Senp1 drives hypoxia-induced polycythemia via GATA1 and Bcl-xL in subjects with Monge's disease

Abstract

In this study, because excessive polycythemia is a predominant trait in some high-altitude dwellers (chronic mountain sickness [CMS] or Monge's disease) but not others living at the same altitude in the Andes, we took advantage of this human experiment of nature and used a combination of induced pluripotent stem cell technology, genomics, and molecular biology in this unique population to understand the molecular basis for hypoxia-induced excessive polycythemia. As compared with sea-level controls and non-CMS subjects who responded to hypoxia by increasing their RBCs modestly or not at all, respectively, CMS cells increased theirs remarkably (up to 60-fold). Although there was a switch from fetal to adult HgbA0 in all populations and a concomitant shift in oxygen binding, we found that CMS cells matured faster and had a higher efficiency and proliferative potential than non-CMS cells. We also established that SENP1 plays a critical role in the differential erythropoietic response of CMS and non-CMS subjects: we can convert the CMS phenotype into that of non-CMS and vice versa by altering SENP1 levels. We also demonstrated that GATA1 is an essential downstream target of SENP1 and that the differential expression and response of GATA1 and Bcl-xL are a key mechanism underlying CMS pathology.

Figure 1.

Production of erythroid cells from human iPSC. (A) Schematic representation of the successive culture steps for production of cultured RBCs (cRBCs) from fibroblasts. Fibroblasts were transformed to iPSCs using the Yamanaka factors. Clumps of undifferentiated iPSCs were cultured in erythroid body (EB) medium for 28 d in the presence of a mixture of cytokines (cytokine profile 1). After 1 wk of culturing in normoxia, erythroid bodies were allowed to differentiate for 3 wk in normoxia/hypoxia, and the populations were analyzed by flow cytometry at day 28 or were allowed to mature further under cytokine profile 2 by sequential sets of cocktails and analyzed at day 58. Bars: (fibroblasts) 200 µm; (iPSC) 200 µm; (EB) 200 µm; (red blood cell, right) 2 µm. (B) CD analysis of various CD markers (CD34, CD45, CD36, CD71, and CD235a) under normoxia for sea-level samples. Each experiment was done in three replicates, and the experiment was done at least three times. The cells are cultured as erythroid bodies in cytokine profile 1 and as single cells in cytokine profile 2 (sequential mixture of cytokines). Error bars represent the mean ± SEM of at least two to three measurements. The experiment was repeated at least three times.

Figure 2.

Hypoxic response of sea-level, non-CMS, and CMS cells: Marked response of CMS samples. (A) Flow cytometric analysis using CD235a (glycophorin A) as a marker after culturing as detailed in Fig. 1 at the EB stage at day 28. Sendai virus results: Representative FACS plots of sea level, non-CMS, and CMS in normoxia (left) and in hypoxia (right). The dot plot represents the live cells as gated through propidium iodide. CD235a⁺ cells are shown in red along the y axis, and CD235a⁻ cells are shown in blue. The percentage in each figure represents the relative proportion of CD235a cells. There are major differences between CMS (bottom) versus the non-CMS (middle) and sea-level (top) samples. The FACS plots are representative of one experiment. Similar results were obtained in all the experimental repeats. (B) Summary of hypoxic response of CMS patients (n = 5 subjects) and non-CMS (n = 4 subjects) and sea-level (n = 3 subjects) control subjects. The graph depicts the relative proportion of CD235a quantified 3 wk after the administration of hypoxia (5% O₂). There is a significantly striking difference between sea level, non-CMS, and CMS under hypoxia. ***, P < 0.001. Error bars represent the mean ± SEM of at least two to three measurements. The experiment was repeated at least three times. (C) Summary of interclonal variability among the subjects: three clones (clones 1, 2, and 3) were tested for three subjects (subjects 1, 2, and 3) for each group: CMS, non-CMS, and sea level. The y axis depicts the relative proportion of CD235a under hypoxia for different clones. Error bars represent the mean ± SEM of at least two to three measurements. (D) Dose response. The graph represents the response (as measured by proportion of CD235a [y-axis]) of CMS, non-CMS, and sea-level cells to 21%, 10%, 5%, and 1.5% O₂ levels (x axis). Each point depicts the mean ± SEM of at least two to three measurements. CMS shows hyperresponsiveness at 10, 5, and 1.5% O₂. The experiment was repeated at least three times.

Figure 3.

Characterization of the erythroid cells under normoxia and hypoxia. (A) CD analysis of various markers (CD45, CD71, and CD235a) for all populations under normoxia (21% O₂). The cells are cultured as described in Fig. 1. Note that the CD profiles are similar for all groups under normoxia. (B) CD analysis of various markers (CD45, CD71, and CD235a) for all populations under hypoxia (5% O₂). The cells were cultured as described in Fig. 1. During week 3, we see significant differences in the proportion of CD235a between CMS and the controls (sea level and non-CMS). (A and B) Each experiment was done in three replicates, and experiments were repeated at least three times. (C) BFU-e assay under hypoxia (5% O₂). The y axis represents the number of BFU-e colonies. The experiment was done in three replicates and repeated twice. (D) Hemoglobin (Hb) analyses and function of erythroid cells with high-performance liquid chromatography profiles of erythroid cells at weeks 6 and 8. Note a shift in hemoglobin from fetal to adult A0. The O2-binding curve also shows a transition of P50 from 14 mmHg (week 6) to 22 mmHg (week 8). Shown is one representative image of data points at weeks 6 and 8. The experiment was repeated at least five times. A.U., arbitrary units; Sat, saturation. (E) Western blot of RBCs (erythroblasts). Note the presence of BAND-3 and GLUT1 in all lanes (normalized to actin). Retic, reticulocytes used as controls. (F) Globin expression by quantitative PCR for sea level and globin switching in time in culture. Similar trends were observed during maturation for all groups. n = 3 and error bars represent SD. Globin percentage is calculated as described by Qiu et al. (2008). The experiment was repeated at least three times. *, P < 0.05.

Figure 4.

Role of SENP1 in CMS polycythemia in the Andean population. (A) The SENP1 region in Andean highlanders. Four known transcripts of human SENP1 with accession nos. (NCBI RefSeq) are shown. Note that SENP1 is transcribed from the negative strand (i.e., right to left). Overlaid above in blue are the genomic positions of 66 SNPs deemed differential by Zhou et al. (2013). Three of our differential SNPs (marked with black arrows) overlap with different regulatory regions such as promoters and enhancers, as described in the Critical role of SENP1 in polycythemia section of Results. These SNPs show a strong signal of frequency differentiation between the non-CMS and CMS highlanders, indicative of strong positive selection in the region. (B) Western blot analysis of SENP1 protein expression under hypoxia (5% O₂) and normoxia (21% O₂) for CMS and non-CMS groups. The representative blot is shown is from one experiment. The relative levels were computed for n = 5 for CMS and n = 4 for non-CMS under hypoxia and normoxia. Data represent at least two to three measurements. The experiment was repeated at least twice for each subject. (C) iPSCs were infected with lentivirus and selected by puromycin. shSENP1#1, shSENP1#2, and shSENP1#3 represent the three clones selected. Each clone showed significant down-regulation of Senp1 expression by quantitative PCR, compared with uninfected iPSCs, as well as an iPSC line infected by control vector. Data represent two measurements in duplicate. (D) Western blot analysis of the loss of SENP1 expression in the shRNA clones. Lanes 1 and 2 show significant reduction in SENP1 levels in shRNA clones 1 and 2 as compared with uninfected and scrambled controls. Each bar represents the mean ± SEM of at least two to three measurements. The experiment was repeated at least two times for each subject. (E) Loss of vigorous erythropoietic response of CMS cells to hypoxia. The first bar represents the response of CMS cells to hypoxia. CMS-Senp1-shRNA represents the CMS cells that were infected by shRNA of Senp1 using lentivirus infection. Four different clones were tested for each line. Data represent three measurements in triplicates. The experiment was repeated three times. An unpaired Student’s t test was used. (F) Overexpression of SENP1 in non-CMS iPSCs. The blot shows a representative image for one experiment. We observed a twofold increase in expression in non-CMS in the cDNA overexpression cell line. Data represent three measurements in triplicates. The experiment was repeated twice. An unpaired Student’s t test was used. (G) SENP1 overexpression (OE) leads to marked increase in RBCs in non-CMS erythroid cell differentiation. The scrambled control overexpression did not change the phenotype of CMS. Each bar represents the mean for each clone measured in triplicates. The experiment was repeated three times. Error bars represent the mean ± SEM. *, P < 0.05; ***, P < 0.001.

Figure 5.

Significant difference in expression between CMS versus non-CMS and sea-level cells. (A) Relative VEGF, GATA1, and Bcl-xL expression by quantitative PCR (normalized with GADPH) in cultures grown in hypoxia in media. CMS cells produced significantly higher levels of VEGF, GATA1, and Bcl-xL. (B) Western blot analysis of GATA1 protein levels. The figure shows a representative blot for one experiment. The bars shows the relative levels for n = 4 for each group (CMS and non-CMS). The experiment was repeated at least three times. An unpaired Student’s t test was performed. (C) Relative mRNA levels of GATA1-inducible (Scl4a and Alas2) and repressive (cMyc and cKit) genes for CMS and non-CMS subjects. (D) Western blot analysis of Alas2 protein levels. The figure shows a representative blot for one experiment. (E) Western blot analysis of cMyc protein levels. The figure shows a representative blot for one experiment. (C–E) The bars show the relative levels for n = 4 for each group (CMS and non-CMS). The experiment was repeated at least three times. An unpaired Student’s t test was performed. (F) Non-CMS cells show significantly high amount of sumoylated GATA1 as compared with CMS and sea-level controls. P < 0.05; unpaired Student’s t test. The results are the summary for n = 4 for both CMS and non-CMS groups. Each bar represents the mean, and error bars represent the SE. For each sample, the levels were measured in triplicates. Each experiment was repeated three times. *, P < 0.05; **, P < 0.01.

Figure 6.

SENP1 desumoylase activity plays a critical role in GATA1 desumoylation and activation. (A) A representative blot showing that non-CMS cells show significantly higher levels of sumoylated GATA1. The sumoylated form of GATA1 was determined by IP with anti-SUMO followed by Western blotting with anti-GATA1. The ratios of SUMO-GATA1/GATA1 were quantified. IB, immunoblot. (B) Non-CMS cells show a significantly high SUMO-GATA1/GATA-1 ratio as compared with CMS. P < 0.05; Student’s t test. The results are the summary for n = 4 for both CMS and non-CMS groups. Each bar represents the mean, and error bars represent SE. For each sample, the levels were measured in triplicates. Each experiment was repeated three times. (C) Fused SUMO-GATA1 cannot restore the remarkable CMS phenotype in a background of GATA1 KO. Non-CMS had a similar result. Each bar represents a mean of three replicates, and the experiment was repeated at least twice. (D) Western blot analysis confirming significantly higher levels of SUMO-fused Gata1 in the overexpressed (OE) line. Lane 1 represents CMS cells lines overexpressing fused SUMO-GATA1 in the GATA1 KO background. Please note the significantly higher levels of SUMO-GATA1 in the GATA1 KO background. Lane 2 shows the relative levels of SUMO-GATA1 and GATA1 in non-CMS cells. Lane 3 shows the relative levels of SUMO-GATA1 and GATA1 in CMS cells. (E) GATAK137R (GATA1KR) mutant causes the polycythemic phenotype in both CMS and non-CMS subjects under hypoxia. Each bar represents a mean of three replicates, and the experiment has been repeated at least twice. (F) Western blot analysis showing overexpression of GATA1K137R in the non-CMS cells. Lane 1 represents non-CMS cell lines overexpressing GATA1K137R in the GATA1KO background. Lane 2 shows the relative levels GATA1 in uninfected CMS cells. Lane 3 shows the relative levels of GATA1 in uninfected non-CMS cells. (C–F) Each bar represents the mean of three replicates, and the experiment was repeated at least twice. Error bars represent SEM. **, P < 0.01.

Figure 7.

GATA1 target Bcl-xL plays a role in CMS polycythemia and erythroid progenitors. (A) Western blot analysis confirming the overexpression of Bcl-xL in the non-CMS levels. Lane 1 represents the non-CMS cell line infected with scrambled control. Lane 2 represents non-CMS cells overexpressing Bcl-xL construct. Lane 3 shows the expression levels on uninfected non-CMS cells. Each bar represents a mean of three replicates, and the experiment was repeated at least twice. Error bars represent SEM. **, P < 0.01. (B) Overexpression (OE) of Bcl-xL partially rescues the blunted response of non-CMS cells (gray bar). Error bars represent mean ± SEM. Measurements were done in triplicates. The experiment was repeated three times. The response of scrambled control (pink bar) as well as sea-level (green), CMS (red), and non-CMS (blue) uninfected cells are also shown. *, P < 0.05; Student’s t test. Scramb, scrambled. (C) Overexpression of Bcl-xL increases the number of BFU-e in non-CMS cells (gray). SENP1KO decreases the number of colonies in CMS cells (brown). *, P < 0.05; Student’s t test. Each bar represents the mean, and error bars represent SE. For each sample, the levels were measured in triplicates. Each experiment was repeated three times.