A new gene associated with a β-thalassemia phenotype: the observation of variants in SUPT5H

TO THE EDITOR:

β-Thalassemia is among the most common autosomal-recessive conditions; it is caused by nucleotide variants and, less commonly, deletions of the β-globin gene (HBB; 11p15.4) or gene cluster,^1,2  disrupting synthesis of the β-globin polypeptide chains of the hemoglobin tetramer HbA (α₂β₂). Heterozygotes usually show reduced erythrocyte indices and elevated HbA₂ levels, and homozygotes or compound heterozygotes have severe hemolytic and dyserythropoietic anemia, usually requiring life-long blood transfusion and iron-chelation therapy.^3,4 

However, rare cases have been observed with a phenotype of β-thalassemia trait but without HBB gene or gene cluster variants. Hematology analysis by standard procedures^5,6  in several members of 2 Dutch families (Figure 1A-B) found microcytic hypochromic anemia and elevated HbA₂, consistent with β-thalassemia trait (Table 1). Molecular analysis, including Sanger sequencing of HBB, HBA1, HBA2, and KLF1, revealed a mild α-thalassemia variant HBA2:c.96-2A>G, which is found at low frequency in the Dutch population, in F1-III.1, F1-IV.1, and F1-IV.2 (F1 indicating Family 1, the Roman number the generation, and the Arab number the individual within 1 generation). No pathogenic variants were found in HBB, and multiplex ligase-dependent probe amplification excluded deletions and rearrangements involving HBB or β-locus control region. Haplotyping excluded linkage between HBB and the β-thalassemia trait in both families. The β/α-globin chain synthesis ratio in F1-II.2 showed a reduction in β-globin chain synthesis (β/α = 0.66), consistent with β-thalassemia trait.⁷  Globin-chain biosynthesis was not possible for other family members, but their erythrocyte parameters and elevated HbA₂ sufficed to establish β-thalassemia trait.

Whole-exome sequencing (WES) was performed on 4 individuals from Family 1 (3 affected: F1-III.3, F1-IV.2, F1-IV.4; 1 nonaffected: F1-IV.5) and 2 individuals from Family 2 (F2-I.2 and F2-II.3) (Figure 1A-B) (SureSelect All-Exon kit [Cre-V2]; Agilent, Santa Clara, CA and analysis using an HiSeq 4000 Sequencing System; Illumina Inc., San Diego, CA), in accordance with the Declaration of Helsinki. Missense, splice, and nonsense variants were filtered with in-house pipelines (LOVDplus). The effect of missense variants was predicted in silico using Align GVGD, PolyPhen-2, SIFT2, and MutationTaster.^8-10  Population frequencies were derived from the Genome Aggregation Database (GnomAD), and known disease associations were derived from the Human Gene Mutation Database.¹¹  Single nucleotide polymorphism array analysis was performed to exclude copy number variations in intergenic regions and to identify minimal shared regions of concordance in individuals segregating with the phenotype.

Considering the β/α-globin ratio, we focused first on pathogenic variants in known transcription factors involved in HBB regulation: GATA1, GATA2, KLF1, KLF3, ERCC2, CCND3, CTSA, PCIF1, PLTP, MMP9, TNNC2, ZFPM1, NFE2, FOG1, SOX6, SP1, and AKL4. No shared founder variant was observed in any single gene between carriers from either family. By excluding shared variants between 1 healthy and 3 affected members of Family 1, a likely pathogenic splice-site variant NM_003169.3 SUPT5H:c.458+1G>A (19q13) was identified in all affected members (Figure 1A). Examining WES data for Family 2 identified a likely pathogenic splice acceptor site mutation NM_003169.3 SUPT5H:c.2259-3C>A in the individual with β-thalassemia trait, whereas it was absent in the nonaffected members (Figure 1B;,Table 1). In addition to the 6 individuals investigated by WES, the remaining members of both families were investigated by Sanger sequencing for the familial SUPT5H variants. For other cases studied, WES was targeted exclusively for the SUPT5H gene.

Four additional independent Dutch individuals (father Wlt.1 and son Wlt.2, Li, and TW) and 2 French individuals (mother S and daughter Cl), showing a β-thalassemia trait phenotype without pathogenic variants in the HBB locus, were examined for SUPT5H variants. This revealed 4 additional pathogenic SUPT5H variants: 3 single-bp deletions and 1 4-bp deletion causing a disrupted reading frame (Table 1).

To examine the effect of splicing variants found in Family 1 and Family 2, messenger RNA (mRNA) was isolated from cultured lymphocytes from F1-IV.3 and F2-I.2 and analyzed by RNA sequencing (NEBNext Ultra II Directional mRNA kit). A significant portion of SUPT5H reads showed an altered splice pattern in both patients, whereas control samples (n = 3) did not show any splicing abnormalities. The patient with SUPT5H:c.458+1G>A showed retention of intron 6 and, consequently, a premature stop codon (TAA) at +2. Likewise the SUPT5H:c.2259-3C>A RNA showed intron 22 retention, resulting in a premature stop codon (TGA) at +46. Both mutations result in truncated Spt5H protein without the functionally important C terminus (supplemental Figure 1, available on the Blood Web site). Based on the above, we propose that haploinsufficiency for the Spt5H protein underlies reduced HBB gene expression. However, given the relatively high frequency of β-thalassemia heterozygotes worldwide, can we expect to observe individuals with digenic inheritance for variants in HBB and SUPT5H? And would they have a pronounced decrease in β-globin gene expression leading to β-thalassemia intermedia or major phenotype?

To test this hypothesis, we analyzed families with known β-thalassemia variants but with family members expressing incompletely resolved moderate β-thalassemia intermedia. In a family from Greece (Crete) with HBB:c.118C>T p.(Gln40*), 2 family members (F3-I.1 and F3-II.2) showed a more pronounced anemia (Figure 1C; Table 1). Likewise in Family 4, from northern Greece, F4-III.1 presented with moderate β-thalassemia intermedia, unexplained by simply being a carrier of HBB:c.92+1G>A. Two novel SUPT5H splice-site mutations, c.1374+2T>C (Family 3) and c.1741_1744dup (Family 4), were identified in these more severe cases; neither was present in the GnomAD database.

Figure 1E presents the genotype-phenotype correlation and hematological data for normal individuals, carriers of SUPT5H variants, and double heterozygotes for SUPT5H and HBB variants. The hemoglobin (Hb), mean cellular volume (MCV), and mean corpuscular hemoglobin (MCH) in SUPT5H carriers are markedly reduced compared with all normal family members, whereas HbA₂ is increased, similar to β-thalassemia trait. Digenic heterozygotes show a phenotype comparable to β-thalassemia intermedia, with moderate anemia, reduced Hb, MCV, and MCH, and markedly elevated HbA₂ (8.5-12.4%).

In conclusion, we report the first observation of the SUPT5H gene as a new trans-acting candidate underlying the phenotypic expression of β-thalassemia. Apart from a de novo mutation in SUPT5H recently reported in congenital heart disease,¹²  no other pathogenic variant has been reported. SUPT5H encodes the Spt5H protein, which dimerizes with Spt4, forming a highly conserved component of the DSIF complex. This complex regulates mRNA processing and transcription elongation by RNA polymerase II. It acts cooperatively with the negative elongation factor complex to enhance transcriptional pausing at sites proximal to the promoter, possibly facilitating assembly of an elongation competent RNA polymerase II complex.^13-15  Although SUPT5H is ubiquitously expressed (highly expressed in testis and bone marrow),¹⁶  studies suggest that DSIF functions in a tissue-specific manner.^17-20  A recent study revealed a fundamental role for polymerase II pausing in specification and proliferation of hematopoietic stem cells by regulating signaling pathways through DSIF.²⁰  Zebrafish experiments have shown that embryonic erythropoiesis is regulated by the transcription elongation factor Foggy/Spt5 through gata1 gene regulation. Erythrocytes were markedly decreased in zspt5KD embryos, suggesting disruption of erythroid differentiation, probably through repressed gata1 expression. In humans, GATA1 also requires FOG1 to activate HBB expression, and the present findings suggest that reduced synthesis of SUPT5H (the human homolog of Spt5) may act similarly.¹⁹  In this study, mRNA was isolated from lymphocytes, and expression of GATA1 and KLF1 was too low to detect differences in expression between SUPT5H variant carriers and noncarriers. Collaborative studies are ongoing to investigate the effects of SUPT5H haploinsufficiency on GATA1 and KLF1 expression in hematopoietic tissue and unravel the mechanism(s) through which SUPT5H haploinsufficiency reduces HBB expression. So far, overall results strongly suggest that haploinsufficient mutations in SUPT5H are associated with downregulation of HBB, acting as a phenocopy for HBB in β-thalassemia.