peripheral blood over time. VAFs of frameshift
indels inasxl1increased, consistent with their
positive selection, whereas other gene variants
exhibited mixed dynamics or no significant
enrichment for frameshift indels (fig. S4B).
Three observations suggested that positive
selection was gene- and mutation type–specific.
First, nondisrupting in-frame indels of these
genes did not change (fig. S4C). Second, VAFs
of control genes in control zebrafish remained
stable (fig. S4D). Third, we observed no change
in VAFs of genes in fin tissue of mutant
zebrafish (fig. S4, E and F). Expanding variants
were invariably found in marrow dominant
clones sorted by color (fig. S4G). We observed
no correlation between clone size and number
of frameshift mutations for the dominant clones
(fig. S4H). These results show that frameshift
mutations in some genes, such asasxl1, exhibit
positive selection over the study time period,
resulting in clonal expansion.
To assess the potential of these mutations to
promote clonal dominance, we analyzed the cor-
relation between VAFs and the Gini coefficient
of clonal skewing. We found that frameshift
indels in the common clonal hematopoiesis
genesasxl1,dnmt8(DNMT3Aortholog), and
tp53positively correlated with the Gini coef-
ficient, indicating that disruption of these
genes was associated with clonal dominance
(Fig. 2A). There was no correlation between in-
frame indels and the Gini coefficient (fig. S4I).
These data show that frameshift mutations in
select clonal hematopoiesis–associated genes
lead to clonal dominance.
TWISTR enables the prospective isolation of
live cells from dominant and nondominant
clones from the same animal, allowing their
direct comparison to identify drivers of clonal
fitness. We sorted myelomonocytes from domi-
nant clusters and compared their mutational
landscape to that of myelomonocytes in smallerclusters from the same mutant zebrafish (Fig.
2B). The dominant clusters harbored frameshift
indels inasxl1significantly more commonly
than did nondominant clusters (Fig. 2C). Single
targeting ofasxl1resulted in clonal dominance
associated with frameshift indels, which sug-
gested that mutantasxl1alone can promote
clonal dominant states (fig. S5). These results
show that frameshift mutations inasxl1confer
clonal fitness.
To identify pathways underlying clonal fit-
ness with acquired mutations, we performed
single-cell RNA sequencing (scRNA-seq) on (i)
unsorted marrow cells from control zebrafish,
(ii) unsorted marrow cells from mutant zebra-
fish with unique sets of mutations, and (iii)
dominant clusters that contributed to >50% of
marrow myelomonocytes in the mutant zebra-
fish (Fig. 3A). All cell types were present in
mutant zebrafish, except for mildly decreased
erythroid precursors ( 12 ) (Fig. 3B, fig. S6, and770 5NOVEMBER2021•VOL 374 ISSUE 6568 science.orgSCIENCE
A Unsorted,
Control
Bcell
MacrophMyelProgMPPT/NKcellHSPCErythrocyteNeutrophil
Vasc
EndoNeut1Ery
ProgBcellPrecLymphProgBUnsorted,
controlUnsorted,
MutantUnsorted,
MutantSorted,
Dominant
colorSorted,
Dominant colorUMAP2
UMAP1Csocs3anr4a1ExpressionHSPC MPP Neutrophil Macrophage Erythroid ProgenitorControl, unsorted (n=2) Mutant, unsorted (n=3) Dominant, sorted (n=3)il1bExpressionExpressionCytokinesInflammatory modulators01230120246012340102402460246024012024024024024012Expressiontnfb01230120246024601ErythPrecFig. 3. Single-cell RNA sequencing reveals an inflammatory state in hemato-
poiesis and differential response of mutant progenitors.(A) Schema of scRNA-
seq experiment. (B) Split uniform manifold approximation and projection (UMAP)
of marrow cells in control zebrafish (n= 2, unsorted), mutant zebrafish (n= 3,
unsorted), and dominant color from these mutant zebrafish (n=3,sorted).(C) Violin
plots showing select gene expression (log-transformed) in the eight samples. Shaded
blue and orange boxes indicate significant difference relative to controls in mutant
and dominant groups, respectively;Padj<5×10Ð^4. MPP, multipotent progenitors.RESEARCH | REPORTS