Introduction

Cripto-1 (CR1) is a transmembrane oncofetal protein that was initially isolated from an embryonic teratocarcinoma and described as an oncogene capable of transforming normal fibroblasts into malignant cells1. The CR1 gene is located on chromosome 3p21-p3 and the protein contains 188 amino acids2. CR1 acts as a membrane co-receptor for Nodal, growth derived factor 1 (GDF1) and GDF3, all of them belonging to the EGF/TGFβ family3.

Numerous studies have demonstrated the influence of CR1 on many cancers, including melanomas4, non-small cell lung cancers5, gastric6 and colorectal7 tumors, breast cancer8, pancreatic cancer9, testicular cancer10, and urological tumors11, among others. Apparently, the mechanism linking CR1 to cancer relies on its function as a growth factor, the ability to induce epithelial to mesenchymal transition, and its influence on tumor stem cells12. This indicates that CR1 may be a good target for anticancer therapies and, indeed, some studies have demonstrated that targeting CR1 reduces tumor growth13,14.

Furthermore, CR1 plays important roles on stem cell regulation. For instance, CR1 is essential for retaining stem cell pluripotency15, is involved in muscle cell regeneration16 and wound healing17, and many of these actions are dependent upon its binding to myosin II18. In addition, CR1 has also been described to modulate tumor stem cell biology19,20.

Regarding developmental functions, a CR1 homolog was described in mice where it was shown to play important roles during early development21. CR1 has been implicated in proper uterine decidualization22, embryo implantation23, correct orientation of the anterior-posterior axis of the embryo24, and correct gastrulation25 and postgastrulation morphogenesis26, among others. Although some studies have shown the expression of CR1 in the placenta27,28, no reports can be found in the literature about the distribution of this protein in the umbilical cord or during fetal development.

Interestingly, several Cripto “pseudogenes” have been described in different regions of the genome29. Among these, only Cripto-3 (CR3) has been shown to be translated into a protein30. The gene for CR3 is located in the X chromosome (Xp21-22) and the protein also contains 188 amino acids31. This sequence differs from that of CR1 by only six amino acids that are dispersed over the whole molecule. This feature implies that most available antibodies32 are not able to discriminate between CR1 and CR3, making it difficult to ascribe specific functions to one protein or the other. To circumvent this issue, our group has recently developed monoclonal antibodies that recognize specifically either CR1 or CR3, with no crossreactivity between them33. These discriminating antibodies open the door to new anatomical and physiological approaches to identify specific roles for these interesting proteins.

The aim of the present study was to characterize the distribution of CR1 and CR3 in the human placenta, umbilical cord, and the tissues of human fetuses.

Materials and methods

Primary antibodies

Primary mouse monoclonal antibodies against CR1 (NCI 5G1-1, RRID: AB_3674310) and CR3 (NCI 5G11-2, RRID: AB_3674311) were selected among a plethora of monoclonal antibodies generated against specific regions of the proteins. They were thoroughly characterized by surface plasmon resonance, Western blotting, ELISA, and immunohistochemical methods33.

Human tissues

The study was performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards.

Placentas (n = 3) and umbilical cords (n = 3) from normal deliveries (40 ± 1.0 weeks of gestational age) were obtained from the Obstetrics Department of the San Pedro University Hospital (Logroño, Spain). The study was approved by the Medical Research Ethics Committee of La Rioja (CEImLAR, protocol number PI-820) and parents signed the informed consent before inclusion. Tissues were fixed for 24 h in 10% buffered formalin, dehydrated, and paraffin embedded. Section (3 μm-thick) were attached to positively charged slides and one of the sections was stained with hematoxylin and eosin for anatomical reference.

In addition, a tissue array containing different human normal fetal organs (typically n = 5 per tissue type), from 13 cases ranging from 3 to 7 months of age, of both sexes, was purchased from US BioMax (Derwood, MD, USA), with catalog number BE01015. The array included 50 cores. Samples from heart, lung, brain, kidney, blood vessels, esophagus, pancreas, liver, sclera, retina and choroid, testicle, and thymus were represented. The Company warrantees that “All tissue is collected under the highest ethical standards with the donor being informed completely and with their consent. We make sure we follow standard medical care and protect the donor´s privacy. All human tissues are collected under HIPPA approved protocols”.

Immunohistochemistry protocol

Tissue sections were dewaxed with xylene and intrinsic peroxidase activity was blocked with 3% H2O2 in methanol. Then, slides were rehydrated through an ethanol series and antigen retrieval was achieved with citrate buffer, pH 6.0, for 20 min at 95 °C. Normal donkey serum (10% in PBS) was added for 60 min and, then, the primary mouse monoclonal antibody (CR1 or CR3) was applied overnight at 4 °C, at a 1:750 dilution in PBS. The next day, after washing, the Novolink Polymer Detection System (Leica Biosystems, Cat# RE7200-CE, RRID: AB_3674357) was used to detect the primary antibodies. This kit provides a ready-to-use rabbit-anti mouse IgG that was added for 60 min. After another set of washes, a polymer that contains both anti-rabbit antibodies and peroxidase molecules, which is also part of the kit, was applied for a further 60 min. Then peroxidase activity was detected with the Liquid DAB + Chromogen System (Agilent, Santa Clara, CA, USA) and a slight nuclear counterstaining was achieved with hematoxylin. Slides were observed and recorded with a Leica DM 6000 B microscope equipped with a digital camera (Leica, L’Hospitalet de Llobregat, Barcelona, Spain).

Specificity controls

The substitution of the primary antibody by PBS and solid-phase preabsorption of the antibody34,35,36 in serial sections were used as specificity controls. Briefly, solid-phase absorption controls were performed with the synthetic peptides used to produce the antibodies33. Polystyrene tubes (VWR International, Radnor, PA, USA) were coated with 2 µg synthetic peptide in 2 mL PBS for 2 h rotating at room temperature, blocked with 1% BSA in PBS for 2 h at room temperature, and then incubated with the optimally diluted antibody overnight at 4 °C. Then, the pre-absorbed antibody was added to the sections. In parallel, antibodies pre-absorbed in BSA-coated tubes were used as a positive control.

Western blotting

Small placenta tissue pieces (n = 3) were lysed in Pierce RIPA Buffer (Invitrogen) supplemented with Complete Protease Inhibitor Cocktail (Roche Diagnostics, Basel, Switzerland). The protein content of the lysates was quantified using the Bradford Protein Assay (BioRad, Hercules, CA, USA). Nupage sample reducing buffer (Invitrogen) was added to the lysates (30 µg of total protein), heated at 70 °C for 10 min, and separated by SDS polyacrylamide gel electrophoresis in 4–12% Bis-Tris gels (Invitrogen). Electrophoresed proteins were transferred to PVDF membranes using the iBlot 2 Transfer Device (Invitrogen). Once transferred, the membranes were blocked using 5% (w/v) non-fat dry milk in Tris-buffered saline-Tween-20 (TBST; 25 mM Tris, pH 7.5, 150 mM NaCl, 0.1% (v/v) Tween-20). After three TBS washes, the membrane was cut into three columns, each one containing the three specimens. The first column was exposed overnight to the CR1 antibody (at 1:1000 dilution), the second to CR3 (at a 1:1000 dilution), and the third to just the vehicle (with no primary antibody as a negative control). On the next day, a secondary anti-mouse antibody linked to horseradish peroxidase (HRP) (715-035-151, Jackson Immunoresearch, West Grove, PA, USA) was used at a 1:20,000 dilution. Then, peroxidase activity was detected with the NZY Advanced ECL Western Blotting Detection Reagent (MB40201, Lisbon, Portugal) and captured in a ChemiDoc MP Imaging System (BioRad). After development, membranes were blocked again and exposed to a 1:10,000 antibody against GAPDH (ab8245, Abcam, Cambridge, UK) as a loading control. The same secondary anti-mouse antibody linked to HRP was used to detect the GAPDH-positive bands. Quantification of the immunoreactive bands was accomplished with the ImageLab software (BioRad).

Results

Placenta and umbilical cord

Placenta

In the term placentas, CR1 immunoreactivity was found in the cytoplasm of endothelial and vascular smooth muscle cells and in the nuclei of all trophoblasts, including cytotrophoblasts, syncytiotrophoblasts, and extravilleous trophoblasts or giant cells. Furthermore, the nuclei of some fibroblasts were also immunoreactive (Fig. 1a). On the other hand, CR3 was located in the cytoplasm of all trophoblasts (Fig. 1b). Interestingly, CR3 immunoreactivity was also found in the decidua´s connective extracellular matrix surrounding the giant cells as well as in the blood serum (Fig. 1b), suggesting that CR3 may play a role as a secretory protein in placenta. Omission of the primary antibody resulted in a total lack of staining, confirming lack of non-specific binding for the detection reagents (Fig. 1c).

Fig. 1
figure 1

Representative microphotographs of a normal human placenta immunostained for CR1 (a), CR3 (b) and in the absence of primary antibody as a negative control (c). After immunostaining, hematoxylin was applied as a counterstain. Scale bar = 200 μm.

Full size image

To further confirm the specificity of the staining, preabsorption controls were performed. Exposure of the antibodies to solid-phase attached peptides resulted in a major reduction of the immunostaining signal for both anti-CR1 and anti-CR3 (Fig. 2).

Fig. 2
figure 2

Preabsorption controls with solid-phased peptides. Representative microphotographs of human placenta immunostained with anti-CR1 (a), anti-CR1 preabsorbed with the CR1 synthetic peptide (b), anti-CR3 (c), and anti-CR3 preabsorbed with the CR3 synthetic peptide (d). Scale bar = 200 μm.

Full size image

In addition, a Western blot was performed on three placenta specimens with anti-CR1 and anti-CR3 (Fig. 3). Different patterns were found for these antibodies with anti-CR1 presenting major bands at about 55 and 66 kDa with two smaller bands of approximately 115 and 125 kDa, whereas anti-CR3 labeled specific bands at 58 and 82 kDa with a fainter band at 32 kDa. Incubation of the membrane in the absence of primary antibody was used as a negative control (Fig. 3). Similar total protein loading was confirmed with an antibody against the house-keeping protein GAPDH (Fig. 3). Full-length blots are shown in Supplementary Figs. S1 and S2.

Fig. 3
figure 3

Western blot of three individual placentas (P1, P2, P3) exposed to antibodies against CR1, CR3 and in the absence of primary antibody as a negative control (Control). A molecular weight ladder can be faintly seen between CR1 and CR3. The weights of the molecular markers are indicated in kDa. The same membrane was exposed to an antibody against the house-keeping protein GAPDH as a loading control.

Full size image

Umbilical cord

In the umbilical cord, CR1 immunoreactivity was restricted to the endothelial cells of the main arteries (Fig. 4a) and vein (Fig. 4d); a few nuclei among the smooth muscle cells of the large vessels (Fig. 4a,d) and the Wharton´s jelly (Fig. 4g); and the cord´s epithelium (Fig. 4g). In the stratified epithelium, most nuclei in the basal and medium layers were positive for CR1 whereas the nuclei in the most apical layer were negative (Fig. 4g inset). A faint cytoplasmic staining was also found throughout the basal and medium layers of the epithelium.

Immunostaining for CR3 appeared also in the endothelial cells (Fig. 4b,e) but, in contrast with CR1, staining in the vascular smooth muscle cells was widespread and clearly cytoplasmic in the main arteries (Fig. 4b). On the other hand, the staining for CR3 in the single vein´s smooth muscle cells was less intense (Fig. 4e) than in the arterial wall. The cytoplasmic staining for CR3 was also very clear among fibroblasts in the Wharton´s jelly and in the epithelium surrounding the organ (Fig. 4h and inset). No nuclear staining was found for CR3. Absence of the primary antibody in serial sections confirmed staining specificity (Fig. 4c,f,i).

Fig. 4
figure 4

Representative microphotographs of a normal human umbilical cord immunostained for CR1 (a,d,g) and CR3 (b,e,h) and in the absence of primary antibody as a negative control (c,f,i). Details of one of the main arteries (a–c), the single vein (d–f), and the umbilical epithelium and Wharton´s jelly (g–i) are shown. Red arrows point to fibroblast nuclei. Scale bar = 200 μm (for a–i); =100 μm (for the insets).

Full size image

Fetal tissues

Furthermore, tissue arrays were immunostained with antibodies against CR1 and CR3. Samples of heart, brain, liver and thymus had no significant staining for either antibody.

Retina

The retinas from 5-month-old fetuses showed a diffuse staining for CR1 that was more intense on the perinuclear cytoplasm of neurons of the ganglion cell layer (GCL) and the inner nuclear layer (INL) (Fig. 5a) and a more intense staining for CR3. Interestingly, the CR3 antibody strongly labeled the cytoplasmic processes of the Müller glia cells (blue arrowheads in Fig. 5b), and especially the outer limiting membrane (OLM), which is also formed by the cytoplasm of the Müller cells (red arrows in Fig. 5b). In addition, and similar to the CR1 pattern, some perinuclear cytoplasm of neurons on the GCL and the INL were also positive for CR3 (Fig. 5b). Staining specificity was demonstrated by lack of labeling in the negative controls (Fig. 5c).

Fig. 5
figure 5

Representative microphotographs of a 5 month-old fetal retina (a–c) and testis (d–f) immunostained for CR1 (a,d), CR3 (b,e) and in the absence of primary antibody as a negative control (c,f). Blue arrowheads signal the cytoplasmic processes of Müller cells. Red arrows indicate the OLM. Some layers of the retina are indicated in (c) for anatomical reference: Retinal pigment epithelium (PE), outer nuclear layer (ONL), inner nuclear layer (INL) and ganglion cell layer (GCL). Scale bars = 100 μm (for a–c); =200 μm (for d–f).

Full size image

Testicle

At 5–6 months of intrauterine life, the testes are still rather immature and are characterized by testicular cords, which are the precursors of seminiferous tubules, surrounded by a thick interstitium with primordial Leydig cells37. The CR1 antibody provided a diffuse staining of the testicular cords and of some endothelial cells (Fig. 5d). The CR3 antibody labeled also the cytoplasm of most cells in the testicular cords but marked strongly the cytoplasm of the germ cells (spermatogonias), including some dividing germ cells. Some of the interstitial cells displayed also a faint immunoreactivity for CR3 (Fig. 5e). The negative control demonstrated staining specificity (Fig. 5f).

Lung

Fetal lungs showed an intense cytoplasmic staining for both CR1 and CR3 in endothelial cells and type II pneumocytes in the parenchyma (Fig. 6a,b). On the other hand, the epithelium of the bronchioles was very faint for CR1 but clearly positive for CR3 (Fig. 6a,b). Some immune cells infiltrating the bronchiolar epithelium were immunoreactive for both CR1 and CR3 as well (Fig. 6a,b). Staining specificity was demonstrated by lack of labeling in the negative control (Fig. 6c).

Fig. 6
figure 6

Representative microphotographs of a 5 month-old fetal lung (a–c) and a 6 month-old fetal esophagus (d–f) immunostained for CR1 (a,d), CR3 (b,e) and in the absence of primary antibody as a negative control (c,f). Red arrows in (e) point out a patch of ciliated epithelial cells. Scale bar = 200 μm.

Full size image

Esophagus

In the esophagus, the CR1 antibody labeled the epithelial cells, and the nuclei of the smooth muscle layer and endothelial cells (Fig. 6d). In the epithelium, the staining was both nuclear and cytoplasmic, being more intense in the outer layers (stratum corneum). On the other hand, the CR3 antibody did not label the smooth muscle cell nuclei but their cytoplasm, and the distribution in the epithelium was different, with an intense cytoplasmic staining in the basal layer (stratum basale) and the stratum spinosum (Fig. 6e). Furthermore, an area of columnar ciliated epithelium showed an intense CR3 cytoplasmic immunoreactivity (red arrows in Fig. 6e). Staining specificity was demonstrated by lack of labeling in the negative controls (Fig. 6f).

Pancreas

In the pancreas, apart from some inflammatory and endothelial cells, CR1 immunoreactivity was found mainly in the cytoplasm of the cells forming the islets of Langerhans. The more intense labeling occurred in cells on the islet´s periphery while the central cells (insulin-producing β-cells) had a fainter staining (Fig. 7a). In contrast, CR3 provided a more general and intense staining pattern throughout the islets, indicating that the β-cells may express more CR3 than CR1 (Fig. 7b). No immunoreactivity was found in the exocrine pancreas, including the acini and the ductal system. The negative control demonstrated staining specificity (Fig. 7c).

Kidney

In the kidneys, the developing convoluted tubules were the structures showing a stronger immunoreactivity for both CR1 and CR3, while the glomeruli were negative (Fig. 7d–f).

Fig. 7
figure 7

Representative microphotographs of a 5 month-old fetal pancreas (a–c) and kidney (d–f) immunostained for CR1 (a,d), CR3 (b,e) and in the absence of primary antibody as a negative control (c, f). Scale bar = 200 μm.

Full size image

Discussion

In this study, we have shown that the immunoreactive patterns for CR1 and CR3 are not identical in normal human placenta, umbilical cord, and fetal tissues, suggesting that these closely related proteins may play different roles during development and fetal growth.

Previous studies have described the distribution of Cripto in human placenta using antibodies that cannot distinguish between CR1 and CR3. These studies focused on the expression of Cripto in extravillous cytotrophoblasts38 and syncytiotrophoblasts27, showing in both cases a cytoplasmic pattern. Other authors have described the expression of Cripto in human placenta through molecular methods such as RT-PCR and Western blotting28. Finally, other researchers have focused on the mouse placenta. For instance, Natale et al. did not find Cripto expression in mouse trophoblasts at any stage of development39. In clear contrast, a recent knockout study has found that eliminating CR1 from the placenta led to smaller implantation sites, reduced placenta thickness, and a reduction in the number of spongiotrophoblasts and syncytiotrophoblats40. In our study, we found abundant immunoreactivity for CR1 and CR3 in all trophoblasts. Nevertheless, in our case, the staining pattern for CR1 was nuclear whereas for CR3 it was clearly cytoplasmic in these cells. This fact suggests that the antibodies used by the Sao Paulo´s Group27,38 were detecting CR3 rather than CR1.

In the endothelial cells of placental blood vessels, we found a clear immunoreactivity for CR1 that was weaker for CR3. This feature was not described in previous studies on the placental distribution of Cripto. The higher expression of CR1 over CR3 immunoreactivity in endothelial cells was already described by our group in tumor tissues33 and clearly supports the hypothesis that the reported angiogenic functions of Cripto41 may rely mainly in the CR1 protein.

Another interesting finding was the strong CR3 immunoreactivity we found in the blood´s serum and the abundant extracellular matrix of the decidua. This was not a staining background problem since the negative controls were clean. Cripto is a cell membrane-associated protein with a short hydrophobic carboxy terminus acting as an anchor42. Removal of this anchor domain by phospholipase D creates a soluble form of biologically-active Cripto43 and, although there are not many studies on the differential functions of attached versus free Cripto, it seems that CR3 may have a similar biology and that the placenta may be a relevant playground for the biological functions of soluble CR3.

To better understand the immunoreactivity of these antibodies, we performed a Western blot with protein extracts from three placentas. Different band patterns were found for the antibodies. These results are in agreement with those previously published in cancer cell lines where a main band of ≈ 50 kDa was described for CR1 and of ≈ 32 kDa for CR333. It seems that the placenta has a much more complex pattern of post-translational modifications of these proteins than the tumor cell lines, something that should be expected from an organ as complex as the placenta.

The distribution of Cripto in the umbilical cord had not been described before. We found a clear difference between the distribution of CR1 that was mainly nuclear and that of CR3 that was cytoplasmic, both in the vascular smooth muscle cells and in the fibroblasts. We are not sure what that differential distribution means from a physiological perspective but it shows that both proteins may have diverse functions in these cells. We also found a big difference between the staining pattern of the umbilical arteries and the single vein, with the latter having a lower CR3 expression in its muscle wall, suggesting that this protein may be involved in the differentiation fate of blood vessels into either arteries or veins. No much is known about the function of Cripto in the development of blood vessels, but an early study found that Cripto expression decreased over time in a human embryoid body model44.

Another distinctive feature of the umbilical cord is its epithelium, which is in contact with the amniotic fluid. Depending on the location, this epithelium may be either single-layered with squamous or cuboidal cells or stratified with 4–5 layers of cuboidal cells45. Here, again, we found mainly nuclear staining for CR1 and a clear cytoplasmic pattern for CR3 that was stronger in the basal layer. Of course, the basal layer contains the epithelium´s stem cells and Cripto has been shown to block TGFβ1-mediated keratinocyte senescence46, so we may expect similar functions in the umbilical cord´s epithelium.

The human retina develops over many months both in utero and postnatally. By the fifth month (week 20) of intrauterine life, the retina already shows five well-defined layers (out of the 10 that it will eventually contain), although the outer plexiform layer (OPL) is very immature compared to the inner plexiform layer (IPL). The OPL would complete its development by no sooner than week 3547. In our study, we found that both CR1 and CR3 are found in the perinuclear cytoplasm of neurons in the GCL and INL. It is well known that TGFβ signaling has an antiproliferative effect on neural cells48 and Cripto may be playing a role in maintaining retinal neurons in their postmitotic state. We also found an intense staining for CR3 in the cytoplasm of the Müller cells, the most abundant glial cell in the retina. Müller cells and the retinal neurons develop from a common precursor49. Müller cells have a high degree of plasticity and retain proliferation potential in response to different insults. This may result in gliosis, a pathological accumulation of Müller cells similar to scar tissue50. Gliosis is just a form of fibrosis and TGFβ signaling has been related to fibrosis formation in many organs51. It is tempting to propose that the high levels of CR3 in these cells may somehow be involved in this process and, perhaps, may offer a novel clinical target to reduce the negative impact of gliosis in vision loss conditions. Another feature related to Müller cells in our study was the intense staining for CR3 in the OLM. The function of the OLM may be related with mechanically maintaining the structure of the retina, but the presence of numerous junction proteins at this level suggest a clear involvement with the formation of the blood-ocular barrier52. TGFβ signaling is closely related to cell adhesion modulation53 and the presence of high levels of CR3 in this important region of the retina may be involved in controlling the permeability of the blood-ocular barrier.

Testis development is a complex process. The genital ridges appear at about 4–5 weeks of gestation as a thickening of the intermediate mesoderm. The proliferating primordial germ cells start migrating at 4 weeks and enter the genital ridges by the 5th week, where they become enveloped by differentiating Sertoli cells within the forming testicular cords. This process is completed by the 7th week of pregnancy54. The fetal germ cells of the testis are known as gonocytes or prespermatogonias, and several subpopulations of fetal germ cells have been identified55,56. Holstein et al. called gonocytes the cells from younger fetuses (8–13 weeks) but in more mature fetuses (13–22 weeks) these cells had different features and were classified as fetal spermatogonias57. In our study, we were working with 5–6 month (20–24 week)-old fetuses, so the cells with an intense staining for CR3 should be identified as fetal spermatogonias. These cells were also stained by the CR1 antibody but with lower intensity, indicating that CR3 may have a more relevant role in spermatogonia development. Nodal, Cripto, and the downstream modulator SMAD2, are activated in testicular germ cells but not in somatic cells of the testis, indicating the existence of an autocrine Nodal signaling in these cells58. On the other hand, this signaling pathway is not activated in female germ cells59. Cripto has been implicated in maintaining stem cell pluripotency and lineage restriction15. These characteristics are very relevant for early spermatogonia function. Also, fetal germ cells are supposed to be the origin of testicular germ cell tumors, and ectopic expression of Cripto has been described in these malignancies60. Up to now, all these functions of Cripto were assumed to be led by CR1 but our current results indicate that CR3 may play a more fundamental role in male germ cell development. This concept merits further investigation.

TGFβ signaling regulates multiple cellular processes in the lungs, including growth suppression of epithelial cells, alveolar epithelial cell differentiation, fibroblast activation, and extracellular matrix organization61. In our study, we found that both CR1 and CR3 are present in endothelial cells and type II pneumocytes where they may be involved in cell differentiation. On the other hand, the epithelial cells of the bronchioles were positive for CR3 but not for CR1, suggesting that the function of blocking epithelial cell growth in this organ may be exclusively dependent on CR3. The same distribution was also described in the adult lung33, indicating that this pattern is established early during fetal development and stays that way through adult life.

In the esophagus, we also found a distinct staining for CR1 and CR3 in both the smooth muscle cells of the muscularis mucosae and in the squamous epithelium. It should be noted that the epithelial cells closer to the surface (more mature) had more CR1 immunoreactivity whereas the more basal cells were predominantly positive for CR3. As already discussed above, the presence of CR3 in the basal layer may be due to some regulation of the stemness function of these cells. We also found an area of ciliated cells on the surface of the epithelium. During development, the human esophageal mucosa exhibits a stratified columnar ciliated epithelium that appears at about 10 weeks. This epithelium is substituted by the regular stratified squamous epithelium at about 20–25 weeks. However, the replacement is progressive and it is not uncommon to find patches of ciliated epithelium for some time62. These ciliated cells were strongly positive for CR3 and this may be related with their early developmental stage.

In fetal pancreas, we found immunoreactivity for CR1 and CR3 in the developing islets of Langerhans and some blood vessels, but not in the ductal or acinar structures. This is in agreement with a previous study that described a lack of Cripto in human cell lines derived from normal pancreatic ducts63. The presence of Cripto in adult pancreatic islets is somewhat controversial, with some authors describing its immunoreactivity64 while others found no Cripto expression65. There is another reference in mice, where the authors describe Cripto staining in the fetal but not in the adult pancreas. They also describe a strong Cripto expression when they induce pancreatic regeneration66. Interestingly, it has been shown that Nodal, which is expressed by β- (insulin producing) and α- (glucagon producing) cells of the islets, is involved in β-cell proliferation, differentiation and viability65. Thus, these facts indicate that the Nodal-Cripto pathway may be critical for the correct development of the endocrine pancreas. It is worth noting that CR1 was more strongly expressed in peripheral cells of the islets that probably represent α-cells. In human pancreatic fetal development, the first insulin expression is detected by the 6th −7th week of gestation and expression of the other hormones appears shortly thereafter67. It is already known that Cripto has a relevant function in the development of pancreatic cancers64 and pancreatitis9, but nothing is known about its potential link to diabetes. Given the specific pattern of CR1 and CR3 in the fetal pancreas, it would be interesting to investigate this relationship.

In the fetal kidney, we found that both CR1 and CR3 were located in the epithelial cells of the convoluted tubules whereas the glomeruli were devoid of staining. To the best of our knowledge, no detailed description of the expression of Cripto in normal adult human kidney is available, but it has been shown that kidney tumors express higher levels of Cripto that normal tissues, with high Cripto levels being a biomarker for poor survival68. It has been demonstrated that EGF has direct physiological effects on the convoluted tubules69, so the presence of Cripto proteins in these structures is not unexpected. In any case, further studies are needed to better understand the functional relevance of Cripto proteins in the kidney, especially during the fetal period.

In this study, we have used a tissue array to perform immunohistochemical staining for two specific antibodies. The use of tissue arrays has both advantages and disadvantages70. A clear advantage is the fact that a single slide allows for the simultaneous staining of many specimens with exactly the same exposure to all reagents. On the other hand, this simultaneity may be counterproductive if some of the sections may have needed a longer exposure to DAB or other reagents, thus resulting in some false negative results. Therefore, we cannot confirm that the fetal tissues that did not stain for either CR1 or CR3 in our study (fetal heart, brain, liver and thymus) are completely devoid of these molecules, and further studies are needed to ascertain this point.

Cripto has been described to be active both as a membrane-bound protein and as a soluble form that can be found and quantified in diverse bodily fluids42. In this study we found that CR1, but not CR3, can be also found in the cell nucleus in a very cell-type dependent manner, something that had not been reported before, even in our previous paper on tumor tissues where no nuclear staining was found33, suggesting that this feature may not be present in tumors. Proteins that translocate to the nucleus possess nuclear locator sequences (NLS), which consist of short stretches of positively charged amino acids71. Both CR1 and CR3 have a putative NLS at amino acids 47–49 (RDD) and, intriguingly, the epitope used to generate our CR1 specific antibody (L44-R66) includes this region33. Furthermore, one of the six amino acid differences between CR1 (40-RGYLAFRDD−50) and CR3 (40-RGDLAFRDD−50) occurs nearby and this local molecular interaction may be the basis for the different intracellular distribution of CR1 versus CR3. Although molecular studies are needed to confirm the implication of these motifs in the nuclear translocation of CR1, this observation opens new possibilities, such as a potential function of CR1 acting as a nuclear transcription factor, that merit further investigation.

In summary, our study shows that, despite having a very similar amino acid sequence, CR1 and CR3 display a very diverse anatomical distribution based on immunohistochemical analyses with highly discriminating monoclonal antibodies. Of course, this diversity needs to be confirmed by molecular and physiological studies to conclude that CR1 and CR3 have fully independent functions. One of the best approaches to identify the functional properties of novel proteins is to generate knockout models in rodents. Conditional knockout models are available for the mouse homolog of CR122 but, unfortunately, Cripto “pseudogenes” are not well studied in mice. When available, mouse CR3 knockouts will be great tools to elucidate the functional aspects of this understudied protein.