(702b) Tissue Architectural Cues and Differential Extravasation Patterns Drive the Non-Random Trafficking of Tumor Cells in Larval Zebrafish

Clinical patterns of metastasis are
non-random, with certain types of cancers preferentially metastasizing to
certain secondary organs. However, the lack of models to visualize the early
stages of metastasis in multiple tissues has made it difficult to delineate how
organ selectivity emerges during the metastatic cascade. We used bone- and
brain-tropic subclones of breast cancer cell lines injected into the
circulation of embryonic zebrafish as a model xenograft system of metastatic
dissemination and early organ colonization (Fig. 1A-C). The zebrafish contains
vascular vessels on the scale of human capillaries, a brain with structural and
cellular similarity to the mammalian brain, and hematopoietic tissue in the
caudal vein plexus (CVP) analogous to mammalian bone marrow. Breast cancer cells that home to specific murine organs
(brain and bone) ultimately colonized analogous tissues (brain and
hematopoietic tissue in the CVP) in larval zebrafish (Fig. 1B-C). This pattern
was conserved across human MDA-MB-231 brain- and bone-targeting subclones
(231BR and 231BO, respectively; Fig. 1D), a brain-seeking human Bt474-m1 cell
line, and murine 4T1 bone- and brain-seeking variants. We exploited this model
to delineate rate-limiting steps in organ selectivity during early metastasis. Using a
combination of live-cell imaging, mechanical mapping of tissue properties, and quantification
of vascular architecture, we determined that initial organ selectivity was
non-random but largely cell autonomous, with both 231BO and 231BR cells
preferentially arresting in the complex vasculature of the CVP. Intravital cell
tracking confirmed this observation, with both cell types exhibiting significantly
longer residence times in the CVP than the brain during the first 12 h after
injection (Fig. 1E). However, differential arrest profiles between the
organ-targeting cell types was not observed during bloodborne dissemination.
Instead, differences in organ selectivity between 231BO and 231BR cells was an
emergent phenomenon developing after cell arrest. By 5 days post-injection
(dpi), a significantly higher fraction of 231BO cells had extravasated in the
CVP compared to 231BR cells (Fig. 1F). Importantly, intravascular cells were
cleared over time, leading to a higher proportion of 231BR cells in the brain and
the observed organ selectivity. Mass spectrometry and RNAseq of organ-targeting
MDA-MB-231 cell lines revealed that integrin signaling pathways were
upregulated in bone-seeking compared to brain-seeking subclones, with β1
integrin serving as a key integration point in these signaling pathways. Therefore, we
treated cells with a function-blocking antibody against integrin β1, which did not
change the initial arrest dynamics of 231BO cells in the vasculature of the CVP.
However, siRNA-mediated knockdown of integrin β1 reduced the
extravasation of 231BO cells in the CVP following arrest. Our results show
that organ selectivity is driven by both occlusion and extravasation at the
tumor-endothelial interface and provide important insights into the early
stages of metastasis.

Fig. 1. Non-random trafficking of
organ-targeting cancer cells in embryonic zebrafish. (A) Overview image
of 7 days post-fertilization (dpf) transgenic zebrafish. Blood vessels are
displayed in red, lymphatic vessels in green, and disseminated cancer cells in
blue. The brain and caudal vein plexus (CVP) are outlined. Detail of (B) 231BR
and (C) 231BO cells at 5 days post-injection (5 dpi) in the brain and CVP.
Boxes show positions of insets, and arrows indicate human cancer cells. (D)
Ratio of 231BR and 231BO cell numbers in the brain to CVP at 5 dpi following
circulatory injection (mean ± SEM). *, p<0.05 by Mann-Whitney test. (E)
Average residence times of 231BR and 231BO cells trafficking through the brain
and CVP. **, p<0.01; ****, p<0.0001 by two-way ANOVA with Tukey’s multiple
comparisons post-test. (F) Fraction of cells present that were extravasated in
the CVP at 5 dpi for 231BR and 231BO cells. *, p<0.05 by unpaired t-test.