Novel Teleost CD4-Bearing Cell Populations Provide Insights into the Evolutionary Origins and Primordial Roles of CD4⁺ Lymphocytes and CD4⁺ Macrophages

The largest subset of T cells in jawed vertebrates expresses a TCR bearing α and β-chains that recognizes Ags bound to MHC molecules (1). Such T cells use two main coreceptors, CD4 and CD8, which show mutually exclusive expression on naive Th and cytotoxic T cells, respectively, in mammals (2). CD4⁺ Th cells can further differentiate into a variety of effector Th cell subsets that produce cytokines involved in the regulation of inflammation and immune responses against different types of pathogens (3). Mammalian CD4 is also expressed on cell types other than T lymphocytes (4). CD4 expression pattern shows species disparity but often defines functionally distinct subsets in a certain myeloid cell population. Significantly, the majority of human and rat monocytes/macrophages express CD4, whereas in mouse and birds CD4 appears to be absent in these cells (5, 6). Whether CD4-expressing monocytes/macrophages play any particular role in immunity remains for the most part a mystery.

Although sharks contain TCR-expressing lymphocytes (7, 8), recent genome sequence and transcriptome analyses of cartilaginous fish (elephant and nurse shark) have failed to identify a molecule with classical CD4 features, thus making teleosts the oldest living species with bona-fide CD4 coreceptors (9–12). In contrast with the situation of tetrapods, which possess a single CD4 gene, bony fish contain two CD4 genes, cd4-1 and cd4-2, which share low amino acid identity (∼20%) (13, 14). Like tetrapod CD4, teleost CD4-1 contains four Ig domains: two V (D1 and D3) and two C (D2 and D4) domains. In contrast, CD4-2 may contain two or three Ig domains (1V and 1C or 2V [D1 and D3] and 1C) (13, 14). Similarities between D1 and D3 and between D2 and D4 of tetrapod CD4 have led to the hypothesis that the four-domain CD4 emerged through the duplication of the gene encoding an ancestral two-domain (V-C) receptor (14–18). Hence, it has been hypothesized that the teleost two-domain CD4-2 might be reminiscent of this ancestral gene (14, 18). In line with these hypotheses, a CD4-like gene encoding only two Ig domains (V-C) has been identified in lamprey, and it was proposed to represent the prototypic two-domain CD4 coreceptor in vertebrate (19). It is worth noting that lymphocyte-specific protein tyrosine kinase (LCK) in catfish binds to both CD4-1 and CD4-2, thus suggesting that both teleost CD4 coreceptors are functionally active and play a role in T cell development and activation (20).

Thus far the concurrent surface expression of CD4-1 and CD4-2 on teleost leukocytes has not been determined because of the lack of reagents capable of detecting both molecules in a single species. In the absence of those, transcript levels have been used to assess expression patterns of cd4-1 and cd4-2, but results in teleosts have been inconclusive. For example, flounder cd4-1 and cd4-2 transcripts are expressed in mutually exclusive cell types (21), whereas transcript analysis of sorted CD4-1⁺ cells in zebrafish, ginbuna carp, and fugu suggests the existence of cells expressing CD4-1 and/or CD4-2 (22–24). Because of the poor characterization of teleost CD4⁺ leukocytes, very few functional studies have been carried out on these cells. More specifically, it has been reported that CD4-1⁺ cells in ginbuna carp and zebrafish undergo Ag-specific proliferation (25–27). With regard to cytokine expression, zebrafish and fugu CD4-1⁺ cells can produce T cell–related cytokines in response to stimulation with TLR ligands and specific Ag (22, 24); however, the expression of such cytokines upon pathogen challenge remains to be studied. The only study performed on teleost CD4-2⁺ cells suggests that these leukocytes represent a regulatory T cell (Treg)–like phenotype in pufferfish (28). In the context of these functional similarities between tetrapod CD4⁺ T cells and teleost CD4-1⁺ or CD4-2⁺ cells, it has been well documented that teleost fish contain most of the critical genes (i.e., cytokines and their receptors) involved in T cell function, thus supporting the potential presence of effector T cell subsets in teleost fish (29, 30).

Because two-Ig-domain CD4 molecules are reminiscent of the prototypic CD4 coreceptor (14, 18, 19), we hypothesized that by gaining understanding into the cell types bearing CD4 with either two- or four-Ig domains, we could shed light into the evolutionary history of CD4-bearing cells, as well as their primordial roles in immunity. To this end, we phenotypically and functionally characterized CD4-1– and CD4-2–expressing cells in rainbow trout, a model species in the field of evolutionary and comparative immunology (31). Our studies represent the first comprehensive phenotypic and functional characterization of two- and four-domain CD4-bearing lymphoid and myeloid cells in a vertebrate species and reveal critical insights into the evolutionary origins and functionally conserved roles of vertebrate CD4⁺ T cells and CD4⁺ monocytes/macrophages.

Rainbow trout (Oncorhynchus mykiss) were provided by the National Center for Cool and Cold Water Aquaculture (NCCCWA), and Clear Springs Foods Fish were maintained in the laboratory of J.O.S. as previously described (32). Animal procedures were approved by the Institutional Animal Care and Use Committees of the University of Pennsylvania.

RNA extraction and gene expression analyses of sorted trout leukocytes were performed as described previously by us (32–34). Real-time PCR was performed with Power SYBR Green Master Mix (Thermo Fisher Scientific) and gene-specific primers in a 7500 Fast Real-Time PCR System or ViiA 7 Real-Time PCR System (Thermo Fisher Scientific) depending on sample numbers analyzed. Primer sequences are shown in Supplemental Table I.

Rainbow trout CD4-1, CD4-2a, CD4-2b, and LAG-3 were cloned from cDNA generated from rainbow trout thymocytes. The full length of the coding DNA sequences except the signal peptide of CD4-1 and the cDNA sequences encoding Ig domains for CD4-2a, CD4-2b, and LAG-3 were further subcloned into pDisplay vector (Thermo Fisher Scientific) containing a hemagglutinin A (HA) tag (for both CD4s and LAG-3) and a PDGFR transmembrane region (for CD4-2a, CD4-2b, and LAG-3). The aforementioned DNA constructs were further subcloned into a pMXs-IRES-Puro Retroviral Expression Vector (Cell Biolabs). Plat-E Retroviral Packaging Cell Lines (Cell Biolabs) were thereafter transfected with the pMXs vectors by using FuGENE HD Transfection Reagent (Promega). NRK-52E (Normal Rat Kidney cells; ATCC CRL-1571) were infected with the retrovirus produced from the transfected Plat-E cells. The resulting stable NRK cells expressing rainbow trout CD4-1 (CD4-1/NRK), CD4-2a (CD4-2a/NRK), CD4-2b (CD4-2b/NRK), and LAG-3 (LAG-3/NRK) were thereafter selected with puromycin (Invivogen). The expression of CD4-1, CD4-2a, CD4-2b, and LAG-3 on NRK cells was confirmed by flow cytometry with a murine anti-HA mAb (Sigma-Aldrich). With the goal to produce soluble CD4-2b, the cDNA sequence encoding the Ig domains of CD4-2b were also subcloned into pINFUSE-mIgG2b-Fc2 vector (Invivogen). CD4-2b/Fc fusion protein was produced with FreeStyle 293 Expression System (Thermo Fisher Scientific) according to the manufacturer’s instructions and then purified with HighTrap Protein G HP column (GE Healthcare). Rats were immunized with either 20 million CD4-1/NRK cells or 100 μg CD4-2b/Fc-fusion protein emulsified in TiterMax Gold (Sigma-Aldrich). Both preparations were injected into the tail-base muscle of different Sprague–Dawley rats. Two weeks after immunization, lymph node cells from the rat iliac lymph node were isolated as previously described (25, 35). Cell fusion of SP2/0 myeloma cells with rat lymph node cells and hybridoma culture were performed in the Cell Center Service Facility at the University of Pennsylvania. Of all hybridomas stably secreting Abs to CD4-1 or CD4-2, we chose three different hybridomas to characterize CD4-1– and CD4-2–bearing leukocytes in trout: clone 4.1.1 (recognizing CD4-1; rat IgG1 isotype), clone 4.1.2 (recognizing CD4-1; rat IgG2a isotype), and clone 4.2.12 (recognizing CD4-2b; rat IgG2b isotype). The isotypes of rat mAbs were determined with the Rat Ig Isotyping ELISA Kit (BD Pharmingen). For further validation of mAb specificity, we used the Amaxa Cell Line Nucleofector Kit L (Lonza) to transfect the trout embryo cell line (STE-137) with the same pDisplay vectors encoding rainbow trout CD4-1, CD4-2a, CD4-2b, and LAG-3. The transfection conditions were found according to the Amaxa Cell Line Nucleofector Kit instructions. Transfected STE-137 cells expressing the aforementioned molecules were then tested by flow cytometry for the reactivity of the mAbs to rainbow trout CD4-1 and CD4-2b. The expression of aforementioned molecules on the transfected STE-137 cells was confirmed with the simultaneous staining of the murine anti-HA mAb and rat anti–CD4-1 or anti–CD4-2b mAbs. Alexa Fluor 647–conjugated goat anti-mouse IgG1 and Alexa Fluor 488–conjugated donkey anti-rat IgG (H+L) (2 μg/ml; Jackson Immunoresearch) were used as secondary Abs.

The positive hybridomas for trout CD4-1 and CD4-2b were injected i.p. into nude mice to produce ascites (Cocalico Biologicals, Reamstown, PA). Alternatively, mAbs were also produced in Hybridoma-SFM (Thermo Fisher Scientific) in the Cell Center Service Facility at the University of Pennsylvania. The IgG fraction from the ascites and supernatant of the serum-free medium was purified using a HiTrap protein G and L column (GE Healthcare) according to the instructions of the manufacturer.

Trout leukocytes from lymphoid organs and blood were isolated as previously reported (32–34). In brief, blood was drawn from the caudal vein with a heparinized syringe and diluted in a DMEM (Life Technologies) supplemented with 1% FBS (Serum Source International) and 1% penicillin-streptomycin (Life Technologies). Trout organs (spleen, head kidney [HK], and thymus) were removed and pressed through a 100-μm cell strainer (Corning Life Sciences) and suspended in DMEM. The diluted blood and cell suspensions from spleen, HK, and thymus were layered onto 34/51% discontinuous Percoll (GE Healthcare) density gradients. After centrifugation (400 × g, 30 min), cells lying at the interface of the gradient were collected and washed with DMEM twice. The cell suspensions were kept on ice until further use. To initially identify trout CD4-bearing cells, we stained HK leukocytes with biotinylated rat anti–CD4-1 (clone 4.1.1; 5 μg/ml) and rat anti–CD4-2b (clone 4.2.12; 5 μg/ml) primary mAbs. Stained cells were detected with Brilliant Violet 421 Streptavidin (BioLegend; 1 μg/ml) and mouse anti-rat IgG2b-PE mAbs (Southern Biotech; 1 μg/ml). To elucidate whether trout CD8α⁺ cells expressed CD4-1⁺ and/or CD4-2⁺, we stained leukocytes with biotinylated rat anti–CD4-1 (clone 4.1.1; 5 μg/ml), rat anti–CD4-2b (5 μg/ml), and rat anti-CD8α (clone 13.2D; rat IgG2a isotype; 5 μg/ml) (35) primary mAbs. Stained cells were detected with Brilliant Violet 421 Streptavidin (1 μg/ml; BioLegend), mouse anti-rat IgG2b-PE (1 μg/ml; Southern Biotech), and mouse anti-rat IgG2a-Alexa Fluor 647 (1 μg/ml; Southern Biotech) mAbs. To elucidate whether trout B cells expressed CD4-1⁺ and/or CD4-2⁺, we stained leukocytes with rat anti–CD4-1 (clone 4.1.2; 5 μg/ml), rat anti–CD4-2b (5 μg/ml) primary mAbs in combination with either biotinylated mouse anti-IgM (clone 1.14; mouse IgG1 isotype; 1 μg/ml) (36) or biotinylated mouse anti-IgT (clone 41.8; mouse IgG2b isotype; 1 μg/ml) (32) primary mAbs. Stained leukocytes were thereafter detected with mouse anti-rat IgG2a PE (1 μg/ml; eBioscience), mouse anti-rat IgG2b eFluor 660 (1 μg/ml; eBioscience) mAbs, and Brilliant Violet 421 Streptavidin. Cell suspensions were incubated on ice with primary mAbs and corresponding secondary conjugates for 30 and 15 min, respectively, and then were washed twice with DMEM after each respective staining step. As controls, we used biotinylated rat IgG1 (clone RTK2071; 5 μg/ml; BioLegend), rat IgG2a (clone eBR2a; 5 μg/ml; eBioscience), rat IgG2b (clone RTK4530; 5 μg/ml; BioLegend), biotinylated mouse IgG1 (clone MOPC-21; 1 μg/ml; BioLegend), and biotinylated mouse IgG2b (clone MPC-11; 1 μg/ml; BioLegend) isotype-matched control mAbs. Flow cytometry was performed using a BD LSRFortessa cell analyzer (BD Biosciences). Alternatively, stained cells were sorted for cytology and gene expression analyses with BD FACSAria II flow cytometer (BD Biosciences) as previously reported (32, 33). Data on flow cytometry were analyzed using FlowJo software (Tree Star).

Sorted cells from HK leukocytes were spun onto poly-l-lysine–coated slides (Newcomer Supply) by using a Shandon Cytospin. Cytospin preparations were stained with Wright–Giemsa-like (WG; Hema 3 from Thermo Fisher Scientific), myeloperoxidase (MPO), Sudan Black B (SBB), naphthol AS-D chloroacetate esterase (NCAE), and β-glucuronidase (BG) staining (Sigma-Aldrich) according to the manufacturer’s instructions. The cells were imaged with Eclipse E600 (Nikon).

Spleen leukocytes were stained with biotinylated rat anti–CD4-1 (clone 4.1.1; 5 μg/ml) and rat anti–CD4-2b (5 μg/ml) primary mAbs followed by labeling with Alexa Fluor 488 Streptavidin conjugates (2 μg/ml; Jackson Immunoresearch) and mouse anti-rat IgG2b Alexa Fluor 647 mAbs (2 μg/ml; Southern Biotech) as described earlier. Stained cells were then fixed with fixation buffer (BioLegend), counterstained with DAPI (BioLegend), and mounted on slides with Fluoromount-G (Southern Biotech) according to the manufacturer’s instructions. The cells were imaged with Eclipse E600 (Nikon).

NRK cells and NRK transfectants were lysed with Laemmli sample buffer. The lysed samples were resolved on 4–15% SDS-PAGE Ready Gel (Bio-Rad) under reducing conditions. The gels were transferred onto Sequi-Blot PVDF membranes (Bio-Rad). The membranes were blocked with 8% skim milk and incubated with anti–CD4-1 (clone 4.1.2; 2 μg/ml) or anti–CD4-2b (2 μg/ml) mAbs followed by incubation with peroxidase-conjugated anti-rat IgG (GE Healthcare). Immunoreactive bands were visualized using the HyGLO Chemiluminescent HRP Ab Detection Reagent (Denville Scientific Products).

For the repertoire analysis of CD4⁺ T cell populations, we isolated whole HK leukocytes of healthy adult rainbow trout (∼50 g) and sorted their CD4 double-positive (DP) and CD4-2 single-positive (SP) lymphocytes. Total RNA from whole HK leukocytes was purified and DNase treated using the RNeasy Mini Kit (Qiagen). RNA (2 μg) was reverse transcribed into cDNA using Superscript II Reverse Transcriptase (Thermo Fisher Scientific) with 2.5 mM oligodT₂₅ primer in a final volume reaction of 20 μl. For sorted CD4-2 SP and CD4 DP cells, total RNA was extracted from each cell population using RNeasy Micro Kit (Qiagen). Full-length cDNA was generated and amplified using the SMARTer PCR cDNA synthesis kit (Clontech Laboratories), following the manufacturer’s instructions. The optimized protocol of this kit preferentially enriches for full-length cDNAs and retains true gene representation of genes in the original sample. The optimal number of PCR cycles determined for double-stranded cDNA synthesis and amplification of CD4-2 SP and CD4 DP samples was 21 cycles.

The spectratyping of TCR β V region (TRBV) CDR3 length (Immunoscope analysis) was performed as previously described (37). Primer sequences for the spectratyping of TRBV CDR3 length are shown in Supplemental Table I. The repertoire diversity can be assessed by a diversity score based on the concept of Shannon entropy that provides a measure of the quantity of information encompassed in the repertoire (37, 38). Repertoire diversity scores were used to perform principal component analysis (PCA) to compare the statistical dispersion of the samples on a multidimensional plan. Statistical and multivariate analyses were performed using R software (http://www.r-project.org/).

Importantly, we verified that low values of diversity index computed for the CD4-2 SP subset, compared with the CD4 DP, were not due to the smaller number of sorted CD4-2 SP cells. To do so, we followed two approaches: 1) cDNA used for spectratyping was synthesized from different amounts of RNA, allowing to compare profiles obtained from similar amount of RNAs from CD4-2 SP and CD4 DP cells; and 2) spectratyping was always repeated from independent initial cDNA amplifications to verify that skewed profiles, mainly observed in the CD4-2 SP subset, were not due to random picking of template molecules. Only reproducible profiles, hence corresponding to templates concentrated enough to avoid such artifact, were considered in our analysis.

To determine the proliferative responses of CD4⁺ lymphocytes to mitogens, we labeled spleen and HK leukocytes with CellTrace Violet (Life Technologies) according to manufacturer’s instruction. Labeled cells were plated in a 96-well round-bottom microplate (Corning) and cultured with either 1 μg/ml leukoagglutinin PHA-L (Sigma-Aldrich), a mix of E. coli 0111:B4 LPS and E. coli 055:B5 LPS (Sigma-Aldrich) (1:1 ratio, dose 100 μg/ml), or DMEM containing 10% FBS, 1% penicillin-streptomycin (Life Technologies), and 0.1% gentamicin (Life Technologies). Leukocytes from blood, spleen, and HK of outbred trout from the ARS-Fp-C strain, which are produced in NCCCWA (39), were isolated as described earlier to examine one-way MLR. With the goal of performing the MLR, we also collected blood leukocytes of outbred trout from the Fish Lake DeSmet strain, which are produced at the Egan State Fish Hatchery, Utah Division of Wildlife Resources (40). Spleen and HK leukocytes from ARS-Fp-C trout were labeled with CellTrace Violet and were used as responder cells. Leukocytes from blood of Fish Lake DeSmet and ARS-Fp-C trout strain were used as allogeneic and autologous stimulator cells, respectively. To neutralize the proliferative ability of these stimulator cells, we incubated leukocytes with 50 μg/ml mitomycin C from Streptomyces caespitosus (Sigma-Aldrich) for 30 min and washed them three times with DMEM. The same cell number (5 × 10⁵ cells) of responder and stimulator cells was cocultured in a 96-well round-bottom microplate. For Ag-specific proliferation of CD4⁺ cells, fish were first immunized with 200 μg hemocyanin, keyhole limpet (KLH), Megathura crenulata (EMD Millipore) emulsified in Freund’s complete adjuvant (Sigma-Aldrich). Three weeks after primary immunization, fish were boosted with the same amount of Ag emulsified in Freund’s incomplete adjuvant (Sigma-Aldrich). Ten days after boosting, spleen and HK leukocytes from immunized fish were isolated and labeled with CellTrace Violet. Labeled cells were incubated with either 100 μg/ml KLH, 100 μg/ml OVA (Sigma-Aldrich) as unrelated control protein, or DMEM alone (as culture medium control). All incubations were carried out in a humidified atmosphere of 5% CO₂ at 20°C for 7 d. For all proliferation experiments, after 7-d incubation of cells with stimulants, proliferation of CD4⁺ cells was assessed by their staining with anti–CD4-1 (clone 4.1.2; 5 μg/ml) and anti–CD4-2b (5 μg/ml) primary mAbs followed by labeling with mouse anti-rat IgG2a PE (1 μg/ml; eBioscience) and mouse anti-rat IgG2b eFluor 660 (1 μg/ml; eBioscience) mAbs as described earlier. 7-Aminoactinomycin D (1 μg/ml; Thermo Fisher Scientific) was added to the cell suspension for the detection and exclusion of dead cell. The percentage of proliferating CD4⁺ cells was analyzed by flow cytometry by the dye-dilution method according to the manufacturer’s instructions (Thermo Fisher Scientific).

Y. ruckeri strain YRNC10 was grown in brain heart infusion broth and agar (Becton Dickinson). Rainbow trout (∼100 g) were challenged by i.p. injection of a dose of ∼2 × 10⁵ CFUs in PBS. Viable CFUs were determined by plate counts. Control fish were injected with the same volume of PBS (200 μl). Four days after challenge, splenic leukocytes were sampled and sorted as described earlier.

Phagocytosis experiments were assessed as previously reported by us (32, 33). In brief, HK leukocytes were isolated as described earlier. Cells (2 × 10⁵ cells/well) were plated and incubated with 1.0 μm Fluoresbrite Yellow Green Microspheres (Polysciences) at 1:10 (cell/bead) ratio. Alternatively, cells were incubated with pHrodo Red E. coli BioParticles conjugates (Thermo Fisher Scientific). The mixture of cells and the phagocytic targets were placed in a humidified atmosphere of 5% CO₂ at 18°C. After 3-h incubation, cells were collected and noningested beads and E. coli were removed by centrifuging cells over a 3% BSA-4.5% glucose gradient. After washing with cold PBS, leukocytes were stained with rat anti–CD4-1 (clone 4.1.2; 5 μg/ml) and rat anti–CD4-2b (5 μg/ml) mAbs followed by labeling with mouse anti-rat IgG2a PerCP-eFluor 710 (1 μg/ml; eBioscience) and mouse anti-rat IgG2b eFluor 660 (1 μg/ml; eBioscience) mAbs as described earlier, and analyzed by flow cytometry in a FACSCanto (BD Biosciences). For immunofluorescence microscopy, HK leukocytes were incubated with 1.0 μm FluoSpheres crimson fluorescent microspheres (Thermo Fisher Scientific) at 1:10 (cell/bead) ratio. After 6-h incubation, noningested beads were removed with the aforementioned method. Leukocytes were labeled with biotinylated anti–CD4-1 mAbs (clone 4.1.1; 5 μg/ml) and Alexa Fluor 488 Streptavidin conjugates (2 μg/ml; Jackson Immunoresearch). Stained cells were then fixed and were imaged as described earlier.

Paired or unpaired Student t test (Excel version 14.0; Microsoft), nonparametric Mann–Whitney U test, and one-way ANOVA with Bonferroni correction (Prism version 6.0; GraphPad) were used for analysis of differences between groups. The p values <0.05 were considered statistically significant.

In addition to CD4-1, salmonid fish contain two cd4-2 genes (cd4-2a and cd4-2b), which are probably derived from salmonid-specific genome duplication, sharing ∼80% amino acid identity (18, 41, 42). To study the surface expression of CD4-1 and CD4-2 molecules in trout leukocytes, we produced mAbs against CD4-1 and CD4-2b that specifically recognized transfected mammalian and trout cells expressing surface CD4-1 and CD4-2b, respectively (Supplemental Fig. 1A, 1C, 1D). The anti–CD4-1 mAbs did not cross-react with CD4-2–transfected rat and trout cell lines, and vice versa, the anti–CD4-2b mAbs did not cross-react with CD4-1–transfected cells (Supplemental Fig. 1A, 1C, 1D). Western blotting with mammalian transfectants further verified the specificity of the anti–CD4-1 and anti–CD4-2b mAbs, which specifically recognized the protein with expected molecular size from CD4-1 and CD4-2b transfectants, respectively (Supplemental Fig. 1B). Moreover, none of these mAbs cross-reacted with trout CD4-2a or LAG-3 (Supplemental Fig. 1A–D). However, because cd4-2a transcripts were exclusively expressed in lymphocytes expressing CD4-2b molecules (see later), the cells expressing surface CD4-2b are hereafter described as CD4-2⁺ cells.

Flow cytometry of whole HK leukocytes double-stained with anti–CD4-1 and anti–CD4-2b mAbs revealed three distinct leukocyte populations: CD4-1⁺/CD4-2⁻ (CD4-1 SP), CD4-1⁻/CD4-2⁺ (CD4-2 SP), and CD4-1⁺/CD4-2⁺ (CD4 DP) subsets (Fig. 1A; isotype-matched control mAbs, Supplemental Fig. 1E). CD4 DP and CD4-2 SP cells were localized within the lymphocyte population (Fig. 1B, top and middle panels) and displayed typical lymphocyte morphology (Fig. 1C, top and middle panels). In contrast, CD4-1 SP cells were found within the myeloid cell population (Fig. 1B, bottom panel) and displayed a morphology resembling that of monocyte/macrophage-like cells (Fig. 1C, bottom panel). Interestingly, we could not detect lymphocytes exclusively expressing surface CD4-1.

We next focused our studies on the two CD4⁺ lymphocyte populations (CD4 DP and CD4-2 SP). Both lymphocyte subsets lacked the expression of surface IgM, IgT, and CD8α, indicating that CD4 DP and CD4-2 SP lymphocytes are neither B nor CD8α⁺ T cells (Fig. 1D, 1E). Importantly, CD4 DP cells comprised the main CD4⁺ lymphocyte subset (Fig. 1D, 1E, bottom panel). To characterize further the phenotype of the CD4 DP and CD4-2 SP lymphocyte subsets and confirm their identity as T cells, we assessed the expression of a variety of B and T cell–related genes in four sorted lymphocyte populations: CD4 DP, CD4-2 SP, CD8α⁺, and CD4-1⁻/CD4-2⁻/CD8α⁻ negative lymphocytes (Neg). In agreement with the staining by the anti–CD4-1 and anti–CD4-2b mAbs (Fig. 1A, 1D, 1E), we confirmed that cd4-1 transcripts were only detected in CD4 DP lymphocytes, whereas cd4-2 transcripts (cd4-2a and cd4-2b) were expressed in both CD4 DP and CD4-2 SP subsets (Fig. 1F). As expected, transcripts defining CD8⁺ T cells (cd8a and cd8b) and B cells (membrane-bound form of ighm and ight) were absent in both CD4 DP and CD4-2 SP lymphocytes (Fig. 1F, 1G). Furthermore, CD4 DP, CD4-2 SP, and CD8α⁺ lymphocytes expressed similar degrees of key T cell marker genes (tcra, tcrb, cd3g/d, and lck1), whereas as expected, the Neg population expressed very low levels of these transcripts (Fig. 1H).

We found CD4 DP and CD4-2 SP lymphocytes in all tissues examined, whereas negligible numbers of CD4-1 SP lymphocytes were detected (Fig. 2A). As expected, there was a large percentage of total CD4⁺ lymphocytes (including CD4 DP and CD4-2 SP cells) among thymic lymphocytes (∼74.5%), whereas a significant percentage was also observed in the lymphocyte gate from spleen (∼10.5%) and HK (∼22.6%) (Fig. 2A, 2D). The frequency of CD4 DP and CD4-2 SP lymphocytes in blood leukocytes was very low (∼1.4%) (Fig. 2A, 2D). CD4 DP cells were the predominant population of CD4⁺ lymphocytes in all tissues examined and accounted for ∼83–91% of the total CD4⁺ lymphocyte count (Fig. 2E). Similar to HK (Fig. 1D, 1E), both CD4 DP and CD4-2 SP cells in blood, spleen, and thymus were devoid of surface IgM and IgT expressions (data not shown). As observed in chicken and most mammals, except for the thymus, no CD8α surface expression was detected on CD4⁺ lymphocytes in any organs analyzed (Fig. 2B, 2C). In contrast, ∼74 and ∼35% of CD4 DP and CD4-2 SP thymocytes, respectively, displayed CD8α expression on their surface (Fig. 2B, 2C, rightmost panel). Overall, eight distinct thymocyte populations could be detected: CD4 DP/CD8α⁺ (∼46.4%); CD4 DP/CD8α⁻ (∼16.3%); CD4-2 SP/CD8α⁺ (∼4.9%); CD4-2 SP/CD8α⁻ (8.9%); CD4-1SP/CD8α⁺ (∼2.1%); CD4-1SP/CD8α⁻ (∼1.3%); CD4 double-negative/CD8α⁺ (∼9.8%); and CD4 double-negative/CD8α⁻ (∼10.3%) thymocytes (Fig. 2F). Immunofluorescence microscopy studies showed that most of the CD4-1 and CD4-2 staining in both splenic CD4 DP (Fig. 2G) and CD4-2 SP (Fig. 2H) lymphocytes appeared punctuated and unevenly distributed in vesicle-like structures not present in cells stained with isotype-matched control mAbs (Fig. 2I). More importantly, for the most part we did not observe colocalization of CD4-1 and CD4-2 staining in CD4 DP cells, thus indicating the lack of a potential association or complex formation between surface CD4-1 and CD4-2 molecules (Fig. 2G).

To gain insight into the TCRβ diversity expressed by the CD4⁺/CD8α⁻ T cell subsets identified in this study, CD4 DP and CD4-2 SP lymphocytes were sorted from HK leukocytes of healthy adult trout, and their TCRβ repertoires were analyzed using CDR3 spectratyping. Similar analyses were also carried out on whole HK leukocytes from the same individual fish to discern the potential TCR peculiarities of the two CD4⁺ T cell subsets. To this end, TCRβ CDR3 length profiles were produced and analyzed for the four trout TCRVβ families that have previously been shown to have the highest expression levels (i.e., TCRVβ-1, -2, -3, and -7) (43).

Overall, bell-shaped profiles typically consisting of five to eight peaks were observed for the four TCRVβ families in CD4 DP T cells, as well as in whole HK leukocytes (Fig. 3A). In contrast, skewed distributions were observed for CD4-2 SP cells, indicating that the repertoires of the sorted CD4-2 SP and CD4 DP subpopulations were distinct. A diversity index based on a modified Shannon index (38) was computed from each Vβ-Jβ profile, and a PCA was performed (Fig. 3B). The PCA projection according to the first two components (40.39 and 15.96% global variability, respectively) clearly distinguishes CD4-2 SP from whole HK leukocytes and CD4 DP lymphocytes. The projection also indicates that interindividual variability was much higher for CD4-2 SP cells. The heat map of diversity index computed from Vβ-Jβ profiles (Fig. 3C) reveals a higher diversity index for TCRβ profiles from whole HK and CD4 DP cells. In that regard, for all four tested Vβ, the left panel of Fig. 3C corresponding to CD4-2 SP cells is significantly lighter (i.e., less red) when compared with the other two panels, thus illustrating the lower diversity of the TCRβ repertoire expressed by this subset.

To further investigate the functional differences between CD4 DP and CD4-2 SP lymphocytes, we studied their proliferative responses to mitogens, MLR, and Ag-specific stimulation. There was a remarkable proliferation of both CD4 DP and CD4-2 SP subsets in spleen and HK in response to the T cell mitogen PHA when compared with the negligible proliferation observed upon stimulation with PBS or the B cell mitogen LPS (Fig. 4A–C). Similarly, significant proliferative responses of both CD4 DP and CD4-2 SP cells were observed in response to alloantigen stimulation (Fig. 4D, 4E). Importantly, the extent of proliferation in response to PHA and alloantigen stimulation was significantly higher in splenic CD4 DP cells than in CD4-2 SP cells (Fig. 4C, 4E). For Ag-specific proliferative responses, whole leukocytes isolated from spleen and HK of fish immunized with KLH were incubated with either KLH or OVA. Both splenic CD4 DP and CD4-2 SP cells showed significant proliferative responses in the presence of KLH, but not OVA or culture medium alone (Fig. 4F, 4G). In contrast, no Ag-specific proliferative responses were detected in the CD4⁺ cell subsets of the HK (Fig. 4G).

To assess the ability of trout CD4 DP and CD4-2 SP cells to produce cytokines, we evaluated the mRNA expression of T cell cytokines in fish infected with Y. ruckeri (44). At the pathogen dose used in this study, maximal inflammation can be detected 3–5 d postinfection (45). Thus, spleen leukocytes from control and infected fish were isolated 4 d postinfection and sorted into four lymphocyte populations: CD4 DP, CD4-2 SP, CD8α⁺, and CD4-1⁻/CD4-2⁻/CD8α⁻ lymphocytes (Neg). In line with previous reports (29, 46), whole spleen leukocytes (WSL) in infected fish showed higher expression levels of il-2, il-10, ifng, il-17a/f-1a, il-21b, and il-22 cytokines when compared with those in control fish, whereas the expression levels of il-4/13a and il-21a remained unchanged (Fig. 5). Transcript levels of both CD4 DP and CD4-2 SP lymphocytes in infected fish showed a similar expression pattern to WSL, although the overall levels (except for il-2) were much higher (Fig. 5). Moreover, except for ifng, the highest expression levels of these upregulated cytokines were consistently detected in the sorted CD4⁺ lymphocyte populations. In fact, most of the analyzed cytokines were not upregulated in either CD8α⁺ or Neg lymphocytes except for il-10a and ifng, which were upregulated in CD8α⁺ lymphocytes. Importantly, no differences in the cytokine expression pattern were found between CD4 DP and CD4-2 SP lymphocytes.

Although the presence of CD4-1 SP lymphocytes was negligible (Figs. 1A–E, 2A–C), we were able to observe a significant population of myeloid cells expressing surface CD4-1, but not CD4-2 (Fig. 1A–C). To further characterize the CD4-1⁺ myeloid cells, we sorted them and compared their expression of myeloid and T cell lineage genes with that of CD4-1⁻ myeloid cells, as well as with CD4⁺ and CD4⁻ lymphocyte populations. Expression analysis of myeloid cell markers revealed that CD4-1⁺ myeloid cells expressed high transcript levels of monocyte/macrophage markers, including mcsfra/b (alias csf1ra/b), mpeg1, and lyz (Fig. 6A). In contrast, the same cells expressed negligible amounts of mpo, a neutrophil marker (alias mpx). Conversely, CD4-1⁻ myeloid cells expressed high levels of mpo while expressing very low to negligible transcript levels of the macrophage markers. Instead, gcsfr (alias csf3r), which is present on a variety of mammalian myeloid cells, was highly expressed in both CD4-1⁺ and CD4-1⁻ myeloid cells. Importantly, the expression of high levels of cd4-1 transcripts was detected in both CD4-1⁺ myeloid cells and CD4⁺ lymphocytes (Fig. 6A), being in agreement with their cell-surface staining by the anti–CD4-1 mAb (Fig. 1A–C). However, CD4-1⁺ and CD4-1⁻ myeloid cells completely lacked the expression of TCR coreceptors (cd4-2a, cd4-2b, cd8a, cd8b, and lag3) and pan T cell markers (tcra, tcrb, cd3g/d, lck1, and lck2) (Supplemental Fig. 2). These expression analyses strongly suggest that CD4-1⁺ myeloid cells are monocytes/macrophages, whereas CD4-1⁻ myeloid cells comprise neutrophils.

Similar to mammalian monocytes/macrophages and neutrophils, trout CD4-1⁺ and CD4-1⁻ myeloid subsets are composed of large cells (high forward scatter [FSC]) with low and high granularity (low and high side scatter), respectively (Fig. 1B, bottom panel). To further validate the identity of these two myeloid cell populations, we sorted CD4-1⁺ and CD4-1⁻ myeloid cells from HK and subjected them to several cytochemical stains, including WG, MPO, SBB, NCAE, and BG stains. Sorted CD4-1⁺ myeloid cells mainly consisted of cells with kidney-shaped nucleus typical of monocytes, as well as cells with typical macrophage morphology (Fig. 6B, WG). Almost all CD4-1⁺ myeloid cells were positive for BG (∼97%) and NCAE (∼95%) staining while negative for MPO and SBB (Fig. 6B, 6D). In contrast, CD4-1⁻ myeloid cells included mostly polymorphonuclear cells that stained positively with MPO and SBB (∼80%), but also contained some cells with round or oval nuclei and dark blue cytoplasm (i.e., immature myeloblast cells) that stained negatively with MPO and SBB (Fig. 6C, 6D). Moreover, the CD4-1⁻ myeloid cells did not stain for BG and NCAE, whereas immature cells (blue cytoplasm) were very weakly positive for these stains (Fig. 6C, 6D). These results are consistent with reports that salmonid monocytes/macrophages are positive for BG and NCAE stains, whereas neutrophils are positively stained with MPO and SBB stains (47, 48). The distribution of CD4-1⁺ myeloid cells was determined by flow cytometry and revealed that significant percentages of CD4-1⁺ myeloid cells were found in the myeloid cell population of blood (∼14%) and HK (∼14.9%), whereas relatively low percentages of these cells were observed in spleen (∼6.4%) and thymus (∼2.8%) (Fig. 6E, gray bars). Because the percentage of all myeloid cells is lower among leukocytes from spleen and thymus when compared with that of blood and HK, the percentage of CD4-1⁺ myeloid cells among whole leukocytes is much higher in the HK (∼4.9%) and blood (∼0.9%) than in the spleen (∼0.14%) and thymus (∼0.12%) (Fig. 6E, white bars).

Both CD4-1⁺ and CD4-1⁻ myeloid cells were able to phagocytose latex beads and E. coli (Fig. 7A, Supplemental Fig. 3A). However, a higher percentage of phagocytic cells was found within the CD4-1⁺ myeloid subset (∼87.5 and ∼77% of the cells ingested beads and E. coli, respectively) when compared with CD4-1⁻ myeloid cells (∼56.9 and ∼42% of the cells ingested beads and E. coli, respectively) (Fig. 7B, Supplemental Fig. 3B). Internalization of beads by CD4-1⁺ myeloid cells was further confirmed by immunofluorescence microscopy (Fig. 7C; isotype-matched control mAbs, Fig. 7D). Moreover, it is worth noting that the phagocytic capacity of CD4-1⁺ myeloid cells was higher than that of the CD4-1⁻ myeloid population because a higher percentage of CD4-1⁺ myeloid cells (∼47.9%) was able to ingest a large number of beads (six or more) when compared with CD4-1⁻ myeloid cells (∼19.7%) (Fig. 7E). Similarly, CD4-1⁺ myeloid cells engulfed more pHrodo E. coli when compared with phagocytic CD4-1⁻ myeloid cells (Supplemental Fig. 3C). In contrast, we found that CD4⁺ lymphocytes had a negligible phagocytic capacity (∼3.3 and ∼2.2% of the cells ingested beads and E. coli, respectively) (Fig. 7A, 7B, Supplemental Fig. 3A, 3B). In conclusion and as summarized in Fig. 8, CD4-1⁺ monocytes/macrophages represent the myeloid population containing the highest percentage of phagocytic cells as well as the cells with the uppermost phagocytic capacity.

CD4⁺ cells are well characterized in mammals and birds; however, the origins and primordial roles of CD4⁺ cells in nontetrapod species are not well defined. In contrast with the single CD4 gene present in tetrapods, two CD4 genes have been identified in teleost fish (13, 14, 30). In this study, we generated and validated mAbs against rainbow trout CD4-1 and CD4-2 that were used to identify three previously unrecognized fish CD4⁺ leukocyte populations in peripheral lymphoid organs: a predominant lymphocyte subset coexpressing surface CD4-1 and CD4-2 (CD4 DP lymphocytes), a minor lymphocyte population uniquely expressing surface CD4-2 (CD4-2 SP lymphocytes), and a significant myeloid population with CD4-1 surface expression. In stark contrast with all other analyzed classes of vertebrates, the concurrent or unique presence of CD4-1 and CD4-2 molecules on trout CD4⁺ T cells defines novel subpopulations of CD4⁺ T cells.

The percentage of total trout CD4⁺ (CD4 DP and CD4-2 SP) cells in the lymphocyte gate was very low in the blood (∼1.4%) but moderate in systemic lymphoid organs (spleen [∼10.5%] and HK [∼22.6%]). At this point, the abundance of trout CD4⁺ T cell subsets cannot be compared with that of other fish because of the lack of reagents to detect both surface CD4-1 and CD4-2 molecules in all other teleost species. In humans, CD4⁺ T cells comprise ∼50, ∼20, and ∼20% of lymphocytes in blood, spleen, and bone marrow, respectively (49), whereas in chicken they represent ∼45 and ∼9% of their blood and spleen lymphoid cells, respectively (5). Thus, when comparing trout and tetrapods, the percentages of CD4⁺ T cells in central lymphoid organs fall within a similar range, but their proportions in blood is significantly larger in tetrapods than in trout (50). Nevertheless, it is worth pointing out that the low percentage of CD4⁺ cells in blood leukocytes shown in this article is in agreement with a report describing the reactivity of an anti-trout CD3ε Ab that stained only ∼2.5% of all trout blood leukocytes (51).

The thymus is the primary lymphoid organ specialized in T cell development in jawed vertebrates (52). In all studied tetrapods, the majority of thymocytes consist of CD4⁺/CD8α⁺ cells, whereas all other lymphoid organs mainly harbor T cells with single CD4 or CD8α surface expression (2, 5, 53). In this article, we identified up to eight trout thymocyte subpopulations on the basis of their CD4-1, CD4-2, and CD8α surface expression. Overall, the subset with surface CD4 (CD4-1⁺ and/or CD4-2⁺) and CD8α represented the majority of trout thymocytes. Therefore, like in mammals, trout CD4⁺/CD8α⁺ cells are likely to represent the main thymic T cell progenitors. In other teleost fish, the presence of CD4⁺/CD8α⁺ cells has only been reported in ginbuna carp, where CD4-1⁺/CD8α⁺ cells comprise 15–35% of all thymocytes (25). Because ginbuna carp thymocytes are also likely to contain CD4-2⁺/CD8α⁺ cells, this would increase the overall percentage of the CD4⁺/CD8α⁺ population.

Teleosts express most of the cytokines involved in Th responses, although some relevant Th cytokines have not yet been found (i.e., IL-9 and IL-25) (29, 30). In support, teleost fish CD4-1⁺ cells have been found to express Th cytokines upon stimulation with TLR ligands, mitogens, or specific Ag (22, 24). However, to our knowledge, our study is the first to show that upon challenge with a pathogen (Y. ruckeri), CD4⁺ lymphocytes produce high transcript levels of cytokines involved in Th1, Th17, and Treg responses. Interestingly, both CD4 DP and CD4-2 SP cells expressed similar levels of cytokine transcripts, suggesting that the activation mechanisms involved in their induction are similar. Importantly, of all leukocyte populations tested, the CD4 DP and CD4-2 SP subsets had the highest expression of Th17, Treg, and il-2 cytokines, whereas ifng was similarly produced by both the CD4⁺ and CD8α⁺ lymphocytes. Thus, as in mammals (3, 54), these data strongly suggest that CD4⁺ lymphocytes in trout are the major producers of the cytokines in response to bacterial challenge.

The few studies assessing the teleost TCR repertoire have generally been undertaken from whole tissues (i.e., spleen), making it impossible to distinguish the respective features of the repertoires expressed by CD4⁺ or CD8⁺ T cells (13). In this study, we performed, to our knowledge, the first repertoire analysis of CD4⁺ T cell populations in a nontetrapod species. Our data show that the repertoire of the CD4-2 SP subset is significantly less diverse than that of CD4 DP T cells. Overall, this observation supports the idea that clones belonging to these two subsets may have been subjected to different selection pressures. More particularly, our repertoire data could be explained by mechanistic constraints linked to the CD4/MHC class II interaction, or by a particular developmental pathway of CD4-2 SP T cells that selects clones with lower diversity and possibly predefined specificities. With regard to the latter possibility, CD4-2 SP lymphocytes may be functionally similar to mammalian T cells with invariant TCRs (i.e., invariant NK T cells and mucosa-associated invariant T cells) (55). Accordingly, a subpopulation of CD4-2 SP T cells might harbor limited or invariant TCRs that can interact with teleost nonclassical MHC class I molecules. In addition, we cannot rule out the possibility that the CD4-2 SP subset consists of a compendium of amplified clones from previous immune responses that have been kept in a nonactivated state. Future work is warranted to analyze the aforementioned hypotheses.

The proliferative responses of CD4⁺ T cells by T cell mitogen, alloantigen, and specific Ag are a common readout of T cell function in tetrapods (e.g., mammals and birds). In this study, we found that both splenic and HK CD4 DP and CD4-2 SP subsets proliferated significantly in response to PHA, alloantigen, and Ag-specific stimulations, with the critical exception of HK CD4⁺ subsets, which were unable to proliferate upon Ag-specific stimulation. Thus, our T cell proliferation data provide strong support for the notion that, as in mammals (56), the trout spleen is the key systemic lymphoid organ where T cell–dependent adaptive immunity develops. In agreement with our finding, the spleen has long been hypothesized to be the main teleost secondary lymphoid organ (52). This belief is supported further by past findings showing that both clonal differentiation of teleost B cells after Ag encounter (38, 57) and early antiviral T cell responses (58) occur in the spleen. The capacity of CD4-2 SP cells to proliferate in an Ag-dependent manner suggests the capability of CD4-2 to interact with trout MHC class II. Future studies are warranted to analyze potential differences in the interaction of trout CD4-1 and CD4-2 with MHC class II. Importantly, in this study, we show that trout CD4 DP cells in spleen showed higher proliferative responses to alloantigen and PHA stimulations than did CD4-2 SP cells. The lower proliferative capacity of CD4-2 SP cells in response to alloantigen might correspond to a distinct state of differentiation, as suggested also by the restricted diversity of their TCRβ V repertoire. Moreover, because of their aforementioned distinct functional properties, it is conceivable that CD4-2 SP cells embody a vestigial CD4⁺ T cell subset, a hypothesis that is in line with their unique expression of CD4-2, a molecule reminiscent of the prototypic two-Ig-domain CD4 coreceptor.

In mammals, CD4 is also expressed on some cell populations of myeloid-lineage cells, although there exist large species-to-species disparities in terms of which myeloid types and what percentage of these cells express CD4 (6, 59, 60). The most consistent expression of CD4 in myeloid subsets across species is that observed on monocytes/macrophages and dendritic cells. Thus, it has been shown that varying percentages of monocytes/macrophages express CD4 in human (∼65–90%) and rat (∼97%), whereas mouse and chicken do not contain these cells (5, 6). In contrast, the presence of CD4⁺ dendritic cells has been shown in several mammals, although they consistently represent a very small percentage of their leukocytes (0.1–1%) (60, 61). Importantly, because CD4 expression in monocytes/macrophages lacks in birds, the question remains whether the presence of CD4 in myeloid-lineage cells is the result of a recent evolutionary event that took place either in mammals or in species preceding the emergence of tetrapods. In this study, cytological and gene expression analyses led to the conclusion that myeloid cells expressing surface CD4-1 were monocytes/macrophages. Interestingly, no myeloid cells with CD4-2 expression were identified. CD4-1 expression in ginbuna carp and fugu was only found in the lymphocyte population, although the possibility still exists that CD4-2 in these species may be found in myeloid cells (22, 23, 25). However, zebrafish cd4-1 and cd4-2 transcripts appear to be expressed at similar levels in lymphocyte and myeloid cell populations (24). To our knowledge, our data represent the first description of monocytes/macrophages with surface CD4 expression in a nonmammalian species, thus suggesting that CD4 expression on the surface of these cells is the result of an ancient evolutionary event preceding the emergence of tetrapods.

To date, the specific roles of mammalian CD4⁺ monocytes/macrophages remain to be fully elucidated. In this article, we found that CD4-1⁺ myeloid cells represented the myeloid population with the highest phagocytic activity and capacity. It will be interesting in the future to analyze whether mammalian CD4⁺ monocytes/macrophages contain also such high phagocytic capabilities. In contrast, it is notable that we found the vast majority of CD4⁺ lymphocytes to be nonphagocytic. Similarly, it has been reported that ginbuna carp CD4⁺ T cells are also nonphagocytic (62). In contrast, we and others have reported large subsets of phagocytic B cells in fish and several tetrapod species, including mammals (31–33, 62–64). Thus, it would appear that from an evolutionary viewpoint, the phagocytic capacity of lymphocytes is mostly restricted to B cells.

In conclusion, this study fills in an important gap in the knowledge of teleost CD4-bearing leukocytes, thus revealing critical insights into the evolutionary origins and primordial roles of CD4⁺ T cells and CD4⁺ monocytes/macrophages (summarized in Fig. 8). The current and future studies on these lymphoid and myeloid CD4-bearing cells are also likely to provide clues for identifying new roles of these cells not only in fish but also in higher vertebrate species, and thus contribute to the development of therapies involving these critical cell types. Importantly, because our knowledge on CD4⁺ T cell responses in teleosts is very scarce, our findings will be critical for the design of more effective vaccines for fish that induce strong effector CD4⁺ T cell responses.

We thank Jeffrey S. Faust and the staff of the Flow Cytometry Facility (Wistar Institute) for the cell-sorting procedures, Sabine Baxter of the Cell Center Service Facility (University of Pennsylvania) for the production and maintenance of hybridoma cell lines, Dr. Gregory D. Wiens (NCCCWA, Agricultural Research Service, U.S. Department of Agriculture) for the provision of rainbow trout and Y. ruckeri, Scott LaPatra (Clear Springs Foods, Inc.) for the provision of rainbow trout, and Dr. Bruce Freedman and Dr. Gordon Ruthel (Penn Vet Imaging Core, University of Pennsylvania) for technical assistance and advice provided for the immunofluorescence microscopy. We thank Dr. Uwe Fischer (Friedrich-Loeffler-Institut) for the provision of the anti-trout CD8α mAb. The Fish Lake DeSmet line of rainbow trout was generously provided to the NCCCWA by Don Bone from Egan State Fish Hatchery, Utah Division of Wildlife Resources.

The authors have no financial conflicts of interest.