BMC Developmental Biology BioMed Central

Background: In addition to erythrocytes, embryonic blood contains other differentiated cell lineages and potential progenitor or stem cells homed to changing niches as the embryo develops. Using chicken as a model system, we have isolated an enriched pool of circulating non red blood cells (nRBCs) from E4 and E6 embryos; a transition period when definitive hematopoietic lineages are being specified in the peri-aortic region.


Background
The isolation and gene expression profiling of embryonic circulating nRBCs would be of great interest to developmental biologists and clinicians alike [1], yet due to limited sample size available from traditionally used model organisms, harvesting a sufficiently large pool of embryonic nRBCs for transcriptome-wide analysis has been difficult. Alternative approaches using perinatal blood have already yielded significant insights [2]. The chick embryo is both large in size and contains a circulatory network of a complexity equal to that of mammals. Herein, we describe the isolation and gene expression profiling of circulating cells during the transition phase of hematopoiesis from primitive or yolk sac associated, to definitive hematopoiesis, at embryonic days 4 and 6. It is during this time that hematopoiesis occurs transiently in the peri-aortic region in the chick embryo (referred to in mammals as the aorta-gonad-mesonephros or AGM), before transitioning to the bone marrow [3,4].

Results and Discussion
Chick blood was isolated from embryos at E4 and E6, using micro-capillaries inserted directly into the heart. Density gradient centrifugation was then employed to isolate the heavier RBCs from a lighter nRBC population from total embryonic blood. Cells within the two populations were analyzed directly by FACS, and by the classical hematological stains Giemsa, benzidine, and Periodicacid Schiff (PAS). Using these techniques, we were able to confirm that two distinct, viable populations; one highly enriched in RBCs, and another population highly depleted of RBCs (nRBCs) had been isolated (Fig. 1).
Further characterization of these populations by RT-PCR demonstrated that nRBCs had high expression levels of the hematopoietic stem cell antigen CD34, whereas the RBC population lacked expression of this gene (Fig.  2B,C). After these preliminary findings had given validity to our technique, gene expression profiling was performed using Affymetrix Gene Expression Arrays. For RNA isolation, handling time was kept to a minimum and cell collection to lysis for RNA extraction was performed in less than one hour. Consequently, cells were not subjected to long incubation periods on ice, or in serum containing medium, which can alter gene expression, as is the case for other commonly used in techniques such as FACS sorting. RNA from both E4 and E6 RBC and nRBC samples were analyzed by duplicate Affymetrix gene chips, from sepa- Array analysis and RT-PCR verification. (A) Heat map generated by TM4 SAM analysis with genes verified by PCR highlighted. (B) Semi-quantitative PCR analysis of candidate genes from array data and control GAPDH and 18S, and embryonic hemoglobin transcripts (pan-globin) in cDNAs from both E4 and E6 RBC and nRBC fractions. (C) QPCR data. All genes tested had significantly higher expression in nRBCs by Student's T-test (except for CD51). Hemoglobins are the only genes with significantly higher expression in RBCs. E4 data: left-hand bars, E6 data: right-hand bars. rate, pooled biological samples (30-100 embryos/array). Comparisons between RBC and nRBC populations were made, and the expression levels of candidate genes were confirmed by PCR ( Fig. 2A-C). The resulting array data has been deposited into NCBI Gene Expression Omnibus (GEO) under the accession number GSE9884. We considered genes to be significantly enriched in the nRBC population by the following two criteria: 1) that they are expressed at higher levels in the nRBC than the RBC pop-ulation by the SAM algorithm; 2) that they are not expressed at high levels in the heart (Fig. 3).

Hematopoietic Stem Cell (HSC) Associated Genes are Upregulated in nRBCs
Many genes known to be associated with HSCs were found to be preferentially expressed in nRBCs, such as the HSC membrane receptor glycoprotein (GP) CD45 (Tables 1 and 2) [5]. Moreover, transcription factors Ets-1 [6], Representative array expression profiles generated by eXintegrator analysis Figure 3 Representative array expression profiles generated by eXintegrator analysis. Example candidate genes kept from TM4 analysis (left) display an observable gradient of low expression in RBC samples and high expression in nRBC samples, and low-to-moderate expression in heart. Example candidate genes discarded from TM4 analysis (right) following examination by eXintegrator display high expression in negative control sample (heart). HEX [7], KLF2 [8] and PU.1 [9], known to be essential for primitive and definitive hematopoiesis, were detected specifically in the nRBC population; as were the signaling molecules BLNK [6], FYN [10], RGS18 [11], Rac2 [12], LYN and SYK [13], VAV3 [14] and the ion channel Slo1 [15]. Additionally, the expression of many integrins, which are known to play an important role in the adhesion and homing of HSCs, were detected in nRBCs. A significant overlapping integrin repertoire was observed between nRBCs and a previous study on adipose derived stromal CD31+ HSCs and includes: CD18 (ITGB2), CD49B (ITGA2), CD49F (ITGA6), CD51 (ITGAV), CD61 (ITGB3/GPIIIa), and the non-integrin cell adhesion molecule CD166 (ALCAM) [16]. In addition to integrins, previous work has established an important role for GPs (e.g. CD34, CD45, CD61) in the adhesion and possibly homing of HSCs, and has demonstrated the expression of the GP receptor CD62L, and the GP Cystatin-7 on HSCs [17,18]. Finally, cell-cell communication required during later differentiation of HSCs in the stroma is known to be mediated by the gap junctional protein connexin 43 (Cx43), another gene detected in the nRBC population [19].

Myeloid Markers Expressed by nRBCs
In addition to the expression of GPs on HSCs, expression of the platelet GP ligand CD62P (P-selectin), important for HSC adherence [20], and the myeloid GPs CD200R [21] and CD36 [22] were detected in the nRBC population. Other markers of the undifferentiated myeloid lineage including gelsolin [23] and PU.1 were both detected in nRBC fraction. Furthermore, many genes detected in nRBCs can be associated with platelet activation pathways such as, Coagulation Factors X and XIII, COX-1, PAI, PDGF, PLCG2, Tissue Factor Pathway Inhibitor, Thrombin Receptor, and VAV3 [24][25][26][27].

HSC (22) Myeloid (17) Lymphoid (10) Germ Line (2)
Coagulation Factor X TLR7 CD62L Coagulation Factor XIII TNFSF13b Cardiac ( trophil chemotaxis [31], demonstrate that a lymphoid differentiation potential is already present at the peri-aortic stage, a finding which has been reported previously for similarly staged AGM derived cells in mice [32]. The only gene expressed in nRBCs which has a role in acquired immunity appears to be the Src-like adaptor molecule, SLAP, whose function is to repress surface IgM expression on B-cells [33]. Interestingly, the heme oxygenase-1 (HMOX1) an essential enzyme in heme catabolism, was the only true erythroid associated gene detected in the nRBCs. However, the expression of HMOX1 has been reported in a variety of primitive and definitive white blood cell types as well, although it does not appear to be essential for their development [34].

Additional Genes Expressed by nRBCs
Finally, a list of genes with unknown functions has emerged from this screen. Although the Interleukin-1 Receptor Associated Protein (IL1RAP) and TNF-alpha Interacting Protein 8 (TNFAIP8) are likely to be involved in the innate immune response, the other 6 genes remain developmentally uncharacterized, and may provide insight into the function and differentiation of HSCs (Table 2). Interestingly, the Peptide Methionine Sulfoxide Reductase gene (MSRA), which was expressed in nRBCs, may help to protect progenitor cells against oxidative stress [41], although conclusive proof to this end remains to be demonstrated.

Further Expression Validation: ISH, IHC and FACS
Following our expression profiling at the in vitro level, we next verified our expression data for certain interesting candidate nRBC expressed genes, for which QPCR had been performed, by wholemount is situ hybridization (ISH) and immunostaining (IHC) of the yolk sac. At E4 In situ hybridization and histological analyses Figure 4 In situ hybridization and histological analyses. From the 12 candidate genes used for QPCR 6 probes were generated that gave some positive signal at earlier embryonic stages; 4 of these stained small numbers of cells in the yolk sac, with the remaining 2 no positive cells were detectable. Embryonic ρ-globin is expressed in the vast majority of circulating cells but is observed to be negative in some rare cells (arrows), whereas CD200R, CD61, HEX, RGS18 and the vitronectin receptor (CD51/61) are expressed in scattered circulating (arrows) and attached cells (arrowheads) throughout the yolk sac vasculature. Red contours outline the vascular lumen. Left and right panels indicate distinct staining of single cells with the indicated probe or antibody. Bars = 20 µm. and E6, the time points analyzed, the yolk sac is highly vascularized and provides a receptacle for circulating cells transiting to and from the embryonic and extraembryonic regions. Reproducible labeling of rare positive circulating cells was observed for CD61, CD200R, and HEX, whereas infrequent clusters of positive cells were found to express RGS18 by ISH (Fig. 4). Furthermore, HEX expression was observed in large numbers of mesenchymal cells neighboring blood vessels either containing, or devoid of HEX positive circulating cells. CD200R occasionally labeled cells with an endothelial morphology. IHC for the CD51/ CD61 heterodimer or vitronectin receptor (alpha3betaV integrin) revealed expression in rare circulating small rounded cells which were either clumped and associated with the endothelium, or singular. FACS analysis using this antibody demonstrated that 4% of nRBCs are positive for this antigen, whereas RBC staining was negligible (Fig.  5). ρ-globin, which is still prominently expressed in both E4 and E6 RBCs [42], was used as a positive control for yolk sac ISH (Fig. 4). ρ expression was observed in the majority of circulating cells, but was negative in certain infrequent cells presenting non-RBC morphology.

Conclusion
In summary, gene expression profiling of nRBCs in the chick embryo has revealed the expression of a set of genes indicative of a wide range of hematopoietic stem cells and progenitors primarily of either the erythroid or myeloid lineages present in early circulation. It has indeed been postulated that cells with an "erythromyeloid" potential constitute the first subset of HSCs with potential for liver engraftment and eventual long-term hematopoiesis in the bone marrow [43]. Lastly, the identification of several previously undescribed genes may prompt closer examination of their functions in chick and other model organisms. We, however, do not observe a prominent difference in expression profiles between E4 nRBCs and E6 nRBCs, during which period the second wave of HSC generation is actively taking place in the peri-aortic region, transiting from the initial appearance of intra-aortic clusters at about E4 to the formation of para-aortic foci at E6 [4]. It is therefore unclear whether the nRBCs we detect in E4-6 circulation, with the profiles of hematopoietic cells and progenitors, represent those from yolk sac or peri-aortic cells.

Blood Isolation
Blood was collected from the embryonic ventricles using fine glass microcapillaries. Cells were washed in PBS-EDTA, centrifuged at 1500 g, and separated on a Redi-Grad:NaCl (9:2.5) density gradient (Amersham) by centrifugation for 20 minutes at 10,000 g. Upper nRBC and lower RBC populations were collected by pipette and placed in RNA lysis buffer or assayed using chemical stains or FACS.

RNA Isolation and RT-PCR
Cells were lysed using QIAshredder spin columns and total RNA was extracted using the RNeasy total RNA extraction kit (Qiagen). Equal amounts of total RNA from each sample were converted to first-strand cDNA in parallel with the Superscript III reverse transcriptase synthesis system (Invitrogen). Real-time QPCR was performed using Quantitect SYBR PCR master mix (Qiagen) in a 7900 HT Fast Real-Time PCR System (Applied Biosystems). All PCRs were performed in duplicate with at least 2 biological samples at an annealing temperature of 60°C, using between 30 and 45 amplification cycles. Analysis of the melting curve excluded the amplification of unspecific products. In each QPCR run, a standard curve was generated using duplicate 6-log spanning serial dilutions. PCR products for standard curves were columnpurified, measured for DNA concentration, sized by agarose gel electrophoresis, sequenced, aliquoted and stored at -80°C for a maximum of 2 months. Standard curves were calculated by SDS software, and test samples were fitted to the generated curve (Applied Biosystems). Primer sequences are available in Table 3. Statistical analyses of the complexity of RBC and nRBC transcriptomes Figure 6 Statistical analyses of the complexity of RBC and nRBC transcriptomes. (A) Summary of present (P) calls for all arrays used in this study as a percentage of all calls (either present, P; absent, A; or missing, M). nRBCs have 10% more P calls than RBCs. Nevertheless, 1/4 of the RBC transcriptome is scored as present. (B) Graphical representation of SAM analysis showing significantly different genes between nRBCs and RBCs in red, which were examined in this study. (C) Grouped T-test between RBCs and nRBCs shows that for a confidence level of 95%, 98% of the hybridization values for probe sets between these two groups are not significantly different.