**3. Results**

#### *3.1. The Trophoblast of Chorionic Villi in Human Placentae Expressed E-cadherin and a Wide Array of TJ Proteins*

In this study, we analyzed the expression of TJ proteins in the placentae from control and ZIKV-infected women. The clinical data and perinatal outcomes of control and ZIKV-infected women during pregnancy and their newborns are summarized in Tables 1 and 2. With regards to the newborn of ZIKV-infected donor D8, that carries a 22q11 deletion, all the reported phenotypic characteristic of the child conform with 22q11 deletion syndrome.


**Table 1.** Clinical data of women with ZIKV during pregnancy and physical findings of their newborns.

> GW, Gestational week; HC, Head circumference; F, Female.

**Table 2.** Clinical data of control women during pregnancy and physical findings of their newborns.


GW, Gestational week; F, Female; M, Male; HC, Head circumference.

In the trophoblast of chorionic villi in placentae derived from control and ZIKV-infected women during pregnancy, we explored the expression of the AJ protein E-cadherin and of the following TJ proteins: claudins -1, -2, -3, -4, -5, -7, and -10; JAMs -A, -B, and -C; occludin, ZO-1, and ZO-2 (Table 3). All these proteins, with the exception of claudins -2, -10, and JAM-A (data not shown), were expressed in the STB of chorionic villi.


**Table 3.** Expression of E-cadherin and TJ proteins in the trophoblast of placental chorionic villi derived from control and ZIKV-infected women.

BD, barely detected; ND, not detected. −, absent; ±, barely detected; +++ highly abundant; =, no change; ↓, decrease.

#### *3.2. E-cadherin Stained the Basal Membrane of the STB and Its Expression was not A*ff*ected in ZIKV-Infected Placentae*

In chorionic villi of human placentae, E-cadherin was observed not at the apical surface of the STB layer in contact with the intervillous space, which in situ is occupied by maternal blood, but in the basal membrane of the STB in contact with remnants of the CTB layer and the chorionic mesenchyma (Figure S1a). In this and most of the subsequent images, the trophoblast layer was identified with cytokeratin 18, a canonical marker of epithelial cells [28]. The intensity of the fluorescent signal of E-cadherin in the STB layer was not affected in placentae derived from ZIKV-infected women (Figure S1b). Likewise, by Western blot, no difference in the amount of E-cadherin was observed in placental lysates from control and ZIKV-infected women (Figure S1c,d).

#### *3.3. The Expression of Claudin-1 Slightly Increased in Placentae Derived from ZIKV-Infected Women*

By immunofluorescence, we observed that claudin-1 stained fetal endothelia within the chorionic villi parenchyma as well as the STB layer. The relative mean fluorescence intensity of the signal in the STB layer slightly increased in placentae from ZIKV-infected women in comparison to control placentae, albeit not at a significant level (Figure S2a,b). By Western blot, we observed that the amount of claudin-1 was not significantly higher in placentae derived from ZIKV-infected women in comparison to control placentae (Figure S2c,d).

#### *3.4. The Expression of Claudin-3 was Strong in Placental Vessels and Faint at the STB Layer of Both Control and ZIKV-Infected Placentae*

*Claudin-3* was strongly expressed in the vessels of the parenchyma in chorionic villi, while staining at the STB cell layer was weak in both control and ZIKV-infected placentae (Figure S3a,b). By Western blot, we detected no difference in the amount of claudin-3 between control and ZIKV-infected placentae (Figure S3c,d).

#### *3.5. The Expression of Claudin-4 at the Basolateral Membrane of STB Diminished in ZIKV-Infected Placentae*

In the chorionic villi of control human placentae, claudin-4 was observed at the basolateral membrane of the STB in contact with the chorionic parenchyma. The same pattern was observed in the placentae from ZIKV-infected women; however, in this case, the fluorescent signal at the STB layer was less intense (Figure 1a,b). Western blot analysis confirmed the decreased expression of claudin-4 in ZIKV-infected placentae in comparison to those of the control (Figure 1c,d).

**Figure 1.** Claudin-4 present at the basal surface of the STB layer diminished in ZIKV-infected placentae. (**a**) Frozen sections of human placentae derived from women infected or not (control) with ZIKV were processed for immunofluorescence with a rabbit antibody against claudin-4 (Cl-4) and a mouse antibody anti cytokeratin 18 (K18). DNA of nuclei was stained with DAPI. Apical surface of STB cell layer (arrow); basolateral surface of STB cell layer (arrowhead); intervillous space (asterisk). Bar, 50 μm; magnification bar, 25 μm. (**b**) Measurements of mean fluorescence intensity of the trophoblast layer were done on three independent images from each condition. Since the variances were equal among the two groups, we applied a Student's *t*-test. To include effect size estimation, since both groups have similar standard deviations, we used Cohen's *d*. The three values obtained per donor are represented by dots with the same color. \*\* *p* = 0.0054, Cohen's *d* = 3 indicating that 99.9% of the values from the ZIKV group are below the mean value of the control group. (**c**) Representative Western blot of three independent experiments. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was employed as loading control. (**d**) Densitometric analysis of Western blots. The three values obtained per donor are represented by dots with the same color. \*\* *p* = 0.00092 Cohen's *d* = 2.67 indicating that 99.9% of the values from the ZIKV group are below the mean value of the control group.

#### *3.6. Claudin-5 Strong Expression in the Vessels of Chorionic Villi and Faint Stain in STB Cells Did not Change with ZIKV Infection*

Claudin-5 was previously identified as an endothelial claudin [29], and accordingly, strongly stained the vessels within the chorionic parenchyma. In addition, claudin-5 faintly stained the STB cell layer in both ZIKV-infected and in control placentae (Figure S4a,b). Western blot analysis also revealed that ZIKV infection induced no change in the amount of claudin-5 expressed in the placentae (Figure S4c,d). The Western blot was done with a placental lysate, and hence, contains claudin-5 from both the vessels and the STB. In the case of claudin-5, this situation is particularly relevant, as this claudin is more abundant in vessels than in the STB cell layer.

#### *3.7. Claudin-7 Expression at the Basal Surface of STB Cells was not Altered by ZIKV Infection*

In the chorionic villi of control and ZIKV-infected placentae, claudin-7 was preferentially expressed in the basal membrane of STB, although a faint staining was also detected at the apical surface (Figure S5a,b). Immunofluorescence quantitation of claudin-7 at the STB cell layer and Western blot analysis showed no change in claudin-7 abundance in ZIKV-infected placentae (Figure S5c,d).

#### *3.8. Occludin Expression in the STB was Low and not A*ff*ected by ZIKV Infection*

In chorionic villi, occludin was strongly expressed in parenchymal vessels and faintly stained the STB cell layer in both control and ZIKV-infected placentae (Figure 2a,b). Western blot analysis showed a decrease in occludin expression in response to ZIKV infection (Figure 2c,d). These Western blots, however, most likely reflect the amount of occludin present in endothelial TJs rather than in the STB cell layer.

#### *3.9. The Expression of ZO-1 was Higher in Chorionic Vessels than in the STB and was not A*ff*ected by ZIKV Infection*

ZO-1 strongly stained the vessels in the chorionic villi and delineated the STB layer in a moderate manner, although in some placentae from both control (D19) and ZIKV-infected women (D10), labeling at the STB was more intense (Figure S6a,b). Western blot analysis revealed a similar amount of ZO-1 in both group of placentae, and most likely reflects the content of ZO-1 at endothelia, due to the higher expression of this protein in vessels in comparison to the STB (Figure S6c,d).

#### *3.10. In Chorionic Villi, the Expression of ZO-2 in the STB was Much Higher than in the Mesenchymal Vessels*

*ZO-2* in the chorionic villi of human placentae was clearly expressed in the STB, while it was barely present in the vessels of the mesenchyme. By immunofluorescence, only in one ZIKV-infected placentae (D8), the amount of ZO-2 increased in comparison to the rest of the placentae (Figure S7a,b) and the Western blot analysis showed no change in ZO-2 expression induced by ZIKV infection (Figure S7c,d).

#### *3.11. ZIKV Infection Had no Impact on the Expression of JAMs -B and -C in the STB of Chorionic Villi*

JAM-B stained with a similar intensity the STB cell layer of both control and ZIKV-infected placentae. Dots of JAM-B were also present in the chorionic mesenchyme, but not along the TJs of chorionic vessels, which were conspicuously stained with ZO-1 (Figure S8a,b). By Western blot, similar amounts of JAM-B were found between control and ZIKV-infected placentae (Figure S8c,d). In contrast, JAM-C was not observed at the STB cell layer and profusely stained the endothelial vessels in the chorionic mesenchyme. In the control and ZIKV-infected placentae, the same JAM-C staining pattern was observed (Figure S9a,b), and no change in the amount of JAM-C was detected by Western blot after ZIKV infection (Figure S9c,d).

In summary, the exploration of E-cadherin and a wide variety of TJ proteins revealed that E-cadherin and claudins -4 and -7 were present in the basal membrane of the STB, while all the other TJ

proteins studied localized in both the apical and basal membranes of the STB. The pattern of expression of these proteins was preserved in ZIKV-infected placentae, although in this pathological condition, the amount of claudin-4 diminished in the STB.

**Figure 2.** Occludin expression at chorionic vessels is stronger than in the cell layer. (**a**) Frozen sections of human placentae derived from women infected or not (control) with ZIKV were processed for immunofluorescence with a mouse antibody against occludin (Occ) and a rabbit antibody anti E-cadherin (E-cadh). DNA of nuclei were stained with DAPI. Chorionic vessels (arrows); STB (arrowhead); intervillous space (asterisks). Bar, 50 μm. (**b**)Measurements of mean fluorescence intensity of trophoblast layer were done on three independent images from each condition. (**c**) Representative Western blot of three independent experiments. GAPDH was employed as loading control. (**d**) Densitometric analysis of Western blots. \* *p* = 0.0286. The data were further analyzed to include effect size estimation with Glass'delta since each group had a different standard deviation. The value of Glass'delta = 1.955 indicates that 99.9% of the values from the ZIKV group are below the mean value of the control group.

#### *3.12. The STB Layer of Placentae of ZIKV-Infected Women is Permeable to Ruthenium Red*

Since claudin-4 functions as a cationic barrier [30] or an anion pore [31] that increases TER in cationic and anion selective cell lines and decreases paracellular permeability in the cationic selective cell line MDCK [32], we next determined whether the TJs of ZIKV-infected placentae were leaky. For this purpose, placental tissue was fixed and processed for TEM in the presence of ruthenium red, an electron-dense paracellular marker. In all the ZIKV-infected placentae, we observed ruthenium red staining in the paracellular pathway bellow the TJ region in the STB cell layer, in contrast to control placentae, where, as we had previously shown [9], staining was restricted to the apical membrane of the STB (Figure 3).

**Figure 3.** The STB in the placentae of ZIKV-infected women was permeable to ruthenium red. Placental tissue was fixed and processed for TEM in the presence of ruthenium red. Ruthenium red staining in the paracellular pathway (arrows). Bar, 1 μm.

#### *3.13. Hofbauer Cell Hyperplasia, an Increased Diameter of Microvilli and Intravillous Calcifications were Observed in ZIKV-Infected Placentae*

The chorionic villi derived from women infected with ZIKV during pregnancy displayed several histological alterations, including: (1) Hofbauer cell hyperplasia, evaluated by counting the number of CD68+ cells in the parenchyma of floating chorionic villi stained with hematoxylin (Figure 4a,b); (2) intravillous calcifications observed in hematoxylin and eosin-stained samples, which show a tendency to increase in ZIKV placentae in comparison to control, but whose difference is not statistically significant (Figure 4c,d); and (3) a higher diameter of chorionic microvilli in hematoxylin and eosin-stained samples (Figure 4e,f).

**Figure 4.** The chorionic villi of ZIKV-infected placentae display Hofbauer cell hyperplasia and a higher mean diameter. (**a**) Hofbauer cells (arrows) in chorionic villi were detected with an antibody against CD68 in slides stained with haematoxylin. Bar, 100 μm. (**b**) The number of Hofbauer cells was evaluated counting CD68+ positive cells in five optical fields per placenta. Each dot corresponds to the mean value of Hofbauer cells/field present in each placenta. \* *p* = 0.0294. (**c**) Calcification (arrow) present in a chorionic villus detected in a slide stained with haematoxylin and eosin. Bar, 50 μm. (**d**) Calcifications were counted in five optical fields per placenta. Each dot corresponds to the mean value of calcifications/field present in each placenta. (**e**) The diameter of chorionic microvilli is higher in ZIKV-infected placentae than in the control condition. Bar, 200 μm. (**f**) The diameter of chorionic villi was measured using the image analysis software Zen (version ZEN 2.3 lite, Carl Zeiss Microscopy, Jena, Germany) in five optical fields of samples from placentae stained with haematoxylin and eosin. Each dot corresponds to the mean diameter of chorionic villi per placenta. \* *p* = 0.0286.

Some placentae infected with ZIKV also displayed chorionic villi edema, heterogeneous maturation of chorionic villi, characterized by the co-existence of villi with different diameters; increased syncytial knots due to premature aging; and karyorrhexis, the irregular distribution of chromatin in the cytoplasm due to the destructive fragmentation of the nucleus of dying cells (Figures S10–S13). In contrast, tissue sections of chorionic villi from the placentae of control women displayed a homogeneous maturation of chorionic villi, an absence of inflammatory cells, and a parenchyma without abundant Hofbauer cells (Figures S14–S17).

We also performed a Masson's trichrome stain in the chorionic villi of D1 placenta from a ZIKV positive woman in order to detect pathological changes involving the connective tissue. Supplemental

Figure S18 reveals perivascular fibrosis and abundant intravilli collagen, as well as mesenchymal edema and karyorrhexis, which were previously observed with the hematoxylin and eosin stain.

#### *3.14. ZIKV Added to the Basolateral Surface of the Trophoblast-Derived Cell Line BeWo Reduces the Transepithelial Electrical Resistance and Claudin-4 Expression*

To further explore the e ffect of ZIKV on TJs of the trophoblasts, we incubated ZIKV at a MOI of 1 with the trophoblast-derived choriocarcinoma cell line BeWo cultured on Transwell filters. We worked with BeWo cells because they constitute a human cell culture model of placental villous trophoblast, and because the STB layer of human placenta is poorly susceptible to infection by ZIKV [33]. Figure 5a shows that the apical administration of ZIKV immediately increases TER by 34%, reaches 40% above control after 6 h, and diminishes to 27% above control after 24 h. Instead, the administration of ZIKV to the basolateral surface of BeWo cells induced a fast drop of TER of 20% in comparison to control monolayers at 1.5 h. However, after 10 h, the values of TER had recovered and were undistinguishable from those of control monolayers. Hence, these results reveal that ZIKV in contact with the basolateral surface of trophoblast cells is able to reduce the degree of sealing of TJs.

**Figure 5.** ZIKV added to the basolateral surface of BeWo cells transiently diminishes the transepithelial electrical resistance and claudin-4 expression. Confluent monolayers of BeWo cells that had achieved a stable value of TER were incubated with ZIKV (MOI = 1) added to the apical or basolateral surface. (**a**) TER was continuously measured in the cellZscope system in three inserts per condition. Results are shown with the corresponding standard deviation. (**b**) Immunofluorescence for claudin-4, occludin and ZO-1 in BeWo monolayers done after 1.5 h of incubation with ZIKV. Bar, 50 μm. X-Y, en face view; X-Z, lateral view.

By immunofluorescence, we observed no change in the expression of occludin or ZO-1 in BeWo cells, 1.5 h after the addition of ZIKV to the apical or basolateral surfaces (Figure 5b). Instead, the expression of claudin-4 diminishes 1.5 h after ZIKV is added to the basolateral surface. In addition, while occludin and ZO-1 concentrate as spots at the TJ region in immunofluorescence z-sections, claudin-4 distributes along the basolateral membrane, as had been previously observed in intestinal Caco-2 cells [34] These results thus confirm the capacity of ZIKV to diminish the expression of claudin-4 in trophoblasts.
