Characterization of Subpopulations of Chicken Mononuclear Phagocytes That Express TIM4 and CSF1R

Phagocytosis of apoptotic or senescent cells by macrophages is a physiological process for maintenance of cell populations in tissues during embryonic development and adult homeostasis (1, 2). Apoptotic cells are recognized by phagocytes through multiple mechanisms. One mechanism depends upon the exposure of the normal inward-facing phosphatidylserine (PS) of the lipid bilayer to the outer layers of the plasma membrane (3). T cell Ig and mucin domain–containing 4 (TIM4), encoded by the TIMD4 locus, was defined as a plasma membrane PS receptor (4). Timd4 in mice is expressed primarily by subsets of macrophage lineage cells in a restricted set of tissues, notably Kupffer cells in the liver, which mediate clearance of senescent RBCs (5). Resident mouse peritoneal macrophages also express high levels of TIM4, which is essential for their recognition of apoptotic cells. However, in other locations where Timd4 is highly expressed, such as the marginal zone in spleen, TIM4 is not essential for apoptotic cell recognition (6). In mouse liver, Timd4 provides a marker for macrophages of embryonic origin that reside together with, but are distinct from, those recruited from blood monocytes (5). Deficiency of Timd4 in mice produces T and B cell hyperactivity and autoimmunity attributed to the failure to regulate Ag-reactive T cell differentiation (7). Unlike other TIM family members, TIM4 has no tyrosine kinase motif in its cytoplasmic tail (8). Accordingly, other PS receptors or coreceptors, in addition to TIM4, are required to initiate particle uptake and signal transduction. Recognition of PS by TIM4 may also contribute to macropinocytosis of viruses (9, 10), notably in association with TIM1, encoded by the adjacent Havcr1 locus.

We previously identified the chicken TIMD4 locus and produced mAbs against two distinct isoforms of the TIM4 protein (11). Recombinant chicken TIM4 bound to PS and, like its mammalian orthologue, is thereby implicated in recognition of apoptotic cells. A TIM4 fusion protein also had costimulatory activity on chicken T cells, suggesting a function in Ag presentation (11). In birds, as in mammals, macrophage differentiation depends upon signals from the CSF1R, which has two ligands, CSF1 and IL-34 (12). In contrast to the mammalian system, in chickens, TIM4 was highly expressed by macrophages grown in vitro in M-CSF (CSF1).

Anti-CSF1R Abs (13) and transgenic reporter genes based upon control elements of the CSF1R locus (14) provide convenient markers for cells of the macrophage lineage in birds. An emerging view in mammalian macrophage development is that many tissue macrophage populations are maintained by self-renewal of macrophages seeded from yolk sac–derived progenitors during embryonic development independently of blood monocytes (15, 16). This is less evident in chickens, in which intraembryonic transplantation of bone marrow precursors gave rise to donor-derived macrophages throughout the body (17). Nevertheless, the first evidence that macrophages are produced by the yolk sac derived from studies of chicken development, and these cells are involved extensively in the clearance of apoptotic cells (reviewed in Ref. 18). A recent study of the time course of chicken embryonic development based upon cap analysis of gene expression detected expression of both TIMD4 and HAVCR1 around day 2 of development, when the first CSF1R-dependent macrophages are also detected (19, 20). In the current study, we use anti-TIM4 Abs in combination with a CSF1R-mApple reporter to locate and characterize the gene expression profiles of cells that express this receptor. We show that TIM4⁺ Kupffer cells in chicken closely resemble mammalian Kupffer cells and are distinct from TIM4⁺ phagocytes in the bursa. Inter alia, our analysis led to the identification of an abundant population of TIM4^lo/−,CSF1R-mApple⁺ cells in the liver that resemble mammalian Ag-presenting dendritic cells (DCs) but are much more abundant than their counterparts in mouse liver. Their identification and location suggest that the avian liver has a function in the control of T cell–mediated immune responses.

The J line chicken was crossbred from nine lines, originally inbred from Brown Leghorn chickens at the Poultry Research Centre (Edinburgh, U.K.) to study a variety of traits, such as egg laying, plumage, and vigor (http://www.narf.ac.uk/chickens/lines). This strain of chicken expressed multiple TIM4 isoforms (11). CSF1R-mApple/EGFP– transgenic chickens, which carries the chicken CSF1R regulatory sequences directing expression of the red fluorescent protein mApple–enhanced GFP to the cytoplasm of macrophages (14), and commercial NOVOgen BROWN layers were also included in this study. All birds were hatched and housed in premises licensed under a U.K. Home Office Establishment License in full compliance with the Animals (Scientific Procedures) Act 1986 and the Code of Practice for Housing and Care of Animals Bred, Supplied or Used for Scientific Purposes. All procedures were conducted under Home Office project license PPL 70/7860, according to the requirements of the Animal (Scientific Procedures) Act 1986, with the approval of local ethical review committees. Animals were humanely culled in accordance with Schedule 1 of the Animals (Scientific Procedures) Act 1986.

mAb JH9 against chicken TIM4–extracellular domain (aa 1–209) was raised and characterized as described previously (11). JH9 was labeled with Alexa Fluor 647 for flow cytometric analysis, or with Alexa Fluor 568 (Invitrogen, Paisley, U.K.) for immunofluorescent microscopy analysis, as per manufacturer instructions. Anti-chicken CSF1R (ROS-AV170) (13) was also labeled by Alexa Fluor 568 (Invitrogen) for immunofluorescent microscopy analysis. Other primary Abs included anti-chicken CD45–FITC (clone UM16-6; Bio-Rad Laboratories), FITC-labeled anti-chicken monocyte/macrophage marker KUL01 (clone KUL01; SouthernBiotech), anti-Bu-1-FITC (clone AV20; SouthernBiotech), anti-chicken MHC class II (MHC II)–FITC (2G11; SouthernBiotech), IgG1 isotype control, mouse anti-ovine CD335 (GR13.1), rabbit anti-GFP (Alexa Fluor 688 labeled) (Thermo Fisher Scientific rabbit anti-RFP (DsRed) (BioVision; distributed by Cambridge Bioscience, U.K).

Kupffer cells were isolated from livers dissected from 4–6-wk-old NOVOgen BROWN layers (n = 4), based upon a method previously developed for the mouse (21). The liver was chopped into small pieces, transferred into 10 ml of 1 mg Dispase/Collagenase D (Roche Applied Science), and digested by incubation at 37°C for 30 min with occasional gentle mixing. The tissue was passed through a 100-μm cell strainer and the cell suspension was centrifuged at 300 × g for 5 min at 4°C. The pellet was washed twice with 50 ml RPMI 1640 then centrifuged at 50 × g without the brake for 3 min at 4°C to sediment parenchymal cells. This process was repeated. To remove cell debris and RBCs, the nonparenchymal cell pellet was resuspended in 5 ml of RPMI 1640, gently overlaid onto the same volume of Histopaque (1.077), and centrifuged at 400 × g without the brake for 30 min. The cells at the interface between RPMI medium and Histopaque were carefully collected and washed twice with RPMI 1640 medium.

Blood leukocytes were isolated from 4–8-wk-old CSF1R-mApple–transgenic chickens (n = 6) (14) by density gradient centrifugation using Lymphoprep (density 1.077 ± 0.001 g/ml; Alere Technologies, Oslo, Norway) as described previously (22). Nonparenchymal cells in liver tissues from 4–8-wk-old CSF1R-mApple–transgenic chickens (n = 7) were prepared as described above. A similar digestive procedure using Dispase/Collagenase D was also applied to bursal tissues from CSF1R-mApple–transgenic chickens (n = 7); the tissue digest was passed through 70-μm cell strainer for single bursal cell suspension.

For analysis in vitro, isolated Kupffer cells (n = 4) were seeded into 4-well glass chamber slides (Nunc) and cultured in DMEM supplemented with 10% FBS, 1× GlutaMAX, and 100 U/ml penicillin/streptomycin at 41°C in a 5% CO₂ incubator overnight. Nonadherent cells were then removed from the wells. To test the phagocytic activity of the adherent Kupffer cells, zymosan A BioParticles (Thermo Fisher Scientific) or apoptotic chicken thymocytes were diluted in DMEM and added to the wells at ∼10 particles per cell. The cells were then incubated at 41°C for 2 h, washed four times with ice-cold PBS, fixed with 4% paraformaldehyde (PFA) for 20 min, and permeabilized by 1% Triton X-100 for 15 min. The cells were probed by Alexa Fluor 568–conjugated anti-TIM4 mAb JH9, and the nucleus was counterstained with DAPI. Images were taken using a Leica DMLB microscope. Chicken RBCs were isolated from chicken whole blood by Histopaque gradient and aged by incubation at 4°C for 30 h. They were added to adherent Kupffer cells at a ratio of 10:1. The cells were incubated at 41°C overnight. As above, the cells were then washed, fixed, and stained with hematoxylin (Sigma-Aldrich) for 2 min. The results were analyzed using a Nikon Ni-E microscope.

To confirm phagocytic activity in vivo, 4-wk-old CSF1R-mApple–transgenic chickens were injected i.v. with 100 μl of 0.1-μ diameter FluoSpheres (Thermo Fisher Scientific). Birds were culled 3 h after administration of beads by cervical dislocation, and tissues were removed and fixed in 4% PFA. Fixed tissues were processed for immunostaining as detailed below.

Single cell suspensions from embryos, blood, liver, or bursa (n = 6–7) were used for flow cytometry analysis. Cells were washed with FACS buffer (PBS, 0.5% BSA, and 0.05% sodium azide) and incubated with anti-TIM4-AF647 and FITC-conjugated Ab to other markers, including CD45, KUL01, MHC II, or Bu-1 as described above. Cells were incubated at 4°C in the dark for 30 min and washed three times in FACS buffer. Cells were resuspended in 300 μl of PBS with SYTOX Blue Dead Cell Stain (1.0 μM; Invitrogen) 5 min prior to analysis using a BD LRSFortessa (BD Biosciences). At least 100,000 events were acquired. Dead cells were excluded by SYTOX Blue staining, and doublets were discriminated based on signal processing (side scatter–area/height or forward scatter–area/height). Data were analyzed using FlowJo software (FlowJo, Ashland, OR).

Blood leukocytes (n = 4) or isolated cells from liver and bursa (n = 3) were purified into different populations using a BD FACS Aria IIIu, based on their expression of TIM4, CSF1R-mApple, or Bu-1. Blood leukocytes and nonparenchymal liver cells from CSF1R-mApple reporter birds were labeled using anti-TIM4 (AF647 labeled) or bursal cells by anti-TIM4 and Bu-1 (B cell surface marker) for 30 min at 4°C and separated by FACS. All gate settings were based on isotype-matched controls. The separated cell populations (TIM4⁺mApple⁺ and TIM4⁻mApple⁺ from blood; TIM4⁺mApple⁻, TIM4⁺mApple⁺, and TIM4⁻mApple⁺ from liver; TIM4⁺ Bu-1⁻, TIM4⁺ Bu-1⁺, and TIM4⁻ Bu-1⁺ from bursa) were then lysed with TRIzol reagent (Invitrogen). RNA was extracted using an RNeasy Mini Kit (QIAGEN). RNA quality (RIN >7) was assessed by Agilent RNA ScreenTape assay with Agilent 2200 TapeStation and quantified by a Qubit RNA HS Kit (Molecular Probes). cDNA libraries were prepared using NEBNext Ultra RNA Library Prep Kit for Illumina (New England BioLabs), with each sample containing different index primer (NEBNext Multiplex Oligos for Illumina, Index Primer Set 1). Pools of four libraries were sequenced by Edinburgh Genomics using the Illumina HiSeq 4000 Sequencing System. Bone marrow–derived macrophages (BMDM) were prepared by cultivation of bone marrow cells in M-CSF (CSF1) as described previously (12, 17). mRNA was isolated, and gene expression profiles were assayed as described above. For comparative analysis, bursa and spleen mRNA was prepared as described above from 8–12 newly hatched birds as part of a separate project looking at gender and genotype associations with gene expression profiles. For the current analysis, the expression of each transcript was averaged across the whole data set. All of the sequencing data generated for this project are deposited in the European Nucleotide Archive under study accession number PRJEB25788 (http://www.ebi.ac.uk/ena/data/view/PRJEB25788).

Expression level was quantified, as both transcripts per million (TPM) and estimated read counts, using Kallisto v0.42.4 (23). Transcript-level read counts were summarized to the gene level using the R/Bioconductor package tximport v1.0.3 (24) with gene names obtained from the galGal5.0 annotation (via Ensembl BioMart v90). The tximport package aggregates Kallisto output into a count matrix, useable by the R/Bioconductor package edgeR v3.14.0 for differential expression analysis, as well as calculating an offset that corrects for changes to the average transcript length between samples (which can reflect differential isoform usage). Using edgeR, gene counts were normalized using the “trimmed mean of M values” method, with a negative binomial generalized log-linear model fitted and p values corrected for multiple testing according to a false discovery rate.

Tissue slices from at least three birds were fixed in 4% PFA/PBS for 1 h, washed with PBS, equilibrated with 15% sucrose/PBS overnight at 4°C, embedded in OCT compound, and flash frozen in liquid nitrogen. Tissue sections (10 μm) were cut, mounted onto sugar-coated slides, and dried overnight. For immunohistochemistry (IHC), sections were rehydrated with PBS for 5 min; endogenous peroxidase activity was quenched by incubating slides with 0.3% H₂O₂ in PBS. After blocking with normal horse serum, primary Abs were added to sections and incubated at 4°C overnight, followed by secondary Ab biotinylated goat anti-mouse IgG for 30 min, then avidin-biotin-peroxidase complex (GE Healthcare). Vector AEC (Sigma-Aldrich) or NovaDAB substrate was added to reveal bound peroxidase, and sections were counterstained with hematoxylin. The resultant staining was analyzed using a Nikon Ni microscope. For immunofluorescent staining, the sections were rehydrated and blocked with horse serum (14). The fluorescence-labeled Abs were added to sections and incubated overnight. The resulting staining was analyzed using a Leica DMLB microscope.

For whole-mount TIM4 staining of embryonic tissues (n = 4), 4% PFA-fixed tissues were placed in PBS with 10% normal horse serum and 0.1% Triton X-100 for 2 h at 4°C on a rocking platform, followed by overnight incubation with anti-TIM4 Ab. Tissues were washed for 30 min in PBS, incubated for 2 h with donkey anti-mouse Alexa Fluor 647 (Invitrogen) secondary Ab, and washed again for 30 min before imaging. For imaging of whole-mount immunostained tissues, a 35 mm × 10 mm–petri dish was modified by cutting a hole in the lower section and fixing a coverslip over the hole with nail polish. Stained embryonic limb buds were placed on the coverslip, with the lower surface in contact with the cover slip. Samples were imaged using an inverted confocal microscope (Zeiss LSM 710). For three-dimensional rendering, confocal z-stacks were created by obtaining images at 0.45-μm intervals. Images were captured using Zeiss ZEN software and analyzed using Imaris software version 8.2 (Bitplane).

TUNEL staining was carried out as previously described (25), with minor modifications. After rehydration, sections from three birds or embryos were incubated with 20% FCS and 1% BSA in PBS for 1 h. For nick-end labeling, the sections were equilibrated by incubation in TdT buffer (0.2 M potassium cacodylate, 1 mM cobalt chloride, 25 mM Tris, and 0.01% Triton X-100) for 15 min. Labeling reaction mixture (10 U TdT [Thermo Fisher Scientific] and 1 nM biotinylated dUTP [Roche Applied Scientific] in TdT buffer) was added to sections and incubated for 1 h at 37°C in a humidified chamber. The labeling reaction was terminated by washing the slides three times in citrate buffer (sodium chloride 300 mM, sodium citrate 30 mM). To visualize TUNEL-labeled cells, sections were incubated with FITC-conjugated streptavidin (Thermo Fisher Scientific) for 1 h. Sections were then stained with Alexa Fluor 568–conjugated anti-TIM4 for 2 h, and the nucleus was stained with DAPI. The resulting images were visualized using a Leica DMLB microscope.

Bursa of Fabricius, spleen, and liver were taken from 6-wk-old birds (n = 2). Tissues were disrupted using Lysing Matrix D (MP Biomedicals, Loughborough Leicester, U.K.) with lysis buffer (1% NP40/PBS) in a Fastprep-24 homogenizer (MP Biomedicals); the lysates were centrifuged at 13,000 × g for 5 min to remove any cell debris and DNA. Thirty micrograms of lysate protein was separated in 4–15% SDS–PAGE (Bio-Rad Laboratories, Hertfordshire, U.K.). After Trans-Blotting protein onto PVDF membrane (Sigma-Aldrich), the membrane was probed with primary Abs at 4°C overnight, followed by HRP-conjugated secondary Ab for 2 h at room temperature. HRP activity was detected using ECL substrate (Thermo Fisher Scientific).

TIMD4 mRNA was previously found to be relatively abundant in the majority of nonimmune as well as immune-related tissues of the chicken (11). Although macrophages are a major component of all tissues, highlighted using the CSF1R-reporter gene (14), this observation suggests either that the expression of TIMD4 by macrophages is much less tissue-specific than in mammals or that TIMD4 is expressed by nonmacrophage lineage cells. To distinguish these alternatives, sections of tissues from 6-wk-old chicks were examined by IHC using KUL01 (monocyte/macrophage marker) and anti-chicken TIM4 Ab JH9 (11).

To validate the mAb binding, tissue extracts were first analyzed by Western blot. As shown in Fig. 1A, in bursa of Fabricius, the mAb JH9 bound to a single TIM4 product at 100 kDa, the predicted size encoded by the long isoform TIM4L₁ mRNA (11). In the spleen, two TIM4 products at the similar density were detected, the larger band at 150 kDa, predicted by the longer TIM4L₀ mRNA (11). In the liver, the 150-kDa isoform was most highly expressed, but a minor larger isoform was also detected. Fig. 1B shows the patterns of staining of TIM4 in a range of adult tissues, compared with the widely used macrophage marker KUL01, which is encoded by the likely orthologue (MRC1-LB) of the mammalian mannose receptor gene MRC1 (26), itself regarded as a marker of functional polarization in mammalian macrophages (27). In the thymus, TIM4⁺/KUL01⁺ cells were mainly located in the medulla (Fig. 1B). In the caecal tonsil, an important GALT, numerous stellate KUL01/TIM4⁺ cells were scattered throughout the lymphoid tissue within the lamina propria and the submucosa (arrow). Fewer KUL01⁺ and TIM4⁺ cells were also present in the muscularis mucosae (M) (Fig. 1B). In the lung, both Abs labeled numerous cells within the lung parenchyma and the bronchial walls (Fig. 1B). In the bursa of Fabricius, staining patterns with the two Abs were quite distinct. Whereas KUL01 strongly stained interfollicular cells and weakly stained the cells clustered at corticomedullary epithelium, most TIM4⁺ cells were scattered within bursal lymphoid follicles, particularly at the corticomedullary epithelium (Fig. 1B). In the spleen, KUL01 and TIM4⁺ cells were mostly distributed at the interface between periellipsoidal white pulp and red pulp (RP), a region equivalent to the mammalian marginal zone, and fewer KUL01⁺ and TIM4⁺ cells were dispersed in RP (Fig. 1B). In the small intestine (jejunum), both KUL01 and anti-TIM4 stained presumptive macrophages in the lamina propria of the villi and crypts (Fig. 1B). In the liver, KUL01⁺ and TIM4⁺ cells were clearly separate and morphologically distinct. KUL01⁺ cells were relatively small and round, whereas TIM4⁺ cells were stellate and irregular, lining the walls of hepatic sinusoids, consistent with their identity as Kupffer cells (Fig. 1B). In the testis, KUL01⁺ and TIM4⁺ cells appeared similarly located in the connective tissue surrounding seminiferous tubules and within aggregates of lymphoid cells (Fig. 1B). In overview, TIM4 was much more widely expressed in chickens than in mice, apparently restricted to macrophage-like cells but not completely coincident with the widely used KUL01 marker.

Fig. 1C examines the heterogeneity of TIM4 expression by immunofluorescence, with Abs against CSF1R (13) and Bu-1 [expressed by subsets of macrophages as well as B cells (28)] as additional markers. Consistent with the IHC data in Fig. 1B, TIM4 and KUL01 appeared to be mutually exclusive in the liver. By contrast, in the spleen, periellipsoidal macrophages coexpressed TIM4 and KUL01. CSF1R protein was not detectable by immunofluorescence on any TIM4⁺ cells. Costaining of spleen sections using anti-TIM4 and Bu-1 Abs highlighted the architecture of the periellipsoid zone in the spleen surrounded by TIM4⁺ macrophages. TIM4⁺ Bu-1⁺ cells were also evident, scattered in RP. The staining of the spleen revealed the structure of the B cell–rich germinal center, in which TIM4⁺ macrophages not only outlined the germinal center, structurally similar to mammalian metallophilic macrophages, but also intermingled with B cells inside germinal center to engulf B cell debris, equivalent to tingible body macrophages in mammalian germinal centers. Individual fluorochrome images are shown in Supplemental Fig. 1.

In the intestine, the staining of jejunum and ileum indicated that a heterogeneous population of TIM4⁺, KUL01⁺, and TIM4⁺, KUL01⁻ cells localized at the lamina propria of the villi. In the liver, the CSF1R-mApple transgene was detectable on only a small subset of cells, as previously noted (14). By contrast, TIM4 brightly labeled an almost continuous network of cells (Fig. 1D), consistent with the high levels of TIM4 protein detected on Western blots. In the bursa of Fabricius, CSF1R-mApple–positive cells formed an extensive network of stellate cells within the medulla but, again, there was little apparent overlap with TIM4 (Fig. 1D). These results indicate that in chickens, as in mice, TIM4 is a unique marker for Kupffer cells and subpopulations of macrophages in lymphoid organs. However neither the TIM4⁺ Kupffer cells in the liver or TIM4⁺ stellate cells in the bursa showed detectable CSF1R-transgene expression. This indicates that either CSF1R mRNA is not expressed in these cell subsets or that the CSF1R transgene used in this study (14) is not expressed in all CSF1R⁺ cell populations. Finally, in contrast to mice, TIM4 is widely expressed by resident macrophage-like cells throughout adult tissues in the chicken, whereas neither the CSF1R-mApple transgene nor KUL01 was uniformly expressed by all chicken tissue macrophages.

Because they are the largest macrophage population with direct contact to the blood, Kupffer cells contribute to the clearance of particles and damaged or aging erythrocytes as well as potential pathogens (5). Because avian erythrocytes are nucleated, we reasoned that this activity would be detectable in the steady state by TUNEL assay. Fig. 2A shows that this is indeed the case. Most Kupffer cells contained multiple TUNEL-positive nuclei and are TIM4^hi. To examine the phagocytic activity directly, CSF1R-mApple–transgenic chickens were injected with 0.1-μm fluorescent latex beads. As shown in Fig. 2B, after 3 h, these particles were colocated with CSF1R-mApple⁻, TIM4^hi Kupffer cells. The CSF1R-mApple transgene indicates the presence of a population of stellate, CSF1R-mApple⁺, TIM4⁻ cells that were not actively phagocytic. The phagocytic activity of Kupffer cells was examined further following isolation in vitro. These isolated TIM4^hi Kupffer cells exhibited strong phagocytic capability in uptake of fluorescence-labeled zymosan particles (Fig. 2C), apoptotic thymocytes (Fig. 2D), and aged chicken RBCs (Fig. 2E). Although we lack additional markers for the chicken Kupffer cell, the uniformity of cell morphology under light microscopy, phagocytic activity, and high expression of TIM4 (Fig. 2C–E) indicates that our enrichment method was successful, and a pure population of chicken Kupffer cell (>90%) was isolated. We further characterized the TIM4^hi CSF1R-transgene–negative Kupffer cell population ex vivo by flow cytometry (Fig. 2F, 2G). TIM4^hi CSF1R-transgene–negative cells were uniformly CD45⁻ and MHC II⁺, lacked expression of the B cell marker Bu-1, and were largely negative for the T cell marker CD3 (∼6% positive; Fig. 2F), indicating that this is a relatively homogenous population of cells. Low-level expression of KUL01 on Kupffer cells has previously been reported using an enzyme-amplified immunohistochemical detection (29), but TIM4 expression was not detectable in KUL01 cells in the liver in situ by immunofluorescence staining (Fig. 1C). Consistent with these findings, the levels of the macrophage marker KUL01 on TIM4^hi CSF1R-transgene–negative cells detected by FACS (Fig. 2F), were ∼10-fold lower than KUL01⁺ CSF1R-transgene–positive cells (Fig. 2G).

In the bursa of Fabricius, a critical organ involved in avian B cell development, only 1–5% of B cells produced per day leave the organ, whereas many more cells undergo apoptosis in situ (1, 30, 31). To examine the function of TIM4⁺ macrophages in the clearance of apoptotic B cells, we double-stained bursal sections with anti-TIM4 and Bu-1. Apoptotic B cells (Bu-1⁺) with evidence of cell shrinkage and condensed chromatin were apparently phagocytized by TIM4⁺ macrophages in the medulla and around the corticomedullary junction (Fig. 3A). The phagocytosis of apoptotic B cells also occurred in embryonic bursa (Fig. 3B). TUNEL assays confirmed the association of apoptotic B cells with TIM4⁺ macrophages in the bursa in both posthatch birds (Fig. 3C) and the embryo (Fig. 3D).

Macrophages first appear in the yolk sac prior to colonization of the embryonic body (15, 16, 18). An earlier study comparing mouse and chick examined TIMD4 mRNA in the chick semiquantitatively at embryonic days (ED) 4–7 and inferred that TIMD4 mRNA was likely expressed by yolk sac–derived macrophages (32). As noted in the 1Introduction, this conclusion was supported by expression profiling of the chick embryo (19, 20). To confirm the location of TIM4 protein, we first examined the yolk sac and embryos from ED3 (HH19) by whole-mount IHC. By ED3, blood island aggregates already contained numerous TIM4⁺ cells, and these were also clearly visible in vitelline blood vessels but had not yet appeared in the embryo (Fig. 4A). At ED5, embryo-committed progenitors start to give rise to erythroid and monocyte progenitors (13). By this stage, TIM4⁺ cells were clearly detected and distributed in limb buds, the aorta–gonad–mesonephric region, liver, and ventricle (Fig. 4C). TIM4⁺ cells isolated from disaggregated ED5.5 embryos and analyzed by flow cytometry (Fig. 4D) were CD45⁺ (>97%), and the majority also labeled with KUL01 (>67%).

The same CSF1R regulatory elements used in CSF1R-mApple birds have also been used to drive expression of EGFP (14). In the CSF1R-EGFP chick, macrophages were visible throughout the body and were concentrated in areas of programmed cell death, such as the interdigit regions of stage ED8 embryo leg buds (Fig. 4D). TIM4 staining colocalized with these CSF1R-EGFP⁺ macrophages (Fig. 4Dii). Interestingly, in macrophages associated with apoptotic cell remains in the interdigit region, TIM4 staining appeared restricted to intracellular phagolysosomes, whereas in highly ramified macrophages outside this region TIM4 staining clearly delineated the plasma membrane (Fig. 4Diii, iv). This pattern would be consistent with a role for TIM4 in apoptotic cell internalization.

In mice, many tissue macrophage populations are maintained without major input from the blood monocyte pool (33). There is as yet no direct evidence that this is the case in birds, and our earlier data suggested that transplanted bone marrow progenitors can give rise to macrophages that populate most chick organs (17). Even in the mouse, organs such as skin and intestine are constantly replenished with macrophages derived from monocytes (15, 16). We, therefore, examined whether the extensive TIM4⁺ population in tissues might have precursors in the blood in the CSF1R-mApple–transgenic chicken by flow cytometric analysis (Fig. 5A). As noted previously, the entire population of CSF1R-mApple⁺ blood monocytes expressed KUL01 and high levels of MHC II (14). Conversely, TIM4 staining clearly distinguished a subpopulation of positive cells, making up around 50% of the blood monocytes. These cells did not show any differential expression of class MHC II or KUL01 on flow cytometry. Combining the CSF1R-mApple transgene and TIM4 as markers, we separated the TIM4⁺ and TIM4⁻ monocyte populations (Fig. 5A) and assessed their gene expression by RNA sequencing (RNAseq). The primary data, comparing two pooled preparations of TIM4⁺ and TIM4⁻ monocytes separated as shown in Fig. 5A, including relative expression ratios, are provided in Supplemental Table I. TIMD4 mRNA was enriched around 5-fold in the sorted TIM4⁺ population relative to TIM4⁻ monocytes, whereas MRC1LB (ENSGALT000000430910), which encodes KUL01 (26), was not different between the populations. The genes enriched within the TIM4⁺ cells include several globin subunits and spectrin, suggesting that as in the liver TIM4⁺ monocytes may be involved in uptake and destruction of senescent RBCs. Other differences are discussed below.

In single cell suspensions isolated from the liver, we identified three populations of cells separated by expression of the CSF1R-mApple and the level of expression of TIM4 (Fig. 5B). Each population (TIM4^hi, CSF1R-mApple⁻; TIM4^lo, CSF1R-mApple⁺; and TIM4⁻, CSF1R-mApple⁺) was separated by FACS as shown in Fig. 5B. The lack of expression of CSF1R-mApple in the TIM4^hi cells reflects the lack of detection in the active liver phagocytes in Fig. 2. The mRNA expression profiles of the three separated populations were assessed by RNAseq. Pairwise comparisons between the three populations are provided in Supplemental Table I. Surprisingly, the CSF1R-mApple⁻ populations expressed CSF1R mRNA at similar levels to the CSF1R-mApple⁺ cells, indicating that this transgene does not accurately report CSF1R transcription in the liver. The level of TIMD4 mRNA in the three populations was consistent with the expected enrichment based upon the FACS profiles. TIMD4 mRNA was enriched 3-fold higher in the sorted TIM4^lo, CSF1R-mApple⁺ cells relative to TIM4⁻, CSF1R-mApple⁺ cells, and enriched a further 2.5-fold higher in the TIM4^hi, CSF1R-mApple⁻ population.

Finally, in cells isolated from the bursa, we were not able to identify or recover sufficient cells for RNAseq profiling based upon the CSF1R-mApple transgene, despite the apparent prevalence of expressing cells in the tissue. Labeling with TIM4 and the B cell marker Bu-1 identified three populations, which were separated by FACS using the gates shown in Fig. 5C. mRNA expression profiles were again assessed by RNAseq. Pairwise comparisons of the three populations are shown in Supplemental Table I. Consistent with the isolation based upon TIM4, there was a hierarchy of TIMD4 expression. TIMD4 was just detected in the TIM4⁻ Bu-1⁺ population, was >10-fold higher in the TIM4^lo, Bu-1⁺ population, and a further 10-fold higher in the TIM4^hi Bu-1⁻ population. As observed in the liver, despite the absence of detectable CSF1R-mApple, CSF1R mRNA was very highly expressed in the TIM4^hi cells. The expression signatures of the liver and bursal TIM4⁺ populations are discussed further below.

To analyze the relationship of the separated cell populations to each other and to identify transcripts that are strictly coregulated with TIMD4 and CSF1R, we used the network analysis tool Graphia Professional (http://www.kajeka.com), which was developed from BioLayout Express 3D (34). For this purpose, we included averaged expression from RNAseq data sets from hatchling spleen and bursa and from BMDM grown in CSF1 (12) as comparators. The spleen data set we analyzed is derived from 18 male and female birds of diverse genetic background, in which TIMD4 varied around 4-fold between individuals. For the current analysis, the male and female data were separately pooled and averaged. Fig. 6 shows the sample-to-sample clustering. The analysis reveals the clear separation of blood monocytes, BMDM, bursa, and spleen from all cells isolated from liver. The presumptive bursal B cells (Bu1⁺, TIM4⁻) most closely resemble the total spleen and bursa profiles and are clearly separated from macrophages. TIM4⁺ populations associated with their tissue-related cell type rather than with each other, indicating that TIM4 expression is not in itself a differentiation marker. Gene-to-gene clustering identified sets of transcripts that were stringently coexpressed across the whole data set. Annotated clusters with the average expression profile of each cluster are provided in Supplemental Table II. TIMD4 was coexpressed with a small set of transcripts in cluster 87, among which only CX3CR1 was expressed >50 TPM (TIMD4 >800 TPM). CSF1R was part of an even smaller cluster, cluster 188, in which the only other robustly expressed transcript is CLCN5. Clearly, both TIMD4 and CSF1R can be expressed by cells with very divergent cellular phenotypes and cannot be considered as markers of any particular cell type or lineage. Indeed, there were few clusters that were clearly associated with cell types or process. Not surprisingly, one of the largest clusters, Cluster 3, was clearly enriched for phagocyte-associated transcripts and components of the lysosome. A similar coexpression cluster was identified in mice (35); the highest average expression of among members of this cluster was in the chicken BMDM. Otherwise, the only other large expression clusters with evident cell type association were Cluster 7, which contains monocyte-enriched transcripts (including CD14, CCR2, CSF3R, TLR2A) and transcription factors (CEBPB, NFE2L2, PRDM1, and TFEC) and Cluster 9, in which expression was highest in liver CSF1R-mApple⁺ cells. Consistent with the pairwise analysis above, this cluster contains MHC II and transcripts associated with DCs (BLB1, BLB2, CADM1, CIITA, CD74, CD86, FLT3, XCR1, and transcription factors IRF1, IRF5, and IRF8). The Kupffer cell marker MARCO was within a small cluster, cluster 63, with GPR34 and VSIG4.

TIM4 in chickens, as in mammals, is a receptor expressed primarily by macrophages that binds to PS and most likely participates in the recognition and clearance of apoptotic cells. In this study, we used a novel anti-chicken TIM4 Ab to define the location of TIM4⁺ macrophages in adult chicken tissues and in developing chick embryos. In the embryo, macrophages are involved in extensive phagocytosis of dying cells, and the almost uniform coexpression of embryonic TIM4 with other macrophage markers (Fig. 5) is consistent with that function. The data from posthatch birds (Figs. 1, 2) indicate that TIM4 is retained at the highest levels in a subset of tissue macrophages that appear to be associated specifically with the uptake of apoptotic cells and other particles.

We identified a subpopulation of blood monocytes in chickens that express detectable TIM4, around 50% of the total population (Fig. 5A). TIMD4 mRNA is not detected in monocytes in either mouse or human (36). Chicken blood monocytes uniformly expressed the CSF1R-mApple transgene, KUL01 (MRC1, CD206), and high levels of MHC II. In mammals, two subpopulations of blood monocytes have been recognized and referred to as “classical” and “nonclassical” (37). Aside from surface markers (Ly-6C in mouse and CD16 in humans), the monocyte subpopulations in mammals differ especially in expression of the chemokine receptors CCR2 and CX3CR1, which control their extravasation (38). The nonclassical monocytes are derived from the classical monocytes, and their differentiation is controlled by CSF1 (33, 39). We used the combination of the CSF1R-mApple transgene and TIM4 to generate comparative gene expression profiles of chicken blood monocytes (Supplemental Table I). To our knowledge, this is the first such data set generated. The cluster analysis in Supplemental Table II identified a set of transcripts that was strongly enriched in chicken monocytes relative to other macrophage populations. In particular, CSF3R mRNA (encoding the G-CSF receptor) was very abundant. The ligand of this receptor in chickens was originally called myelomonocytic growth factor (40). The level of expression of CSF3R suggests that it does indeed have a function in monocyte as well as granulocyte regulation in birds. The gene annotated as CSF2RA (GM-CSF receptor) was lowly expressed, but a putative paralogue (ENSGALT00000026942) was present at much higher levels in monocytes and on that basis appears more likely to encode the CSF2 receptor. There are few annotated monocyte surface markers in the chick, but MRC1LB (KUL01), CD14, MYD88, TLR4, and TLR2A were each highly expressed and increased around 50% in TIM4^lo cells. In common with mouse monocytes (41), the chicken monocytes expressed high levels of the chemokine receptors CCR2, CXCR4, and CX3CR1. The TIM4⁺ subpopulation had increased expression of CX3CR1 in common with nonclassical mouse monocytes. Other highly expressed and monocyte-associated genes include those encoding a smorgasbord of transcription factors, notably ATF4, CEBPA, CEBPB, CEBPD, EGR1, ELF1, ELF5, ETV6, FLI1, FOS, FOSB, FOSL2, HIC1, HIF1A, ILF3, IRF1, IRF2, IRF5, IRF8, JUN, JUND, KLF4, KLF6, MAF1, MAFB, MEF2D, MITF, NFE2L2, NFKB1, NFKB2, NR4A1, NR4A3, PPARD, RARA, REL, RUNX1, SPI1, STAT1, STAT3, STAT5B, STAT6, and TFEC. Most of these factors have also been implicated in monocyte–macrophage differentiation in mice (1, 41, 42). Most were marginally elevated in the TIM4⁺ chicken monocyte subset, suggesting that these cells are part of a differentiation series rather than a distinct subset (Supplemental Table I). There were few transcripts encoding specific functions that were both highly and selectively expressed in the TIM4⁺ monocytes. Most notable is around a 2-fold increased expression of ACVRL1, C1QA, C1QB, C1QC, CX3CR1, HAVCR1, ITGB5, LRP1, MARCO, MERTK, P2RY3, P2RY6, S1PR2, SMPD1, and STAB1, all of which have been implicated in apoptotic cell recognition and removal (43). The relative overexpression of the major alanine-rich C kinase substrates MARCKSL1 and MARCKS could also represent an adaptation for phagocytic activity (44). Accordingly, we suggest that chicken TIM4⁺ monocytes are adapted for apoptotic cell recognition. Given the extensive populations of TIM4⁺ macrophages in avian tissues, the circulating TIM4⁺ monocytes could provide progenitors to replace them, but equally, as in mammals, the tissue macrophages may self-renew and the TIM4⁺ monocytes may be the resident scavengers of blood.

The histological examination of the liver indicated that the TIM4⁺ positive phagocytes with the characteristic stellate morphology of Kupffer cells lacked expression of the CSF1R-mApple transgene (Fig. 2) but expressed CSF1R mRNA at the same level as the transgene-positive cells (Supplemental Table I). This finding appears at first glance to be distinct from the mouse, in which a CSF1R reporter is expressed at high levels in all Kupffer cells (and indeed in all myeloid cells in the liver) (45). Kupffer cells in mice are rapidly depleted with anti-CSF1R Ab (39). They are the main site of clearance of CSF1 from the circulation (46) and respond to CSF1 with extensive cell proliferation (47, 48). So, mouse Kupffer cells clearly do express functional CSF1R. However, a modified transgene, which lacks a 150-bp distal promoter element but contains the conserved Fms intronic regulatory element (FIRE) enhancer, like the chicken transgene, was active in monocytes and DCs but undetectable in Kupffer cells or many other tissue macrophage populations (49, 50). There are regions of homology across avian species, aside from FIRE, that were not included in the avian CSF1R transgene (14). We infer that these are required for Kupffer cell CSF1R expression in chickens as in mice. Furthermore, most of the monocyte transcription factors noted above were downregulated in Kupffer cells relative to the liver cells (Supplemental Table I) in which CSF1R-mApple was active; among these factors, FOS/JUN, IRF8, STAT1, and RUNX1 all bind to FIRE in mouse macrophages (42). Further consideration of CSF1R transcriptional regulation lies outside the focus of the current study. The key point is that the chicken CSF1R transgene we have produced provides a convenient marker that distinguishes monocytes and DCs (discussed further below) from resident Kupffer cells.

The set of transcripts that was coenriched with TIMD4 in the TIM4^hi, CSF1R-mApple⁻ Kupffer cell population (Supplemental Table I) includes transcripts expressed in hepatocytes (e.g., ALB, TTR) and endothelial cells (CDH5, EDNRB, THBD), which like the Kupffer cells, do not express the transgene. Consistent with the localization of KUL01 in the liver (Fig. 1B) and flow cytometric analysis (Fig. 2F, 2G), the Kupffer cells had very low expression of MRCL1 encoding KUL01.

The Kupffer cell population may also be enriched for platelets. An unannotated gene (LOC101750889) of unknown function was highly expressed and enriched in the Kupffer cells. The most related protein in mouse and human genomes is the platelet membrane protein GP1BA, and GP1BA was itself highly enriched in the TIM4⁺ population. Nevertheless, aside from TIMD4 itself, relative to TIM4^lo, CSF1R-mApple⁺ cells, the isolated TIM4^hi population selectively expressed numerous genes involved in the recognition and elimination of apoptotic cells, including C1Q (A,B,C), CD36, CTSB, CTSD, CTSS, CX3CR1, DNASE2B, GPR34, LGALS1, LGALS3, MARCO, SCARB1, TGM2, and VSIG4 (highlighted in Supplemental Table I). Each of these genes is strongly Kupffer cell–enriched in mice (1, 51, 52) and is induced during embryonic liver differentiation (20). Notably, the scavenger receptor gene MARCO was almost exclusively expressed in the chick liver TIM4⁺ macrophage population (Supplemental Table I). One poorly annotated liver-specific transcript that was even more highly expressed than MARCO, LOC101748207, encodes a soluble scavenger receptor cysteine-rich domain-containing protein and may provide a novel Kupffer cell marker. The cluster analysis (Supplemental Table II) indicates that there are only a few transcripts that share a transcription pattern with MARCO, of which only VSIG4 and GPR34 were highly expressed. The underlying transcriptional regulation is also consistent with data from mice in that genes encoding known regulators of apoptotic receptors in mice, NR1H3 and PPARG (53), were also strongly enriched in the chicken Kupffer cells. In mice, Kupffer cells are required for the elimination of senescent RBCs and recycling of iron (54). Consistent with the conservation of this function and its regulation, the chick Kupffer cells were enriched for expression of the ferriportin gene SLC40A1, the haeme transporter SLC48A1, ferritin H chain FTH1 and haem oxygenase (HMOX1), and the transcription factor MAF, which regulates expression of these genes (55) (Supplemental Table I). In mammals, CD163-mediated endocytosis of hemoglobin–haptoglobin complexes is a major pathway for iron uptake (56). However, CD163 and CD163L1 were barely detected in any of the isolated chicken macrophage populations (Supplemental Table I).

A reciprocal set of genes was enriched in the TIM4^lo, CSF1R-mApple⁺ population of liver macrophages relative to both Kupffer cells and the TIM4⁻ population (Supplemental Table I). The expression of the MHC II genes BLB1 and BLB2 and the class II invariant chain gene CD74 was enriched more than 10-fold in this population relative to Kupffer cells (Supplemental Table I), along with the costimulators CD83 and CD86 and monocyte/DC–associated transcription factors ATF3, BHLHE40, CIITA, FOS, IRF5, IRF8, and NR4A3 and growth factor receptors CSF2RB and CSF3R. Vu Manh et al. (57) have previously isolated conventional DCs from chick spleen on the basis of high MHC II and the absence of KUL01. Like the chicken splenic conventional DC (57), the isolated TIM4^lo liver cells expressed low MRC1LB (encoding KUL01) and, relative to Kupffer cells, were also enriched for BEND5, CADM1, CD40, CD86, CSF2RB, FLT3, LY75 (DEC205), PLEKHA5, XCR1, and ZBTB46. Supplemental Table I identifies additional candidate markers for these cells: LGALS2, LRRK1, and P2RY6. The data in Figs. 2 and 3 demonstrate that these cells are not active phagocytes, but they are not deficient in lysosome-associated transcripts (e.g., LAMP1, CTSB) by contrast to isolated classical splenic DC in mice (35). Unlike the isolated chicken splenic DC, these cells also express CSF1R mRNA at the same level as Kupffer cells. DC-like cells have been identified in mouse liver (58), but they were a minor population and largely concentrated under the capsule. In the chick, they are clearly considerably more numerous. Indeed, based upon the FACS profiles of isolated cells (Fig. 5B) and localization in situ, CSF1R-mApple⁺ liver DC are more abundant than Kupffer cells and colocated in the sinusoids. Their gene expression profiles are related to blood monocytes, but monocytes lack FLT3 mRNA. We speculate that with the absence of lymph nodes and limited lymphatics, the liver might be an important site of Ag recognition and presentation in birds.

The comparison of TIM4⁺ and TIM4⁻, CSF1R-mApple⁺ cells in the liver does not simply recapitulate the comparison in blood. The liver TIM4⁺ populations express TIMD4 mRNA much more highly than TIM4⁺ monocytes. Again, the comparison is compromised by the apparent enrichment of hepatocyte-associated transcripts in the liver TIM4⁺ population. For that reason, the apparent expression of Kupffer cell–associated transcripts, such as MARCO, VSIG4, C1QA, B, and C, could represent a level of contamination with Kupffer cells. The more informative gene set, strongly enriched in the TIM4^- population, includes monocyte-associated markers S100A8, CSF3R, TLR2A, and MRC1LB, inflammatory cytokines (IL-1B, IL-6), and the stress-associated transcription factor NFE2L2. A similar monocyte-like population has also been identified in mouse liver (58). However, the chick TIM4⁻ liver cells also express high levels of FLT3, XCR1, and other DC markers noted above. It may be that they are a heterogeneous mix of cells or an intermediate in differentiation from monocytes. This requires further investigation. Interestingly, the two CSF1R-mApple⁺ DC-like populations also express high levels of SLC11A1, also known as the natural resistance–associated macrophage protein 1 (NRAMP1), relative to much lower expression in TIM4^hi Kupffer cells and blood monocytes, BMDM, and the spleen (see Supplemental Table I and Cluster 9 in Supplemental Table II). The chicken SCL11A1 gene was previously shown to be expressed in the liver, thymus, and spleen. As in mice, SCL11A1 polymorphism was associated with resistance to Salmonellosis (59). Consistent with that report, SLC11A1 was undetectable in bursal cell populations (Supplemental Table I). The enrichment in the DC, which appears to be poorly endocytic, is paradoxical because Salmonella is an intracellular pathogen but suggests that the gene product may have a function in Ag presentation rather than control of intracellular pathogen replication. In overview, we have identified three populations of mononuclear phagocytes in the liver (Fig. 5B), all of which express CSF1R mRNA but differ in transcriptional regulation of the CSF1R reporter gene. Twenty to thirty percent of the isolated cells were TIM4^hi Kupffer cells. The gates in Fig. 5B are arbitrary, and isolation may not be quantitative. However, the relative abundance is consistent with the images in Fig. 2. The remainder of liver mononuclear phagocytes express CSF1R-mApple and appear to be adapted for Ag presentation.

In the bursa, there were few transcripts that were strongly expressed and distinguish TIM4⁺ and TIM4⁻, Bu-1⁺ cells (Supplemental Table I). Those enriched in the TIM4⁺ population include CSF1R, transcription factors SPIC, MAFA, and MAFB, and genes involved in apoptotic cell disposal, including CD274, CD244, and VSIG4. The levels of these transcripts were low, and we cannot eliminate a contribution from small numbers of contaminating macrophages or DC. Based upon high Bu-1 expression and sample-to-sample clustering in Fig. 6, both Bu-1⁺ populations appear to be mainly B cells, and indeed the TIM4⁺ and TIM4^– populations share similar high expression of the BCR subunit CD79B, the B cell kinase BTK, MHC II (BLB1, BLB2, CD74), and B cell transcription factors, IKZF1/IKZF3, PAX5B, POU2F1, and TCF3. Both populations are likely to be proliferative, based upon their shared constitutive expression of cell cycle–associated genes, including regulators BUB1, CDC20, E2F, FOXM1, which are almost absent in the TIM4^hi, Bu-1⁻ cells. The level of TIM4 and TIMD4 mRNA detected on these presumptive B cells was low, which may explain why they were not evident in sections of the bursa (Figs. 1, 2). The TIM4^hi, Bu-1⁻ bursal cells, compared with TIM4^lo, Bu-1⁺ cells, express high levels of TIMD4 mRNA and share many enriched transcripts with Kupffer cells (e.g., ACVRL1, C1QA, CIQB, C1QC, CX3CR1, FTH1, GPR34, HMOX1, LGALS3, LY86, MERTK, TGM2, SLC40A1, SPARC, STAB1), consistent with adaptation for elimination of apoptotic cells and identity with the active phagocytes in Fig. 3. However, they are distinct from Kupffer cells in expressing little MARCO, VSIG4, and CD36 and in expressing very high levels of MHC II (BLB1, BLB2, CD74).

In conclusion, we have shown the specialized functional adaptation and transcriptional profile of liver TIM4^hi Kupffer cells in mammals is conserved in birds, and we have characterized a distinct population of phagocytes that express TIM4 in the bursa and that are also adapted to clear apoptotic cells. We have also identified functional diversity in chicken blood monocytes associated with expression of TIM4 and characterized a surprisingly abundant population of DC in the liver. Recent evidence in mice indicates that blood monocytes can and do differentiate into self-renewing Kupffer cells (60) although this is not a major pathway in the steady state (33). Further studies will be required to determine whether TIMD4-expressing cells in the liver, bursa, and elsewhere and the distinct myeloid populations in the liver derive from monocytes or are entirely self-renewing (61). In this respect, our ability to generate cellular transplantation models, both in ovo and in hatchlings (17), and emerging capacity to generate knockouts (1) may make the chick a unique system for the study of macrophage and DC ontogeny.

This paper is dedicated to the memory of our mentor, friend, and colleague, Prof. Pete Kaiser, who initiated this project. Pete Kaiser died in July 2016. We acknowledge Dr. Taiana Pereira Da Costa at The Roslin Institute and Royal (Dick) School of Veterinary Studies (current address: Wildfowl & Wetlands Trust in Slimbridge, U.K.) for discussion on IHC analysis and Dr. Debiao Zhao for early guidance in embryo analysis.

The authors have no financial conflicts of interest.