FMR1 Protein Expression Correlates with Intelligence Quotient in Both Peripheral Blood Mononuclear Cells and Fibroblasts from Individuals with an FMR1 Mutation

Abstract

Fragile X syndrome (FXS) is the most common heritable form of intellectual disability and is caused by CGG repeat expansions exceeding 200 (full mutation). Such expansions lead to hypermethylation and transcriptional silencing of the fragile X messenger ribonucleoprotein 1 (FMR1) gene. As a consequence, little or no FMR1 protein (FMRP) is produced; absence of the protein, which normally is responsible for neuronal development and maintenance, causes the syndrome. Previous studies have demonstrated the causal relationship between FMRP levels and cognitive abilities in peripheral blood mononuclear cells (PBMCs) and dermal fibroblast cell lines of patients with FXS. However, it is arguable whether PBMCs or fibroblasts would be the preferred surrogate for measuring molecular markers, particularly FMRP, to represent the cognitive impairment, a core symptom of FXS. To address this concern, CGG repeats, methylation status, FMR1 mRNA, and FMRP levels were measured in both PBMCs and fibroblasts derived from 66 individuals. The findings indicated a strong association between FMR1 mRNA expression levels and CGG repeat numbers in PBMCs of premutation males after correcting for methylation status. Moreover, FMRP expression levels from both PBMCs and fibroblasts of male participants with a hypermethylated full mutation and with mosaicism demonstrated significant association between the intelligence quotient levels and FMRP levels, suggesting that PBMCs may be preferable for FXS clinical studies, because of their greater accessibility.

Hypermethylation and aberrant heterochromatinization of the fragile X messenger ribonucleoprotein 1 (FMR1) gene generally occur on expansion of the CGG trinucleotide element to >200 repeats, with the consequent gene silencing being accompanied by the absence or deficiency of its encoded protein, FMR1 protein (FMRP), causing fragile X syndrome (FXS), the leading inherited form of intellectual disability and autism.1, 2, 3, 4, 5

FMRP is an RNA-binding protein6, 7, 8 that regulates the translation of multiple mRNAs, with both activation and inhibitory functions,⁹ transport between cytoplasm and nucleus,¹⁰ and the development and maintenance of synaptic transmission.11, 12, 13 Accordingly, FMRP deficiency causes the neurophysiological features of FXS, including cognitive impairment and behavioral problems, autism-like behaviors, impairments of language and social behaviors, seizures, hyperresponsiveness to sensory stimuli, hyperactivity, problems with concentration, impulsivity, anxiety, and irritability.14, 15, 16, 17 The association between FMRP levels and developmental and cognitive functioning levels in males with FXS has been established such that decreased FMRP levels are strongly correlated with reduced intelligence quotient (IQ) scores18, 19, 20, 21, 22, 23 and with more severe phenotypes.^24,25

The complex association between FMRP levels and severity of FXS phenotypes is further affected by epigenetic methylation and size mosaicism, which is rather common among individuals with the FMR1 full mutation.26, 27, 28 Indeed, approximately 40% of males and 10% of females with FXS are mosaics, having both unmethylated premutation (PM) alleles (55 to 200 CGG repeats) or unmethylated full mutation (FM) alleles (>200 CGG repeats), together with methylated FM alleles within or across cell types. In these cases, FMRP expression is detected and somewhat correlated with the clinical phenotypes.^23,29, 30, 31 To a lesser extent, males with FXS can also present with size mosaicism consisting of FM, PM, normal-sized, or deleted alleles.32, 33, 34, 35 Furthermore, some individuals possess methylation mosaicism in which some cells carry unmethylated alleles ranging from PM to the FM range, combined with fully methylated FM alleles.^29,36, 37, 38 Such unmethylated alleles are regularly transcribed with increased mRNA levels. Translation of FMRP does occur from these mRNAs, albeit at lower-than-normal levels, which depend on the CGG allele size and the methylation status.^39,40 Furthermore, recent reports41, 42, 43 have demonstrated the presence of somatic mosaicism in both male and female carriers of a premutation allele and on their potential biological relevance, as demonstrated for other trinucleotide repeats.44, 45, 46, 47, 48

The methylation profiles for expanded-repeat alleles can diverge between or within various tissues derived from the same individual, leading to intertissue and/or intratissue mosaicism.^41,49 Furthermore, the FMR1 methylation status, together with CGG repeat number, have a profound influence on FMRP expression, which affects the cognitive functions of individuals with FXS, although the mechanisms underlying the somatic instability of CGG allele size are yet to be elucidated.

Several studies have investigated and reported on the relationship between IQ and FMRP levels measured and compared in different tissues; despite the general understanding that absent or low FMRP levels lead to intellectual disabilities in FXS, this relationship remains unclear.^19,22,50

To address this issue, peripheral blood mononuclear cells (PBMCs) and fibroblasts were collected from 66 individuals covering a broad spectrum of FMR1 alleles with a wide range of CGG repeat number (throughout the normal to the FM range) and methylation profile (from unmethylated to a fully methylated status). The FMR1 mRNA and FMRP expression levels were correlated with IQ and adjusted with the percentage of methylation to determine whether PBMCs or fibroblasts could be more effectively related to molecular measures in the clinical setting for the diagnosis of FMR1 mutations, and whether there is a relationship between IQ and FMRP expression levels measured in PBMCs and in fibroblasts. The results affirmed the previously significant positive correlation of FMR1 mRNA expression levels and CGG repeat numbers in PM males.51, 52, 53, 54 However, importantly, IQ demonstrates strong correlations with FMRP expression in both PBMCs and fibroblasts derived from male patients with FM and mosaicism, confirming previous results and suggesting that PBMCs may be a more suitable tissue for the diagnosis of FXS in clinical settings and for outcome measure in clinical trials, given its less invasive collection method compared with fibroblasts.^{18,19,21,23,24,31,49,50,}55, 56, 57

Materials and Methods

Participants

All participants were recruited to the University of California, Davis, Health MIND Institute’s Fragile X Clinical and Research Center. The study protocols and written informed consent forms associated with this research were approved by the Institutional Review Board of the University of California, Davis, in accordance with the Declaration of Helsinki. Written informed assents were collected from parents or guardians of minors. Written consent was obtained from 66 individuals: 18 females (4 with an FM allele, 10 carriers of a PM allele, and 4 with a normal allele); and 48 males (3 with an FM allele, 26 with FM mosaicism, 11 with a PM allele, and 8 with a normal allele).

Molecular Measures

Determination of CGG Repeat Number

Genomic DNA was obtained from PBMCs and fibroblast cell lines using standard procedures (Qiagen, Valencia, CA). CGG number and methylation status were analyzed by Southern blot analysis using isolated genomic DNA digested with EcoR1/Nru1, transferred on a nylon membrane, and hybridized with the specified, dig-labeled, FMR1 StB12.3 probe. Details of the method have been described previously.⁵⁸ The fraction of methylated alleles was measured using an Alpha Innotech (Santa Clara, CA) FluorChem 8800 Image Detection System. CGG repeat number was also assessed using the AmplideX PCR assay (Asuragen Inc., Austin, TX),⁵⁹ and amplified products were visualized by capillary electrophoresis (ABI 3100 Genetic Analyzer; Applied Biosystems, Foster City, CA) following manufacturer’s instructions.

Measurements of FMR1 mRNA Expression Levels

Total RNA from whole PBMCs was isolated using PAX gene tubes, according to manufacturer’s instructions (BD Bioscience, Franklin Lakes, NJ). Total RNA was isolated from cultured fibroblasts in a clean, RNA-designated area using TRIzol reagent (Invitrogen, Carlsbad, CA), followed by spectrophotometric determination of total RNA concentration. FMR1 mRNA levels were measured using quantitative RT-PCR with Assays-On-Demand (Applied Biosystems) and custom TaqMan primers and probe assays, as previously reported.⁵¹

FMRP Expression Levels by FRET Analysis

Fibroblasts (500,000 cells/mL) were thawed and seeded overnight at 100,000 cells per well into 24-well plates with RPMI 1640 medium. The cells were grown overnight in 5% CO₂, ambient O₂. At the 24-hour mark, samples were rinsed with Dulbecco’s phosphate-buffered saline and lysed directly on the plate with 50 μL 1× Cisbio Human FMRP assay lysis buffer + Roche complete Ultra Protease Inhibitor Tablets (MilliporeSigma, Burlington, MA) for 2 hours while rocking. FMRP quantification used the Cisbio Human FMRP assay (63ADK038PEC0; anti-FMRP-K lot number 110514K; anti-FMRP-d2; lot number 110514D; CisbioUS, Bedford, MA) and followed the manufacturer’s protocol. The assay uses homogeneous time-resolved fluorescence technology, as developed for quantification of FMRP.60, 61, 62 The method uses two FMRP-specific monoclonal antibodies conjugated to fluorescent dyes; the donor was labeled with Eu²⁺-cryptate and an acceptor designated d2. The time-resolved aspect of the assay exploits the long fluorescence decay times of the Eu²⁺-cryptate donor fluor, such that fluorescence resonance energy transfer (FRET) measurements do not occur until all sources of short-fluorescence-lifetime fluorescence (including from direct excitation of the acceptor) have decayed. Protein lysate (10 μL) was used in either triplicate or quadruplicate in a 384-well Opti-Plate (Perkin Elmer, Boston, MA), plus 10 μL homogeneous time-resolved fluorescence technology premixed antibodies. Samples were incubated overnight in the dark and then read on the PerkinElmer VictorX5. Total protein concentrations were determined using the Thermo Fisher Micro BCA Assay (Thermo Fisher; Waltham, MA). FMRP values were determined by interpolating FRET expression on a standard curve using a fiducial cell line. The bicinchoninic acid values were used to determine the ratio of interpolated FMRP/total protein.

FMRP Levels by Electrochemiluminescence Enzyme-Linked Immunosorbent Assay

All antibody and lysate dilutions were done in Meso Scale Discovery (MSD) Diluent 100 (Meso Scale Discovery, Rockville, MD). A PBMC lysate, 150 μg/mL FMRP-positive control lysate, 150 μg/mL FMRP-negative control lysate, or a standard curve of recombinant FMRP (Origene, Rockville, MD) was mixed 1:1:1 with custom biotinylated rabbit, polyclonal anti-FMRP ab17722, final concentration 1 μg/mL (Abcam, Cambridge, MA) and mouse monoclonal 6B8/FMRP, final concentration 0.5 μg/mL (Biolegend, San Diego, CA) in a V-bottom, polypropylene, 96-well plate. The plate was sealed and placed on a shaker at 4°C overnight. Overall, 5 μL of the mixture was added to each well of a 384-well avidin-coated MSD plate (Meso Scale Discovery) in quadruplicate. Mixtures were incubated at ambient temperature for 1 hour on a 44 × g shaker. The plate was washed three times using a 50 μL MSD wash buffer per well. The plate was blocked for 1 hour on a 44 × g shaker at ambient temperature in 3% MSD Blocker A in MSD wash buffer, 40 μL per well. The plate was washed three times using MSD wash buffer followed by addition of 5 μL sulfo-tagged anti-mouse (diluted 1:500) to each well of the plate. The plate was then incubated at ambient temperature for 1 hour while shaking at 44 × g, followed by washing three times using MSD wash buffer. MSD read buffer T (4×; Meso Scale Discovery) was diluted in deionized water to give a 2× MSD read buffer T. The 2× MSD read buffer T (40 μL) was added to each well. The applied voltage was optimized by the instrument manufacturer. The peak voltage applied during excitation was approximately 5 V. Chemiluminescence was immediately acquired using the MESO SECTOR S 600 reader (MSD). Data were analyzed using MSD DISCOVERY WORKBENCH software version 3.0. A standard curve in fmol was generated from the recombinant FMRP. The fmol of FMRP for each lysate was calculated from the standard curve. Data were reported as fmol FMRP per μg total protein.

Fibroblast Cell Lines

Established fibroblast cell lines were grown in Gibco (Thermo Fisher, San Jose, CA) AmnioMAX-C100 Basal Medium supplemented with 15% AmnioMAX-C100 Supplement. All reagents were from Invitrogen. Dishes were incubated at 37°C in a humidified 5% CO₂ atmosphere, and medium was replaced every 3 to 4 days. Cells were harvested by trypsinization and transferred into a new dish with a modified Fibroblast Medium (1:1 solution of Gibco AmnioMAX-C100 supplemented with 15% AmnioMAXC100 Supplement and RPMI 1640 Basal Medium supplemented with 10% fetal bovine serum, 1× Primocin, and 1% nonessential amino acids), with medium exchange every 3 to 4 days and allowed to reach 90% confluence before splitting. Fibroblast cultures were passaged no more than three times before collection for DNA, RNA, or protein extract isolation.

Cognitive Measures

The cognitive performance (IQ) of participants was assessed at the time of visit blood draw by standardized IQ tests, including Stanford-Binet Intelligence Scales, Fifth Edition⁶³ and Wechsler Adult Intelligence Scales (III or IV).^64,65 The Stanford-Binet Intelligence Scales, Fifth Edition (2003), is a test of intelligence and cognition for individuals with a mental age of 2 to >85 years. It measures cognitive performance with tasks related to fluid reasoning, knowledge, quantitative reasoning, visuospatial processing, and working memory. Each of the tasks is grouped into one of two domains (verbal and nonverbal). A single composite or full-scale IQ score, a verbal IQ, and a nonverbal IQ are provided. The Wechsler Intelligence Scales determine the intellectual abilities for individuals aged 16 to 90 years through subtests for four composite scores: Verbal Comprehension Index, which measures the application of verbal skills and information to problem solving; Perceptual Reasoning Index, which measures the ability to engage in nonverbal reasoning using visual images; Working Memory Index, which measures working memory, short-term memory, sustained attention, and auditory processing; and Processing Speed Index, which measures visual-motor coordination, attention, concentration, and the speed of mental processing. A full-scale IQ is derived from these composite scores.

Statistical Analysis

FMR1 and FMRP expression were modeled as a function of the number of CGG repeats using linear regression models. To address methylation mosaicism in male participants, percentage of methylated alleles was included in the model as a covariate. For female samples, the activation ratio was accounted for by including it as a covariate in the linear model. FMR1 mRNA, FMRP, and IQ were modeled pairwise using simple linear regression models. To compare the associations between IQ and FMRP in PBMCs and fibroblasts, a linear mixed-effects model was used with FMRP as the response and IQ, tissue (PBMCs or fibroblasts), and their interaction as fixed effects, as well as a random effect for participants. The test comparing the correlation of IQ and FMRP between PBMCs and fibroblasts was the test of the interaction effect from this linear mixed-effects model. Analyses were conducted using R version 4.2.1.⁶⁶

Results

Subjects

The 66 individuals participating in the current study included both males and females, with ages ranging from 2 to 83 years. The CGG repeat number varied from the normal to full mutation range (Table 1). PBMCs and fibroblast cell lines were obtained from the patients and used for the assays performed in this study (Supplemental Table S1). The operators (G.M.E. and J.L.R.) ran and analyzed the MSD and FRET assays blinded to the quantitative RT-PCR FMR1 expression levels, CGG repeat number data, and participants’ IQ. One premutation female was found to be a compound heterozygote (89 and 110 CGG repeat number allele size) and was excluded from the analysis.

Table 1.

Descriptive Statistics of Participant Characteristics by Sex

Variables	Total	Females	Males
N (%)	66	18 (27)	48 (73)
Age, mean (range), years	41 (2–83)	29 (9–59)	43 (2–83)
CGG repeat numbers, N (mean)
Normal	12 (27)	4 (30)	8 (26)
Premutation	21 (96)	10 (73)	11 (116)
Size mosaic	6 (>200)		6 (>200)
Methylation mosaic	17 (>200)		17 (>200)
Size and methylation mosaic	3 (>200)		3 (>200)
Full mutation	7 (>200)	4 (>200)	3 (>200)
IQ, N (mean)	53 (86)	14 (101)	39 (81)

IQ, intelligence quotient.

FMR1 mRNA Levels Correlate with CGG Repeat Number

As expected, a robust positive correlation was observed between FMR1 mRNA expression levels and CGG repeat number in PBMCs derived from males with a PM allele (Figure 1) (P < 0.001, covariate-adjusted correlation = 0.137), but not in fibroblasts, after correcting for percentage of methylation. No correlation between FMRP expression levels, as measured by either MSD or FRET and with CGG repeat number, was observed for either sex or cell type.

Figure 1.

Scatterplots demonstrating a significant positive correlation between FMR1 mRNA expression levels and CGG repeat numbers in peripheral blood mononuclear cells (PBMCs) of premutation males (P < 0.001) compared with a nonsignificant correlation in fibroblast (P = 0.346) from the same individuals. Results from PBMCs are in red, and results from fibroblasts are in blue; closed circles represent normal participants, and triangles serve for premutation participants. The x axis marks the number of CGG repeats, and the y axis marks the FMR1 mRNA expression levels. n = 11 premutation males for PBMCs and fibroblasts; n = 8 normal males for PBMCs; n = 3 normal males for fibroblasts.

FMRP Expression Levels Correlate with IQ in Both PBMCs and Fibroblasts in Males with a Full Mutation

Males with a full mutation and mosaic males (FM, n = 3; mosaic groups, n = 26) exhibited a positive correlation, between FMRP and with IQ, for both PBMCs and fibroblasts (P = 0.0133 and P = 0.0325, respectively) by FRET analysis after correcting for percentage of methylation (Figure 2, A and B, and Table 2). However, IQ increased significantly with increasing FMRP expression levels by FRET analysis in PBMCs (P = 0.008) but not fibroblasts (P = 0.248), in all 48 male participants (3 with an FM allele, 26 with FM mosaicism, 11 with a PM allele, and 8 with a normal allele) (Table 2). Furthermore, in females, the IQ correlated with FMRP by FRET analysis in fibroblast cell lines (P = 0.005) but not in PBMCs (P = 0.347) (Figure 2, C and D, and Table 2). In the smaller sample of female subjects, the effect of FMRP on IQ did not differ significantly between tissues after adjusting for activation ratio (P = 0.065). As the activation ratio is a random process, different values in different tissues (ie, brain) could account for the observed lack of correlation. For the MSD method, IQ was not significantly associated with FMRP expression levels in any tissue, due to insufficient data available (Table 2). FMRP expression levels by sex and mutation categories are provided in Table 3.

Figure 2.

Scatterplots of sex comparison between intelligence quotient (IQ) and FMR1 protein (FMRP) levels quantified by fluorescence resonance energy transfer (FRET) analysis. For male participants, peripheral blood mononuclear cells (PBMCs) were from 19 full mutation, 9 premutation, and 6 normal subjects (A); and fibroblasts were from 22 full mutation, 9 premutation, and 7 normal subjects (B). For female participants, PBMCs were from 1 full mutation, 6 premutation, and 3 normal subjects (C); and fibroblasts were from 4 full mutation, 4 premutation, and 3 normal subjects (D). Results from PBMCs are in red, and results from fibroblasts are in blue; closed circles represent normal participants, triangles serve for premutation participants, and squares depict full mutation participants. The x axis depicts the FMRP expression levels, and the y axis shows the IQ score in all four plots.

Table 2.

P Value by Pairs of Correlation, Sex, and Tissues

Pairs of correlation	Females		Males
	PBMCs	Fibroblasts	PBMCs	Fibroblasts
FMR1 mRNA by CGG repeats	0.167	0.899	<0.001∗	0.346
FMRP (FRET) by FMR1 mRNA	0.429	0.133	0.476	0.612
FMRP (MSD) by FMR1 mRNA	0.462	0.832	0.731	0.419
IQ by FMRP (FRET)	0.347	0.005∗	0.008∗	0.248
In full mutation group			0.013∗	0.032∗
IQ by FMRP (MSD)	0.498	0.471	0.386	0.455
IQ by FMR1 mRNA	0.025∗	0.004∗	0.065	<0.001∗
Methylation by FMR1 mRNA			<0.001∗	<0.001∗
IQ by methylation			<0.001∗	<0.001∗

IQ, intelligence quotient; FMRP, FMR1 protein; FRET, fluorescence resonance energy transfer; MSD, Meso Scale Discovery; PBMC, peripheral blood mononuclear cell.

^∗Significant correlation.

Table 3.

Mean (SD) of FMRP Levels by Sex and Category from FRET Analysis

Category	Females			Males
	N	PBMCs	Fibroblasts	N	PBMCs	Fibroblasts
Normal	4	0.351 (0.072)	2.01 (1.39)	8	0.666 (0.239)	1.08 (0.384)
Premutation	9	0.560 (0.232)	2.55 (1.68)	11	0.423 (0.237)	1.44 (0.694)
Full mutation	4	0.222 (NA)	0.989 (0.814)	29	0.172 (0.209)	0.434 (0.465)
Overall	17	0.476 (0.221)	1.99 (1.49)	48	0.317 (0.287)	0.780 (0.669)

FMRP, FMR1 protein; FRET, fluorescence resonance energy transfer; PBMC, peripheral blood mononuclear cell.

FMR1 mRNA Expression Levels Correlate with IQ in Both PBMCs and Fibroblasts

IQ increased significantly with increasing FMR1 mRNA expression levels by FRET analysis in fibroblasts (P < 0.001) but not PBMCs (P = 0.065) (Figure 3, A and B) in all male participants (n = 48) (Table 2). The regression slope of IQ by FMR1 mRNA is significantly larger for fibroblasts than for PBMCs (tissue-by-FMRP interaction effect P = 0.05). However, for females, IQ correlated with FMR1 mRNA expression levels in both PBMCs (P = 0.025) and fibroblast cell lines (P = 0.004) (Figure 3, C and D).

Figure 3.

Scatterplots of sex comparison between intelligence quotient (IQ) and FMR1 mRNA levels quantified by fluorescence resonance energy transfer analysis. For male participants, peripheral blood mononuclear cells (PBMCs) were from 19 full mutation, 9 premutation, and 6 normal subjects (A); and fibroblasts were from 22 full mutation, 9 premutation, and 7 normal subjects (B). For female participants, PBMCs were from 1 full mutation, 6 premutation, and 3 normal subjects (C); and fibroblasts were from 4 full mutation, 4 premutation, and 3 normal subjects (D). Results from PBMCs are in red, and results from fibroblasts are in blue; closed circles represent normal participants, triangles serve for premutation participants, and squares depict full mutation participants. The x axis depicts the FMR1 mRNA expression levels, and the y axis shows the IQ scores in all four plots.

The Percentage of Methylated Alleles Is Significantly Correlated with FMR1 mRNA Expression Levels and IQ

For all male participants, the percentage methylation significantly correlated with FMR1 mRNA expression levels. The negative correlation was observed with both PBMCs and fibroblast cell lines using simple linear regression. FMR1 mRNA expression levels significantly increased with decreased percentage of methylation (P < 0.001) (Table 2 and Figure 4). Similarly, IQ was significantly negatively correlated with percentage of methylation in both PBMCs and fibroblasts using simple linear regression. In addition, IQ decreased significantly with increased percentage of methylation in both PBMCs and fibroblasts (P < 0.001) (Table 2 and Figure 5).

Figure 4.

Scatterplots of correlations between percentage of methylation and FMR1 mRNA expression levels in both peripheral blood mononuclear cells (PBMCs) and fibroblasts derived from male participants. For PBMCs, there were 27 full mutation, 11 premutation, and 8 normal subjects; and for fibroblasts, there were 19 full mutation, 11 premutation, and 3 normal subjects. Results from PBMCs are in red, and results from fibroblasts are in blue; closed circles represent normal participants, triangles serve for premutation participants, and squares depict full mutation participants. The x axis marks the FMR1 mRNA expression levels, and the y axis shows the percentage of cells carrying methylated alleles.

Figure 5.

Scatterplots of correlations between intelligence quotient (IQ) and percentage of methylation in both peripheral blood mononuclear cells (PBMCs) and fibroblasts of male participants (P < 0.001). For PBMCs, there were 23 full mutation, 9 premutation, and 7 normal subjects; and for fibroblasts, there were 23 full mutation, 9 premutation, and 7 normal subjects. Results from PBMCs are in red, and results from fibroblasts are in blue; closed circles represent normal participants, triangles serve for premutation participants, and squares depict full mutation participants. The x axis shows the percentage of cells carrying methylated alleles, and the y axis shows the IQ scores.

CGG Repeat Size Is Unstable in Premutation Alleles

PCR analysis showed that most males with PM alleles (9 of 11; 81%) (Supplemental Table S1) demonstrated allele CGG size instability, with the presence of several allele populations (peaks) visualized on the capillary electrophoretograms (Figure 6, A and B). Females with a PM allele also demonstrated instability (4 of 10; 40%) (Figure 6, C and D), in agreement to recent reports.⁴¹

Figure 6.

Electropherograms showing differences of CGG allelic instability between peripheral blood mononuclear cells (PBMCs; A) and fibroblasts (B) of a premutation male and PBMCs (C) and fibroblasts (D) of a premutation female, as illustrated by the presence of serial peaks, each representing single distinct alleles. The x axis marks the size of the alleles in base pairs, and the y axis marks the fluorescence intensity of each allele. The red lines represent the internal ROX1000 size ladders, which allow automated data analysis and attainment of a precise sizing DNA fragment, by capillary electrophoresis.

Discussion

This study determined whether PBMCs or fibroblasts may better represent molecular factors, including FMRP levels, that correlate with cognitive abilities. Significant positive correlations were found between FMRP levels and IQ levels in the FM group (hypermethylated full mutation and mosaics) for both PBMCs and fibroblasts, in agreement with previous studies.^{18,19,21,23,24,31,49,50,}55, 56, 57 However, only PBMCs yielded FMRP levels that strongly correlated with IQ when all male participants were included in the analysis. As expected, PM males demonstrated a significant association between FMR1 mRNA expression levels and CGG repeat number, but only in PBMCs, not in fibroblasts, after correcting for percentage of methylation, as previously reported.^40,51,52, 53, 54^,67 The results indicated that PBMC is the preferred tissue to investigate the molecular measures at the FMR1 locus, including FMR1 mRNA expression levels, CGG repeat number, and the FMRP levels, which are strongly related to cognitive functions identified by IQ levels.

FXS is a complex and heterogeneous neurodevelopmental disorder for which FMRP levels have a demonstrated relationship with cognitive deficits.^21,23,68 FMRP itself can be modulated by the FMR1 gene activity, which is affected by multiple molecular determinants, including CGG repeat number, FMR1 mRNA expression levels, percentage of methylation, and tissue specificity.

Using the FRET method, Kim et al⁵⁰ investigated dermal fibroblasts of 184 participants with CGG repeats ranging from normal to FM in both sexes to determine which FMRP level or range is associated with normal intellectual function. Results suggested that borderline/normal cognitive function (IQ = 85) can be achieved with FMRP levels that are only approximately one-third of the normal mean level; increasing FMRP levels more than one SD above the normal mean level are not associated with any further increases in IQ. Although the study by Kim et al⁵⁰ supports the use of peripheral tissue to evaluate the relationship between FMRP and IQ levels, further studies with larger cohort sizes are needed to validate their conclusions. However, this study may provide a guide to establish the proper FMRP levels that could sustain the normal cognitive function.⁵⁰

Roth et al²² optimized two methods using an electrochemiluminescence immunoassay and a multiparameter flow cytometry assay to quantify the FMRP levels in PBMCs from 27 male participants spanning from normal to FM alleles. Their findings showed that FMRP levels in PBMCs positively correlated with IQ. Despite the fact that use of PBMCs is minimally invasive and practical,²² there is still a debate over what is the proper surrogate cell type to measure the molecular markers, including CGG repeat numbers, FMR1 mRNA, methylation status, and particularly FMRP levels that best represent the cognitive impairments, a core symptom in FXS.^69,70 The authors suggested that blood is a practical tissue to measure FMRP deficits in patients with FXS and a favorable choice for a prospective biomarker of treatment efficacy.

Most recently, Boggs et al⁵⁷ used a Luminex-based assay to quantify FMRP in dried blood spots from 187 participants with CGG repeats spanning normal, PM, mosaic, and FM size ranges. They observed that IQ scores and FMRP levels positively correlated in males with FM alleles. However, when they excluded six males with mosaicism, the correlation was no longer significant, suggesting that a threshold of FMRP presence is essential for typical cognitive functions. The work of Boggs et al⁵⁷ further supports the benefit of using PBMCs as a practical, non–central nervous system surrogate for defining the severity of cognitive dysfunction in individuals with FXS.

Recently, Dionne et al⁷¹ investigated a rate of protein synthesis from freshly extracted PBMCs and platelets, comparing 13 males with an FM allele with 14 age-matched healthy controls using a radiolabeled methionine and cysteine assay. They observed a 27% reduction in the rate of protein synthesis in fresh PBMCs of males with FXS compared with controls, whereas the rate in platelets was barely detectable. They concluded that the decreased rate of protein synthesis may reflect the neuronal deficits exhibited in FXS, as both PBMCs and neurons are non-proliferative cells, in which protein synthesis is restricted to maintenance purpose. This finding is in agreement with previous studies using L-[1-⁽¹¹⁾C]leucine positron emission tomography method, which found a diminished rate of cerebral protein synthesis in individuals with FXS.^72,73

Another study compared buccal cells with PBMC-based profiles from 42 participants with a wide range of FMR1 expansions, evaluated the relationship between all molecular measures, including FMRP levels, and the severity of FXS phenotypes, including IQ levels and autism spectrum disorder symptoms,²⁴ and demonstrated concordance between PBMCs and buccal specimens across all molecular measures.

Taken together, these findings establish PBMCs as an appropriate surrogate for measuring FMRP levels, lack or deficiency of which leads to cognitive impairment and FXS. More broadly, lowered FMRP levels may be an important contributor for both autism spectrum disorders and psychiatric disorders,74, 75, 76, 77, 78, 79 which could lead to the development of targeted therapeutic approaches in the future and represents an outcome measure in future clinical trials to predict the efficacy of response to targeted treatment.

Different tissues also demonstrated diverse allele profiles and propensity to CGG repeat expansion; for instance, expansions of PM alleles have been shifted from a unimodal peak PCR profile to bimodal peaks over time, as also illustrated by animal tissues, particularly of tail, testes, and liver; whereas heart tissues and PBMCs from older mice exhibited slight expansions and remain similar to younger ones.^80,81 PM alleles in PBMCs show only modest instability over time in animal model.⁸¹ In this study, divergent allele profiles and different patterns of PCR repeat profiles were observed, indicating the presence of somatic mosaicism, which has also been reported in other trinucleotide repeat disorders.44, 45, 46, 47, 48 Importantly, in FMR1 PM females,⁴² a positive correlation with attention-deficit/hyperactivity disorder was recently reported.⁴³ The potential role of the observed somatic mosaicism in PBMCs and fibroblast cells derived from both PM males and females (Figure 6 and Supplemental Table S1) in the phenotype of these individuals is currently unclear.

Although these findings apparently affirmed PBMCs as a preferred surrogate over fibroblast cell lines, there are several limitations in this study. First, the number of patients in each category of alleles is relatively small, particularly the normal and FM groups, which may not fully represent the characteristics of those categories and may explain why some associations between molecular measures did not reach significance. Furthermore, the number of females was only half of male patients, which may explain a lack of statistically significant correlations in the female groups. Finally, PBMCs and IQ values from several participants were not available for all assays and correlations.

Conclusion

Although PBMCs and fibroblasts are both beneficial for investigating molecular measures, including FMRP levels, and demonstrate strong significant correlations with IQ, PBMCs appear to be a better surrogate suitable to use in clinical settings because they are practical, easy to manage, and less invasive to obtain than fibroblasts.

Disclosure Statement

R.J.H. and F.T. received funding from the Azrieli Foundation and Zynerba Pharmaceuticals to perform treatment studies in fragile X syndrome.