BLVRB redox mutation defines heme degradation in a metabolic pathway of enhanced thrombopoiesis in humans

Abstract

Human blood cell counts are tightly maintained within narrow physiologic ranges, largely controlled by cytokine-integrated signaling and transcriptional circuits that regulate multilineage hematopoietic specification. Known genetic loci influencing blood cell production account for <10% of platelet and red blood cell variability, and thrombopoietin/cellular myeloproliferative leukemia virus liganding is dispensable for definitive thrombopoiesis, establishing that fundamentally important modifier loci remain unelucidated. In this study, platelet transcriptome sequencing and extended thrombocytosis cohort analyses identified a single loss-of-function mutation (BLVRB(S111L)) causally associated with clonal and nonclonal disorders of enhanced platelet production. BLVRB(S111L) encompassed within the substrate/cofactor [α/β dinucleotide NAD(P)H] binding fold is a functionally defective redox coupler using flavin and biliverdin (BV) IXβ tetrapyrrole(s) and results in exaggerated reactive oxygen species accumulation as a putative metabolic signal leading to differential hematopoietic lineage commitment and enhanced thrombopoiesis. These data define the first physiologically relevant function of BLVRB and implicate its activity and/or heme-regulated BV tetrapyrrole(s) in a unique redox-regulated bioenergetic pathway governing terminal megakaryocytopoiesis; these observations also define a mechanistically restricted drug target retaining potential for enhancing human platelet counts.

Figure 1

Identification of candidate gene/SNVs in the thrombocytosis cohort. (A) Workflow schema detailing phenotypic cohorts that were genetically studied (sample numbers in parentheses; BLVRB^S111L [462^C→T] mutant number in brackets) and output SNVs at critical validation and genetic association steps. NL, healthy controls. (B) Schema delineating computational pipeline with algorithmic filtering analyses at critical steps of genetic bioinformatics processing. (C) Heat map was generated using expression profiles for the candidate gene subset (N = 29) encompassing oligonucleotide probes on the Affymetrix HG_U133AAofAv2 array (scale bar on right). (D) Hematopoietic lineage schema is displayed as a Wilcoxon signed-rank test of the t statistic (−log₁₀ P) calculating the likelihood that the 29 member gene subset is more greatly expressed relative to all other genes expressed in ≥1 time point by lineage (scale bar on right).

Figure 2

Gene/phenotypic characterization within thrombocytosis cohorts. (A) Distribution of the 5 nsSNVs among the ET cohort by platelet JAK2^V617F allelic burden. (B) ORs with confidence intervals (CIs) were calculated by thrombocytosis phenotype (ET, RT) using genotyped validation controls (N = 208); only BLVRB^S111L remains a strong risk allele irrespective of thrombocytosis etiology (RT OR = 10.2; CI, 1.96-53.6; P = .005). (C-D) Hemoglobin and platelet counts plotted by JAK2^V617F allelic burden (substratified by zygosity for BLVRB 462^C→T [S¹¹¹L] mutation) were fit by linear regressions.

Figure 3

Heme degradation pathway characterization during E/Meg development. (A) E/Meg-restricted expression of the heme degradation pathway genes using in silico data. A log₂ signal intensity <6 is considered background expression. Cell lineage abbreviations are provided in supplemental Table 2. (B-C) Expression patterns of heme degradation pathway genes were quantified by qPCR using human CD34⁺ cells differentiated along (B) erythroid (Epo) or (C) megakaryocytic (Tpo) lineages; Tpo-directed cultures include Platelet Factor 4 (PF4) and c-MPL transcripts as lineage markers. All results are expressed as actin-normalized means ± standard error of the mean (SEM) from triplicate experiments. (D-E) RNA levels (normalized reads per kilobase of transcript per million reads) of heme degradation pathway genes from (D) ET (N = 7) and normal (N = 5) platelets or as (E) quantified by qPCR (ET, N = 5; normal, N = 5); *P < .05.

Figure 4

Biliverdin IXβ reductase S¹¹¹L characterization. (A) BLVRB expression in gel-filtered platelets (10 μg/lane) from representative cohorts that are wild type (^−/−) or heterozygous (^+/−) for mutant BLVRB^462C→T. (B) Schematic globular structure displays BLVRB^S111L mutation (green) within the single BV/NAD(P)H binding fold, with higher-resolution ribbon structures (insets) modeled to predict mutant Leu¹¹¹ or native Ser¹¹¹ on BV or NADPH interactions; note that Ser¹¹¹ is uniquely positioned for recognition and/or proton transfer within the Rossmann binding fold, with predicted proximity interference by the hydrophobic Leu¹¹¹ aliphatic isobutyl side chain. Models were generated using PYMOL software, based on NADP/mesobiliverdin IVα ternary (Protein Data Bank ID code 1HE3) and NADP/flavin mononucleotide ternary (Protein Data Bank ID code 1HE4) complexes. (C) Immunoblot of purified recombinant enzymes after thrombin cleavage and glutathione affinity depletion of the GST (glutathione-S-transferase) carrier (50 ng/lane). (D-E) Recombinant BLVRB^WT and BLVRB^S111L were used for flavin reductase (100 μM flavin mononucleotide) or BVR (20 μM BV dimethyl esters) specific activity determinations (N = 6, expressed as mean ± SEM); **** P < .00001. (F) Solubilized lysates from Lv-infected HEK293 cells were used for BVR activity in the presence of 20 μM BV dimethyl esters (N = 3); *P < .05; immunoblot of Lv-infected HEK293 cells (20 μg/lane) is shown. (G) BVR activity assays using recombinant BLVRB^WT or commercially available BLVRA.

Figure 5

Differential patterns of ROS handling in genetically modified CD34⁺ iPS cells. (A) Genetically modified CD34⁺ NCRM1 iPSCs (1 × 104 cells per well) were loaded with 0.1 (v/v) resazurin for time-dependent spectrofluorimetric detection (530-nm excitation, 590-nm emission) of reduced resorufin as a quantitation of cellular redox coupling; data expressed as mean ± SEM of relative fluorescent units (N = 6); cellular BLVRB expression is shown, 20 μg lysates/lane; ****P < .0001. (B) Genetically modified NCRM1 iPSCs (1 × 10⁵/mL) were treated (or not) with 200 μM TBHP for 1 hour at 37°C, followed by flow cytometric quantification of ROS-expressing cells after a 60-minute loading with cell-permeable 500 nM CellROX green as indicator (N = 6); *P < .05; **P < .01; ***P < .001; ****P < .0001.

Figure 6

BLVRB^S111L hematopoietic effects on ROS accumulation and lineage fate. (A-B) CD34⁺ HSCs transduced with Lv/BLVRB^WT, Lv/BLVRB^S111L, or Lv/Control were puromycin-selected and expanded for 48 hours prior to terminal differentiation (day 0). Genetically modified cells were assessed for viability using (A) trypan blue exclusion (N = 4) or (B) cell quantification and flow cytometry. Cell quantification (N = 6) is expressed as relative fold-increase to control, and flow cytometric quantification was determined using gated, live (7-AAD negative) CD34⁺ cells (N = 4); flow cytometry gate (forward scatter [FSC], size vs side scatter [SSC], side scatter as a measure of complexity) used for quantification of 7-AAD negative CD34⁺ cells and for ROS tracking is shown. (C-D) Genetically modified CD34⁺ HSCs were puromycin selected and plated for determination of lineage fate using (C) multipotential progenitor assays (N = 4 experiments) or (D) CFU-MK determinations (N = 3 experiments). (E) Representative flow cytometric analysis of developing MKs dually labeled for ROS accumulation and CD41 at discrete time points (LV/control-infected CD34⁺ HSCs are shown). (F) Cumulative distribution plots across ROS^high subsets generated by flow cytometric quantification of MK phenotypic parameters (ROS and CD41 [Tpo cultures], or CD41 [CD41⁺/GlyA⁻ [bilineage Tpo/Epo cultures]) are displayed by culture conditions and time points for genetically-modified CD34⁺/BLVRB^WT, CD34⁺/BLVRB^S111L, and CD34⁺/Control; results are from a single representative experiment repeated on 3 occasions; P values for pairwise comparisons were calculated over 200 bootstrapped samples. (G) Percentage of CD41⁻/GlyA⁺ erythroid cells in day 10 bilineage cultures (N = 3). (H) UV-visible spectroscopy of day 10 bilineage cultures (normalized to the peak Soret absorbance of oxyhemoglobin [λ₄₁₄]) demonstrates no differences in methemoglobin (λ₆₃₀) accumulation across the genotypes. For all panels except F, results are expressed as mean ± SEM; *P < .05, **P < .01, ***P < .001, N.S., not significant.