The FMR1 gene plays a critical role in the expression of a range of clinical phenotypes, including both neurodevelopmental and neurodegenerative disorders. Located in the 5′ untranslated region (5′ UTR) on the long arm of the X chromosome, FMR1 encodes the Fragile X Mental Retardation Protein (FMRP), which is highly expressed in the brain and plays a role in synaptic plasticity (1–4). A full mutation of the FMR1 gene (>200 cytosine-guanine-guanine [CGG] trinucleotide repeats) causes methylation and subsequent silencing of the gene, inhibiting production of FMRP and causing fragile X syndrome (FXS), a rare condition (~1 in 4,000 males and ~1 in 8,000 females) that is the most common inherited cause of intellectual disability and monogenic cause of autism spectrum disorder (ASD) (5–7). The FMR1 premutation (PM; 55–200 CGG repeats), occurs in roughly 1 in 150–250 women, and is less prevalent in men (~1 in 430–460) (8–12). Though the PM was once believed to have limited phenotypic expression, a number of clinical and subclinical phenotypes have since been identified, including PM-specific disorders (i.e., fragile X-associated primary ovarian insufficiency [FXPOI], fragile X-associated tremor/ataxia syndrome [FXTAS], and fragile X-associated neuropsychiatric disorders [FXAND]) (4, 13, 14). FMR1-related molecular genetic variation has been associated with these phenotypes, including CGG repeat length, such as mid-range vulnerability (90–110 repeats), toxic gain-of-function production of mRNA, RAN translation, and FMRP variation (15–23). As such, detailed phenotypic characterization of the PM is important from a public health perspective, with potential to connect complex human traits to known genetic variation.

An important body of work has described a number of clinical-behavioral traits among carriers of the FMR1 PM, including ASD and subclinical ASD-related traits (24–27), anxiety and depression (28–33), and differences in social cognition and executive function (EF) (34–38). Because most prior work has examined these phenotypes in separate study samples, a key remaining question concerns whether such phenotypes co-occur, together constituting a phenotypic profile associated with the FMR1 PM. Examining the co-occurrence of key phenotypes within individuals can also help to address whether such features may interrelate causally. It could be, for instance, that personality traits previously reported in PM groups reflect underlying differences in social cognition or EF. This study attempted to build on prior work to address these gaps by studying a range of clinical-behavioral, social-cognitive, and executive phenotypes associated with the FMR1 PM within a relatively large sample of female PM carriers and controls.

Consistent with the strong phenotypic overlap observed between FXS and ASD (where most individuals with FXS exhibit at least some ASD symptoms, and many meet full diagnostic criteria for ASD) (24, 39–42), a number of studies have documented elevated rates of ASD among PM carriers, particularly males (~14%) (24, 25, 43). Subclinical ASD traits have also been noted in the PM more generally, including personality styles such as social reticence (28) and rigid and perfectionistic traits (27, 34). Collectively known as the broad autism phenotype (BAP), this constellation of subclinical personality and language traits mirror the central features of ASD and are thought to index genetic liability to the disorder as they are observed at higher rates among first degree relatives of individuals with ASD relative to the general population (44–46) and associated with increased polygenic burden for ASD (47).

Losh et al. (27) evaluated personality and pragmatic language features of the BAP in PM carriers. Using direct assessment measures, including the Modified Personality Assessment Scale (MPAS) (48) and Pragmatic Rating Scale (PRS) (49), they found similar profiles of BAP personality traits and increased pragmatic language violations in PM carriers and mothers of individuals with ASD compared to controls. Further, within-family associations were detected in the PM group, showing that children with FXS whose mothers exhibited BAP traits had more severe ASD symptoms. Such co-segregation of ASD-related phenotypes within a subgroup of families is intriguing, particularly when considering known interactions between a number of ASD risk genes and the FMR1 gene that might underlie these phenotypes (50, 51).

Beyond ASD-related risk, female PM carriers may display elevated symptoms of anxiety and depression, though rates vary considerably between studies (30, 32, 33, 52, 53). For instance, whereas Jiraanont et al. (53) reported higher diagnostic rates of depression (50%) and anxiety (33%) compared to controls (8.3 vs. 4.2%, respectively), Gossett et al. (52) reported no significant differences, but noted that the majority of the control group (~54%) had clinically elevated symptoms. Studies including younger PM carriers and adults without children have also reported elevated mood and anxiety disorders, suggesting that these symptoms may arise early on and are not merely related to the stress of parenting a child with FXS (25, 32). Interestingly, there is some evidence to suggest that mood and anxiety symptoms may co-occur with other phenotypes, such as FXPOI and EF deficits (e.g., working memory), among subgroups of PM carriers (54, 55), underscoring the importance of examining how these symptoms may co-occur with other phenotypes, such as social-cognitive deficits and ASD-related features.

Subtle differences in social cognition (i.e., understanding mental states and feelings essential for supporting fluent interpersonal interactions) have also been observed among PM carriers. Studies of male PM carriers reported lower social-cognitive performance, including decreased theory of mind (34, 56) and reduced social reward processing, which related to lower FMRP (37). Male PM carriers have been reported to exhibit reduced neural activation in the amygdala compared to controls when viewing faces with fearful expressions (35). Far less is known about the social-cognitive profiles of females and potential biological correlates. Klusek et al. (57) found that, compared to controls, female PM carriers displayed reduced visual attention to others' direct gaze, which can impact the ability to interpret others' intent. These results were hypothesized to reflect difficulty with interpreting ambiguous social information or with recognizing direct gaze as an important social cue.

Even subtly impaired social-cognitive abilities have been associated with increased pragmatic language violations in ASD and FXS and the FMR1 PM (58–62). For example, subgroups of parents of individuals with ASD who displayed reduced social-cognitive skills tended to show elevated pragmatic language violations during semi-structured conversation (60). In FXS, increased pragmatic language deficits have been reported to cosegregate with more severe impairments in social cognition (62), and some evidence suggests that differences in social cognition among PM carriers may also relate to pragmatic language. In a study of visual attention, Winston et al. (63) reported atypical visual scanning patterns of faces among female PM carriers, but found that these differences were associated with better social cognition and pragmatic language, perhaps suggesting that some female PM carriers may employ alternative strategies for deducing meaningful information in social exchanges. Such findings suggest that subgroups of female PM carriers may exhibit social-cognitive or attentional patterns that facilitate pragmatic language, whereas for others, these domains may not be related. Consistent with this possibility, Winston et al. (64) identified a subgroup of PM carriers who demonstrated social viewing patterns characteristic of those found among parents of individuals with ASD and greater co-occurring pragmatic difficulties, and also had children with more severe ASD symptoms.

Numerous studies have also identified EF differences among female PM carriers without FXTAS (36, 65–67), including differences in working memory across both visual and verbal modalities (68, 69), inhibition (23, 54, 70), and attention (71). PM carriers also exhibit differences in language domains that are thought to reflect underlying EF difficulties, such as verbal disfluencies (72–74). Additionally, a recent study reported inefficient language processing and eye-voice coordination on a task of rapid naming of familiar objects; difficulties were particularly evident during the latter portions of the task when executive demands have been shown to be the greatest (22). The overlap between these language tasks and EF could suggest that aspects of the PM phenotype, such as subtle differences in pragmatic language, could cluster together with EF difficulties. For instance, Kraan et al. (54) found that EF deficits co-occurred with neuropsychiatric symptoms (such as depression and anxiety) in female PM carriers, though far less is known about the co-occurrence of executive dysfunction and other clinical-behavioral or cognitive correlates.

The present study aimed to further characterize the phenotypic spectrum associated with the FMR1 PM by examining performance across a battery of clinical-behavioral, social-cognitive, and EF tasks in females with the FMR1 PM in comparison to controls. Further, we utilized latent profile analyses to examine whether homogenous phenotypic subgroups within the PM might be identified based on distinct constellations of these phenotypes. Finally, following prior evidence that FMR1-related molecular-genetic variation is often associated with clinical-behavioral and cognitive phenotypes in the PM (21, 23, 30, 33, 37, 56, 75–77), we examined associations between the phenotypic battery and CGG repeat length and FMRP.

This study characterized clinical-behavioral, social-cognitive, and EF features in females with the FMR1 PM. Consistent with prior work, results revealed elevated rates of subclinical ASD-related personality and language features among women with the PM, as well as increased self-reported difficulties in executive functioning, but no differences in mood and anxiety symptoms compared to controls. Some differences in social-cognitive tasks were also observed, including differences in complex social-emotional judgements of trustworthiness of faces, and in accuracy identifying basic emotions when viewing complex scenes. Latent profile analysis revealed three subgroups within the PM group who exhibited distinct phenotypic profiles across clinical-behavioral, social-cognitive, and EF measures, which together with group differences, and some associations with FMR1-related variation, may provide insights into phenotypic profiles associated with the FMR1 PM.

Replicating prior work, personality and pragmatic language differences that define the BAP were observed at elevated rates in the PM carrier group. Direct measurement tools, scored by raters blind to group status, identified approximately half of the PM carrier group as displaying personality traits consistent with the BAP, compared with published rates among individuals without a family history of ASD or FXS ranging from ~5 to 10% [e.g., (46, 80)]. Pragmatic language differences have been repeatedly observed among PM carriers (27, 57, 100–102), and were evident in this study as well, with additional patterns noted across the types of pragmatic language violations occurring more frequently in the PM group, who tended to display a more dominant conversational style (e.g., overly detailed, tangential language) than controls. Differences in pragmatic language have consistently emerged as a phenotypic marker associated with the PM in females (27, 100–102), and among individuals with FXS, particularly those who meet criteria for ASD (62, 103–109) and may be of clinical importance. For instance, prior work has shown that pragmatic language violations among mothers of children with FXS were associated with reduced self-reported quality of life for mothers (102), as well as elevated ASD symptoms and weaker expressive and receptive language in their children (27, 100). Together, this suggests that pragmatic language may be relevant to the well-being of both mothers and their children.

In contrast to previous work demonstrating higher rates of mood and anxiety disorders in PM carriers (25, 28, 30), we found no significant differences in depression scores between PM carriers and controls and only marginal differences from controls for anxiety symptoms, with both groups self-reporting elevated anxiety symptoms (>90% of participants). Rates of anxiety and depression were far lower in both groups using the MINI, a standardized psychiatric interview (ranging from 0 to 25%), and also comparable to rates reported in prior work (31, 33, 110). In follow-up analyses, no differences were observed on depression and anxiety features according to maternal status, or the number of affected children in each family, suggesting these findings are not related to parenting factors.

Unique to the present study was the inclusion of multiple measures of social cognition within a single sample, permitting characterization of social cognitive strengths and weaknesses across different types of stimuli. Differences in social-cognitive profiles have been reported among PM carriers compared to controls, particularly among males (35, 37). This study identified key social-cognitive differences in female carriers of the PM in specific tasks, which echo findings observed among individuals with FXS (62, 111). PM carriers differed in rating trustworthiness in response to faces ranging in emotional valence, a task that draws on social-cognitive and social decision-making skills. Differences on this task have been reported in ASD and the BAP in clinically unaffected relatives (61), and also among patients with bilateral amygdala damage, implicating this brain region in atypical performance in these groups (87, 89, 112). In studies of the PM, differences in amygdala volume and activation have been linked to aspects of social cognition among male carriers (35, 37), and these findings may suggest similar relationships among women with the PM that will be important to investigate in future work.

PM carriers were also less accurate in inferring emotions from faces when viewing still movie scenes. These results may indicate that PM carriers use different strategies in making social judgments relative to controls when viewing faces. Prior work that used eye tracking to examine looking patterns in response to faces indicates that PM carriers use different visual strategies from controls to inform social judgments (63). In the present study, marginal differences were also observed on the task involving reading complex thoughts and emotions from the eye region of the face. Thus, sensitivity to gaze and emotion expression, as measured by these social-cognitive tasks, may be an objective behavioral marker of underlying neural processes associated with interpreting emotion within the PM (35).

Consistent with prior reports of EF impairment associated with the PM (113), the PM group exhibited significantly higher executive functioning difficulty than controls, with approximately one third of the PM group reporting clinically significant EF difficulties. Characterization of the cognitive phenotype associated with the PM is especially important for understanding the manifestation of the neurodegenerative disorder, FXTAS, which may reveal subclinical phenotypic markers that are evident in a subgroup of PM carriers who go on to develop the disorder, as dysexecutive symptoms are a hallmark of FXTAS, particularly in males (114). Nonetheless, females with the PM exhibit differences in EF symptoms even without a diagnosis of FXTAS (68, 69, 115, 116). This study highlights the value of using self-report measures of EF in studies of the PM, as prior work has typically used online behavioral measures [e.g., see (113) for review]. The use of the BRIEF-A among females with the PM has been limited to only a handful of prior studies (23, 117), but is useful in that clinical significance may be easily determined and a standardized self-report measure can enable cross-cohort comparisons. Of note, however, such self-report measures could lead to over-reporting of symptoms (118, 119), and so may best be interpreted within the context of results from studies employing direct assessment of executive skills.

Complementing results from group comparisons, latent profile analyses revealed three distinct groups of PM carriers who displayed different profiles of performance across the various domains. Profile 1 comprised the largest subgroup, and represented those scoring at the mean of the sample across measures, and who largely contributed to the group differences observed in pragmatic language and select social cognitive tasks and EF. Profile 2 included 17% of the sample, and was characterized by elevated and co-occurring mood and anxiety symptoms, mild expression of personality features of the BAP, and higher suprasegmental violations (e.g., atypical variation in intonation, volume, or rate of speech). This specific co-occurrence of features is not surprising in the context of social-emotional patterns commonly observed among individuals who are more anxious or depressed, in that their symptoms can interfere with their social relationships, and vice versa (120). Allen et al. (55) also reported distinct clusters of PM carriers who reported different mood and anxiety symptoms, and other work has reported the co-occurrence of mood/anxiety features with executive dysfunction among PM carriers (54). Consistent with prior work (121), this subgroup also exhibited slightly elevated BAP traits, which raises the possibility that the co-occurrence of such traits may also be common to a subgroup of family members of individuals with ASD (of note, individuals showing Profile 3, discussed below, demonstrated elevated BAP features in the absence of elevated mood/anxiety symptoms). Although suprasegmental speech violations committed by PM carriers in this subgroup does not constitute clinical impairment, among individuals with ASD, where suprasegmental atypicalities are more pronounced, such variation can pose a significant obstacle to social interactions (122). First-degree relatives of individuals with ASD also demonstrate subtle differences in suprasegmental aspects of language (e.g., prosody) (123, 124).

Profile 3 represented the smallest subgroup, comprised of ~5% of PM participants who exhibited starker differences across all clinical-behavioral, social-cognitive, and executive domains relative to other PM carriers. This subgroup displayed notable differences on social-cognitive tasks together with elevated BAP features, and high rates of pragmatic language violations in particular. This pattern of performance is markedly similar to features described in prior investigations of parents of individuals with ASD who display the BAP (60, 61). Given the large number of ASD risk genes known to interact with FMR1 (50, 51), it may be that this phenotypic profile reflects an increased genetic liability for ASD among this subgroup. However, the small size of Profile 3 warrants cautious interpretation. We found no group differences in age, IQ, or FMR1-related variation between the Profile subgroups, likely due in part to some missing data in Profile 3 and unequal sample sizes across groups, and further investigations in larger samples will be important for confirming these patterns. Importantly, membership within any of the profile groups was not associated with differences in age, IQ, or parenting stress-related factors (including number of affected children, or severity of child symptoms), suggesting the phenotypic profiles identified are likely reflective of inherent traits, rather than systematic environmental differences between subgroups.

Finally, some associations were detected between phenotypic profiles and FMR1-related molecular genetic variability in the PM carrier group. We found linear associations between CGG repeat length and pragmatic language, indicating greater violations at lower ends of the CGG continuum. The CGG range reflected in the participants studied extended from 59 to 126 CGGs; thus, it may be that inclusion of more PM carriers with higher repeats (i.e., over 120) might have altered the findings observed here. Prior work has observed curvilinear links between language and CGG length, with different patterns noted in the mid-range (90–110 repeats) as compared to those with CGG repeats beyond 120 (21). Interestingly, the linear CGG association became marginal after factoring in participants' activation ratios, suggesting the importance of considering the second, healthy X allele in phenotype-genotype associations of females with the FMR1 PM. Additionally, we found that higher levels of FMRP predicted poorer performance on one social-cognitive task. This finding is somewhat in contrast to those from Hessl et al. (37), who found associations between reduced FMRP and poorer performance on social processing tasks in male PM carriers. Interestingly, a prior study based on the PM sample of participants studied here reported increased FMRP related to poorer performance on a language fluency task that taps into executive functioning (22), which together may suggest that the findings observed here may be specific to our sample of PM carriers (and thus may not be replicable), or that FMRP from blood is not analogous to FMRP in the brain. The links between FMR1 and the phenotypes included in the present study are likely not straightforward, and it is possible that other FMR1-related factors (e.g., mosaicism, mRNA) could also help to elucidate FMR1-associated patterns not explored in this study (125, 126).

This study provided a comprehensive assessment of clinical and subclinical phenotypes associated with the FMR1 PM. We identified differences from controls on pragmatic language features of the BAP, executive functioning, and some aspects of social cognition, but did not observe differences in mood and anxiety. Using LPA, we found subgroups within the PM sample characterized by unique patterns of performance on these measures. This study adds to a growing literature suggestive of important phenotypic heterogeneity among PM carriers, and provides further insight into FMR1-associated phenotypes.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://nda.nih.gov/edit_collection.html?id=1958.

The studies involving human participants were reviewed and approved by Northwestern University Institutional Review Board. The patients/participants provided their written informed consent to participate in this study.

NM helped to conceptualize the project, drafted the initial manuscript, and analyzed the data. JG, KN, MW, and JL assisted in preparing the dataset and provided valuable feedback on manuscript drafts. GM contributed to manuscript preparation. LB, SP, and MLe contributed to data preparation and processing. JS provided further consultation and verification of statistical methods and results. DH advised on PM participant evaluation of neurocognitive functioning. LZ, KS, and EB-K provided resources for genetic analysis of the FMR1 gene, helped to draft the Materials and Methods section pertaining to genetic analyses, and helped with interpretation of results. MLo secured funding for the projects from which these data were drawn, conceptualized the project, helped to interpret data, and develop the manuscript. All authors reviewed and approved the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.