Gene-editing technologies have made it feasible to create nonhuman primate models for human genetic disorders. Here, we report detailed genotypes and phenotypes of TALEN-edited MECP2 mutant cynomolgus monkeys serving as a model for a neurodevelopmental disorder, Rett syndrome (RTT), which is caused by loss-of-function mutations in the human MECP2 gene. Male mutant monkeys were embryonic lethal, reiterating that RTT is a disease of females. Through a battery of behavioral analyses, including primate-unique eye-tracking tests, in combination with brain imaging via MRI, we found a series of physiological, behavioral, and structural abnormalities resembling clinical manifestations of RTT. Moreover, blood transcriptome profiling revealed that mutant monkeys resembled RTT patients in immune gene dysregulation. Taken together, the stark similarity in phenotype and/or endophenotype between monkeys and patients suggested that gene-edited RTT founder monkeys would be of value for disease mechanistic studies as well as development of potential therapeutic interventions for RTT.

In a previous study, we reported the successful application of TALEN-mediated mutagenesis to produce MECP2 mutant monkeys (), which opened up great opportunities for using monkey models to better understand RTT and to develop potential treatments for the disease. In this study, using the same technology, we generated additional MECP2 mutants, from which we recapitulated embryonic lethality of mutant males and physiological and behavioral abnormalities of mutant females, including primate-unique eye tracking, which are seen with RTT patients. Moreover, brain imaging by MRI, MeCP2 protein measurements by semiquantitative western blotting, and transcriptome profiling of peripheral blood all demonstrated that these phenotypes (and/or endophenotypes) of the MECP2 mutant monkeys were strikingly similar to RTT clinical manifestations, which can be used for disease mechanistic studies as well as development of potential therapeutic interventions for RTT.

As RTT is a monogenic disorder, genetic modification technologies have made it possible to develop animal models for further study. RTT animal models were first generated in mice and recently in rats (). It is interesting that RTT-related neurological phenotypes mostly occur in adult male rodents, which is different from the human disease (). It is therefore conceivable that gene-edited nonhuman primates (NHPs) would serve as a better choice for modeling genetic neurological disorders including RTT (). Although monkeys and humans are different in many ways given that evolution has separated the Old World primate lineage into different paths, NHPs, such as rhesus and cynomolgus monkeys, still share high levels of genetic, physiological, social behavioral, and CNS developmental similarities with humans, making them suitable for the study of neurodevelopmental disorders ().

Rett syndrome (RTT) is a progressive neurodevelopmental disorder that mostly manifests in girls with a morbidity rate of 1:10,000–1:15,000 (). Almost 95% of RTT is believed to be caused by mutations of an X-linked gene methyl-CpG-binding protein 2 (MECP2) (). MECP2 mutations are most often embryonic lethal for boys, except for very few, who are born with severe encephalopathy leading to death before 2 years of age (). RTT girls seem to have normal development for up to 6–18 months but manifest a series of symptoms associated with intellectual disability, loss of acquired language, and compromised cognitive, social, and motor skills, etc. ().

Gene ontology (GO) term enrichment of biological process revealed that genes in blue and brown modules were heavily associated with immune responses, which was downregulated in mutant monkeys. This indicates potential deficits in immune functions, which is associated with MECP2 deficiency. In contrast, genes involved in protein translation and RNA processing (turquoise module) were upregulated in mutants ( Figures 6 C–6H). To assess the similarity between mutant monkeys and RTT patients, we analyzed the blood transcriptome sequencing in humans from six patients and five healthy age-matched controls. RTT patients’ blood samples also revealed modules that were either downregulated and associated with “immune functions” or upregulated and associated with “translation” and “RNA processing” ( Figure S4 ). Moreover, we found a list of genes that showed similar changes in monkeys and human correlated with MECP2 deficiency ( Figures 6 I and 6J and S4 ). These gene expression changes could be potentially used as biomarkers for RTT.

Due to high-level expression of MECP2 in neuronal cells, RTT is mainly a nervous system disease. Nevertheless, RTT patients also have many peripheral symptoms including cardiac, pulmonary, muscle, bone, gastro-intestinal, and immune system abnormalities, which may or may not be a direct link to dysfunctions of the nervous system. The peripheral symptoms above could also arise from MeCP2 loss of function in peripheral tissues, because MeCP2 is also expressed in many non-neural cells including fibroblasts. Since peripheral blood from RTT patients is more easily accessible than other tissues, analyses of blood samples for biomarkers may facilitate diagnosis and/or serve as outcome measures for potential clinical interventions. We thus profiled blood transcriptomes of monkeys to uncover potential endophenotype or biomarkers. Through weighted gene coexpression network analyses (WGCNA) (), we isolated three interconnected gene modules/clusters, which were expressed at different levels between mutant and WT monkeys ( Figure 6 A). Module-trait correlation matrix demonstrated that these gene clusters were either downregulated (brown and blue modules) or upregulated (turquoise module) in MECP2 mutant animals with statistical significance ( Figure 6 B).

(C) Eye-tracking pattern for emotional faces, including aggression, submission, and neutral. (a) Lines show the differences between the two groups in preference of emotional charged faces. The mutant monkeys preferred looking at neutral faces compared with aggressive or submissive faces, but no significant differences were found between the two groups in both looking counts (F = 1.730, p = 0.199, η p 2 = 0.126) and looking duration (F = 2.307, p = 0.121, η p 2 = 0.161). (b) Histograms show that differences between two groups in looking counts and looking duration at monkey emotional faces. The mutant monkeys looked fewer times at aggressive (F = 13.919, p = 0.006) and submissive faces (F = 10.714, p = 0.011), and shorter duration at aggressive (F = 8.473, p = 0.020) and submissive faces (F = 14.965, p = 0.005). When presented with neutral faces, no significant difference was observed between the two groups, including looking counts (F = 3.652, p = 0.092) and looking duration (F = 0.408, p = 0.541). The measurements of looking counts and looking duration were obtained from eye-tracking paradigm. Bars and error bars represent mean ± SEM of replicate measurements. ns p > 0.05, ∗ p < 0.05 (two-way ANOVA for interaction test and one-way ANOVA for pairwise test).

(B) Eye-tracking pattern for face-object stimuli. (a) Lines show the differences between the two groups in preference of monkey faces or objects. The mutant monkeys preferred looking at monkey faces as compared to WT controls, including looking counts (F = 8.747, p = 0.009, η p 2 = 0.353) and looking duration (F = 15.443, p = 0.001, η p 2 = 0.491). (b) Histograms show that differences between the two groups in looking counts and looking duration at monkey faces versus objects. No significant differences were found between two groups in looking counts, but the mutants spent more time looking at monkey faces than WT controls (F = 9.695, p = 0.014).

(A) View patterns of mutants (n = 5) and WT control monkeys (n = 5) in paired face-object task and paired emotional faces task. (a) Flow chart of eye-tracking task. The apostrophe indicates more test cycles. (b) An example of looking at one paired object-face stimuli. (c) An example of looking at one paired emotional face stimuli. The looking circle with the number “1” at the center represents the first look at this image. Subsequent looking numbered two traces the location and order of the next looks. Looking duration was in proportion to the diameter of the looking circles.

Using primate-unique eye-tracking measures, five mutant female monkeys and five age- and gender-matched WT controls were assessed for social preferences (when presented with monkey faces versus objects) and emotional processing (when presented with monkey faces in different emotional states). Mutant monkeys exhibited a preference for socially weighted stimuli, as they spent more time looking at monkey faces than inanimate objects ( Figure 5 A; Movie S5 ). However, when presented faces with aggressive or submissive emotions versus neutral faces, the mutants showed less interest in emotionally charged expressions but preferred neutral faces, as compared with WT monkeys ( Figure 5 B; Movie S5 ). These results are consistent with eye-tracking features of RTT patients who show more interest in socially weighted stimuli but have difficulties recognizing emotional expressions ().

Social interaction behaviors were video-recorded and analyzed to assess active interactions. In this test, we found that mutant monkeys exhibited less active social contact, less environmental exploration, but more stereotypical behaviors, compared to WT controls. Moreover, mutant monkeys did not show aggressive behaviors as WT controls did ( Figure 4 B, d–g; Movies S3 and S4 ). These results demonstrated stark similarity between mutant monkeys and RTT patients.

Monkeys were under 24-hr continuous monitoring for assessment of activity levels. Compared to WT controls, mutants showed no significantly lower levels of activity during the 24 hr monitoring (WT versus mutant: 19.412 versus 16.522, p = 0.665). During light time (8:00–20:00), no significant difference was observed between mutants and WT controls in levels of activity (mutants versus WT = 30.504 versus 37.920, p = 0.574), and there was also no significant difference in dark time (20:00–8:00) activities (mutants versus WT = 2.542 versus 0.906, p = 0.299) ( Figure 4 B, c).

Active avoidance test was conducted with a high decibel level noise (120 dB noise) that was used as a source to trigger fear responses in monkeys according to a previous report (). The mutant monkeys could hear the auditory stimuli ( Movie S2 ), but they covered their ears and hid themselves in the corner of the cage rather than escaping as WT controls did when the noise was presented ( Figure 4 B, b, Movie S2 ). These behaviors are very similar to those found in RTT patients with retarded reactions () or in mouse models ().

We examined monkeys’ responses to heat stress and found that WT monkeys moved their feet frequently and constantly in response to rising temperatures, while mutants made no attempt to escape from the heating plate ( Movie S1 ). When temperature rose further to certain threshold, both mutant monkeys and WT controls exhibited nocifensive behaviors involving withdrawal and/or licking of either hind- and/or forelimbs, but mutants had a higher threshold for pain sensation toward higher temperatures ( Figure 4 B, a).

(B) Responses to environments. (a) Pain perception of mutants and WT monkeys. The mutants’ retarded responses to heat indicated they had a higher threshold for pain than WT controls (mutants versus WT = 48.192: 45.037°C, p = 0.030). (b) Active avoidance to noise exhibited by mutants and WT. The mutant monkeys showed delayed responses to noise, as it took longer for mutants to escape than the WT controls did (mutants versus WT = 124.067: 1.267 s, p = 1.345 × 10 −5 ). (c) Activity. Compared with WT control monkeys, mutants showed no significantly lower levels of activity during the 24 hr monitoring (mutants versus WT = 16.522: 19.412, p = 0.665). During light time (8:00–20:00), no significant difference was observed between mutants and WT controls in levels of activity (mutants versus WT = 30.504: 37.920, p = 0.574), and there was also no significant difference in dark time (20:00–8:00) activities (mutants versus WT = 2.542: 0.906, p = 0.299). (d) Aggressive behaviors. Mutant monkeys exhibited no aggressive behaviors, which was significantly different from WT controls (mutants versus WT = 0: 10.200 f/h, p = 7.772 × 10 −6 ). (e) Environmental exploration. Mutant monkeys exhibited less frequencies of environmental exploration than WT controls (mutants versus WT = 2.933: 12.800 f/h, p = 0.035). However, no significant difference was found in duration of environmental exploration (mutants versus WT = 22.200: 112.867 s/h, p = 0.053). (f) Stereotypical behaviors. Mutant monkeys exhibited more stereotypical behaviors than WT controls not only in frequencies (mutants versus WT = 34.667: 2.933 f/h, p = 0.021) but also in duration (mutants versus WT = 423.600: 16.667 s/h, p = 0.009). (g) Social contact. Mutant monkeys exhibited less social contact than WT controls in both frequencies (mutants versus WT = 17.267: 49.400 f/h, p = 0.001) and duration (mutants versus WT = 922.867: 1688.800 s/h, p = 0.037). All the behavioral tests above were conducted in five mutants and five age-matched WT monkeys. The “f/h” means frequencies per hour, and the “s/h” means seconds per hour.

(A) Sleep pattern. (a and b) The mean value of awake (a) and asleep (b) durations (minutes per night) of mutant and WT monkeys. Mutant monkeys were in a longer state of awake (mutants versus WT = 248.800: 213.467 min per night, p = 0.050) and a shorter state of asleep (mutants versus WT = 531.200: 566.533 min per night, p = 0.050) than WT monkeys, but the differences have no significance (p = 0.05). (c) Sleep fragment (frequencies per night). The data suggested the mutant monkeys had trouble sleeping, as referred from the substantial number of bouts compared with WT controls (mutants versus WT = 41.333: 26.200 numbers per night, p = 0.018). (d) Video recording of an overnight alternation of sleep and wake of a random selected mutant and an age-matched WT monkey (12 hr from 7:00 p.m. to 7:00 a.m.). The patterns indicated there were more frequent naps for the mutant monkey.

Sleep pattern was classified into awake and asleep phases (including relaxed and transitional sleep). Frequency and duration will be an important index to evaluate the sleep pattern. Thus, we recorded and analyzed these two indexes in MECP2 mutants and WT control monkeys. The total awake durations of mutants were found to be significantly longer while the total sleep durations were shorter than those of WT controls ( Figure 4 A, a). Remarkably, sleep in mutants was more fragmented than that in control monkeys, with more bouts of awaking and sleep ( Figure 4 A, b–d). It seemed difficult for mutant monkeys to have a continuous and longer-hour sleep, which is similar to RTT patients with abnormal circadian rhythms as well as Mecp2 mutant mice ().

According to clinical diagnostic criteria for RTT (), behavioral symptoms such as loss of acquired purposeful hand skills, loss of acquired spoken language, development of hand stereotype, inappropriate reactivity to environments, impaired locomotion, disruption in social interaction, social withdrawal, sleeping disorder, and reduced response to pain are often used to diagnose the disease before genetic testing for MECP2 mutations. Based on aforementioned symptoms, we used a battery of behavioral assays to analyze the phenotype of MECP2 mutant monkeys.

In addition, to investigate whether MECP2 mutant monkeys have any differences in growth and development compared to age- and gender- matched WT controls, basic parameters including body weight, body length, head circumference, and biparietal diameter were measured every 3 months. There was no difference between mutants and WT monkeys in these parameters ( Figure S3 ).

The heart rate (HR) and QTc of five mutants and five WT monkeys. The mean value of HR in mutants was lower than that in WT monkeys, while QTc measurements in mutants were significantly prolonged compared with WT monkeys. The normality of data for heart rate and electrocardiogram was analyzed by Kolmogorov-Smirnov test (p > 0.05), and one-way ANOVA was used to compare mutant monkeys with WT monkeys. f/m, frequencies per min. Bars and error bars represent mean ± SEM of replicate measurements. ∗ p < 0.05 (one-way ANOVA).

Abnormalities of electrocardiogram (EKG) records including QT interval (QTc) value and heart rate (HR) were found in patients (). We tried EKG recording on five mutant monkeys and five WT controls. The data were analyzed by both Kolmogorov-Smirnov test and one-way ANOVA (p > 0.05). We found that mutant monkeys exhibited lower HR and longer QTc ( Figure 3 ), which resembled RTT patients.

(A) Based on the findings of the dynamic changes in volumes shown in Figure 2 , the volume of the right posterior parahippocampal gyrus and the right cingulated at 8 M, and that of the right corpus callosum at 20 M decreased significantly in mutant monkeys. (B) Based on the findings of the dynamic changes in thickness shown in Figure S1 , there were significant decrease in the right inferior temporal gyrus, the left annectant gyrus, the right posterior parahippocampal gyrus, the right fusiform gyrus, the right and left precuneus at 8 M in mutant monkeys. (C) Based on the findings of the dynamic changes in surface area shown in Figure S1 , there were significant reduction in the right occipital gyrus, the left occipital gyrus, the right inferior occipital gyrus, and the left posterior parahippocampal gyrus at 20 M in mutant monkeys.

Bars and error bars represent mean ± SEM of replicate measurements.p > 0.05,p < 0.05 (rank-sum test). For more information and further analysis, see also Figures S1 and S2 . General growth and development of mutants and WT monkeys were also monitored; for the result, see also Figure S3

(A) Regional measurements on cortical volume in mutant monkeys showed significantly smaller values in several regions (A), including in bilateral parahippocampal gyrus at 8 months, left lingual gyrus at 8 months, left occipital gyrus at 8, 15, and 20 months, right lateral orbital gyrus at 8 months, right inferior occipital gyrus at 8 months, right inferior temporal gyrus at 8 and 20 months, right occipital gyrus at 15 and 20 months, right annectant gyrus at 20 months, and right lingual gyrus at 20 months (a–j).

Non-invasive neuroimaging is a very useful approach to assist diagnosis and monitoring longitudinal disease progression over time, which is particularly suitable for neural developmental disorders including RTT (). We used MRI scanning to measure global or regional structural changes in brains of MECP2 mutant monkeys. Similar MRI scanning has been done with RTT patients and reported (). In our study, MRI scanning was conducted three times with the first scanning performed at 7–8 months after birth, the second at 15 months, and the third at 20 months. A total of 60 subregions of each monkey’s brain was segmented, and data were analyzed by both rank-sum test and false discovery rate (FDR) to correct for multiple comparisons. Mutant monkeys showed significantly reduced cortical gray matter (GM) volumes in 16 regions at least at one stage when compared to normal controls ( Figure 2 , a–j). Younger (8 months old) mutant monkeys had more cortical areas showing statistically significant reductions as compared to WT controls, whereas at the 20month of age only bilateral occipital gyri, right inferior temporal gyrus, and right annectant gyrus displayed statistically significant reduction in volumes. In contrast, the difference in right annectant gyrus only became significant at the 20month but not at younger ages ( Figure 2 , a–j). Widespread reductions in GM volumes of specific sub-cortical areas and region-specific white matter (WM) volumes were also apparent in MECP2 mutant monkeys ( Figure 2 , k–p). Total brain GM and WM volumes, as well as total cortical volumes, total cortical surface areas, and average cortical thicknesses, were consistently smaller (with mean values) at all time points (i.e., 8, 15, and 20 months) in mutant monkeys, compared with WT controls, but the differences did not reach statistical significance ( Figure S1 ). When these parameters were measured in a regional-specific manner, statistically significant alterations in mutant monkeys could be observed ( Figures S1 and S2 ). Given that all five MECP2 mutant monkeys have different repertoires of MECP2 mutations, these results suggest that abnormal brain development in mutant monkeys, including reductions of subregional GM and WM volumes, specific cortical surface area, and cortical thickness, are likely due to MeCP2 deficiency. This is further supported by consistent abnormalities reported in previous MRI studies on human subjects with RTT ().

Concerning potential off-target effects, we used a genome-wide TALEN off-target site prediction tool named TALENoffer (). The statistical model of TALENoffer assumes that possible nucleotides on a target site can be traced from the repeat-variable di-residue of the corresponding repeat, which determines DNA binding specificity. Using a definite cutoff (TALENoffer score > −1.8) to define potential off-targets in the whole genome, a total of 41 exonic loci were predicted to be potential off-target sites ( Table S3 ). The fragments of all the potential off-target loci from three pairs of TALENs were PCR amplified and subjected to Sanger sequencing. No mutations were found on these sites ( Table S3 ). These data indicate that we did not detect off-target mutations on all 41 potential off-target sites from the four new monkeys and 11 WT control monkeys ( Table S3 ).

MeCP2 protein levels in tissues (placentae and fibroblasts) of live monkeys and in brains (cortices) of aborted fetuses were overall lower in mutants as compared with those for WT controls ( Figure 1 G). Reduction of protein levels is shown in all five mutant monkeys ( Figure 1 G, a and b). Since it is not feasible to sample from cerebral cortices from live female monkeys due to technical and ethical concerns, we only compared MeCP2 levels in cortices from aborted fetuses from mutants and WT monkeys. Data from aborted male monkeys’ cerebral cortices showed substantial reductions in MeCP2 levels ( Figure 1 G, c), which indirectly suggested that brain MeCP2 levels were also likely to be reduced in live mutant female monkeys. These data support the previous report that MeCP2 is a key regulator for neural development, which affects brain functions and behaviors ().

Mutations of the MECP2 gene from live monkeys and aborted fetuses were verified through T7EN1 cleavage, Sanger sequencing, and genome alignment. Live female mutants were further analyzed using Sanger sequencing of PCR-based clones from accessible tissues and compared with the reference for detection of missense, nonsense mutations, and/or indels on exon 3 of the MECP2 gene ( Table S2 and Figures 1 C–1F). Distributions and frequencies of mutations for the #1 female MECP2 mutation-carrying monkey (ID: 130958) can be found in the previous report (). As for the other four newborn mutant females, mutations detected in placenta, blood, and skin were plotted in Figure 1 C. All detected point mutations were plotted based on their positions and mutation rate in the entire targeted 351 bp of exon III.

(G) Comparison of protein levels between the mutants and WT controls via western blots. (a and b) Western blots and statistical analysis of MeCP2 protein expressions in lysate of placenta (a) and skin fibroblasts (b) from alive mutant monkeys and WT controls; (c) western blots and statistical analysis of cortices from aborted WT and mutant fetuses. Western blot image data were averaged from five repeats with ImageJ software. Compared with WT monkeys (n = 3, relative MeCP2 value is set as 100%), all of the mutant monkeys exhibited lower MeCP2 protein expression. For MeCP2 expression levels in placenta of monkeys #130958, 142002, 142004, 152004, and 152014 were 44.71%, 16.47%, 55.26%, 54.28%, and 63.57%, respectively, lower than those in WT; in fibroblasts, the numbers were 61.92%, 25.79%, 53.63%, 57.10%, and 9.46%, lower than WT. In the cortices, male monkeys Mut1 and Mut2 expressed MeCP2 at levels 51.36% and 54.74%, respectively, lower than WT (n = 3, relative MeCP2 value is set as 100%).

(D) T7EN1 cleavage analysis and mutations of the placenta and skin tissue of five alive female monkeys. Mutation rates of five female monkeys calculated from Sanger sequencing on exon 3 of MECP2 (a–e). From total clones sequenced, perfect match (WT) and mismatch (mutant) sequences were identified comparing with reference sequences and counted for calculating mutation rates (f). For more mutant information, see also Tables S1 and S2

In our previous report, one live female mutant monkey and three spontaneously aborted male mutant fetuses were generated using TALENs (). After that, we generated another six live monkeys (four female mutants and two non-affected wild-type [WT] males) and five spontaneously aborted fetuses (two male mutants, two female mutants, one non-affected WT female). The fact that all male mutant monkeys were embryonic lethal recapitulated the human disease () but different from Mecp2 knockout mice. In this study, we gathered all available animals, i.e., five female mutants and five spontaneously aborted male fetuses from 41 surrogate recipients, and carried out the following studies ( Figures 1 A and 1B; Table S1 ).

Discussion

RTT is a genetic neurodevelopmental disorder, currently with no cure. A lack of good animal models could potentially contribute to the slow progress of better understandings of the disease and finding cures. Although MECP2 mutant male mice exhibit phenotypes that resemble some of the symptoms of RTT female patients, it has been difficult or impossible to model more sophisticated disease symptoms in rodents. For example, Mecp2 knockout rodents show “limb-clasping” phenotype, but this phenotype is shared by neurodegeneration conditions in mice, which is not characteristic for RTT. In contrast, monkey models showed complicated behavioral features, such as fragmented sleep, increased stereotypy, reduced active avoidance of noisy or heat stimuli, and reduced environmental exploration, all of which resemble symptoms of patients. These activities are not widely set up for detection in mouse models.

Some of the characteristics of RTT displayed only in monkey models include male embryonic lethality, social withdrawal, and eye-tracking features, which have never been reported in rodents. Also, MRI of mutant monkeys provided information on longitudinal changes of brain structures with time, which could be very helpful for dissecting mechanisms of RTT. Although it is difficult to use MRI in clinical settings to track early-stage neurodevelopmental changes in patients, limited human RTT patient MRI data showed striking similarity with the monkey data and further attest to the validity of our monkey models. Obviously, longitudinal MRI studies in rodents will be close to impossible, and, even if they are done, the short lifespan and rapid brain development of rodents as well as dramatic differences in brain structures between human and rodents will potentially make rodent models even less compelling. Moreover, in peripheral blood transcriptome sequencing, the genetic changes of monkey models overlap with that of RTT patients, which could be important molecular indicators to reflect RTT.

Liu et al., 2016 Liu Z.

Li X.

Zhang J.T.

Cai Y.J.

Cheng T.L.

Cheng C.

Wang Y.

Zhang C.C.

Nie Y.H.

Chen Z.F.

et al. Autism-like behaviours and germline transmission in transgenic monkeys overexpressing MeCP2. This work is also different from a previous report on monkey models of MECP2 duplication syndrome (). In that report, monkeys with overexpressed MECP2 were generated via virus-induced gain-of-function expression of exogenous MECP2, which shares little similarity in terms of disease mechanisms with RTT. The ectopic pan-neuronal expression of MECP2 transgene may generate ectopic phenotypes, which are different from clinical MECP2 duplication mutations, where increased MECP2 expression only happens in neurons that originally express the gene, as not all neurons express MECP2. MECP2 mutant monkeys in our case were generated through TALEN-mediated gene editing that induced MECP2 deficiency, which is believed to be direct cause for RTT and thus making it possible to study the disease mechanism.

Great progress in neuroscience has been made and will continue by studying rodents and other simpler species, or even directly by clinical cases. NHP research will surely not replace these approaches and should be considered only in cases in which monkeys are the best and only available models for research. It seems RTT is a typical case in which monkey models may serve as a great system to study the mechanism and progression of the disease and to deliver new and effective treatments, since brains of other species are just too dissimilar to human brains. Non-primate models certainly are not able to mimic the sophisticated features of cognition, social interactions, etc. Our study showed that a valid monkey model might offer robust phenotype or endophenotype, which could be implemented as outcome measures in human clinical trials in the future to facilitate drug development.