Stemness screen and AML data sets implicate C3orf54/INKA1 as LSC regulator. (A-B) Both plots represent the same pool from the stem screen, as indicated by the legend on the right. (A) From individual vector cultures predicted (inferred from %BFP⁺) and actual pool composition (ddPCR, day 6 in vitro) is shown according to total %BFP⁺ next to pool composition analysis from total BM per individual mouse (m). The corresponding percentage CD45⁺ and BFP⁺ is indicated above each bar (mouse). (B) Pool composition in CD34⁺BFP⁺ cells sorted from BM of m59 is shown in comparison with in vitro input and total BM pool composition as percent vector copies. (C-D) Same as panels A-B for the second pool containing the C3orf54 expressing vector. (D) BM CD34⁺BFP⁺ sorted fraction of m19 was subjected to ddPCR analysis. (E) C3orf54/INKA1 expression in 89 LSC⁻ and 138 LSC⁺ fractions from 78 patients with AML (GSE30377, microArray, Mann-Whitney U test).³  (F) C3orf54/INKA1 expression (RNAseq), in paired diagnosis-relapse samples that are split into relapse origin committed-like (ROc) and primitive-like (ROp), as determined functionally or by clusters from perturbation deconvolution analysis, using gene expression data from normal hematopoietic cell subsets.⁹  Shown are the log2 ratios of expression levels at relapse vs diagnosis of each individual sample (Student t test). (G-H) CD34⁺ sorted cells from 2 primary AML samples (AML1, AML2) were transduced with shCTRL or INKA1-KD vectors (pooled shINKA1-1 and shINKA-2; supplemental Figure 14) and intrafemorally injected into NSG at a cell dose of 2.4 × 10⁵ and 4.4 × 10⁵/mouse, respectively. After 4 and 8 weeks, injected femurs were flushed and BM cells were analyzed for CD34 and BFP expression. The relative decline of percentage BFP⁺ cells within the CD34⁺ population (output) is plotted as the relative log₂ ratio to %BFP⁺ at input (4 days posttransduction; supplemental Figure 14) with the FC of the means for shINKA1 vs shCTRL indicated above each point.