Two female patients presented typical LGMD symptoms

The proband (II1) was born from healthy Chinese parents. She had normal motor developmental milestones, but her sports performance was worse than that of her peers. Physical examination showed normal intelligence and cranial nerve signs. Neurological examination revealed muscle atrophy of the bilateral proximal and distal lower extremities. Muscle strength (Medical Research Council grade scale) is shown in Table 1. She had a positive Gowers’ sign and a mixture of stoppage and waddling gait. She could not perform toe walking or heel walking. The knee reflex and Achilles tendon reflex were absent on both sides. The serum CK level was mildly elevated to 610 IU/L (normal range 2–178 IU/L). Nerve conduction studies (NCSs) showed normal results. Electromyography (EMG) revealed myopathic changes with the duration of motor unit action potentials (MUAPs) of the biceps brachialis, vastus medialis, and tibialis anterior decreased by 21%, 30%, and 27% respectively, and mean amplitude of MUAPs of the above muscles being 305 μV, 363 μV, and 332 μV respectively (reference value 308 μV, 350 μV, and 381 μV). Abnormal spontaneous activities were not inspected. Muscle biopsy showed myopathic features, including muscle fiber size variation, atrophic and hypertrophic fibers, increased interstitial fat and connective tissue, increased internalized nuclei and scattered vacuolar changes inside some muscle fibers (Fig. 1B). Thigh and lower leg magnetic resonance imaging (MRI) revealed symmetric changes in muscle atrophy and fatty infiltration, which were more remarkable in the biceps femoris, semitendinosus, semimembranous, and gastrocnemius, with the gracilis, tibialis posterior, and flexor digitorum longus markedly spared (Fig. 1C).

Table 1 Clinical and genetic information of family members

Patient 2 (II2) is the younger sister of patient 1. She has similar poor performance in sports as her sister. Both sisters have a short and thin habitus. Muscle strength examination revealed similar results to her sister, with the proximal and distal muscles of the upper and lower limbs affected (Table 1), and she had a positive Gowers’ sign and identical stoppage and waddling gait. Serum CK was mildly elevated to 427 IU/L. EMG revealed myopathic MUAPs in the biceps brachialis, vastus medialis, and tibialis anterior. The other family members (I1, I2, and II3) all showed a normal phenotype (Table 1).

Two patients (II1 and II2) both experienced progressive weakness and fatigue during pregnancy. The proband was pregnant for the first time at the age of 18. During this pregnancy, she felt generally fatigued, and she unusually lost 4 kg during the pregnancy without any known reason, but she did not showed typical LGMD symptoms at that time. Then, months after the delivery, she returned to her pre-pregnancy status. She was pregnant for the second time at the age of 23, and the fetus died at the seventh month of pregnancy without an apparent cause. During her third pregnancy at age 24, she felt weakness of all limbs, which was more severe in the lower limbs and she had to push her knees with hands when standing up from a squatting position. Her weakness slowly worsened during the latter part of the pregnancy until it was terminated by cesarean section, and she gave birth to a healthy boy. The weakness did not improve after delivery, nor did it get worse until she was pregnant for the fourth time at age 29. Then, she felt the deterioration and weakness along with chest tightness and palpitations. The pregnancy was terminated by induced abortion. Her sister had an uneventful first pregnancy at age 23 and experienced progressive weakness and fatigue during her second pregnancy at age 27. After delivery at age 28, her symptoms reached a plateau. The two patients have been followed up for 5 years up to present. Neither of them got pregnant again during this period, and their motor function remains stable.

A novel missense mutation and a novel deletion of the TRIM32 gene were identified

Analysis of neuromuscular disease V3-panel NGS showed that the proband was a homozygote of a novel missense mutation (TRIM32:c.1700A > G, p.H567R), and copy number analysis of the TRIM32 gene by RT-PCR revealed that one allele was deleted. Therefore, the proband (II1) should be a compound heterozygote of a missense mutation and a deletion. Genetic analysis showed that the missense mutation was inherited from her father (I1), and the deletion was inherited from her mother (I2). Another female patient (II2) had the same genotype as II1, and her brother (II3) only inherited the deletion from the mother (Fig. 1A).

WGS analysis was used to predict the range of the deletion. Bioinformatics analysis based on WGS results predicted it to be approximately 46 kb in length. According to the relatively precise range estimation, primers on both sides of the breakpoints were designed, and unique gap-PCR products were obtained in II1 (Fig. 2A). Sanger sequencing revealed that the deletion (chr9.hg19:g.119431290_119474250del) was approximately 43 kb in size, with an additional insertion of nucleotide C. The deletion resulted in the removal of the whole TRIM32 gene in addition to a portion of ASTN2 downstream region. The transcription of ASTN2 gene was not affected because the promoter and exons were maintained intactly.

Fig. 2

Molecular genotyping and expression analysis of mutated TRIM32 at the RNA and protein levels in vivo and in vitro. A Detection of the deletion in TRIM32 by multiplex gap-PCR and the exact deletion breakpoint of TRIM32 in the proband (chr9.hg19:g.119431290_119474250del). B Sanger sequencing of TRIM32 in the proband (c.1700A > G). C Multiplex sequence alignment in various species revealed that the variable base (567, labeled in red) was relatively conserved. D The predicted 3D models of the wild-type (right) and mutated (left) TRIM32 protein. An area different between the two was marked after contrastive analysis by PyMOL. E HEK 293 T cells transfected with a wild-type plasmid (pEGFP-TRIM32-WT) or mutant plasmids (pEGFP-TRIM32-D487N; pEGFP-TRIM32-H567R) under the same conditions showed that the mutant constructs exhibited diffuse cytoplasmic staining with loss of characteristic aggregates. Scale bar = 100 µm (× 40 magnification). F Western blot analysis of TRIM32 protein showed that the upper band corresponding to TRIM32/ubiquitin appeared in the wild-type and not in the mutant types (p.D487N and p.H567R). G Quantitation of TRIM32 gene transcription levels in the family

The novel missense mutation detected by NGS analysis was also confirmed by Sanger sequencing (Fig. 2B). This mutation caused the TRIM32 protein to change from histidine (H) to arginine (R) at position 567. The position was fully conserved across multiple species by species conservativeness analysis (Fig. 2C), and sequence analysis based on PolyPhen-2 predicted a score of 0.92, defined as possibly damaging. This mutation is located in the NHL domain of the TRIM32 protein. Structural modeling showed that the mutation changed the tertiary structure of the protein, causing the loss of portions of the alpha-helix.

Functional analysis of the novel missense mutation

To evaluate the effect of the missense mutation, we transfected the plasmids into HEK 293 T cells and attempted to analyze the self-association ability and monoubiquitination of TRIM32 protein by confocal microscopy and western blot. The wild-type protein (TRIM32–WT) was observed with many aggregated spots in the cytoplasm, also called cytoplasmic bodies, and was usually located around the nucleus (Fig. 2E). Meanwhile, we found the transfected cells with cytoplasmic bodies contained monoubiquitinated TRIM32 by western blot (Fig. 2F). In contrast, the known pathogenic TRIM32-D487N mutant protein was observed to have diffused cytoplasmic staining with loss of characteristic aggregates and no ubiquitin. The TRIM32-H567R mutant caused by the novel missense mutation in this study also exhibited diffuse cytoplasmic staining similar to that of the TRIM32-D487N mutant, with the protein evenly dispersed in the cytoplasm and few characteristic aggregates(Fig. 2E). In addition, similar to the known TRIM32-D487N mutant, the TRIM32-H567R mutant also did not contain the protein ubiquitination that found in the wild-type protein (Fig. 2F).

Finally, we extracted RNA from the peripheral blood of the proband (II1) and her family members and then quantitatively analyzed the mRNA expression of TRIM32. Compared with normal controls, the TRIM32 transcript level was decreased by 80% in the proband, by 70% in her sister, and by only 50% in her brother, but her parents also decreased by approximately 60% (Fig. 2G).

TRIM32 mutation spectrum and genotype–phenotype correlation in LGMD R8

After searching the literature for TRIM32 mutations plus the newly identified mutations in this study, a total of 7 deletions (Fig. 3A) and 37 point mutations (Fig. 3B) were reported. All deletions resulted in the removal of the open reading frame of the TRIM32 gene. The shortest deletion was approximately 2 kb, and the longest deletion was approximately 336 kb. Among the 37 point mutations, there were 25 missense mutations, 11 shift mutations, and only 1 nonsense mutation. These mutations can be classified into five groups based on their location on the TRIM32 gene, of which one is located in the RING structural domain, two in the B-box structural domain, six in the coiled-coil structural domain, 22 in the NHL repeats, and the remaining 6 in intermediate regions outside the structural domain. We found that the vast majority of point mutations leading to LGMD R8 were clustered within the C-terminal NHL repeats (63% [22/35]), while those leading to BBS were located in the B-box structural domain.

Fig. 3

Literature review of TRIM32 mutations and phenotypes. A Schematic diagram of TRIM32 gene deletions (length and location). B Schematic diagram of the variants in the TRIM32 gene and the protein structure. TRIM32 is located on chromosome 9(q33.1), inlaid in the ASTN2 gene. All variants are arranged horizontally in five regions by location and vertically in 3 categories by disease. The mutation and deletion described in this article are marked in red

From the published literature plus the two cases in this study, we collected 86 TRIM32 mutation-associated cases with detailed phenotype and genotype data (Supplementary Table S1). There were 22 male cases and 23 female cases, and the sex of the remaining 41 cases was not reported. Except for the hutterites, most patients were European, followed by Asian, and no patients of African were found. There were two cases of LGMD R8-merged BBS and one case of BBS, with a common feature of mutations that damage the B-box structure domain of TRIM32. The remaining 83 patients had LGMD R8, and five patients carrying monoallelic mutations showed only mild symptoms. The most common genotype was p.D487N homozygote, which was mainly distributed in the Canadian Hutterite population (51% [44/86]). The same genotype has significant phenotypic heterogeneity, with CK levels ranging from a low of 56 U/L to a high of 5556 U/L and cases ranging from asymptomatic to more severe clinical symptoms (such as being in a wheelchair).

We divided the cases into two groups according to sex. It was observed that the age of onset in males was significantly lower than that in females (mean value: male 23 years old, female 33 years old, Fig. 4A, p < 0.05). Similarly, the CK levels of males were significantly higher than those of females (mean value: male 1400 U/L, female 519 U/L, Fig. 4B, p < 0.05). In addition, since most mutations causing LGMD R8 are clustered in the NHL repeat domain, we divided the cases into six groups according to the genotype of the TRIM32 gene, including heterozygote, NHL repeats/NHL repeats, non-NHL repeats/NHL repeats, non-NHL repeats/non-NHL repeats, NHL repeats/deletions, and deletions/deletions. As shown in Fig. 4C, we found that the cases in the NHL/NHL group had a significantly lower age of onset than those in the NHL/non-NHL (p < 0.01) and NHL/DEL groups (p < 0.05), but compared with the non-NHL/non-NHL group, the difference was not significant (p > 0.05). In terms of CK levels, compared to cases in the NHL/DEL group, cases in the NHL/NHL group had significantly higher CK levels (p < 0.05), but the cases with CK level data in the NHL/non-NHL group and non-NHL/non-NHL group were too few to compare (Fig. 4D).

Fig. 4

Cases carrying TRIM32 mutations reported by age of onset and CK level. A, B All TRIM32 alleles were classified into two types: male and female. C, D All TRIM32 alleles were classified into six types: HET (heterozygote), NHL/NHL (NHL repeats/NHL repeats), NHL/non-NHL (NHL repeats/non-NHL repeats), non-NHL/non-NHL (non-NHL repeats/non-NHL repeats), NHL/DEL (NHL repeats/deletion), and DEL/DEL (deletion/deletion). The dotted line shows the normal boundary line. Details and references for the cases in the figure can be seen in Supplementary Table S1