Human and mouse studies establish TBX6 in Mendelian CAKUT and as a potential driver of kidney defects associated with the 16p11.2 microdeletion syndrome

Abstract

Congenital anomalies of the kidney and urinary tract (CAKUT) are the most common cause of chronic kidney disease in children. Human 16p11.2 deletions have been associated with CAKUT, but the responsible molecular mechanism remains to be illuminated. To explore this, we investigated 102 carriers of 16p11.2 deletion from multi-center cohorts, among which we retrospectively ascertained kidney morphologic and functional data from 37 individuals (12 Chinese and 25 Caucasian/Hispanic). Significantly higher CAKUT rates were observed in 16p11.2 deletion carriers (about 25% in Chinese and 16% in Caucasian/Hispanic) than those found in the non-clinically ascertained general populations (about 1/1000 found at autopsy). Furthermore, we identified seven additional individuals with heterozygous loss-of-function variants in TBX6, a gene that maps to the 16p11.2 region. Four of these seven cases showed obvious CAKUT. To further investigate the role of TBX6 in kidney development, we engineered mice with mutated Tbx6 alleles. The Tbx6 heterozygous null (i.e., loss-of-function) mutant (Tbx6^+/−) resulted in 13% solitary kidneys. Remarkably, this incidence increased to 29% in a compound heterozygous model (Tbx6^mh/‒) that reduced Tbx6 gene dosage to below haploinsufficiency, by combining the null allele with a novel mild hypomorphic allele (mh). Renal hypoplasia was also frequently observed in these Tbx6-mutated mouse models. Thus, our findings in patients and mice establish TBX6 as a novel gene involved in CAKUT and its gene dosage insufficiency as a potential driver for kidney defects observed in the 16p11.2 microdeletion syndrome.

Graphical ABSTRACT

INTRODUCTION

Congenital anomalies of the kidney and urinary tract (CAKUT) have an estimated prevalence of approximately 3 – 6 in 1000 live births.^6,7 CAKUT clinical manifestations represent a heterogeneous group of developmental anomalies that include renal hypoplasia, renal agenesis, renal hypodysplasia (RHD), and vesicoureteral reflux (VUR).^8,9 Clinically, renal hypoplasia is characterized by abnormally small kidneys with normal morphology and decreased number of nephrons.¹⁰ Renal agenesis is characterized by unilateral or bilateral fetal kidney developmental failure. RHD is described by a small kidney size, reduced nephron number, and/or disorganized renal tissue.¹¹ VUR is caused by abnormal insertion of the ureter within the bladder wall, which can result in susceptibility to acute kidney infection and chronic kidney disease.¹² Importantly, CAKUT are the main cause of end-stage renal disease in childhood.^13,14

Evidence for the genetic contribution to CAKUT is substantive.¹⁵ For example, variants in EYA1, GREB1L, HNF1B, PAX2 and TBX18 can cause CAKUT in humans.^16–19 Nicolaou et al. summarized 208 candidate genes involved in CAKUT.²⁰ Besides gene variants due to single nucleotide variants and indels, genomic copy number variants (CNVs) can also cause CAKUT.^21–25 Of note, the 17q12 deletion was found in 2.2% of RHD patients.^21,26 The 22q11.2 deletion was found in 1% of CAKUT patients.²⁶ Previously, nine CAKUT cases and five population controls with overlapping 16p11.2 deletions were reported in a genome-wide CNV study based on 2,824 CAKUT cases and 21,498 controls (0.32% versus 0.02%; OR 13.7, 95% CI 4.1–52.2; P = 4.39 × 10⁻⁶).²⁷ The human TBX6 gene maps within the 16p11.2 deletion interval. Our previous experiments have suggested that TBX6 dosage insufficiency might be involved in kidney defects associated with the 16p11.2 microdeletion syndrome.²⁷ However, to establish a single-gene causal effect it remains to be demonstrated whether a) TBX6 coding variants are observed in individuals with CAKUT; b) haploinsufficiency is sufficient to cause human kidney defects; and c) TBX6 gene dosage has an effect on the penetrance of renal and extrarenal phenotypes in human CAKUT.

Here we recruited human subjects molecularly diagnosed to be due to 16p11.2 deletion²⁸ from worldwide multi-center cohorts and retrospectively analyzed available clinical data for renal system abnormalities. Our data further suggested 16p11.2 deletion as a genetic factor contributing to CAKUT across different populations. We also identified heterozygous null TBX6 carriers who manifest CAKUT supporting the contention that TBX6 in 16p11.2 is the gene potentially driving the perturbations in kidney and urinary tract system development. Finally, the genetics of TBX6-associated CAKUT were further explored using mouse models of novel Tbx6 allelic series that closely recapitulate the down-regulation of TBX6 expression in humans. Our mouse model experiments provide evidence for Tbx6 gene-dosage alterations and a dosage-dependent risk of CAKUT. In aggregate, these new human and mouse studies establish TBX6 variants as a novel Mendelian cause of CAKUT, and implicate TBX6 as a potential driver of the 16p11.2-associated CAKUT. Moreover, our allelic series support the contention of a role for a compound inheritance model as the modifier of disease penetrance of CAKUT, a genetic model similar to what we previously described in congenital scoliosis (CS) and in lung developmental defects.^1–3,29

RESULTS

Recurrence of CAKUT with 16p11.2 deletion in different ethnicities

Our previous study showed an association between 16p11.2 deletion and CAKUT.²⁷ However, we have also observed that the coexistence of CNV and single nucleotide polymorphisms (SNPs) at a locus can result in distorted calculations of the significance in associating SNPs with disease.³⁰ To further explore this association of 16p11.2 with CAKUT, we conducted reverse phenotyping for renal morphology and function of 24 patients with CS, caused by 16p11.2 deletion.^2,31 These 16p11.2-deleted subjects were collected from the Peking Union Medical College Hospital (PUMCH, China). Detailed clinical information was available from 12 deletion carriers (Table 1). Three (25%) subjects (XH265, XH300, and XR636) had renal hypoplasia; reductions in kidney volumes of > 2 standard deviations below the mean in controls were considered as renal hypoplasia (Tables 1 and Supplementary Table S1 and Figure S1).^32,33 Thus, the renal phenotypes of CAKUT were observed in at least 3/12 (25%; subjects XH265, XH300 and XR636 underlined in Table 1) Chinese 16p11.2 deletion carriers available for objective evaluation of kidney morphology and function.

Table 1|.

Genetic and renal characteristics of Chinese subjects with 16p11.2/TBX6-associated CS

			TBX6 gene		Renal evaluation
Human subject	Sex	Age (yr)a	First null allele	Second hypomorphic alleleb	Kidney volume (cm3)		Kidney symmetry	Renal hypoplasiac	Renal parenchyma damage	Phenotype
					Left	Right
XH004	F	4	16p11.2 del	Yesd	55.25	55.96	0.99	Left: no Right: no	NA	Isolated CS
XH186	M	4	16p11.2 del	Yesd	37.27	33.21	0.89	Left: suggestive Right: suggestive	No	Isolated CS
XH265	M	3	16p11.2 del	Yesd	47.99	27.82	0.58	Left: no Right: yes	No	Syndromic (CAKUT+CS)
XH292	F	12	16p11.2 del	Yesd	103.94	115.53	0.9	Left: no Right: no	No	Isolated CS
XH300	M	3	16p11.2 del	Yesd	32.26	25.85	0.8	Left: yes Right: yes	No	Syndromic (CAKUT+CS)
XR330	M	4	16p11.2 del	Yese	45.58	36.09	0.79	Left: no Right: suggestive	No	Isolated CS
XR402	M	13	16p11.2 del	Yesd	116.96	90.55	0.77	NA	No	Isolated CS
XR439	F	7	16p11.2 del	Yesd	61.67	52.13	0.85	Left: no Right: no	No	Isolated CS
XR468	M	3	16p11.2 del	Yese	38.41	32.61	0.85	Left: suggestive Right: suggestive	No	Isolated CS
XR522	M	8	16p11.2 del	Yese	104.28	85.19	0.82	Left: no Right: no	Persistent proteinuria	Isolated CS
XR529	M	14	16p11.2 del	Yese	129.39	129.45	1	NA	No	Isolated CS
XR636	F	9	16p11.2 del	Yese	43.92	53.39	0.82	Left: yes Right: suggestive	No	Syndromic (CAKUT+CS)
XH122	M	8	c.266dupC (p.V91Gfs*80)	Yesd	91.09	58.78	0.65	Left: no Right: yes	Bilateral delayed renal tissue tracer transit	Syndromic (CAKUT+CS)
XH170	M	4	c.704dupG (p.M236Hfs*44)	Yesd	31.28	40.21	0.78	Left: yes Right: no	No	Syndromic (CAKUT+CS)
XR341	M	2	c.1179_1180delAG (p.G395Lfs*91)	Yesd	28.14	27.37	0.97	Left: yes Right: suggestive	No	Syndromic(CAKUT+CS)
XR625	M	7	c.933C>A (p.C311*)	Yese	50.2	59.85	0.84	Left: suggestive Right: no	No	Isolated CS

Note: The renal phenotypes potentially involved in CAKUT are underlined.

Abbreviations: CAKUT, congenital anomalies of the kidney and urinary tract; CS, congenital scoliosis; del, deletion; F, Female; M, Male; NA, not available; yr, years.

^aSubject age (years) at the time of evaluating the renal volume.

^bThe hypomorphic haplotype of human TBX6 was defined by the mutated alleles of 3 TBX6 variants: rs2289292, rs3809624, and rs3809627.

^cThe left and right kidneys were investigated separately. Differences in kidney volumes of > 2 standard deviations below the means in age-matched and ethnically matched controls were considered to reflect renal hypoplasia.¹⁰ Differences of > 1 standard deviation below the means in control subjects were considered to be suggestive of renal hypoplasia. Please refer to Supplementary Table S1 for details.

^dGenotyping of these subjects was conducted in our previous study.³¹

^eGenotyping of these subjects was conducted in our recent studies.^2,3

Furthermore, in terms of kidney symmetry, two additional subjects XR330 and XR402 had symmetry values (i.e., volume ratio of the small kidney to the contralateral large kidney in each patient) < 0.8, and the value of the aforementioned subject XH265 was < 0.6, also suggesting unilateral renal hypoplasia (Table 1). Thus, the rate of kidney defects in 16p11.2 deletion carriers would be further increased to 5/12 (42%; subjects XH265, XH300, XR330, XR402 and XR636 in Table 1), when considering kidney asymmetry.

We next investigated 78 anonymized subjects of Caucasian/Hispanic ancestry with 16p11.2 deletion from Baylor College of Medicine (BCM) in Houston, USA. Imaging and clinical data assessing the renal system were available for 25 subjects. Interestingly, five (20%) subjects (BCM-1 to BCM-5) with 16p11.2 deletions were identified with various renal or urinary tract anomalies (Figure 1a, Table 2), including solitary kidney (subject BCM-5), VUR (subjects BCM-1 and BCM-2), hydroureteronephrosis (subject BCM-5), complex cystic structure in the left kidney (subject BCM-2), trace bilateral pelvicalyceal dilation (subject BCM-4), and bilateral small kidneys indicating renal hypoplasia (subject BCM-3). Since the exact data of kidney volumes of subject BCM-3 are not available, this case is not categorized into the CAKUT group in this study (shown as “suggestive” in Table 2). Thus, renal phenotypes consistent with CAKUT were observed in at least 4/25 (16%) carriers (BCM-1, BCM-2, BCM-4 and BCM-5) of 16p11.2 deletion available for objective evaluation (Figure 1a, Table 2). The 16p11.2 deletion associated renal phenotypes in Caucasian/Hispanic subjects are consistent with our observations in the Chinese population.

Figure 1|. The 16p11.2/TBX6 deletions and TBX6 frame-shift variants in the BCM cohort.

(a) Common proximal 16p11.2 deletions (29.6 – 30.2 Mb, the genome reference consortium human build 37 [GRCh37]) were identified in subjects BCM-1, BCM-2, BCM-3, and BCM-5. Blue shadowing indicates the common deletion region. The deletion in subject BCM-4 is larger (21.7 – 30.2 Mb) than the common deletion. The red circle indicates the location of TBX6. The microarrays applied with samples from these individuals were based on different versions of customized oligonucleotide CGH arrays with different numbers of oligonucleotide probes. (b) A heterozygous frame-shift variant of TBX6 (c.1018_1019delCT; p.L340Vfs*7) in subject BCM-6. The red box indicates the deleted nucleotides. (c) A heterozygous frame-shift variant of TBX6 (c.138_141dupGGAT; p.C48Gfs*30) in subject BCM-7. Red shadowing indicates the duplicated nucleotides. (d) A heterozygous splice-site variant of TBX6 (c.118+1G>C) in subject BCM-8. The red arrow indicates the mutated nucleotide. Abbreviation: WT, wide type.

Table 2|.

Renal system evaluation of the 16p11.2/TBX6-associated subjects in the BCM cohort

				TBX6 gene
Human subject	Sex	Age at diagnosis	Ancestry	First null allele	Second hypomorphic allelea	Urinary tract phenotype	CAKUT	Vertebral malformations
BCM-1	M	2 mo	Hispanic	16p11.2 del	No	Left VUR, bilateral mild increased renal echogenicity	Yes	Hemivertebrae between L2 and L4
BCM-2	M	4 yr	Hispanic	16p11.2 del	Yes	Bilateral VUR and complex cystic structure in left kidney	Yes	NA
BCM-3	F	10 mo	Hispanic	16p11.2 del	Yes	Bilateral kidneys smaller than normal size	Suggestive	Hemivertebrae at T10 and T11
BCM-4	M	6 yr	Caucasian	16p11.2 del	Yes	Bilateral increased renal echogenicity with trace bilateral pelvicalyceal dilation, enlarged right kidney	Yes	NA
BCM-5	M	3 mo	Hispanic	16p11.2 del	No	Right solitary kidney with hydroureteronephrosis with markedly dilated and tortuous right ureter to the level of the pelvis	Yes	Three hemivertebrae at L1, L3, and L5
BCM-6	M	1 yr	Hispanic	c.1018 1019delCT (p.L340Vfs*7)	No	Right kidney agenesis and contralateral compensatory hypertrophy	Yes	No scoliosis
BCM-7	F	2 yr	Hispanic	c.138 141dupGGAT (p.C48Gfs*30)	No	Recurrent urinary tract infections	Suggestive	NA
BCM-8	F	12 yr	Hispanic	c.118+1G>C	No	No obvious abnormality	No	No scoliosis

Note: Renal system evaluation was performed by ultrasound except BCM-5 (by MRI).

Abbreviations: CAKUT, congenital anomalies of the kidney and urinary tract; del, deletion; F, Female; M, Male; L, lumbar; mo, months; NA, not available; T, thoracic; VUR, vesicoureteral reflux; yr, years.

^aThe mild hypomorphic haplotype of human TBX6 was defined by the mutant alleles of 3 TBX6 SNPs: rs2289292, rs3809624, and rs3809627.

In our recent study on 42 subjects with 16p11.2 deletion from the Children’s Hospital of Philadelphia (CHOP), clinical data assessing the kidney and urinary tract were available for 15 cases, and 6 (40%) of them showed CAKUT phenotypes.²⁷

Collectively, our observations on the Chinese, BCM and CHOP cohorts revealed CAKUT-related phenotypes in 13/52 (25%) cases, strongly implicating a potential gene dosage effect at this 16p11.2 locus as an important susceptibility factor for CAKUT.

Heterozygous loss-of-function variants of TBX6 constitute a novel form of human monogenic CAKUT

Among genes mapping within 16p11.2 (Supplementary Figure S2), TBX6 is a well-known transcription factor, in which genetic and genomic variants at the locus have been associated with multiple developmental disorders.³⁴ A recent study showed Tbx6-expressing cells in the mesonephric duct and in the cells that would later give rise to the metanephric mesenchyme, the latter being the forerunner of renal tubules, of mouse embryos.³⁵ We previously reported that TBX6 in 16p11.2 is a candidate gene responsible for kidney developmental anomalies.²⁷ However, no loss-of-function variants of TBX6 in CAKUT patients have been reported to date.

Here we reviewed the renal morphology and function of human subjects with heterozygous loss-of-function variants in TBX6. Seven Chinese cases were previously collected from PUMCH and 3 Caucasian/Hispanic cases were newly collected from BCM.^2,31 Imaging and clinical data assessing the renal system of 4 subjects from PUMCH and 3 subjects from BCM were available for review of the data.

In total, by reverse phenotyping of these 7 TBX6 variant allele carriers (XH122, XH170, XR341, XR625, BCM-6, BCM-7 and BCM-8), we identified clear CAKUT phenotypes in 4 (57%) subjects (XH122, XH170, XR341 and BCM-6 in Tables 1 and 2). Two additional variant carriers (XR625 and BCM-7) had milder abnormalities suggestive of CAKUT (Tables 1 and 2).

Specifically, among Chinese patients with heterozygous TBX6 loss-of-function variants (Table 1, Supplementary Tables S1 and S2), subject XH122 (c.266dupC, p.V91Gfs*80) had unilateral renal hypoplasia and bilateral delayed renal tissue tracer transit in ^99mTc-diethylenetriamine pentaacetic acid (^99mTc-DTPA) renal dynamic imaging suggesting a urinary obstruction consistent with ureteropelvic junction obstruction, another form of CAKUT. Subjects XH170 (c.704dupG, p.M236Hfs*44), and XR341 (c.1179_1180delAG, p.G395Lfs*91) had evidence for renal hypoplasia by volume measurements and in comparison to age-matched control values.^31–33 Subject XR625 (c.933C>A, p.C311*) had asymmetric kidney volume but the smaller left kidney measured to the lower limit or normal size.

Among Caucasian/Hispanic cases with heterozygous TBX6 loss-of-function variants (Figure 1b, Table 2), subject BCM-6 (c.1018_1019delCT, p.L340Vfs*7) had CAKUT characterized by right renal agenesis with contralateral compensatory hypertrophy. Subject BCM-7 (c.138_141dupGGAT, p.C48Gfs*30) had recurrent urinary tract infections suggestive of CAKUT (in particular VUR) but no definitive imaging studies were available for confirmation.

Thus, in aggregate, our data strongly implicate TBX6 haploinsufficiency as a cause of CAKUT phenotypes across human populations of different race and ethnicity.

Reduced gene dosage of Tbx6 induced CAKUT phenotypes in mice

No homozygous null alleles have been reported in humans while homozygosity for a null allele of Tbx6 is embryonic lethal in mice.^2,36 We recently demonstrated that severe reduction of Tbx6 gene dosage in the ‘rib-vertebrae’ (rv) mutant mice resulted in high penetrance of CAKUT.²⁷ In order to study a model that more closely recapitulates the effect of human heterozygous 16p11.2 deletions or TBX6 loss-of-function variants, we sought to study mouse models of milder reduction of Tbx6 gene dosage. We previously generated a Tbx6 null allele (Tbx6^‒) and a mild hypomorphic allele of Tbx6 (Tbx6^mh) in mice using the CRISPR-Cas9 technology.² The Tbx6^mh allele resulted in the reduction of Tbx6-expression levels to ~ 65% of the wild-type (WT) allele, as determined using a gene-expression reporter system.² The Tbx6^mh allele mimics the human TBX6 hypomorphic allele in function (Supplementary Figure S3).²

To explore the phenotypic consequences of combinations of Tbx6 alleles for renal system development, mice bearing different combinations of alleles were studied. In a pilot experiment we dissected 10 compound heterozygous Tbx6^mh/‒ adult mice, and identified solitary kidneys in 20% (2/10) as compared to 0/10 WT mice. This incidence of solitary kidney was much greater than expected in mice and also significantly higher than that in humans (~ 1/1000 found in humans at autopsy; Figure 2a–c).³⁷ Five different mouse strains with distinct combinations of genotypes at the Tbx6 locus (Tbx6^+/+, Tbx6^+/mh, Tbx6^mh/mh, Tbx6^+/−, and Tbx6^mh/‒), aged 35–45 days, were obtained and examined. This allelic series allowed examination of diminishing gene dosages at the Tbx6 locus. The strain Tbx6^+/mh had the smallest reduction in gene dosage, and the strain Tbx6^+/− represented 50% reduction at the locus.

Figure 2|. CAKUT phenotypes were frequently observed in the mice with Tbx6 gene-dosage insufficiency.

(a) Normal kidneys. (b) Solitary kidney (renal agenesis). (c) Asymmetric kidneys (renal hypoplasia). (d) The frequencies of solitary kidney in the Tbx6^+/+, Tbx6^+/mh, Tbx6^mh/mh, Tbx6^+/−, and Tbx6^mh/‒ mice. (e) Kidney symmetry in mice with different genotypes (excluding mice with solitary kidneys). Bar heights indicate mean values. Error bars indicate 95% confidence intervals. * P < 0.05; **** P < 0.0001; ns, not significant.

Significantly more solitary kidneys were observed in Tbx6^+/− mice (7/55, 13%; P = 0.013 versus WT, Fisher’s exact test) and Tbx6^mh/‒ mice (16/56, 29%; P = 1.0 × 10⁻⁵ versus WT, Fisher’s exact test). But no solitary kidneys occurred in Tbx6^+/+ (n=52), Tbx6^+/mh (n=58), or Tbx6^mh/mh (n=58) mice (Figure 2d).

In addition, we directly and quantitatively measured mouse ‘kidney symmetry’ calculated as the weight ratio of the small kidney to the large kidney for each mouse. The theoretical symmetry values range from 0 to 1 (Figure 2e). We also excluded mice with a solitary kidney and investigated kidney symmetry and potential renal hypoplasia only in mice with both kidneys. The symmetry values of the Tbx6^+/− (0.86±0.12) and Tbx6^mh/‒ (0.84±0.10) groups were significantly less than that of the WT group (P = 2.8 × 10⁻⁶ and P = 7.4 × 10⁻⁷ by Wilcoxon’s rank-sum test, respectively; Figure 2e), indicating unilateral renal hypoplasia. No significant difference in kidney symmetry was observed in the Tbx6^+/mh or Tbx6^mh/mh groups (P = 0.58 and P = 0.19 by Wilcoxon’s rank-sum test, respectively) when compared with the WT group (Figure 2e).

Given that the mice with kidney symmetry values of < 0.7 are suggestive of renal hypoplasia, we studied the frequencies of renal hypoplasia across genotypes with diminished Tbx6 gene dosage. Renal hypoplasia and solitary kidneys was found in none of 52 WT mice (Tbx6^+/+), in none of 58 Tbx6^+/mh mice, in only 1/58 (2%) Tbx6^mh/mh mice, in 12/55 (22%) Tbx6^+/− mice (P = 2.6 × 10⁻⁴ versus WT, Fisher’s exact test), and in 19/56 (34%) Tbx6^mh/‒ mice (P = 7.6 × 10⁻⁷ versus WT, Fisher’s exact test) (Supplementary Figure S4). Our renal phenotype observations in the Tbx6 allelic series of mouse models show a statistically significant correlation of Tbx6 gene dosage levels with the rates of renal hypoplasia and agenesis.

Severe renal phenotypes may result in mouse death before 35–45 days old. Therefore, we also investigated renal phenotypes in mouse E18.5 embryos. The Tbx6^mh/- embryos showed unilateral renal hypoplasia and renal agenesis (Fig. 3a–c), which are similar to those in adult mice of 35–45 days old. Hematoxylin-eosin stained renal sections showed no obvious abnormality in renal cortex and kidney medulla (Fig. 3e).

Figure 3|. Analysis of CAKUT phenotypes in E18.5 Tbx6^mh/– embryos.

(a–c) Whole mounts of urogenital tracts. (a) Normal kidneys in a WT mouse embryo. Renal hypoplasia (b) and solitary kidney (c) were observed in Tbx6^mh/– embryos. (d-f) Hematoxylin and eosin stained sagittal sections of the right kidneys of samples in a-c. (d) Renal section of the WT embryo. (e) Renal section of the Tbx6^mh/– hypoplasia kidney. (f) Renal section of the Tbx6^mh/– solitary kidney. The arrow points to the undeveloped kidney. Abbreviations: B, bladder; K, kidney; m, kidney medulla; rc, renal cortex; U, ureter.

Altered expression of CAKUT-related genes in mouse embryos with Tbx6 dosage insufficiency

RNA sequencing was conducted to analyze differential expression of the genes potentially involved in renal development between WT and Tbx6 low-dosage (Tbx6^mh/‒) embryos (Supplementary Table S3).²⁰ Mouse ureteric buds branch into the metanephric mesenchyme at E11.5, and renal vesicles and S-shaped bodies develop from E12.5 to E16.5.^38,39 Because both stages are crucial for renal hypoplasia etiology, we harvested mouse embryos at E11.5 and E14.5, and collected renal tissues for RNA sequencing. Hoxd11 was significantly up-regulated at E11.5, and 5 genes involved in kidney development (Cdh6, Dach1, Frem1, Lhx1 and Slit2) were significantly down-regulated in Tbx6^mh/‒ embryos, when compared with the expression levels in WT mice (Q < 0.05 with false-discovery-rate correction) (Supplementary Table S4). Because Tbx6 expression was very low in renal tissues from embryos at E11.5 and E14.5, altered expression of the above genes could be an indirect consequence of kidney development in Tbx6 mutants, rather than a direct regulation by Tbx6. Ectopic activation of mouse Hoxd11 altered the morphology and differentiation of mesonephric tubules.⁴⁰ Furthermore, Lhx1 is a marker of the anterior intermediate mesoderm, essential for mammalian kidney development.⁴¹ Experimental evidence from animal models suggests that Tbx6 acts upstream of Lhx1.^42,43

DISCUSSION

The 16p11.2 deletion has a population frequency of ~ 0.03% worldwide.⁴⁴ Our previous studies only revealed CS in the carriers of 16p11.2 deletion.^1–3,45 In a large genome-wide study for CNV, we recently showed a significant enrichment of this deletion in CAKUT cases as compared to population controls (0.32% versus 0.02%; OR 13.7, P = 4.39 × 10⁻⁶), and showed that 40% of North Americans with 16p11.2 deletion had evidence for CAKUT.²⁷ Here, we found that at least 25% of 16p11.2 deletion carriers of Chinese ancestry and 16% of carriers of Caucasian/Hispanic ancestry had clinical features consistent with CAKUT. These experimental observations, combined with our previous work²⁷ indicate an overall prevalence of CAKUT in at least 25% (13/52) of 16p11.2 deletion carriers, further supporting the contention that this deletion is an important genetic risk factor contributing to CAKUT across world populations of different ethnicities.²⁷

Tbx6 is expressed in the mesonephric duct of mouse embryos,^35,46 suggesting that Tbx6 is associated with renal development. However, it remained unclear whether TBX6 variants represent a monogenic form of CAKUT in subjects without 16p11.2 deletion. Our previous work also showed non-coding variants of TBX6 in CAKUT;⁴⁷ however, the causality of TBX6 coding variants in CAKUT was not established in that study. Here we provide evidence that heterozygous TBX6 loss-of-function variants can also cause monogenic CAKUT on their own with incomplete penetrance. In fact, by screening worldwide populations, we found that at least 3 Chinese subjects (XH122, XH170 and XR341) and 1 Caucasian/Hispanic subject (BCM-6) carrying TBX6 frame-shift variants also had clinical phenotypes consistent with CAKUT. Moreover, an additional analysis of our new mouse models with milder reduction of Tbx6 gene expression (hence more closely recapitulating the gene dosage effect of 16p11.2 deletion in humans) provide strong evidence that TBX6 is a genetic driver of the renal system birth defect phenotypes observed in patients with the 16p11.2 deletion syndrome. Also, we cannot exclude the genetic contributions of other 16p11.2 genes to CAKUT. It has been recently reported that MAZ in 16p11.2 is involved in CAKUT.⁴⁸

Pathogenic variants can have pleiotropic effects. For example, heterozygous JAG1 null variants lead to heart, liver, skeletal, and eye malformations in humans.^49–51 Heterozygous NOTCH2 null variants cause human renal and vertebral malformations.^52,53 Homozygous null variants in HAAO or KYNU cause cardiac, vertebral, and renal defects in humans and mice.⁵⁴ Similarly, TBX6 variants are also pleiotropic.

The relationship between CAKUT phenotypes and CS in the TBX6-mutated subjects as well as the genetic model underlying these traits was unclear until now. TBX6 gene-dosage reduction can lead to CS through a compound inheritance model.^2,3,31,55 Here, we found that subject BCM-6 carrying TBX6 loss-of-function variants only showed CAKUT without any symptoms of CS. Our previous study showed that among the 15 CAKUT patients with 16p11.2 deletions, only two had CS, suggesting different penetrance for CAKUT and CS, where CAKUT manifestation requires haploinsufficiency at the 16p11.2 locus and, mainly TBX6.²⁷ Consistently, our Tbx6^+/− mice also showed frequent renal hypoplasia with no CS-related phenotypes. Therefore, the consistent renal phenotypes observed in humans and mice harboring TBX6 variants were not dependent on CS, but were associated with TBX6 dosage reduction. Our study will facilitate a more comprehensive molecular diagnosis of the disorder and genetic counseling for human subjects carrying 16p11.2 deletion or a TBX6 null variant.

Tbx6^rv/rv mice, with a 182-bp insertion at 240 bp upstream of the translational initiation site of Tbx6,⁵⁶ manifested high penetrance of kidney and urinary tract malformations in our previous study.²⁷ Nevertheless, the ‘rv’ mutant is a severe hypomorphic allele that can strongly reduce Tbx6 expression to much less than 50% of that in WT. In contrast, the common variant hypomorphic allele at human TBX6 is mild, which reduces TBX6 expression level to ~ 70% of the WT allele.³¹ Furthermore, the human mild hypomorphic TBX6 allele is very common in human populations, and its homozygous carriers had no clinical phenotypes.³¹ Therefore, the severe rv mutant in mice does not readily mimic the common variant mild hypomorphic TBX6 allele in humans. Here, we generated a mild hypomorphic mutant allele (Tbx6^mh) in mice. The expression of Tbx6^mh, which was ~ 65% that of the WT allele, closely matched the expression level of human hypomorphic TBX6 allele.² Here we showed that the compound heterozygous alleles Tbx6^mh/‒ caused a further reduction in Tbx6 gene dosage and led to a higher rate of solitary kidney and renal hypoplasia when compared with the Tbx6^+/− mice (Figure 2). This observation suggests that the common mild hypomorphic TBX6 allele in human populations may be an important genetic modifier in human CAKUT.

Our previous study suggested that the combination of a rare null allele with a hypomorphic common variant allele of TBX6/Tbx6 resulted in CS in humans and mouse.^1–3 Such a compound inheritance at the same locus can result in gene dosage reduction and modify penetrance of diseases. In addition to TBX6, more loci are potentially involved in the compound inheritance model.⁵⁷ While haploinsufficiency in TBX6 clearly contributes to the penetrance of CAKUT, it may be that common variation at a downstream gene contributes to a compound inheritance model. Such a digenic mutational burden in a developmental pathway has been proposed for penetrance of craniosynostosis.⁵⁸

Our data emphasize the gene-dosage effect and its important contribution to human birth defects. The molecular mechanism whereby TBX6 causes CAKUT and CS showed different dosage sensitivities. Here, we found that half of normal TBX6 gene dosage can result in high-to-moderate penetrance of CAKUT. However, this half dosage of TBX6 was not sufficient to manifest CS. Further reduction in TBX6 gene dosage to levels less than haploinsufficiency can cause CS.^2,3,31 Furthermore, we found that Tbx6 gene dosage was related to the penetrance of CAKUT in mouse models. With a further reduction of Tbx6 dosage, an increasing proportion and increasing asymmetric severity of CAKUT occurred in mice. It was reported that different VUR percentages in different mouse strains.⁵⁹ Therefore, the Tbx6 mouse models were not assayed for VUR in this study.

In short, our study revealed three major findings. Firstly, we found that at least 25% of 16p11.2 deletion carriers of Chinese and Caucasian/Hispanic ancestries had clinical features consistent with CAKUT, a different phenotype from well-known 16p11.2-associated CS.^1–3,45 This study extends what was newly reported that 16p11.2 deletion is an important genetic risk factor of CAKUT.²⁷ Secondly, this current study provided human and mouse experimental evidence to support that TBX6 is a potential genetic driver of CAKUT in cases with 16p11.2 deletion. Thirdly, using a novel mouse allelic series, we provide evidence for Tbx6 gene dosage insufficiency as the main mechanism for CAKUT determination and for a Tbx6 dosage-dependent penetrance of CAKUT.

MATERIALS AND METHODS

CAKUT cohort of patients

Han Chinese cases were collected from PUMCH. The institutional review boards at Fudan University and PUMCH approved this work. Informed consents were obtained from their guardians (for participants < 18 years of age). Quantitative polymerase chain reaction (qPCR) was conducted to screen for 16p11.2/TBX6 deletions in human subjects, and comparative genomic hybridization (CGH) microarrays (Agilent, USA) were used to verify 16p11.2/TBX6 deletions. The experimental details of qPCR, Sanger sequencing, and CGH microarrays were provided in our previous works.^2,31 Three-dimensional kidney reconstructions were conducted using computed tomography and magnetic resonance imaging (MRI). The methodological details were provided in Supplementary Figure S1.

Data from additional anonymous subjects with TBX6 deletions and loss-of-function variants were obtained from BCM (Houston, USA). The institutional review board at BCM approved this work. These subjects were initially referred for clinical chromosomal microarray analysis or exome sequencing, due to various medical problems.^60,61 Most subjects in the BCM cohort are of Caucasian/Hispanic ancestry.

Renal hypoplasia analysis

For renal hypoplasia analysis in human subjects, three-dimensional kidney reconstructions were conducted using computed tomography (CT) and MRI (Supplementary Figure S1). Volume=0.5233×length×width×thickness.⁶² Normal parameters referred to the reported kidney volume of Chinese children without a history of renal disease or pathological abnormalities.⁶³ The left and right kidneys were investigated separately. Differences in kidney volumes of > 2 standard deviations below the means in age-matched and ethnically matched controls were considered to reflect renal hypoplasia.¹⁰

As for renal phenotypic analysis of adult mice, mice were sacrificed by cervical dislocation at aged 35–45 days before kidney images were taken. Mouse kidneys were removed from individual cadaveric animals, and the weights were determined between each kidney with the ratio used to determine renal hypoplasia. As for mouse renal hypoplasia analysis, mouse ‘kidney symmetry’ was calculated as the weight ratio of the small kidney to the large kidney for each mouse. The mice with kidney symmetry values of < 0.7 were suggestive of unilateral renal hypoplasia in this study.

The kidneys of E18.5 mouse embryos were taken images and fixed overnight in 4% paraformaldehyde and dehydrated through a graded ethanol series (75–85–90–95–100%), then washed with xylene. The embryos were incubated in two changes of paraffin wax at 55°C under vacuum. Then the embryos were embedded in paraffin wax and sectioned at 4 μm. Sections were dewaxed in xylene, rehydrated through ethanol (100–95–90–80–70%), then rehydrated with deionized water, stained with hematoxylin and eosin and dehydrated before mounting in Cytoseal.

Sanger sequencing and TBX6 haplotyping in human subjects

The entire TBX6 gene and its ~ 1 kb upstream region were amplified by long-range PCR, and the PCR products were analyzed by Sanger sequencing. Clone sequencing was performed to verify common TBX6 variants. Supplementary Tables S5 and S6 show the amplification and sequencing primers. The experimental details were provided in our previous work.³¹

Generation of Tbx6-mutated mice

Animal experiments were approved by the institutional review board at Fudan University. The FVB/NJ mouse strain (the Jackson Laboratory, #001800) was used for animal experiments. CRISPR-Cas9 technology was employed to edit Tbx6 in mice (Supplementary Table S7). Mouse models and strain construction were described in our previous work.² Mouse genotyping was performed by PCR and Sanger sequencing, using DNA from toe clips. The amplification and sequencing PCR primers are shown in Supplementary Table S8. The experimental details were provided in our previous work.²

RNA sequencing

Female Tbx6^+/mh mice were crossed with male Tbx6^+/− mice to produce Tbx6^+/+ and Tbx6^mh/‒ embryos. Mouse embryos were harvested at E11.5 and E14.5, three of which were randomly selected for the Tbx6^+/+ and Tbx6^mh/‒ genotypes. The renal region of E11.5 embryos and the left kidney tissue from E14.5 embryos were collected for total RNA extraction using the RNeasy Mini Kit (Qiagen). Total RNA (2 μg) was used to construct a cDNA library for each embryo. DNA-sequencing libraries were prepared using the TruSeq Nano DNA HT Library Prep Kit (Illumina). All cDNA libraries were sequenced using the Illumina HiSeq X-TEN platform. At least 10 Gb of raw data were obtained for each library. The expression levels of candidate genes were normalized and presented as fragments per kilobase of transcript per million fragments mapped to provide estimated gene-transcript abundances.⁶⁴

Statistical analysis

We used the Fisher’s exact test to investigate the different prevalence of CAKUT phenotypes between WT and Tbx6-mutated mice. The Wilcoxon rank-sum test was performed using R software for statistical analysis of kidney symmetry. Gene expression level changes between Tbx6^+/+ and Tbx6^mh/‒ embryos were analyzed using the Cuffdiff program in the Cufflinks software.⁶⁵ The P values calculated for each gene were adjusted to Q values using the false-discovery rate method for multiple testing corrections.⁶⁶

Supplementary Material

1

Figure S1. Three-dimensional reconstruction of the kidneys in TBX6-associated human subjects using MRI images.

Figure S2. The human genomic region affected by proximal 16p11.2 deletion and the locations of human TBX6 variants.

Figure S3. Generation of Tbx6 null and mild hypomorphic mutants in mice using CRISPR-Cas9.

Figure S4. Renal hypoplasia and agenesis in the mice with different genotypes, namely Tbx6^+/+, Tbx6^+/mh, Tbx6^mh/mh, Tbx6^+/− and Tbx6^mh/−.

Table S1. Clinical records and kidney parameters of the TBX6-associated Chinese subjects.

Table S2. Urinalysis and renal function parameters of the TBX6-associated Chinese subjects with CS.

Table S3. Mouse genes implicated in solitary and asymmetric kidneys.

Table S4. Candidate genes with significant differences (Q < 0.05) in expression levels between the Tbx6^mh/− and wild-type mouse embryos.

Table S5. Primers for amplifying the human TBX6 gene.

Table S6. Sequencing primers for the human TBX6 gene.

Table S7. Target sequences and locations of single-guide RNA (sgRNA) in the mouse genome.

Table S8. Primers for mouse Tbx6 gene amplification and sequencing.