BRIEF COMMUNICATIONS Figure 2 Live imaging of hematopoietic cell pui: gfp/galaTa:dsred egfp/wild type behavior and homing in parabiotic zebrafish embryos and larvae (a-h)Fluorescence images of pul:gfp//gatala: dsred parabiotes with separate trunks and tails. GFP+ and DsRed+ myeloid cells from either embryo are seen in the yolk sac(a, inset, arrowheads )and tail (f, arrowheads)of the conjoined embryo; DsRed+ erythrocytes circulate in the yolk sac(a, b). 48 h-pf.d 48 h.p.f. heart(b, arrowhead)and tail of the pul: gfp(c, g)or atala: dsred(d, h) parabiotes. In contrast, nonmigratory GFP+ muscle cells(e) and DsRed* epidermal mucous cells(d, blue arrowheads) are present in only the respective bomb∥cd4gp transgenic embryo. This was observed in 24 of 24 embryos, pooled from three experiments (i-k)Fluorescence images of karl gfp/wild-type parabiotes. Fluorescent HSPCs from the kdr: p aorta are seen in the caudal hematopoietic tissue(i,j, arrowheads), thymus(k, arrowhead) and kidney(k, inset, dotted region) of the kdrtgfp//wild type(wild-type tail) 3 nontransgenic wild-type parabiote. This was 18 h.p.f. 72 h.p. g two experiments.(I)Fluorescence images of cd41: gfp//mib parabiotes. cd41:gfp GFP+ pro-thymocytes are seen in the thym rudiments of both the cd41: gfp(pink box and set)and the nontransgenic, noncirculating mind bomb mutant(blue box and inset). This was observed in 3 of 3 embryo pairs, in which the mind bomb mutant partner lacked circulation. Zebrafish parabiotes were imaged with a stereomicroscope(a-j, )or confocal microscope(k), under transmitted light(l) fluorescence(c-f l, insets)or both(a, b, g-k). Scale bars: 250 um(a, b, k, ) 100 um(c-j: k, inset; L, pink inset), 50 um(a, inset; L, blue inset) n, notochord; t, thymus; y, yolk sac:*, urogenital opening: WT, wild type. See also Supplementary Videos 2 and 3 backgrounds at the 128-cell blastula stage(Online Methods). of the embryos overnight( Supplementary Fig. 1). Therefore, in Using a Pasteur pipette, we collect both embryos and transfer them all subsequent experiments, blastulae were fused between the gether into a methylcellulose drop covered with high-calcium 256-to 512-cell stage and just before 30% epiboly Ringer's solution containing antibiotics. With a fine paintbrush, we We quantified the effect of blastula orientation on fusion s then gently move the methylcellulose around both embryos, bring- success and the resulting parabiosis patterns(Fig. le-g and (Fig. la and Online Methods; detailed protocol in Supplementary mostly gave rise to viable but malformed embryo pairs(10 of 17 Note). Using a pulled glass micropipette, we detach a few cells from Supplementary Fig. 2a, d, e). Blastulae fused by their blastoderm both blastulae at their contact point(Fig. Ib and Supplementary margins or by margin to intermediate region mostly separated Video 1). The methylcellulose is then moved again around the during the experiment (27 of 38 pairs). Among the 1l successful blastulae so as to press the wounds against each other and pro- fusions, 4 were malformed or died and 8 survived. Of these eight mote fusion(Fig. 1c). The next day, we replace the high-calcium to survive to parabiosis, five gave rise to parabiotes with separate nger's solution containing dissolved methylcellulose with heads(Fig. If and Supplementary Fig 2c-e). In contrast, fusions mbryo water containing antibiotics(Online Methods). Between between the animal pole of one blastula and the intermediate or 17%and 56% of the blastula pairs, depending on the develop- marginal region of the other led to a high rate of viable, well- mental stage(Supplementary Fig. 1), developed into viable and formed parabiotes by 2 d p f (22 of 51); of these, 19 were fused partially fused embryos that shared a common blood circulation by their heads( Fig. le and Supplementary Figs. 2b, d, e and 3) Cparabiotes)and that otherwise displayed normal overall mor- Blastulae fused by their intermediate regions also yielded a high phology and spontaneous movement(Fig. 1d-g) proportion of viable parabiotes by 2 d p f. (43 of 95), most of We found that the developmental stage at which blastula fusion which(31 of 43)had separate heads(Fig. 1g and Supplementary is performed is critical for successful parabiosis. Blastula fusion Fig 2d, e). Thus, by selecting the proper orientation upon blastula must be performed between the 256-cell and the 30% epiboly fusion, one can favor the generation of parabiotic pairs fused in stages to result in viable parabiotic embryos. Before the 256-cell patterns appropriate to the biological question to be addressed stage, fusion usually led to a single embryo of both genetic back- To assess the frequency of shared circulation of the fused grounds surrounding the two yolk sacs(data not shown). Blastula embryos, we fused transgenic gatala: dsred blastulae, in which fusions performed at the 256-to 512-cell stages led to 40% viable erythroid cells and early primitive myeloid cells are labeled, with parabiotic embryos by 1 d post-fertilization(d p f ),49% when nontransgenic partners. We analyzed 24 parabiotes displaying performed at 1,000-cell/high stages and 56% when done at sphere/ different fusion patterns(fused heads or fused tails)at 2 d p f, dome stage. The success rate then decreased abruptly to 17% by and all of them displayed shared blood circulation as evidenced by 2 I ADVANCE ONLINE PUBLICATION I NATURE METHODS© 2013 Nature America, Inc. All rights reserved.  |  ADVANCE ONLINE PUBLICATION  | nature methods brief communications First, we remove the chorions from embryos of two different backgrounds at the 128-cell blastula stage (Online Methods). Using a Pasteur pipette, we collect both embryos and transfer them together into a methylcellulose drop covered with high-calcium Ringer’s solution containing antibiotics. With a fine paintbrush, we then gently move the methylcellulose around both embryos, bring￾ing them closer to each other in the proper orientation for fusion (Fig. 1a and Online Methods; detailed protocol in Supplementary Note). Using a pulled glass micropipette, we detach a few cells from both blastulae at their contact point (Fig. 1b and Supplementary Video 1). The methylcellulose is then moved again around the blastulae so as to press the ‘wounds’ against each other and pro￾mote fusion (Fig. 1c). The next day, we replace the high-calcium Ringer’s solution containing dissolved methylcellulose with embryo water containing antibiotics (Online Methods). Between 17% and 56% of the blastula pairs, depending on the develop￾mental stage (Supplementary Fig. 1), developed into viable and partially fused embryos that shared a common blood circulation (‘parabiotes’) and that otherwise displayed normal overall mor￾phology and spontaneous movement (Fig. 1d–g). We found that the developmental stage at which blastula fusion is performed is critical for successful parabiosis. Blastula fusion must be performed between the 256-cell and the 30% epiboly stages to result in viable parabiotic embryos. Before the 256-cell stage, fusion usually led to a single embryo of both genetic back￾grounds surrounding the two yolk sacs (data not shown). Blastula fusions performed at the 256- to 512-cell stages led to 40% viable parabiotic embryos by 1 d post-fertilization (d.p.f.), 49% when performed at 1,000-cell/high stages and 56% when done at sphere/ dome stage. The success rate then decreased abruptly to 17% by the 30% epiboly stage, mainly because of separation and/or death of the embryos overnight (Supplementary Fig. 1). Therefore, in all subsequent experiments, blastulae were fused between the 256- to 512-cell stage and just before 30% epiboly. We quantified the effect of blastula orientation on fusion success and the resulting parabiosis patterns (Fig. 1e–g and Supplementary Fig. 2). Blastulae fused by their animal poles mostly gave rise to viable but malformed embryo pairs (10 of 17; Supplementary Fig. 2a,d,e). Blastulae fused by their blastoderm margins or by margin to intermediate region mostly separated during the experiment (27 of 38 pairs). Among the 11 successful fusions, 4 were malformed or died and 8 survived. Of these eight to survive to parabiosis, five gave rise to parabiotes with separate heads (Fig. 1f and Supplementary Fig. 2c–e). In contrast, fusions between the animal pole of one blastula and the intermediate or marginal region of the other led to a high rate of viable, well￾formed parabiotes by 2 d.p.f. (22 of 51); of these, 19 were fused by their heads (Fig. 1e and Supplementary Figs. 2b,d,e and 3). Blastulae fused by their intermediate regions also yielded a high proportion of viable parabiotes by 2 d.p.f. (43 of 95), most of which (31 of 43) had separate heads (Fig. 1g and Supplementary Fig. 2d,e). Thus, by selecting the proper orientation upon blastula fusion, one can favor the generation of parabiotic pairs fused in patterns appropriate to the biological question to be addressed. To assess the frequency of shared circulation of the fused embryos, we fused transgenic gata1a:dsred blastulae, in which erythroid cells and early primitive myeloid cells are labeled, with nontransgenic partners. We analyzed 24 parabiotes displaying different fusion patterns (fused heads or fused tails) at 2 d.p.f., and all of them displayed shared blood circulation as evidenced by pu1:gfp//gata1a:dsred kdrl:gfp//wild type (wild-type tail) a Eye 25 h.p.f. Eye y 48 h.p.f. 48 h.p.f. 48 h.p.f. 72 h.p.f. * * * * * * * * y d f h c e g 48 h.p.f. y gata1a:dsred gata1a:dsred pu1:gfp pu1:gfp pu1:gfp tail gata1a:dsred tail b c d e f g h i j kdrl:gfp//wild type kdrl:gfp Eye Eye y n y 4.75 d.p.f. Ear WT 72 h.p.f. k mind bomb//cd41:gfp mind bomb y 72 h.p.f. t t Eye y y t t Ear Ear Eye Eye cd41:gfp l Figure 2 | Live imaging of hematopoietic cell behavior and homing in parabiotic zebrafish embryos and larvae. (a–h) Fluorescence images of pu1:gfp//gata1a:dsred parabiotes with separate trunks and tails. GFP+ and DsRed+ myeloid cells from either embryo are seen in the yolk sac (a, inset, arrowheads) and tail (f, arrowheads) of the conjoined embryo; DsRed+ erythrocytes circulate in the yolk sac (a,b), heart (b, arrowhead) and tail of the pu1:gfp (c,g) or gata1a:dsred (d,h) parabiotes. In contrast, nonmigratory GFP+ muscle cells (e) and DsRed+ epidermal mucous cells (d, blue arrowheads) are present in only the respective transgenic embryo. This was observed in 24 of 24 embryos, pooled from three experiments. (i–k) Fluorescence images of kdrl:gfp//wild-type parabiotes. Fluorescent HSPCs from the kdrl: gfp aorta are seen in the caudal hematopoietic tissue (i,j, arrowheads), thymus (k, arrowhead) and kidney (k, inset, dotted region) of the nontransgenic wild-type parabiote. This was observed in 7 of 7 parabiote pairs pooled from two experiments. (l) Fluorescence images of cd41:gfp//mib parabiotes. cd41:gfp GFP+ pro-thymocytes are seen in the thymic rudiments of both the cd41:gfp (pink box and inset) and the nontransgenic, noncirculating mind bomb mutant (blue box and inset). This was observed in 3 of 3 embryo pairs, in which the mind bomb mutant partner lacked circulation. Zebrafish parabiotes were imaged with a stereomicroscope (a–j,l) or confocal microscope (k), under transmitted light (l), fluorescence (c–f; l, insets) or both (a,b,g–k). Scale bars: 250 µm (a,b,k,l), 100 µm (c–j; k, inset; l, pink inset), 50 µm (a, inset; l, blue inset). n, notochord; t, thymus; y, yolk sac; *, urogenital opening; WT, wild type. See also Supplementary Videos 2 and 3