Defective FGFR1 Signaling Disrupts Glucose Regulation: Evidence From Humans With FGFR1 Mutations

Abstract Context Activation of fibroblast growth factor receptor 1 (FGFR1) signaling improves the metabolic health of animals and humans, while inactivation leads to diabetes in mice. Direct human genetic evidence for the role of FGFR1 signaling in human metabolic health has not been fully established. Objective We hypothesized that individuals with naturally occurring FGFR1 variants (“experiments of nature”) will display glucose dysregulation. Methods Participants with rare FGFR1 variants and noncarrier controls. Using a recall-by-genotype approach, we examined the β-cell function and insulin sensitivity of 9 individuals with rare FGFR1 deleterious variants compared to 27 noncarrier controls, during a frequently sampled intravenous glucose tolerance test at the Reproductive Endocrine Unit and the Harvard Center for Reproductive Medicine, Massachusetts General Hospital. FGFR1-mutation carriers displayed higher β-cell function in the face of lower insulin sensitivity compared to controls. Conclusion These findings suggest that impaired FGFR1 signaling may contribute to an early insulin resistance phase of diabetes pathogenesis and support the candidacy of the FGFR1 signaling pathway as a therapeutic target for improving the human metabolic health.

Reproduction and metabolism have been known to be intricately linked for decades, yet the underlying precise pathomechanisms that define this intersection remain to be fully elucidated.One potential pathway that could merge those two domains is the fibroblast growth factor (FGF) signaling pathway, which has been implicated in developmental patterning and cell proliferation of many organs, including the hypothalamic neuroendocrine system and the pancreas.Attenuated signaling through the fibroblast growth factor receptor 1 (FGFR1) leads to diabetes in mice [1].In addition, activation of FGFR1 signaling has been shown to improve the metabolic health of rodents, nonhuman primates, and humans [2][3][4].Common and rare FGFR1 variants have been implicated in metabolic traits, including waist-to-hip ratio, body mass index (BMI), obesity, type 2 diabetes (T2D), and fasting glucose levels in population-based genetic epidemiology studies [5][6][7][8][9][10].These observations strongly implicate FGFR1 signaling as an essential determinant of glucose homeostasis in humans.
Rare naturally occurring FGFR1 mutations contribute to approximately 12% of the genetic etiology of isolated hypogonadotropic hypogonadism (IHH) with or without anosmia, a rare inherited disorder of infertility that is caused by deficiency in the hypothalamic secretion of the gonadotropinreleasing hormone (GnRH) [11].The mechanism by which FGFR1 mutations cause IHH is haploinsufficiency (heterozygous inactivating/missense alleles).These "experiments of nature" mutations provide a framework to study the pleiotropic effects of haploinsufficient loss of function (LoF) of FGFR1, and patients harboring such mutations offer a unique opportunity to directly test the affect of altered FGFR1 signaling on glucose metabolism.In this study, we hypothesized that humans with rare LoF single-nucleotide variations (SNVs) in FGFR1 will display defects in β-cell function and insulin action.To test this hypothesis, we examined the β-cell function and insulin sensitivity in IHH individuals with rare deleterious FGFR1 SNVs compared to healthy noncarrier controls during a frequently sampled intravenous glucose tolerance test (FSIGT).In addition, to corroborate the findings from the rare variant studies, we examined the role of common FGFR1 variants in the metabolic health of participants of the large hospital-wide cohort of the Massachusetts General Brigham Biobank (MGBB).

Study Approval
This research was reviewed and approved by the Massachusetts General Brigham Institutional Review Board (IRB protocol 2009P002349).All participants provided written informed consent.Participants were recruited from 2 recallable genetics studies.FGFR1 mutation carriers were participants with IHH in a longstanding and ongoing genetics study at Massachusetts General Hospital (MGH), Reproductive Endocrine Unit (IRB protocols 1999P006955 and 2020P000762), which has been reviewed and approved by the MGH Partners IRB. Noncarrier controls were participants in the MGBB, which has been reviewed and approved by the MGB IRB (protocol 2009P002312).Common variant association analysis was conducted as secondary use of clinical/research data under the MGH IRB protocol 2020P000762, which has been reviewed and approved by the MGB Institutional Review Board.

Case Selection and Recruitment of the FGFR1m Cohort From Massachusetts General Hospital Reproductive Endocrine Unit
Using a recall-by-genotype approach, we searched wholeexome sequencing (WES) data within the MGH Reproductive Endocrine Unit cohort (n = 1394) for individuals with confirmed IHH and rare SNVs in FGFR1-NM 0231103 (FGFR1 mutation carriers-FGFR1m cases).WES was performed as previously described [11].Rare SNVs were defined by a minor allele frequency of less than 0.1% in the control database-gnomAD [12].Among the 1394 IHH participants with WES data available, 175 IHH probands were found to carry SNVs in FGFR1 (58 probands harbored proteintruncating variants [PTVs] and 117 probands harbored missense SNVs, Supplementary Table 1 [13], ClinVar accession Nos.SCV003932463 and SCV0039332605).Individuals with IHH were prioritized for recruitment if they harbored rare SNVs in FGFR1 that disrupt FGFR1-c isoform that were (1) predicted PTVs (ie, nonsense, frameshift, essential splice site) or, also recruited, (2) missense variants that were predicted to be deleterious by in silico and in vitro analyses .In silico analysis included the utilization of the following prediction programs: CADD, Polyphen 2, SIFT, Mutation Taster, MutPred Score, REVEL, and Eve (Supplementary Table 2 [13]).Among those probands, 9 participants (7 men and 2 women) with heterozygous rare SNVs in FGFR1 were enrolled and all participated in the study visits.Tyr701Cys), with the majority being predicted to be pathogenic and likely pathogenic based on the American College of Medical Genetics criteria and absent from the control cohort of gnomAD (see Supplementary Table 2 [13]).All variants had a CADD score greater than 20, and missense SNVs were predicted to be deleterious based on in silico analyses (see Supplementary Table 2 [13]).The FGFR1 SNVs were the primary genetic cause for IHH in the enrolled participants.The participants with rare FGFR1 SNVs were diagnosed with either normosmic IHH or IHH with anomia (Supplementary Table 3 [13]), that was defined based on formal smell testing using the University of Pennsylvania Smell Identification Test (UPSIT) [14] or self-reported complete inability to smell and were all supplemented with sex steroids at the time of the study.FGFR1 rare SNV carriers demonstrated additional nonreproductive phenotypes that have previously been linked to mutations in the IHH-related FGF-pathway genes, including hearing loss, bone abnormalities, teeth defects, and eye disorders (see Supplementary Table 3 [13]).

Control Selection and Recruitment of FGFR1c Cohort From the Massachusetts General Brigham Biobank
The MGBB cohort (n = 65 247) was used to identify participants with no FGFR1 mutations (noncarrier controls-FGFR1c).For the recruitment, a recontact letter was sent from the biobank staff along with an invitation letter, cosigned by the Biobank principal investigator and the principal investigator of this study, and an opt-in or opt-out letter to patients.Ten business days after the biobank sent the recontact letters to patients, each patient was directed to be contacted for recruitment, and a list of the response status for each patient was sent back to the biobank.Among the 808 MGBB participants who were contacted by our study team, 45 were enrolled in the study.Sixteen were excluded from participation after failing to meet one or more of the inclusion criteria for the study (discussed later).Twenty-seven healthy controls (16 men and 11 women) were enrolled.WES data of the MGBB participants showed no SNVs in FGFR1.

Inclusion Criteria for the Entire Study Cohort
Inclusion criteria included the following: (i) no history of bleeding or thromboembolic disorders (ie, thrombocytopenia, deep vein thrombosis, pulmonary embolism, cerebrovascular disease, warfarin treatment, hypercoagulability syndromes); (ii) no history of illicit drug or heavy alcohol use (>4 g of alcohol per day); (iii) stable weight for previous 3 months; (iv) not currently pregnant; all female participants were administered a urinary pregnancy test to rule out pregnancies; (v) serum hemoglobin greater than or equal to 10 g/dL and baseline hematocrit greater than or equal to 38%; (vi) no current or previous diagnosis of type 1 or T2D as defined by the American Diabetes Association criteria: fasting glucose greater than 126 mg/dL or random blood glucose greater than 200 mg/dL on 2 occasions; and (vii) not currently taking medications that may influence glucose metabolism (eg, corticosteroids, thiazide diuretics).
Age, Sex, Gender, and Ethnicity Background of the Enrolled Participants The age, sex, and gender of the enrolled participants is provided in Supplementary Table 3 [13].The ethnicity and race background of the enrolled participants was as follows: In the FGFR1m group, 5 participants identified as not-Hispanic White, 2 as Asian, 1 as Black, and 1 as having more than 1 ethnicity background, while the ethnicity of the enrolled FGFR1c cohort was noted as follows: not-Hispanic White (n = 20), Hispanic White (n = 2), Hispanic Unknown (n = 2), Asian (N = 2), and Black (N = 1).[15] and is less complex compared to glucose clamp studies.Even though an oral glucose tolerance test could be used as an alternative approach, the FSIGT was chosen to help define precise effects during the first and second phases of insulin secretion and estimate both glucose effectiveness and the disposition index (D I ) [16].The FSIGT began with baseline sampling for plasma insulin and glucose at −10 and −1 minutes.These measurements were followed by the administration of a 0.

MinMod Analysis
The MinMod Millennium software was used for this analysis [17].This software is used to estimate critical indices of glucose-insulin dynamics during an FSIGT.The minimal model analysis allows quantitation of the ability of insulin to enhance glucose disposal by using rate constants characterizing glucose flux.The metabolic measures estimated in this fashion, for example, insulin sensitivity and glucose effectiveness, have been shown to correlate highly with the goldstandard euglycemic glucose clamp technique [15].The glucose minimal model uses insulin observations to drive the glucose response and attempts to portray glucose dynamics in terms of 2 key parameters, glucose responsiveness and insulin sensitivity.The insulin minimal model uses glucose observations to drive the insulin response, and it attempts to portray insulin dynamics in terms of another set of key parameters, first-phase responsivity and second-phase responsivity.
There are 4 reasonably distinct phases with this model: (1) a mixing phase, immediately following the glucose injection, and lasting for about 10 minutes, as glucose equilibrates in the circulation; (2) a phase of steady and almost constant decline as the concentration imbalance between circulating glucose and cellular glucose leads to glucose, largely, mediating its own disposal into the cell (minutes 12-20); (3) an extension of the glucose decline beyond that which would be anticipated based on "2" and due, primarily, to the action of exogenous interstitial insulin to prolong the disposal of glucose (minutes 22-50); and (4) a phase of recovery of glucose back to its basal value (minutes 60-180) [17].

Indices of Insulin Resistance
The indices of insulin resistance were estimated on: (i) the fasting insulin and C-peptide levels; (ii) the glucose response to its own mediated action, and insulin infusion was calculated as the area under the curve (AUC) for glucose (AUC glucose ) during the FSIGT: The AUC glucose during 0 to 10 minutes represents the acute response to glucose load; the AUC glucose during 12 to 20 minutes represents the ability for glucose to mediate its own disposal into the cells; the AUC glucose during 22 to 50 minutes represents the glucose response to the action of exogenous interstitial insulin; and the AUC glucose during 60 to 180 minutes is the phase of recovery of glucose back to its basal value; (iii) the insulin sensitivity (S I ) and glucose effectiveness (S G ): S I and S G were calculated with the minimal model software [17] (discussed earlier).S I is defined as fractional glucose disappearance per insulin concentration unit.S G is defined as the ability of glucose per se to promote its own disposal and inhibit hepatic glucose production in the absence of an incremental insulin effect; (iv) the homeostatic model assessment of insulin resistance (HOMA-IR): HOMA-IR correlates well hyperinsulinemic-euglycemic clamps and minimal model estimates of insulin resistance in prior studies [18,19]; and (v) the homeostatic model assessment of insulin sensitivity (HOMA-SI) [18,19].

Indices of Pancreatic β-Cell Function (Insulin Secretary Function)
The indices of pancreatic β-cell function were estimated based on the following: (i) The first and second phase of insulin secretion: During the FSIGT, the first phase of insulin secretion or acute insulin response to glucose (AIRg) is represented by insulin being released after glucose injection, and was calculated using the AIRg calculation with the minimal model analysis [17] (discussed earlier) and the AUC of insulin and C-peptide concentration (AUC insulin and AUC Cpeptide ) during the first 10 minutes.The second phase of insulin secretion was calculated as the AUC for both parameters during 12 to 20 minutes after glucose bolus (prior to exogenous insulin administration); (ii) the HOMA of β-cell function (HOMA-β) is a static assessment of β-cell function using basal values of glucose and insulin or C-peptide [19] and was calculated based on fasting glucose, insulin, and C-peptide levels using the HOMA2 calculator [18]; (iii) D I is expressed by calculating the product of insulin secretory capacity and insulin sensitivity [17] and is an indicator of β-cell function.

Mass General Brigham Biobank Common Variant Associations
Common FGFR1 variants (ie, SNVs) that have previously been linked to clinical phenotypes were identified via the GWAS catalog [20].Using the MGBB portal, the associations of those common FGFR1 variants were tested with already the curated clinical phenotype of T2D: 65 247 MGBB participants with genotyping data: rs9657190 rs881301, rs881299, rs7828172, rs6984358, rs66504003, rs62505473, rs60527016, rs59498392, rs57709857, rs4739558, rs4647906, rs4647903, rs4082204, rs3925, rs36061954, rs34036147, rs328301,  [21].Given that the incidence of diabetes and obesity increases with age, we ensured that despite the age difference between the 2 groups, none of the enrolled participants were diagnosed with prediabetes or diabetes, which was reflected by the absence of fasting hyperglycemia and no differences in their HbA 1c levels (see Table 1).In addition, the BMI was also not different between the 2 groups (mean BMI of 28.5 in the FGFR1m patient cases vs 26.1 in the FGFR1c controls; P = .233).All IHH participants were maintained on their appropriate sex steroid treatment at the time of the study.All 7 FGFR1m male participants were treated with either transdermal or intramuscular testosterone, and their testosterone levels did not differ from the testosterone levels of the controls (385 ng/dL vs 448 ng/dL; P = .558).Both female FGFR1m participants were treated with hormone therapy at the time of the study (one participant with an estradiol patch and progesterone and the second with a combined oral contraceptive pill).Dual-energy x-ray absorptiometry scan assessment showed that the FGFR1m patient cases demonstrated a higher percentage of total body fat (37.8%) compared to noncarrier controls (30.8%,P = .008;see Table 1).No differences were observed in the lean mass of participants in those 2 groups (see Table 1).To investigate any differences in insulin sensitivity between the 2 groups, the glucose response to its own mediated action and to an insulin bolus were calculated using the AUC for glucose (AUC glucose ) during the FSIGT.FGFR1m patient cases demonstrated higher AUC glucose during minutes 12 to 20 of the study compared to noncarriers (Fig. 1).This phase represents their glucose-mediated glucose disposal.They also displayed a higher AUC glucose during minutes 22 to 50, the phase that represents glucose's response to the action of exogenous insulin, compared to controls (see Fig. 1).While no differences in S G were noted between the groups, S I , that is, the fractional glucose disappearance per insulin concentration unit, was lower in the FGFR1m patient cases compared to the healthy noncarrier controls (Table 2).These results were concordant with the higher fasting insulin and C-peptide levels, higher HOMA-IR, and the lower HOMA-SI calculated scores in the FGFR1-mutation carrier group compared to noncarrier controls (see Table 2), which remained significant after adjusting for BMI and percentage of total body fat (see Table 2).

FGFR1m Patient Cases Demonstrate Higher β-cell Function Compared to Controls (FGFR1c)
During the FSIGT, FGFR1m patient cases demonstrated an increased AUC insulin and AUC Cpeptide for the total duration of the study, as well as during the first-and second-phase insulin secretion compared to healthy noncarriers (Figs. 2  and 3, respectively).These findings were consistent with the observed higher AIRg and the higher HOMA-β in the FGFR1m patient cases compared to controls (see Table 2).These findings suggest that the FGFR1m patient cases mounted a significantly higher β-cell response to maintain normal blood glucose levels.The D I , which is expressed by calculating the product of insulin secretory capacity and insulin sensitivity, was similar between the 2 groups.This finding reflected the compensatory higher β-cell function in the face of lower insulin sensitivity that the FGFR1m patient cases demonstrated compared to controls (see Table 2).Similar to the indices of insulin sensitivity/resistance, most of the indices of β-cell function remained significant when adjusting for BMI and total fat mass (see Table 2).

FGFR1 Common Alleles Are Associated With Differential Prevalence of Type 2 Diabetes in the Large Cohort of the Massachusetts General Brigham Biobank
To further evaluate the role of FGFR1 signaling on glucose metabolism beyond the rare variant spectrum, we examined the association between FGFR1 common variants and T2D in the MGBB.Among the consented participants in the MGBB with genomic data (N = 65 247), 6768 individuals were diagnosed with T2D based on curated phenotypic data (positive predictive value = 0.99).As shown in Fig. 4, the risk of T2D diagnosis increased with 2 common FGFR1 genotypes: rs3925 and rs10101096.Specifically, for the genotype rs3925, T2D prevalence increased significantly for individuals with the homozygous GG alleles and the heterozygous GA alleles compared to the individuals carrying the homozygous the AA alleles.Similarly, the prevalence of T2D increased significantly for participants with the homozygous AA alleles at the rs10101096 locus compared to the carriers of the homozygous CC alleles.

Discussion
Genotype-first approaches and targeted recall-by genotypes studies represent powerful tools in genomic medicine.Such approaches allow the early diagnosis of individuals with mutations in disease-causing genes and permit the discovery of novel genotype-phenotype associations.Using such an approach, we investigated whether rare FGFR1 deleterious SNVs impaired glucose metabolism in humans.We show that participants with rare heterozygous deleterious FGFR1 SNVs demonstrate higher insulin/C-peptide response to glucose and higher insulin resistance compared to noncarriers.Further, supported by the higher prevalence of T2D in individuals harboring common FGFR1 risk alleles, this study now provides robust human genetic evidence of the critical role of FGFR1 signaling in glucose homeostasis.
The importance of impaired insulin release and insulin resistance in the pathogenesis of T2D is well known and has been evaluated in numerous prior studies [22][23][24].Insulin sensitivity appears to decrease approximately 5 years prior to the development of T2D, while insulin secretion increases 3 to 5 years prior to the diagnosis, likely as a compensatory mechanism, to then decrease as individuals get closer to the development of T2D [22][23][24][25].Compensatory insulin secretion by the pancreatic β cells initially maintains normal plasma glucose levels, but β-cell function progressively worsens over time [26].As a result, increased insulin secretion and resistance are considered independent risk factors for T2D [22].This is also true for high-risk patients, that is, patients already diagnosed with prediabetes, in whom a combination of decreased baseline insulin sensitivity and secretion appears to act additively to increase the risk for T2D development over time [27].The hyperinsulinemia and lower insulin sensitivity observed in our study participants in the absence of overt hyperglycemia suggest that FGFR1 signaling may be temporally implicated in the early insulin hypersecretory phase and insulin resistance seen in initial stages of T2D pathophysiology.Despite the observed dysregulated metabolic indices, due to the study design wherein the presence of overt diabetes was considered an exclusion criterion, none of the FGFR1m patient cases carried a diagnosis of T2D.To further clarify the role of FGFR1 variants in T2D within a population setting, we examined the association of common FGFR1 variants with the prevalence of T2D within the MGBB.In line with prior reported studies [5,6], common FGFR1 genotypes were linked to T2D prevalence among MGBB participants.All common variant risk genotypes harbored noncoding possibly regulatory variants that mapped to the FGFR1 gene locus.The precise mechanisms by which these genotypes alter the T2D risk remains unclear and will require further interrogation.Taken together, the results of this report strongly suggest that carriers of deleterious FGFR1 SNVs and those possibly those harboring FGFR1 common risk alleles should also be considered at higher risk of T2D and should be monitored for the development of the disease.
While BMI was not different between the 2 groups of this study, FGFR1m patient cases demonstrated a higher percentage of total body fat mass compared to controls.The differences in the fat distribution between the 2 groups could potentially explain the insulin resistance seen in the IHH FGFR1m participants, as abdominal obesity is often linked to ectopic fat deposition (eg, in muscle and liver) and may underlie the resistance to the effects of insulin on peripheral glucose and fatty acid utilization observed in patients with low insulin sensitivity, prediabetes, and diabetes [22,28,29].However, β-cell function, insulin sensitivity, and insulin resistance remained significantly different in FGFR1m participants even after controlling for these adiposity indices.This suggests that the underlying defects in the FGFR1 signaling per se is likely to contribute to the glucose dysregulation observed in this study rather than their body composition differences.This assertion is also supported by prior studies showing that mice with dominant-negative loss of FGFR1 demonstrate abnormal β-cell differentiation, fasting, and nonfasting hyperglycemia (ie, diabetes-like phenotypes), despite being lean [1].Possible alternate mechanisms that could explain the differences observed between the 2 groups may include altered thermoregulation and hypothalamic or peripheral regulation of browning of the white adipose tissue that is mediated via the FGFR1 signaling [30][31][32].In addition, FGFR1 is highly expressed in the human adipose tissue, and specific knockout of FGFR1 eliminates the beneficial effects of FGF21, including weight loss and energy expenditure in obese rodent models [33], suggesting of an important role of FGFR1 expression in white adipose tissue.It is likely that the human carriers of deleterious FGFR1 variants may have defective signaling within the adipose tissue.Obtaining an adipose tissue biopsy in the enrolled participants and assessment of visceral adiposity measures should be addressed in future studies.
While the inactivation of the FGFR1 signaling pathway worsened insulin resistance in this study, prior studies have shown the beneficial effect of the activation of the FGFR1 signaling pathway on the metabolic health of animals and humans [2][3][4].Hence a key biologic question arising from these observations relates to the identity of the specific FGF ligand(s) that may underlie the observed metabolic phenotype resulting from altered FGFR1 signaling.Among the multiple FGF ligands, several have been implicated in glucose regulation through different mechanisms: (i) Mice lacking fgf1 develop marked hyperglycemia and insulin resistance when challenged with a high-fat diet and in ob/ob and db/db mice or diet-induced obesity (DIO) models, peripheral delivery of a single dose of recombinant FGF1 can normalize blood glucose levels within hours, without inducing hypoglycemia, making Fgf1 a promising therapeutic agent of diabetes and insulin resistance [34,35]; (ii) FGF2b may transform the nonendocrine human pancreatic cells into endocrine insulinsecreting cells [36]; (iii) FGF7 has been shown to enhance islet engraftment and improve metabolic control following islet transplantation in diabetic mice [37]; (iv) administration of FGF19 to obese diabetic mice leads to beneficial effects on metabolism [38]; and (v) exogenous FGF21 treatment of animal models reduces hyperglycemia by preventing islet destruction and improving glucose clearance and decreases their body weight and circulating lipids [39][40][41][42], while administration of FGF21 molecules in humans leads to a decrease in body weight, and an improvement of dyslipidemia [43][44][45][46][47][48][49][50].
Of the aforementioned FGF ligands, data from humans strongly suggest that FGF21 may be the primary driver of metabolic health relating to FGFR1 signaling.FGF21 is an endocrine FGF ligands that acts through the FGFR1-β-klotho complex and appears to have a crucial role in metabolic regulation.Specifically, higher serum FGF21 concentrations have been described in humans with obesity and diabetes, suggesting the potential presence of FGF21 resistance in addition to insulin resistance [51][52][53][54].Further support for the role of FGF21 in improving metabolic health comes from studies   that used agonistic FGFR1 receptor antibodies that mimic the action of FGF21.Those studies resulted in improvement of hyperglycemia, hyperinsulinemia, hyperlipidemia, and hepatosteatosis in obese diabetic mice [4], and significant weight loss in obese monkeys [3].Further, in a recent clinical study, overweight human participants who received a single dose of a similar-acting antibody showed a transient weight reduction, improvement in cardiometabolic parameters, and reduction in carbohydrate intake [2].Furthermore, FGF21 analogues have been studied both in animal and human models as therapeutics for obesity, T2D, hyperlipidemia, and nonalcoholic steatohepatitis [31,[44][45][46][47][55][56][57][58][59][60].The present study used an FSIGT method to assess glucose metabolism and hence potential incretin effects (eg, glucagon-like peptide-1 [GLP-1] axis) on glucose metabolism would be missed.While no direct relationship between FGFR1 signaling and the GLP-1 axis has been reported previously, GLP-1 receptor analogues appear to stimulate hepatic FGF21 production and inhibit gluconeogenesis [61].This latter observation suggests that at least part of the FGF21's effects on glucose metabolism may be incretin mediated.In summary, while FGF21 was not directly studied in the present report, prior studies strongly implicated FGF21-FGFR1 signaling as the primary pathway for metabolic regulation in humans.The potential mechanisms for the role of FGF21-FGFR1 signaling in metabolic health include an improvement of β-cell function and insulin sensitivity, decreased glucagon release, increased thermogenesis due to central activation of the sympathetic nervous system, and induction of energy expenditure via brown fat activation [54,62].
The intersection between genes implicated in IHH and their potential role in metabolic regulation is intriguing.FGFR1plays an important role in the IHH architecture.Mice homozygous for hypomorphic Fgfr1 alleles display a reduction in hypothalamic GnRH [63], while expression of a dominantnegative FGF receptor in mouse GnRH neurons results in a decreased number of GnRH neurons in the forebrain and late pubertal onset [64].Similar to mice, our group and others have shown that rare deleterious genetic FGFR1 variants are a leading cause of IHH [65,66].Our recent analysis of nextgeneration sequencing data from a large cohort of IHH patients showed that 12% of them carry rare variants in FGFR1, demonstrating the major role of this signaling pathway in the genetic etiology of IHH [11].The FGF21-FGFR1 signaling complex also includes β-klotho, an essential coreceptor that has also been implicated in IHH as well as metabolic phenotypes.Specifically, a broad spectrum of metabolic phenotypes, including obesity/overweight, fasting hyperglycemia, and insulin resistance, has previously been reported in IHH individuals harboring rare damaging heterozygous SNVs in KLB, the gene encoding for β-klotho [67], and rare digenic LoF variants in FGFR1 and KLB have been reported in patients with severe insulin resistance [68].Interestingly, while the binding site of β-klotho to FGFR1 is independent of the common site that β-klotho uses to bind to FGF19 and FGF21, it overlaps with the binding site for ligands of the FGF8 subfamily and prevents the formation of the FGF8-FGFR1 complexes [69].Mutations in FGF8 contribute to the genetic etiopathogenesis of IHH, and specific FGFR1 mutations decrease the signaling of FGF21 and FGF8 by impairing the association of FGFR1 with β-klotho, leading to both IHH and metabolic phenotypes [67,[69][70][71][72].In addition, central and peripheral administration of FGF21 induces GnRH neuronal growth and stimulates the release of GnRH from the hypothalamus [67].Thus, the regulation of both the GnRH neuronal and pancreatic cell function by the FGF21-β-klotho-FGFR1 signaling pathway raises the hypothesis of shared regulatory genetic networks between GnRH and β-cell pancreatic development, with the FGF21-β-klotho-FGFR1 signaling pathway linking metabolism to reproduction.This study's main strength is the utilization of a genotypefirst approach that capitalizes on naturally occurring FGFR1-damaging variants to study the effect of FGFR1 signaling on human metabolic health.Heterozygous dominantnegative and recessive hypomorphic mutations in FGFR1 have been implicated in causing dominant and recessive forms of Hartsfield syndrome, which is characterized by holoprosencephaly, ectrodactyly, cleft lip/palate, intellectual disability, and hypogonadotropic hypogonadism [73], and transgenic mice expressing a dominant-negative version of the FGFR1 demonstrate overt diabetes [1].Hence, future studies in humans with dominant-negative FGFR1 mutations may help validate the observations of this study.Similarly, metabolic studies in other FGFR1-related human diseases, such as Pfeiffer syndrome, a disease caused by FGFR1 gainof-function variants [73,74], may also provide valuable information.In addition, hypogonadism, the main feature of IHH, has been previously linked to insulin resistance, with prior studies showing that repletion of hypogonadal individuals with sex steroids improves their metabolic health [75][76][77].Thus, hypogonadism may represent a potential confounder contributing to the metabolic dysregulation observed in this study.However, all IHH participants were supplemented with sex steroids at the time of this study, suggesting that the metabolic abnormalities observed is unlikely to be related to their current sex steroid milieu.However, potential longterm metabolic sequalae stemming from pubertal hypogonadism, which could be contributing to the current observations, cannot be excluded.In addition, while FGFR1m participants demonstrated increased first-and second-phase insulin and C-peptide secretion, future cellular model studies are required to examine the direct effect of FGFR1 on islet cell function.Moreover, T2D is a polygenic disease and known to be regulated by common variants in multiple loci.While polygenic risk scores were not calculated for the enrolled participants, none of them were diagnosed with T2D or prediabetes at the time [78].However, potential confounding of the study findings by polygenic genetic risk cannot be fully excluded.In conclusion, using a genotype-first approach, we showed that FGFR1 human deleterious variants lead to insulin resistance and may increase the risk for T2D.Our findings provide direct human genetic evidence to support prior studies investigating the beneficial effect of the FGFR1 signaling pathway as a therapeutic target of diabetes and obesity in humans.
(grant Nos.P50 HD104224 and R37 HD043341 to S.B.S.; R01 DE031452 and R01 HD096324 to R.B.; and F32 HD108873 to M.I.S.), and the Food and Drug Administration (grant R01 FD007843 to S.B.S.).The project described was supported by grant numbers 1UL1TR001102 and 1UL1TR002541-01.The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Center For Research Resources, the National Center for Advancing Translational Science, or the National Institutes of Health.

Figure 1 .
Figure 1.FGFR1m cases demonstrate higher glucose levels during the frequently sampled intravenous glucose tolerance test (FSIGT) phase of glucose-mediated and insulin-mediated glucose disposal compared to controls (FGFR1c).During the 4-hour FSIGT, baseline sampling for plasma glucose occurred at −10 and −1 minutes.These measurements were followed by administrating 0.3 g/kg bolus of glucose within 2 minutes.At 20 minutes, participants received a 0.03-U/kg infusion of regular human insulin over 45 seconds to enhance the insulin level to better help assess the effect of insulin on glucose uptake.FGFR1m cases (in white) demonstrated higher levels of glucose during the glucose-mediated (12-20 minutes) and insulin-mediated glucose disposal (22-50 minutes) compared to controls (ie, FGFR1c participants [in black]).*P less than .05;**P less than 001.

Figure 2 .
Figure 2. Fgfr1m cases demonstrate higher insulin response to glucose during the frequently sampled intravenous glucose tolerance test compared to controls (FGFR1c).The FGFR1m cases (shown in white) demonstrate higher insulin response to glucose compared to FGFR1c controls (shown in black), which was manifested during the first (0-10 minutes) and second phase (12-20) of the study.*P less than .05;**P less than 001.

Figure 4 .
Figure 4. Common FGFR1 alleles are linked to type 2 diabetes (T2D) prevalence in the Mass General Brigham Biobank (MGBB) population.MGBB individuals with the GG and GA genotypes in the rs3925 locus and individuals with the AA genotype in the rs10101096 locus demonstrated an increased risk of T2D compared to individuals with the AA and the CC genotype, respectively.*P less than .012(0.05/number of variants tested); **P less than .001.

Figure 3 .
Figure 3. FGFR1m cases demonstrate higher C-peptide response to glucose during the frequently sampled intravenous glucose tolerance test compared to controls (FGFR1c).The FGFR1m cases (shown in white) demonstrate a higher C-peptide response to glucose compared to FGFR1c controls (shown in black), which was manifested during the first (0-10 minutes) and second phase (12-20) of the study.*P less than .05;**P less than 001.
Logarithmic transformation was applied for nonnormally distributedvalues by which the normality of all transformed variables was tested again with a Shapiro-Wilks test.P values less than .05wereconsideredstatisticallysignificant.For the common variant analysis, a Fisher exact test was used, and a statistical significance was defined by a P value with a cutoff of less than .012(multipletestingcorrectionfor the identified variants).Nine FGFR1m patient cases and 27 noncarrier controls were enrolled and completed the study.The average age of the FGFR1m cases was 32 years, consistent with the young age at which individuals with IHH usually present at the clinic due to pubertal failure (Table1).The average age of the FGFR1c controls was 47 years, which is in line with the fact that 46% of individuals enrolled in the MGBB are older than 60 years and only 24% of the MGBB participants are younger than 40 years Distributions of the outcomes and clinical characteristics were reported using mean and SD for normally distributed values and median and first and third quartile for nonnormally distributed values.The normality of data distribution was tested with a Shapiro-Wilks test.Statistical comparisons were performed with a t test and nonparametric Wilcoxon rank sum test for normally and nonnormally distributed values, respectively.Values that were missing (44/3242 time points) or were found to be more than above and below the mean ±3 × SDs (outliers) were excluded from the analysis.General linear model was applied to adjust the effect of FGFR1 mutation carrier status for BMI and percentage of fat mass.

Table 1 . Baseline characteristics and adiposity indices of enrolled participants
FGFR1-mutation carriers demonstrate a higher percentage of their total body fat compared to noncarrier controls, despite nondifferent BMI levels between the two groups.Abbreviations: BMI, body mass index; FGFR1, fibroblast growth factor receptor 1; HbA 1c , glycated hemoglobin A 1c ; TSH, thyrotropin.aReported as median (first and third quartile) due to nonnormal distribution of the data.

Table 2 . Indices of insulin resistance and secretary function in FGFR1-mutation carriers vs noncarrier controls Indices of insulin resistance/secretion FGFR1-mutation carriers (N = 9) Mean (SD) Noncarrier controls (N = 27) Mean (SD)
FGFR1-mutation carriers demonstrated higher fasting insulin and C-peptide levels compared to noncarrier controls.In addition, insulin sensitivity, as calculated by the S I and HOMA-SI, was lower and insulin resistance, as shown by the HOMA-IR, was higher in the FGFR1-mutation carriers compared to noncarrier controls.
Abbreviations: AIRg, acute insulin response to glucose; D I , Disposition Index; FGFR1, fibroblast growth factor receptor 1; HOMA-B, homeostatic model assessment of β-cell function; HOMA-IR, homeostatic model assessment of insulin resistance; HOMA-SI, homeostatic model assessment of insulin sensitivity; S G , glucose effectiveness; S I , insulin sensitivity.a Reported as median (first and third quartile) due to nonnormal distribution of the data.