RNA binding domain is unique to modern humans. To study its physiological effects, we generated mice carrying the human-specific I197V variant (Nova1hu/hu) and analyzed the molecular and behavioral consequences. While the I197V substitution had minimal impact on NOVA1’s RNA binding capacity, it led to specific effects on alternative splicing, and CLIP revealed multiple binding peaks in mouse brain transcripts involved in vocalization. These molecular findings were associated with behavioral differences in vocalization patterns in Nova1hu/hu mice as pups and adults. Our findings suggest that this human-specific NOVA1 substitution may have been part of an ancient evolutionary selective sweep in a common ancestral population of Homo sapiens, possibly contributing to the development of spoken language through differential RNA regulation during brain development.
Fossil records indicate that modern humans (Homo sapiens) emerged 200,000–300,000 years ago as the predominant species from a common ancestral population1,2. Humans differ significantly from their closest living relatives, the great apes, particularly in their ability to communicate through complex learned vocal communication, a necessary component of spoken language3. This complexity is driven by some anatomical adaptions of the vocal tract and intricate neural networks linking various brain regions3,4,5,6,7. However, the genetic basis underlying these specialized human traits remains to be fully identified.
The closest evolutionary relatives of modern humans are two extinct lineages: Neanderthals and Denisovans. Genome sequencing from fossilized remains of these archaic humans has identified distinct genetic differences between them and modern humans, which may be relevant to recent human evolution8,9,10,11. Additionally, the availability of extensive human genome data over the past few decades initially focused on European populations, has significantly expanded the scope of evolutionary studies12,13,14.
The transcription factor forkhead box P2 (FOXP2) is of particular interest as a potential driver of human language function, as it harbors two amino acid substitutions present in humans but not in chimpanzees and many other mammalian genomes. Families with FOXP2 mutations exhibit severe speech defects15,16 while FOXP2 disruption in mice leads to vocalization abnormalities17,18 suggesting a role in spoken language function. Studies on mice with the two amino acids substituted to the human version have reported vocal changes both in the neonatal19 and adult stages20,21. While Hammerschmidt et al. observed minimal vocal changes, von Merten et al. reported qualitative changes under a more natural vocalization paradigm20,21, suggesting the involvement of these two amino acids in vocalization. However, these substitutions are also present in archaic humans, and comprehensive analyses using diverse human genome datasets have found no evidence of recent selection. This suggests that the FOXP2 substitutions occurred earlier than initially thought12,22. Similarly, the TKTL1 gene contains a human-specific amino acid thought to influence greater neurogenesis in humans than Neanderthal frontal cortex, though this finding is based on European ancestry genome datasets23. Broader analyses of modern human genomes reveal that 0.03–0.2% of individuals possess the ‘putative Neanderthal variant’, indicating its presence in a significant portion of the population14. These findings underscore the importance of incorporating diverse human samples to identify and validate the genetic background of modern human traits through genomic comparisons.
Genomic comparisons between archaic humans, ape genomes, and the broader human population have identified 61 human-specific nonsynonymous coding variants that are fixed or nearly fixed in modern humans13. Notably, one of the genes includes an isoleucine to valine substitution at position 197 (I197V) in the RNA binding protein neuro-oncological ventral antigen1 (NOVA1)8,13. NOVA1 is highly expressed in neurons of the central nervous system (CNS) in both mice and humans24, with its expression also noted in cultured human and rat cells25,26,27. NOVA1 was first identified as an autoantigen targeted in the paraneoplastic neurologic disorder (PND) opsoclonus-myoclonus ataxia (POMA)24. PNDs develop when tumor cells ectopically express proteins normally restricted to the nervous system, triggering an anti-tumor immune response that breaches the blood-brain barrier, leading to autoimmune neurologic disease28,29. In POMA, a robust immune response is mounted against NOVA1 and its paralog, NOVA230. The autoimmune disorder is characterized by motor dysfunction due to the failure midbrain neuron inhibition, resulting in the hyperactivity associated with opsoclonus-myoclonus ataxia24. In mice, homozygous deletion of the Nova1 gene results in an early postnatal lethal phenotype due to abnormal motor function31. Therefore, NOVA1 plays a crucial role in neural development and neuromuscular control in mammals. On a molecular level, NOVA proteins directly bind RNA in the mouse brain32,33 to regulate pre-mRNA processing31,34,35, translation36 and neurophysiology37. Genetic studies mapping NOVA target RNAs in mice and humans have also linked it to autism38. Interestingly, a human patient with a heterozygous deletion of NOVA1 presented with delay of language development, learning disabilities, motor hyperactivity and behavioral dysregulation36.
Studies have explored the role of the NOVA1 I197V variant by reverting the ancestral isoleucine 197 variant back into human iPSC-derived organoids reveal morphological and electrophysiological changes in vitro13,39. While these effects were not observed in a study that reintroduced the same substitution in different iPSCs40, technical concerns continue to make definitive conclusions about the nature of the NOVA1 I197V variant in brain challenging. Therefore, we generated humanized mice harboring this variant to study its consequences for RNA regulation and behavior in vivo.
In this work, we used gene-editing to substitute the NOVA1 isoleucine (I) isoform present in most mammals and archaic hominids (Neanderthals and Denisovans) with the human-specific valine (V) variant at position 197 in mice. Comparison of these humanized NOVA1 mice (Nova1hu/hu) with wild-type mice carrying the ancestral Nova1 gene (Nova1wt/wt) revealed specific transcriptomic and behavior differences related to vocalization. Taken together, the unique role of NOVA1 in neurons, its association with human disease, and evidence that the human-specific amino acid 197 variant confers vocalization changes in humanized mice suggest a role for NOVA1 in the evolution of human-specific language.
In line with studies of genetic variants that have played a role in the evolution of modern humans19,109,110, we investigated the biological effect of a single amino acid substitution, I197V in NOVA1, which is unique to modern humans. By analyzing Nova1hu/hu mice carrying this allele, we identified molecular changes in alternative splicing in the brain, including brain regions associated with vocal behavior, and identified changes in vocalization patterns in pups and adult mice. These findings suggest that during human evolution, the I197V substitution in NOVA1 protein may have contributed to the development of neural systems involved in more complex vocal communication.
The importance of NOVA1 in mammals is evident from the lethal phenotype of Nova1 knockout mice31 and the neurological symptoms caused by NOVA1 haploinsufficiency in humans36. This significance is further highlighted by the high conservation of the NOVA1 protein in mammals. Interestingly, the NOVA1 gene harbors an Ultra Conserved Element (UCE; uc.359) at the end of the 3’ UTR111,112 with additional high conservation extending upstream from the NOVA1 UCE to most of the 3’ UTR and terminal exon encoding NOVA1 KH2 and KH3 domains. This underscores the unique nature of the I197V variant, which occurred within a region of the genome resistant to change.
Previous studies have confirmed the evolutionary restriction of NOVA1 variants, including the I197V variant (termed I200V in one study13). We support this analysis and have expanded upon it with larger human sequence datasets across diverse ethnic groups, and from methods that infer selection coefficients from ancient samples48,49. These results confirm that NOVA1 has undergone strong positive selection and that the I197V variant is part of an evolutionary selective sweep in the emergence of Homo sapiens.
The observation that the I197V NOVA1 allele is nearly fixed across human populations (Supplementary Fig. 1a) suggests that it emerged and increased in frequency well before the divergence of ancient human lineages. The earliest split among modern human groups—that of the San - is currently estimated to have mostly occurred by roughly 200 kya, well before migration of modern humans out of Africa and the Near East to Eurasia around 50kya1,9,113. Unlike more recent selective sweeps, such as the LCT locus, which is dated around 10 kya, and is population-specific, the NOVA1 variant is part of an older, more widespread sweep. These older sweeps, shared across modern human populations, may leave subtler genetic signatures that require novel detection methods. This suggests that the ancient NOVA1 selective sweep may represent part of a broader set of undiscovered ancient sweeps.
One possible explanation for the changes in vocal behavior observed in Nova1hu/hu mice could be molecular changes in midbrain and brainstem vocal pathways, which express high levels of NOVA1 and are involved in regulating innate vocalizations (USVs), including breath coordination, timing, and amplitude114,115,116. An alternative possibility is that changes occurred in more recently evolved cortical vocal regions, which control pitch, frequency modulation, and duration (the Kuypers/ Jürgens hypothesis84,117 and the volitional articulatory motor network87,11
. Given that NOVA1 is expressed in the mouse cortex, predominantly in inhibitory neurons61, it is plausible that the I197V substitution affects cortical regulation of vocalization.
Notably, Nova1hu/hu mice exhibit qualitative changes in vocal characteristics compared to control mice both in pups and adults, despite producing a similar number of calls. These findings suggest that the vocalization changes in Nova1hu/hu mice are not simply the result of alterations in general motor performance. This idea is supported by other observations showing that Nova1hu/hu mice perform similarly to control mice in motor function tests, such as the rotarod, and display comparable locomotion activity levels in the Y-maze test (Supplementary Fig. 17). Additionally, the Y-maze test results indicated that Nova1hu/hu mice had spatial working memory comparable to that of control mice.
The changes in vocalization in Nova1hu/hu mice varied in a development- or context-dependent manner. It has been reported that high-frequency (Fq) USVs are emitted more frequently by male adult mice during social interactions119 and that female mice are attracted to male mice that emit more complex USVs96. Given these reports, the increased the proportion of higher Fq USVs in pups, and the increased complexity of these USVs in adults may potentially offer social advantages in mice. However, since auditory frequency resolution in mice has been reported to be limited120 and our preference experiment using humanized NOVA1 pup USVs showed no significant preferences in the mother’s responses, it remains unclear to what extent other mice can recognize these vocalization changes. It should be noted that we were unable to address the effect of the I197V substitution on the vocalizations of adult female mice, as our study focused on a courtship-induced vocalization paradigm that predominantly elicits USVs from male mice121. However, recent studies suggest that female mice also vocalize under certain experimental conditions or social contexts21,122,123. Future studies will be necessary to investigate the effects of the I197V substitution on USVs in female mice, as well as adult female preferences to USVs in adult Nova1hu/hu male mice.
Interestingly, the vocalization changes observed in Nova1hu/hu mice share some similarities with those observed in humanized Foxp2 mice (with two human-specific substitutions). In both cases, the changes were developmental or context-dependent and included a decrease in peak frequency in simple syllables and modulation of high-frequency regions in complex syllables19,20,21 (Supplementary Data 17). Conversely, male mice with a humanized Foxp2 mutation produced simpler song bouts with more “s” syllables17,18. These observations may indicate a common or related molecular alteration in the neural circuits involved in the USV production between humanized Nova1 mice and humanized Foxp2 mice. Future studies should aim to identify the molecular and neural basis of these alterations, as well as the physiological significance of these vocalization changes in the context of social behavior.
Our molecular analysis showed that the sequence-specific RNA binding of NOVA1 was unaffected by the human substitution and that steady-state gene expression levels in the brains of Nova1hu/hu mice were nearly identical to those of wild type mice. However, we detected alternative splicing changes in several transcripts associated with vocalization. The expression pattern of NOVA1 in the brain and the enrichment of its target transcripts to specific biological pathways support a link between NOVA1 function and vocal behavior. Uncovering the precise molecular mechanisms underlying the phenotypes in Nova1hu/hu mice will require further study of the neural circuits for vocalization, as well as on regulatory factors influencing NOVA protein function. This study sets the groundwork for understanding molecular mechanisms driving the evolution of human vocal communication.
Biochemically, NOVA proteins harbor three KH domains responsible for sequence-specific RNA-binding41,44,45, with amino acid 197 located in the KH2 domain. Although the I197V substitution in NOVA1 alters the hydrophobic core of the KH domain, it does not lead to a loss of RNA-binding capacity or functional attenuation, in contrast to other KH domain point mutations51,52,53. This is supported by the unchanged global gene expression levels observed in the brains of Nova1hu/hu mice, while NOVA1 knockout mice (which exhibit postnatal lethality31) show significant expression changes of key neuronal genes in the midbrain at E18.5 (Supplementary Fig. 5e, f, Supplementary Data 1. KH domains harbor three alpha-helices (H) and three beta-sheets (S) (S1-H1-H2-S2-S3-H3; Supplementary Fig. 9a, b); the first and second alpha helices (H1-H2) determine single-stranded RNA binding specificity44,45, and may also be involved in protein-protein dimerization62 (Supplementary Fig. 9c, d). Protein structure predictions suggest that the addition of a single carbon atom in valine 197 extends its ability to interact with several nearby amino acids (185Ile in H1; 232Ala, 235Leu, 236Ile, 239Lys in H3), allowing the KH2 domain to gain contact with H1 and 239Lys in H3. Thus, while the I197V substitution does not change sequence specificity or binding affinity of NOVA1 with RNA (Fig. 2i, j), it may affect KH domain dimerization or may have undiscovered effects on protein-protein interactions62. Interestingly, the amino acid corresponding to NOVA1 amino acid 197 is also an isoleucine in the related proteins FMR1 and hnRNP E1/E2/K, but is a valine in both human and mouse NOVA244 (Supplementary Fig. 7). Functional differences between NOVA1 and NOVA2 in mice43,61 may reflect structural and functional differences in their respective KH domains.
In summary, we analyzed a single amino acid unique to modern humans in the RNA binding protein NOVA1 and examined its biological effects in vivo by introducing this amino acid in mice. NOVA1 is highly intolerant to changes in amino acid sequences during evolution with the exception of this single amino acid change in humans. We propose that this change was part of an evolutionary sweep associated with specific changes in the neuronal transcriptome and vocal communication.
Article follows.
Archive | live (nature)
Abstract
NOVA1, a neuronal RNA-binding protein expressed in the central nervous system, is essential for survival in mice and normal development in humans. A single amino acid change (I197V) in NOVA1’s second RNA binding domain is unique to modern humans. To study its physiological effects, we generated mice carrying the human-specific I197V variant (Nova1hu/hu) and analyzed the molecular and behavioral consequences. While the I197V substitution had minimal impact on NOVA1’s RNA binding capacity, it led to specific effects on alternative splicing, and CLIP revealed multiple binding peaks in mouse brain transcripts involved in vocalization. These molecular findings were associated with behavioral differences in vocalization patterns in Nova1hu/hu mice as pups and adults. Our findings suggest that this human-specific NOVA1 substitution may have been part of an ancient evolutionary selective sweep in a common ancestral population of Homo sapiens, possibly contributing to the development of spoken language through differential RNA regulation during brain development.
Introduction
Fossil records indicate that modern humans (Homo sapiens) emerged 200,000–300,000 years ago as the predominant species from a common ancestral population1,2. Humans differ significantly from their closest living relatives, the great apes, particularly in their ability to communicate through complex learned vocal communication, a necessary component of spoken language3. This complexity is driven by some anatomical adaptions of the vocal tract and intricate neural networks linking various brain regions3,4,5,6,7. However, the genetic basis underlying these specialized human traits remains to be fully identified.
The closest evolutionary relatives of modern humans are two extinct lineages: Neanderthals and Denisovans. Genome sequencing from fossilized remains of these archaic humans has identified distinct genetic differences between them and modern humans, which may be relevant to recent human evolution8,9,10,11. Additionally, the availability of extensive human genome data over the past few decades initially focused on European populations, has significantly expanded the scope of evolutionary studies12,13,14.
The transcription factor forkhead box P2 (FOXP2) is of particular interest as a potential driver of human language function, as it harbors two amino acid substitutions present in humans but not in chimpanzees and many other mammalian genomes. Families with FOXP2 mutations exhibit severe speech defects15,16 while FOXP2 disruption in mice leads to vocalization abnormalities17,18 suggesting a role in spoken language function. Studies on mice with the two amino acids substituted to the human version have reported vocal changes both in the neonatal19 and adult stages20,21. While Hammerschmidt et al. observed minimal vocal changes, von Merten et al. reported qualitative changes under a more natural vocalization paradigm20,21, suggesting the involvement of these two amino acids in vocalization. However, these substitutions are also present in archaic humans, and comprehensive analyses using diverse human genome datasets have found no evidence of recent selection. This suggests that the FOXP2 substitutions occurred earlier than initially thought12,22. Similarly, the TKTL1 gene contains a human-specific amino acid thought to influence greater neurogenesis in humans than Neanderthal frontal cortex, though this finding is based on European ancestry genome datasets23. Broader analyses of modern human genomes reveal that 0.03–0.2% of individuals possess the ‘putative Neanderthal variant’, indicating its presence in a significant portion of the population14. These findings underscore the importance of incorporating diverse human samples to identify and validate the genetic background of modern human traits through genomic comparisons.
Genomic comparisons between archaic humans, ape genomes, and the broader human population have identified 61 human-specific nonsynonymous coding variants that are fixed or nearly fixed in modern humans13. Notably, one of the genes includes an isoleucine to valine substitution at position 197 (I197V) in the RNA binding protein neuro-oncological ventral antigen1 (NOVA1)8,13. NOVA1 is highly expressed in neurons of the central nervous system (CNS) in both mice and humans24, with its expression also noted in cultured human and rat cells25,26,27. NOVA1 was first identified as an autoantigen targeted in the paraneoplastic neurologic disorder (PND) opsoclonus-myoclonus ataxia (POMA)24. PNDs develop when tumor cells ectopically express proteins normally restricted to the nervous system, triggering an anti-tumor immune response that breaches the blood-brain barrier, leading to autoimmune neurologic disease28,29. In POMA, a robust immune response is mounted against NOVA1 and its paralog, NOVA230. The autoimmune disorder is characterized by motor dysfunction due to the failure midbrain neuron inhibition, resulting in the hyperactivity associated with opsoclonus-myoclonus ataxia24. In mice, homozygous deletion of the Nova1 gene results in an early postnatal lethal phenotype due to abnormal motor function31. Therefore, NOVA1 plays a crucial role in neural development and neuromuscular control in mammals. On a molecular level, NOVA proteins directly bind RNA in the mouse brain32,33 to regulate pre-mRNA processing31,34,35, translation36 and neurophysiology37. Genetic studies mapping NOVA target RNAs in mice and humans have also linked it to autism38. Interestingly, a human patient with a heterozygous deletion of NOVA1 presented with delay of language development, learning disabilities, motor hyperactivity and behavioral dysregulation36.
Studies have explored the role of the NOVA1 I197V variant by reverting the ancestral isoleucine 197 variant back into human iPSC-derived organoids reveal morphological and electrophysiological changes in vitro13,39. While these effects were not observed in a study that reintroduced the same substitution in different iPSCs40, technical concerns continue to make definitive conclusions about the nature of the NOVA1 I197V variant in brain challenging. Therefore, we generated humanized mice harboring this variant to study its consequences for RNA regulation and behavior in vivo.
In this work, we used gene-editing to substitute the NOVA1 isoleucine (I) isoform present in most mammals and archaic hominids (Neanderthals and Denisovans) with the human-specific valine (V) variant at position 197 in mice. Comparison of these humanized NOVA1 mice (Nova1hu/hu) with wild-type mice carrying the ancestral Nova1 gene (Nova1wt/wt) revealed specific transcriptomic and behavior differences related to vocalization. Taken together, the unique role of NOVA1 in neurons, its association with human disease, and evidence that the human-specific amino acid 197 variant confers vocalization changes in humanized mice suggest a role for NOVA1 in the evolution of human-specific language.
Discussion
In line with studies of genetic variants that have played a role in the evolution of modern humans19,109,110, we investigated the biological effect of a single amino acid substitution, I197V in NOVA1, which is unique to modern humans. By analyzing Nova1hu/hu mice carrying this allele, we identified molecular changes in alternative splicing in the brain, including brain regions associated with vocal behavior, and identified changes in vocalization patterns in pups and adult mice. These findings suggest that during human evolution, the I197V substitution in NOVA1 protein may have contributed to the development of neural systems involved in more complex vocal communication.
The importance of NOVA1 in mammals is evident from the lethal phenotype of Nova1 knockout mice31 and the neurological symptoms caused by NOVA1 haploinsufficiency in humans36. This significance is further highlighted by the high conservation of the NOVA1 protein in mammals. Interestingly, the NOVA1 gene harbors an Ultra Conserved Element (UCE; uc.359) at the end of the 3’ UTR111,112 with additional high conservation extending upstream from the NOVA1 UCE to most of the 3’ UTR and terminal exon encoding NOVA1 KH2 and KH3 domains. This underscores the unique nature of the I197V variant, which occurred within a region of the genome resistant to change.
Previous studies have confirmed the evolutionary restriction of NOVA1 variants, including the I197V variant (termed I200V in one study13). We support this analysis and have expanded upon it with larger human sequence datasets across diverse ethnic groups, and from methods that infer selection coefficients from ancient samples48,49. These results confirm that NOVA1 has undergone strong positive selection and that the I197V variant is part of an evolutionary selective sweep in the emergence of Homo sapiens.
The observation that the I197V NOVA1 allele is nearly fixed across human populations (Supplementary Fig. 1a) suggests that it emerged and increased in frequency well before the divergence of ancient human lineages. The earliest split among modern human groups—that of the San - is currently estimated to have mostly occurred by roughly 200 kya, well before migration of modern humans out of Africa and the Near East to Eurasia around 50kya1,9,113. Unlike more recent selective sweeps, such as the LCT locus, which is dated around 10 kya, and is population-specific, the NOVA1 variant is part of an older, more widespread sweep. These older sweeps, shared across modern human populations, may leave subtler genetic signatures that require novel detection methods. This suggests that the ancient NOVA1 selective sweep may represent part of a broader set of undiscovered ancient sweeps.
One possible explanation for the changes in vocal behavior observed in Nova1hu/hu mice could be molecular changes in midbrain and brainstem vocal pathways, which express high levels of NOVA1 and are involved in regulating innate vocalizations (USVs), including breath coordination, timing, and amplitude114,115,116. An alternative possibility is that changes occurred in more recently evolved cortical vocal regions, which control pitch, frequency modulation, and duration (the Kuypers/ Jürgens hypothesis84,117 and the volitional articulatory motor network87,11
. Given that NOVA1 is expressed in the mouse cortex, predominantly in inhibitory neurons61, it is plausible that the I197V substitution affects cortical regulation of vocalization.Notably, Nova1hu/hu mice exhibit qualitative changes in vocal characteristics compared to control mice both in pups and adults, despite producing a similar number of calls. These findings suggest that the vocalization changes in Nova1hu/hu mice are not simply the result of alterations in general motor performance. This idea is supported by other observations showing that Nova1hu/hu mice perform similarly to control mice in motor function tests, such as the rotarod, and display comparable locomotion activity levels in the Y-maze test (Supplementary Fig. 17). Additionally, the Y-maze test results indicated that Nova1hu/hu mice had spatial working memory comparable to that of control mice.
The changes in vocalization in Nova1hu/hu mice varied in a development- or context-dependent manner. It has been reported that high-frequency (Fq) USVs are emitted more frequently by male adult mice during social interactions119 and that female mice are attracted to male mice that emit more complex USVs96. Given these reports, the increased the proportion of higher Fq USVs in pups, and the increased complexity of these USVs in adults may potentially offer social advantages in mice. However, since auditory frequency resolution in mice has been reported to be limited120 and our preference experiment using humanized NOVA1 pup USVs showed no significant preferences in the mother’s responses, it remains unclear to what extent other mice can recognize these vocalization changes. It should be noted that we were unable to address the effect of the I197V substitution on the vocalizations of adult female mice, as our study focused on a courtship-induced vocalization paradigm that predominantly elicits USVs from male mice121. However, recent studies suggest that female mice also vocalize under certain experimental conditions or social contexts21,122,123. Future studies will be necessary to investigate the effects of the I197V substitution on USVs in female mice, as well as adult female preferences to USVs in adult Nova1hu/hu male mice.
Interestingly, the vocalization changes observed in Nova1hu/hu mice share some similarities with those observed in humanized Foxp2 mice (with two human-specific substitutions). In both cases, the changes were developmental or context-dependent and included a decrease in peak frequency in simple syllables and modulation of high-frequency regions in complex syllables19,20,21 (Supplementary Data 17). Conversely, male mice with a humanized Foxp2 mutation produced simpler song bouts with more “s” syllables17,18. These observations may indicate a common or related molecular alteration in the neural circuits involved in the USV production between humanized Nova1 mice and humanized Foxp2 mice. Future studies should aim to identify the molecular and neural basis of these alterations, as well as the physiological significance of these vocalization changes in the context of social behavior.
Our molecular analysis showed that the sequence-specific RNA binding of NOVA1 was unaffected by the human substitution and that steady-state gene expression levels in the brains of Nova1hu/hu mice were nearly identical to those of wild type mice. However, we detected alternative splicing changes in several transcripts associated with vocalization. The expression pattern of NOVA1 in the brain and the enrichment of its target transcripts to specific biological pathways support a link between NOVA1 function and vocal behavior. Uncovering the precise molecular mechanisms underlying the phenotypes in Nova1hu/hu mice will require further study of the neural circuits for vocalization, as well as on regulatory factors influencing NOVA protein function. This study sets the groundwork for understanding molecular mechanisms driving the evolution of human vocal communication.
Biochemically, NOVA proteins harbor three KH domains responsible for sequence-specific RNA-binding41,44,45, with amino acid 197 located in the KH2 domain. Although the I197V substitution in NOVA1 alters the hydrophobic core of the KH domain, it does not lead to a loss of RNA-binding capacity or functional attenuation, in contrast to other KH domain point mutations51,52,53. This is supported by the unchanged global gene expression levels observed in the brains of Nova1hu/hu mice, while NOVA1 knockout mice (which exhibit postnatal lethality31) show significant expression changes of key neuronal genes in the midbrain at E18.5 (Supplementary Fig. 5e, f, Supplementary Data 1
. KH domains harbor three alpha-helices (H) and three beta-sheets (S) (S1-H1-H2-S2-S3-H3; Supplementary Fig. 9a, b); the first and second alpha helices (H1-H2) determine single-stranded RNA binding specificity44,45, and may also be involved in protein-protein dimerization62 (Supplementary Fig. 9c, d). Protein structure predictions suggest that the addition of a single carbon atom in valine 197 extends its ability to interact with several nearby amino acids (185Ile in H1; 232Ala, 235Leu, 236Ile, 239Lys in H3), allowing the KH2 domain to gain contact with H1 and 239Lys in H3. Thus, while the I197V substitution does not change sequence specificity or binding affinity of NOVA1 with RNA (Fig. 2i, j), it may affect KH domain dimerization or may have undiscovered effects on protein-protein interactions62. Interestingly, the amino acid corresponding to NOVA1 amino acid 197 is also an isoleucine in the related proteins FMR1 and hnRNP E1/E2/K, but is a valine in both human and mouse NOVA244 (Supplementary Fig. 7). Functional differences between NOVA1 and NOVA2 in mice43,61 may reflect structural and functional differences in their respective KH domains.In summary, we analyzed a single amino acid unique to modern humans in the RNA binding protein NOVA1 and examined its biological effects in vivo by introducing this amino acid in mice. NOVA1 is highly intolerant to changes in amino acid sequences during evolution with the exception of this single amino acid change in humans. We propose that this change was part of an evolutionary sweep associated with specific changes in the neuronal transcriptome and vocal communication.
